global_warming_impact_harte_john_430

合集下载

Epic Systems Corporation

Predictive Analysis of a Wavefront Application Using LogGP*David Sundaram-Stukel Epic Systems Corporation5301 Tokay BlvdMadison, WI 53711 dsundara@Mary K. Vernon University of Wisconsin-Madison Computer Sciences Dept.1210 W. Dayton StreetMadison, WI 53706-1685 vernon@ABSTRACTThis paper develops a highly accurate LogGP model of a complex wavefront application that uses MPI communication on the IBM SP/2. Key features of the model include: (1) elucidation of the principal wavefront synchronization structure, and (2) explicit high-fidelity models of the MPI-send and MPI-receive primitives. The MPI-send/receive models are used to derive L, o, and G from simple two-node micro-benchmarks. Other model parameters are obtained by measuring small application problem sizes on four SP nodes. Results show that the LogGP model predicts, in seconds and with a high degree of accuracy, measured application execution time for large problems running on 128 nodes. Detailed performance projections are provided for very large future processor configurations that are expected to be available to the application developers. These results indicate that scaling beyond one or two thousand nodes yields greatly diminished improvements in execution time, and that synchronization delays are a principal factor limiting the scalability of the application. KeywordsParallel algorithms, parallel application performance, LogP model, particle transport applications.1.INTRODUCTIONThis paper investigates the use of the parallel machine model called LogGP to analyze the performance of a large, complex application on a state-of-the-art commercial parallel platform. The application, known as Sweep3D, is of interest because it is a three-dimensional particle transport problem that has been identified as an ASCI benchmark for evaluating high performance parallel architectures. The application is also of interest because it has a fairly complex synchronization structure. This synchronization structure must be captured in the analytic model in order for the model to accurately predict application execution times and thus provide accurate performance projections for larger systems, new architectures, or modifications to the application. One question addressed in this research is which of variants of the LogP model [4] is best suited for analyzing the performance of Sweep3D on the IBM SP system. Since this version of Sweep3D uses the MPI communication primitives, the LogGP model [2] which includes an additional parameter, G, to accurately model communication cost for large pipelined messages, turned out to provide the requisite accuracy. Possibly due to the blocking nature of the MPI primitives, the contention at message processing resources is negligible and thus recent extensions to LogP for capturing the impact of contention [7,12] are not needed.In previous work [4,6,7], the LogP models have been applied to important but fairly simple kernel algorithms, such as FFT, LU decomposition, sorting algorithms, or sparse matrix multiply. Two experimental studies have applied the model to complex full applications such as the Splash benchmarks [9, 11]. However, in these studies, the effects of synchronization on application performance and scalability were measured empirically rather than estimated by the model. Many other previous analytic models for analyzing application performance are restricted to simpler synchronization structures than Sweep3D (e.g., [8]). One exception is the deterministic task graph analysis model [1], which has been shown to accurately predict the performance of applications with complex synchronization structures. The LogGP model represents synchronization structures more abstractly than a task graph. A key question addressed in this research is whether the more abstract representation is sufficient for analyzing a full, complex application such as Sweep3D.We construct a LogGP model that not only captures the synchronization structure but also elucidates the basic synchronization structure of Sweep3D. Similar to the approach in [2], we use communication micro-benchmarks to derive the input parameters, L, o, and G. However, as we show in section 3, deriving these parameters is somewhat more complex for MPI communication on the SP/2 than for the Meiko CS-2; thus explicit models of the MPI-send and MPI-receive primitives are developed. Although the LogGP input parameters are derived from four-processor runs of Sweep3d, the LogGP model projects performance quite accurately up to 128 processors, for several fixed total problem sizes and several cases of fixed problem size per processor. The model also quickly and easily projects performance for the very large future processor configurations that are expected to be available to the application developers.*This research is supported in part by DARPA/ITO under contract N66001-97-C-8533.Computer Sciences Technical Report #1392, University of Wisconsin-Madison, February 1999. To appear in Proc. 7th ACM SIGPLAN Symp. on Principles and Practice of Parallel Programming (PPoPP ’99), Atlanta, GA, May 1999.We show several interesting results that can be derived from the analysis.Section 2 provides a brief overview of the Sweep3D application.Section 3 derives the models of MPI-send and MPI-receive and the parameter values that characterize communication cost.Section 4 presents the LogGP equations for Sweep3D, as well as the modifications that are needed when the application utilizes the multiprocessor SMP nodes of the SP/2. In the latter case, there are two types of communication costs: intra-cluster and inter-cluster. Section 5 provides model validation results as well as performance projections for future systems. Section 6 provides the conclusions of this work.2. Sweep3DSweep3D is described in [10]. A detailed task graph showing the complex synchronization among the tasks in the version of the code that is analyzed in this paper, is given in [5]. Here we give a simple overview of this version of Sweep3D, including only the aspects that are most relevant to the LogGP model. The structure of the algorithm will be further apparent from the LogGP model presented in section 4.As its name implies, the Sweep3D transport calculations are implemented as a series of pipelined sweeps through a three dimensional grid. Let the dimensions be denoted by (i,j,k). The 3D grid is mapped onto a two-dimensional array of processors, of size m ×n , such that each processor performs the calculations for a partition in the i and j dimensions of size it ×jt ×k , as shown in Figure 1. Note that, due to the problem mapping in Figure 1, the processors in the processor grid of Figure 2 will be numbered p i,j where i varies from 1 to n and indicates the horizontal position of the processor.A single iteration consists of a series of pipelined sweeps through the 3D grid starting from each of the 8 corners (or octants) of the grid. The mapping of the sweeps to the two dimensional processor grid is illustrated in Figure 2. If mo denotes the number of angles being considered in the problem, then each processor performs it ×jt ×k ×mo calculations during the sweeps from each octant. To create a finer granularity pipeline, thus increasing parallelism in the computation, the block of data computed by a given processor is further partitioned by an angle blocking factor (mmi ) and a k-plane blocking factor (mk ). These parameters specify the number of angles and number of planes in the k-dimension, respectively,that are computed before boundary data is forwarded to the next processor in the pipeline. Each processor in the interior of the processor grid receives this boundary data from each of two neighbor processors, computes over a block based on these values, and then sends the results of its calculations to twoneighbor destination processors, determined by the direction of the sweep.In the optimized version of Sweep3D that we analyze, once all blocks at a given processor are calculated for the sweeps from a given pair of octants, the processor is free to start calculating blocks for sweeps from the next pair of octants. For example, the lower left corner processor can start to compute the first block of the sweep for octant 7 after it has computed its last block of the sweep originating from octant 6. This will be shown in greater detail in the LogGP model of Sweep3D in section 4.The pipelined sweep for octant 8 completes one iteration of the algorithm for one energy group. In the code we analyze, twelve iterations are executed for one time step. The target problems of interest to the ASCI program involve on the order of 30 energy groups and 10,000 time steps, for grid sizes on the order of 109(1000×1000×1000) or twenty million (e.g., 280×280×255). We can scale the model projections to these problem sizes, as shown in section 5.3. Communication Parameters: L, o, GBefore we present the LogGP model of Sweep3D for the SP/2, we derive models of the MPI-send and MPI-receive communication primitives that are used in the application. The MPI-send/receive models are needed in the LogGP model of Sweep3D, and are also needed to derive two of the communication parameters values,namely the network Latency (L ), and the processing overhead (o )to send or receive a message. The communication structure of Sweep3D is such that we can ignore the gap (g ) parameter, as the time between consecutive message transmissions is greater than the minimum allowed value of inter-message transmission time.Below we give the roundtrip communication times for MPI communication on the IBM SP, which are measured using simple communication micro-benchmarks. The value of G (Gap per byte )is derived directly from these measurements. We then discuss how we modeled the SP/2 MPI-send and MPI-receive primitives using the L, o, and G parameters, followed by a description of how the values of L and o are derived. A significant result is that we derive the same values of L and G (but different values of o )from the Fortran and the C micro-benchmark measurements. This greatly increases our confidence in the validity of the MPI communication models.Figure 1: Partitioning the 3D Grid in the i and j Dimensions Figure 2: The Sweeps for each OctantkNSEW(Processor Grid)3.1 Measured Communication TimesThe roundtrip communication time as a function of message sizefor a simple Fortran communication micro-benchmark is given in Figures 3 (a) and (b). For each data point, a message of the given size is sent from processor A to processor B, received by a process on processor B, and immediately sent back to A. The roundtrip time is measured on A by subtracting the time just before it calls MPI-send from the time that its MPI-receive operation completes.Each figure also includes the results of our model of the roundtrip communication, which is used to derive the L and o parameters,as discussed below.As can be seen in the figures, the measured communication time increases significantly with message size. Hence, the G parameter is required to accurately model communication cost. Two further points are worth noting from the Figure:•The communication cost changes abruptly at message size equal to 4KB, due to a handshake mechanism that is implemented for messages larger than 4KB. The handshake is modeled below.•The slope of the curve (G ) changes at message size equal to 1KB.The message processing overhead (o ) is also different for messages larger than 1 KB than for messages smaller than 1KB,due to the maximum IP packet size. Thus, we will derive separate values of G s / G l and of o s / o l for "small" (<1KB) and "large"(>1KB) messages.3.2 Models of MPI-send and MPI-receiveThe models developed here reflect a fairly detailed understanding of how the MPI-send and MPI-receive primitives are implemented on the SP/2, which we were able to obtain from the author of the MPI software. It might be necessary to modify the models for future versions of the MPI library, or if Sweep3D is run on a different message-passing architecture or is modified to use non-blocking MPI primitives. The models below illustrate a general approach for capturing the impact of such system modifications.Since the SP/2 system uses polling to receive messages, we can assume the overhead to send a message is approximately the same as the overhead to receive a message, o.For messages smaller than 4KB, no handshake is required, and the total end-to-end cost of sending and receiving a message is modeled simply as:Total_Comm = o + (message_size × G) + L + o (1)where the values of G and o depend on whether the message size is larger or smaller than 1KB.For messages larger than 4KB, the end-to-end communication requires a "handshake" in which just the header is initially sent to the destination processor and the destination processor must reply with a short acknowledgment when the corresponding receive has been posted. If the receive has been posted when the header message is sent, the end-to-end cost is modeled as follows:Total_Comm = o s + L + o s + o s + L + o l+ (message_size × G l ) + L + o l (2)Note that the processing overhead for receiving the ack is modeled as being subsumed in the processing overhead for sending the data. If the corresponding receive has not yet been posted, a additional synchronization delay will be incurred. This delay is modeled in the next section.In addition to the total cost for communication given above, the LogGP model for Sweep3D requires separate costs for sending and receiving messages. For message size less than 4KB:Send = o (3a) Receive = o (3b)where the value of o depends on the message size. For message size greater than or equal to 4KB:Send = o s + L + o s + o s + L + o l (4a)Receive = o s + L + o l + (message_size × G l ) + L + o l (4b)The receive cost includes the time to inform the sending processor that the receive is posted, and then the delay for the message to arrive.3.3 Communication Parameter ValuesUsing the above equations for Total_Comm and the measured round-trip communication times, we can derive the values of L,o s , o l , G s , and G l , which are given in Table 1. The values of G s and G l are computed directly from the slope of the curve (in Figure 3)for the respective range of message sizes. To derive L and o , we solve three equations for Total_Comm (for message sizes less than 1KB, between 1-4KB, and greater than 4KB, respectively) in three unknowns (L , o s , and o l ). Applying this method to the roundtrip time measurements obtained with C micro-benchmarks yields the same values of L and G as for the measurementsFigure 3: MPI Round Trip Communication Times.1101001000100001000001E+06Message SizeT i m e (u s e c )obtained with Fortran benchmarks, although the value of o is different, as shown in Table 1. This greatly increases our confidence in the validity of the above models of the MPI communication primitives. Using the parameter values derived in this way, the measured and modeled communication costs differ by less than 4% for messages between 64-256KB, as shown in Figure 3. Note that although the measured and modeled values seem to diverge at message size equal to 8KB in figure 2(a),figure 2(b) shows that the values for message sizes above 8KB are in good agreement.4. The LogGP Model of Sweep3DIn this section we develop the LogGP model of Sweep3D, using the models of the MPI communication costs developed in section 3. We first present the model that assumes each processor in the m ×n processor grid is mapped to a different SMP node in the SP/2. In this case, network latency is the same for all communication. We then give the modified equations for the case that 2×2 regions of the processor grid are mapped to a single (four-processor) SMP node in the SP/2. The round-trip times and parameter values computed in section 3 were for communication between processors in different SMP nodes. The same equations can be used to compute intra-node communication parameters.4.1 The Basic ModelThe LogGP model takes advantage of the symmetry in the sweeps that are performed during the execution, and thus calculates the estimated execution time for sweeps from one octant pair and then uses this execution time to obtain the total execution time for all sweeps, as explained below.During a sweep, as described in section 2, a processor waits for input from up to two neighbor processors and computes the values for a portion of its grid of size mmi × mk × it × jt. The processor then sends the boundary values to up to two neighbor processors,and waits to receive new input again. Using costs associated with each of these activities, we develop the LogGP model summarized in Table 2, which directly expresses the precedence andsend/receive synchronization constraints in the implemented algorithm.The time to compute one block of data is modeled in equation (5)of Table 2. In this equation, W g is the measured time to compute one grid point, and mmi , mk , it and jt are the input parameters,defined in section 2, that specify the number of angles and grid points per block per processor.Consider the octant pair (5,6) for which the sweeps begin at the processor in the upper-left corner of the processor grid, as shown in Figure 2. Recall that the upper-left processor is numbered p 1,1.To account for the pipelining of the wavefronts in the sweeps, we use the recursive formula in equation (6) of Table 2 to compute the time that processor p i,j begins its calculations for these sweeps,where i denotes the horizontal position of the processor in the grid. The first term in equation (6) corresponds to the case where the message from the West is the last to arrive at processor p i,j . In this case, the message from the North has already been sent but cannot be received until the message from the West is processed due to the blocking nature of MPI communications. The second term in equation (6) models the case where the message from the North is the last to arrive. Note that StartP 1,1 = 0, and that the appropriate one of the two terms in equation (6) is deleted for each of the other processors at the east or north edges of the processor grid.The Sweep3D application makes sweeps across the processors in the same direction for each octant pair. The critical path time for the two right-downward sweeps is computed in equation (7) of Table 2. This is the time until the lower-left corner processor p 1,m has finished communicating the results from its last block of the sweep for octant 6. At this point, the sweeps for octants 7 and 8(to the upper right) can start at processor p 1,m and proceed toward p n,1. Note that the subscripts on the Send and Receive terms in equation (7) are included only to indicate the direction of the communication event, to make it easier to understand why the term is included in the equation. The send and receive costs are as derived in section 3.2.The critical path for the sweeps for octants 7 and 8 is the time until all processors in the grid complete their calculations for the sweeps, since the sweeps from octants 1 and 2 (in the next iteration) won’t begin until processor p n,1 is finished. Due to the symmetry in the Sweep3D algorithm, mentioned above, the time for the sweeps to the Northeast is the same as the total time for the sweeps for octants 5 and 6, which start at processor p 0,0 and move Southeast to processor p n,m . Thus, we compute the critical path time for octants 7 and 8 as shown in equation (8) of Table 2.Equation (8) represents the time until processor p n,m has finished its last calculation for the second octant pair. The processorW i,j = W g × mmi × mk × it × jt (5) StartP i,j = max (StartP i –1,j + W i −1,j + Total_Comm + Receive, StartP i,j −1 + W i,j −1 + Send + Total_Comm) (6) T 5,6 = startP 1,m + 2[(W 1,m + Send E + Receive N + (m-1)L) × #k-blocks × #angle-groups] (7) T 7,8 = startP n-1,m + 2[(W n-1,m + Send E + Receive W + Receive N + (m–1)L+ (n-2)L) × #k-blocks × #angle-groups]+ Receive W + W n,m (8) T = 2 ( T 5,6 + T 7,8 ) (9)Table 2 LogGP Model of Sweep3DMessage Size:≤ 1024> 1024L 23 µsec 23 µsec o (Fortran)23 µsec 47 µsec o (C)16 µsec 36 µsec G0.07 µsec0.03 µsecTable 1. SP/2 MPI Communication Parametersdirectly to its East, p n-1,m, must start computing, calculate and communicate all needed results from the blocks for both octants, and then wait for processor p n,m to receive the results from the last block of these calculations and compute the results based on this block.Due to the symmetry between the sweeps for octants 1 through 4 and the sweeps for octants 5 through 8, the total execution time of one iteration is computed as in equation (9) of Table 2.The equation for T5,6 contains one term [(m–1)L], and the equation for T7,8contains two terms [(m–1)L and(n-2)L], that account for synchronization costs. These synchronization terms are motivated by the observation that measured communication times within Sweep3D are greater than the measured MPI communication cost discussed in section 3. The (m-1)L term in T5,6 and T7,8 captures the delay caused by a send which is blocked until the destination processor posts the corresponding receive. This delay accumulates in the j direction; thus the total delay at p1,m depends on the number of processors to its North (m-1). Furthermore, this synchronization cost is zero for the problems with message sizes smaller than 4KB, since in this case, the processor sends the message whether or not the corresponding receive has been posted. The second synchronization delay in T7,8, (n-2)L, represents difference between when a receive is posted, and when a message is actually received from the sending processor. Since a processor receives from the North after the West on a southeast sweep, it is more likely to wait for the message from the West. Since this delay is cumulative over all processors in the i dimension, at processor p(n-1),m we model this delay as (n-2)L. Notice that this receive synchronization term is 0 for processors on the west edge of the processor grid since there are no processors to its West from which to receive a message. This is why it was not included in the T5,6 expression above.4.2The Model for the Clustered SMP NodesA few modifications to the above model are needed if each 2×2 region of processor grid is mapped to a single four-processor SMP cluster in the IBM SP/2, rather than mapping each processor in the grid to a separate SMP node. These changes are outlined here, in anticipation of the next generation of MPI software for the SP that will support full use of the cluster processors.Let L local denote the network latency for an intracluster message, L remote denote the latency for an intercluster message, and L avg= (L local+ L remote)/2. In the following discussion, o and G are assumed to be the same for intra-cluster and inter-cluster messages, but the equations can easily be modified if this is not the case. Let L and R be subscripts that denote a model variable (e.g,. TotalComm, Send, or Receive) that is computed using L local or L remote, respectively. Using this notation, the modified equations that compute the execution time of Sweep3D are given in Table 3 and described below.Recall that processor numbering starts from 1 in both the i and j dimensions. Also recall that, for processor p i,j, i denotes its horizonal position in the processor grid. If both i and j are even, then all incoming messages are intra-cluster and all outgoing messages are inter-cluster. The vice versa is true if both i and j are odd. This means that StartP i,j is computed with TotalComm L, Receive L, and Send L(for the incoming messages) in the former case, and with TotalComm R, Receive R, and Send R in the latter case. For i odd and j even, the variables in the first term of StartP i,j are for inter-cluster communication and the communication variables in the second term are for intra-cluster communication. The vice versa is true for i even and j odd.The Send and Receive variables in the equations for T5,6 and T7,8 are all intra-cluster variables, assuming that the number of processors in each of the i and j dimensions is even when mapping 2×2 processor regions to the SMP clusters. The synchronization terms in T5,6and T7,8are computed using L avg. These are the only changes required in the model.The modified model has been validated against detailed simulation [3]. However, since we cannot yet validate them with system measurements (because efficient MPI software for intra-cluster communication doesn’t yet exist), only results for the case that each processor is mapped to a separate SMP node are given in this paper. Nevertheless, the changes to the model for full cluster use are simple and illustrate the model’s versatility. Furthermore, these equations can be used to project system performance for the next generation MPI software.4.3Measuring the Work (W)The value of the work per grid point, W g, is obtained by measuring this value on a 2x2 grid of processors. In fact, to obtain the accuracy of the results in this paper, we measured W g for each per-processor grid size, to account for differences (up to 20%) that arise from cache miss and other effects. Since the Sweep3D program contains extra calculations (“fixups”) for five of the twelve iterations, we measure W g values for both of these iteration types. Although this is more detailed than the creators of LogP/LogGP may have intended, the increased accuracy is substantial and needed for the large scale projections in section 5. Furthermore, our recursive model of Sweep3D only represents thei even, j even: StartP i,j = max (StartP i –1,j + W i−1,j + Total_Comm L + Receive L , StartP i,j−1 + W i,j−1 + Send L + Total_Comm L) i odd, j odd: StartP i,j = max (StartP i –1,j + W i−1,j + Total_Comm R + Receive R , StartP i,j−1 + W i,j−1 + Send R + Total_Comm R)i odd, j even: StartP i,j = max (StartP i –1,j + W i−1,j + Total_Comm R + Receive R , StartP i,j−1 + W i,j−1 + Send L + Total_Comm L) i even, j odd: StartP i,j = max (StartP i –1,j + W i−1,j + Total_Comm L + Receive L , StartP i,j−1 + W i,j−1 + Send R + Total_Comm R)T5,6 = startP1,m+ 2[(W1,m+ Send E+ Receive N + (m-1)L avg) × #k-blocks × #angle-groups]T7,8 = startP n-1,m+ 2[(W n-1,m+Send E+Receive W+ Receive N+ (m–1)L avg+ (n-2)L avg) × #k-blocks × #angle-groups]+ Receive W + W n,m Table 3: Modified LogGP Equations for Intra-Cluster Communication on the SP/2sweeps of the Sweep3D code. In addition, we measure the computation time before and after this main body of the code (i.e.,between the iterations for a time step). These computation times,denoted W before and W after , are measured during a single processor run of a specific problem size. All model parameters are thus measured using simple code instrumentation and relatively short one, two, and four-processor runs. In the next section we investigate how accurately the model predicts measured execution time for the Sweep3D application.5. Experimental ResultsIn this section we present the results obtained from the LogGP model. We validate the LogGP projections of Sweep3D running time against measured running time for up to 128 processors and then use the LogGP model to predict and evaluate the scalability of Sweep3D to thousands of processors, for two different problem sizes of interest to the application developers. Unless otherwise stated the reported execution times are for one energy group and one time step with twelve iterations in the time step.In Figure 4 we compare the execution time predicted by the LogGP model to the measured execution time for the Fortranversion of Sweep3d on up to 128 SP/2 processors, for fixed total problems sizes (150×150×150 and 50×50×50), and k-blocking factor, mk , equal to 10. As the number of processors increases, the message size and the computation time per processor decrease,while the overhead for synchronization increases. For these problem sizes and processor configurations, the message sizes vary from over 16KB to under 1KB; there is remarkably high agreement between the model estimates and the measured system performance across the entire range. Figure 5 shows that the larger problem size achieves reasonably good speedup (i.e., low communication and synchronization overhead) on 128 processors while the smaller problem size does not. Note that the model is highly accurate for both cases.In Figure 6, we show the predicted and measured application execution time as a function of the number of processors on the SP/2, for two different cases of fixed problem size per processor .In Figure 6(a) each processor has a partition of the three-dimensional grid that is of size 20×20×1000. In Figure 6(b), each processor has a partition of size 45x45x1000. In these experiments, the total problem size increases as the number of processors increases. The agreement between the model estimates(a) Problem size: 150×150×150 (b) Problem size: 50×50×50Figure 4: Validation of the LogGP Model for Fixed Total Problem Size(Fortran Code, mk=10, mmi=3)(a) Up to 128 processors(b) Up to 2500 processorsFigure 5: Sweep3D Speedups for Fixed Total Problem Sizes in Figure 4(Fortran Code, mk=10, mmi=3)020*********120140050100150ProcessorsS p e e d u p0501001502002503003504004500100020003000ProcessorsS p e e d u p。

模糊云资源调度的CMAPSO算法

模糊云资源调度的CMAPSO算法作者：李成严，宋月，马金涛来源：《哈尔滨理工大学学报》2022年第01期摘要：针对多目标云资源调度问题，以优化任务的总完成时间和总执行成本为目标，采用模糊数学的方法，建立了模糊云资源调度模型。

利用协方差矩阵能够解决非凸性问题的优势，采取协方差进化策略对种群进行初始化，并提出了一种混合智能优化算法CMAPSO算法（covariance matrix adaptation evolution strategy particle swarm optimization，CMAPSO ），并使用该算法对模糊云资源调度模型进行求解。

使用Cloudsim仿真平台随机生成云计算资源调度的数据，对CMAPSO算法进行测试，实验结果证明了CMAPSO算法对比PSO算法（particle wwarm optimization），在寻优能力方面提升28%，迭代次数相比提升20%，并且具有良好的负载均衡性能。

关键词：云计算;任务调度;粒子群算法; 协方差矩阵进化策略DOI：10.15938/j.jhust.2022.01.005中图分类号： TP399 文献标志码： A 文章编号： 1007-2683（2022）01-0031-09CMAPSO Algorithm for Fuzzy Cloud Resource SchedulingLI Chengyan，SONG Yue，MA Jintao（School of Computer Science and Technology， Harbin University of Science and Technology， Harbin 150080，China）Abstract：Aiming at the multiobjective cloud resource scheduling problem， with the goal of optimizing the total completion time and total execution cost of the task， a fuzzy cloud resource scheduling model is established using the method of fuzzy mathematics. Utilizing the advantage of the covariance matrix that can solve the nonconvexity problem， adopting the covariance evolution strategy to initialize the population， a hybrid intelligent optimization algorithm CMAPSO algorithm （covariance matrix adaptation evolution strategy particle swarm optimization，CMAPSO） is proposed to solve the fuzzy cloud resource scheduling model. The Cloudsim simulation platform was used to randomly generate cloud computing resource scheduling data， and the CMAPSO algorithm was tested. The experimental results showed that compared with the PSO algorithm （particle swarm optimization）， the optimization capability of CMAPSO algorithm is increased by 28%， the number of iterations of CMAPSO algorithm is increased by 20%， and it has good load balancing performance.Keywords：cloud computing; task scheduling; particle swarm algorithm; covariance matrix adaptation evolution strategy0引言云計算是一种商业计算的模型和服务模式[1]，而云计算资源调度的主要目的是将网络上的资源进行统一的管理和调式，再给予用户服务调用。

模拟退火免疫粒子群算法在皮肤电信号情感识别中的应用

ｅｐｒｍｅｔｅｕｔｓｏｈｔｃｍｐｄｗｉＰＯ，ＳＩＳａｃｉｖｅａｉｅｙｈｇｅｏｎｔｎｒｔｔｅｒｆａｕｅ，ｘｅｉｎａｒｓｌｈｗｔａｏ￣ｅｔＩＳｌｓｈＡ－ＰＯｃｎａｈｅｅｒｌｔｌｉｈｒｃｇｉｏａｅｗｉｆｗｅｔｒｓｖｉｈｅｔｅａｐｉａｉｎｏｉｌｔｄａｎａｉｇｍｅｈｎｓｃｎｈｌｔｈｐｉｚｔｎｏｅｔｒｅｅｔｎａｄｔｅｍｐｏｅｈｐｌｔｆｓｍｕａｅｎｅｎｃａｉｍａｅｐｗｉｔｅｏｔｃｏｌｈｍｉａｉｆｆａｕｅｓｌｃｉ，ｎｈｉｒｖｄｏｏａｇｒｈａｓｅｆｒｓｗｌｏｌｂｏｖｒｅｃ．ｌｏｔｍｏｐｒｍｅｌｎｇｏａｃｎｅｇｎｅｉｌｏｌ
ｎｈＳａｄｔｉｌｅｎｅｉｇＩｍｎａｉｅＳａｍＯｔｉｔｎ（ＡＩＳｅｍｕａｄＡｎａｎ－ｕｅＰｒｃｗｒｐｉｚｉｔｌｍｔｌｍａｏＳ — Ｏ）ｗｓａｏｔｒｆａｒｅｅｔｎｔｅＰａｄｐｄｆｅｔｅｓｌｉ．ｈｅｏｕｃｏ
第３卷第１１０期
２１０１年１０月
计算机应用
ＪｕｎｌｏｏｕｅｐｉａｉｎｏｒａｆＣｍｐｔｒＡｐｌｔｓｃｏ
Ｖ０＿Ｎｏ０Ｉ３１．１
０ｃ．０１ｔ２１

高炉煤气产生量预测方法对比研究

ENERGY FOR METALLURGICAL INDUSTRYMay. 2021222018年全球共生产生铁12. 65亿t L1J 。

高炉*国家自然科学基金(51704069)收稿日期=2020-07 -28赖 «(1999 -),本科生;110819辽宁省沈阳市。

通讯作者:孙文强，副教授;E - mail ：sunwq@ mail. neu.edu. cn.高炉煤气产生量预测方法对比研究*赖茜1王子豪V 孙文强2(1.东北大学冶金学院热能工程系，2.中国科学技术大学纳米科学技术学院，3.国家环境保护生态工业重点实验室)摘要为了对比回归分析法、移动平均法和人工神经网络法(ANN)对高炉煤气产生量的预测效果，采用某企业实际数据，探讨了不同工况下各种预测方法的预测误差。

研究表明：回归分析法能保持较高的预测精度和稳定性，但是对鼓风量的精度要求较高；移动平均法在平稳运行期和过渡期的性能优于ANN,平稳运行期和过渡期的预测误差分别为44. 47m 3/min 和69. 30m 3/min ； ANN 在工况转换期的性能优于移动平均法，在减风期和休风期的预测误差分别为 1442. 93m 3/min 和 1220. 57 m 3/min o关键词高炉煤气预测回归分析移动平均人工神经网络文献标识码：A 文章编号：1001 -1617 ( 2021) 03 - 0022 - 05Comparison of different methodsfor predicting the generation of blast furnace gasLai Xi 1 Wang Zihao 1,2 Sun Wenqiang 1,3(1. Department of Thermal Engineering , School of Metallurgy , Northeastern University ,2. Nano Science and Technology Institute , University of Science and Technologyof China,3. State Environmental Protection Key Laboratory of Eco - industry)Abstract In order to compare the performances of different methods , including regression analysis ,moving average method and artificial neural network ( ANN ) , for predicting the generation of blast fur nace gas , the prediction errors under various working conditions were investigated based on real data. Results show that the regression analysis is more accurate and robust , however , it requires even more accurate input of the volume of blast. Regarding the moving average method and ANN , the averagemoving method is better than ANN during the smooth operation and transition periods , with the predic tion error values of 44. 47 and 69. 30 m 3/min , respectively. ANN performs better than moving averagemethod during the switch period , with errors of 1442. 93 and 1220. 57 m 3/min, respectively , during blast attenuation and blowdown periods.Keywords blast furnace gas prediction regression analysis moving average artificial neuralnetwork (ANN)煤气是高炉炼铁过程中的副产能源S3】，产生量约为1500 ~ 2000 m 3/t,且波动频繁⑷，给高炉煤气的智能化调度造成了困难[5_6],导致能源浪费与环境污染，甚至威胁整个煤气管网和设备的安全运行因此，高炉煤气的预测研究吸引了很多学者的兴趣。

融合注意力机制的孪生网络目标跟踪算法研究

2021578近年来，目标跟踪逐渐成为计算机视觉领域的热点课题，它被广泛地应用在视频监控、自动驾驶、人机交互和医学诊疗等众多领域。

目前，目标跟踪面临诸多挑战，比如跟踪目标的尺度变化、跟踪漂移以及背景杂乱等。

因此，设计一个准确率和稳健性较高的算法成为目标跟踪研究的重点。

主流的跟踪算法分为相关滤波类算法和深度学习类算法。

在相关滤波算法中，KCF （Kernelized Correla-tion Filter ）[1]使用循环矩阵扩大样本容量，提高分类器的准确性。

成悦等[2]提出使用加权方法融合多种特征，增强算法的鲁棒性。

在深度学习算法中，Bertinetto 等[3]提出基于相似度匹配的目标跟踪算法SiamFC （Fully-Convolutional Siamese Networks ）。

SiamFC 使用离线训练的网络模型，通过比较输入图像和模板图像的相似度大小来得到跟踪结果。

以SiamFC 为基础，RepresentationLearning for Correlation Filter （CFNet ）[4]将相关滤波器作为一个网络层嵌入到孪生网络中，加强对深度特征的学习。

在SiamFC 中，模板图像由简单裁剪得到，背景也被当作正样本来和检测图像匹配。

这可能会导致：（1）当模板图像中背景与前景外观较为相似的时候，背景可能获得更大的相似度评分从而导致跟踪漂移。

（2）在目标运动过程中背景发生多次变化时，算法可能会跟踪到与模板中背景相似的目标上。

此外，SiamFC 使用最深层的卷积特征，缺乏对目标底层颜色及纹理信息的学习。

最近，注意力思想被融合到目标跟踪中，用来提高算法的准确率。

其中，CSR-DCF （Discriminative Corre-融合注意力机制的孪生网络目标跟踪算法研究王玲，王家沛，王鹏，孙爽滋长春理工大学计算机科学技术学院，长春130022摘要：在全卷积孪生网络跟踪算法（SiamFC ）的基础上，提出一种融合注意力机制的孪生网络目标跟踪算法。

AquaMaster 4 软件 FEW410 430 和 FET410 430 软件版本 3KXF2

—A B B M E A SU R EM ENT & A N A LY TI C S | R ELE A S E NOTE | R N/FE W410/FE W430/FE T410/FE T430/001-EN R E V BAquaMaster4 softwareFEW410/430 and FET410/4301 Software change detailsNew features:• Diagnosis for detecting default sensor setting due to NV memory issues.Bug fixes:• Totalizer reset option enabled with “Advance Access Level” for Non - Metrology approved devices.• Excessive power consumption in sensor.• Sensor firmware update.2 RecommendationsThis upgrade is advised for:• Devices with the above-mentioned issues • Devices that have software/firmware part code as mentioned in the release note.3 Product supportEmail: **********************.com—AquaMaster 4software FEW410/430 and FET410/430Software version:3KXF208402U0113/03.04.00 Release date: 15th June 2023Measurement made easyR N /F E W 410/F E W 430/F E T 410/F E T 430/001-E N R e v . B 06.2023—We reserve the right to make technical changes or modify the contents of this document without prior notice. With regard to purchase orders, the agreed particulars shall prevail. ABB does not accept any responsibility whatsoever for potential errors or possible lack of information in this document.We reserve all rights in this document and in the subject matter and illustrations contained therein. Any reproduction,disclosure to third parties or utilization of its contents – in whole or in parts – isforbidden without prior written consent of ABB. ©ABB 2023—ABB LimitedMeasurement & Analytics Oldends Lane,Stonehouse, GL10 3TA United KingdomTel: +44 (0) 1453 826661 Fax: +44 (0) 1453 829671email: **********************.com/measurement。

一种求解TSP问题的改进人工免疫算法

体的蛋白质多肽链结构并将优势肽植入新生抗体，
基酸分子组成的肽称二肽，由三个氨基酸分子组成的肽称三肽，由三个以上氨基酸分子组成的肽称多
模拟了免疫系统的浓度调节机制，并定义了一种新
的抗体相似性判断方法。改进后的人工免疫算法
肽，抗体和其他蛋白质一样，由许多氨基酸分子是借肽键连接起来的多肽链结构体。当有抗原入侵机体时，抗原会刺激那些与之匹配较好的抗体进行克隆繁殖，克隆体通过变异从而产生与抗原匹配更好的抗体。匹配得越好，克隆得越多。同时，免疫调节机制会控制抗体的克隆数量。当系统内某种抗体的浓度过高时，它会受到抑止。而减少克隆从
既具有理论价值又有广泛的工程应用价值的组合法，越来越受到人们的关注。免疫算法通过模拟自
优化问题。因此，人们不断地寻找能够有效地解决ＴＰ的各种方法…。除了局部搜索法、Ｓ贪婪算法、动
态规划等传统方法外，各种智能优化算法，如神经搜索、优化、机器学习等领域Ｊ。网络算法、拟退火算法、模禁忌搜索算法、遗传算在自然免疫系统中存在着大量的抗体用于保
解问题的目标函数、候选解、候选解对目标函数的
生抗体补充进入Ａｓｂ替换亲和力最低的抗体。新细
胞的数量是Ａｓｂ规模 Ⅳ的５％一ｌ％。０
维普资讯
１期
黎湖广，：等一种求解ＴＰ问题的改进人工免疫算法Ｓ
６１
规模。另外，免疫系统会不断产生新的抗体补充进
Ｓｐ：ｔ通过变异算子对克隆抗体群体Ａｓ进ｅ６ｂ

17-4PH不锈钢热老化的磁多参数无损评估

试验研究________________________N B T 8D O I ： 10.11973/w s jc 20210500717-4P H 不锈钢热老化的磁多参数无损评估褚英杰、孙琦\李乾武\史芳杰2,黄飞2,杨广宇2(1.福建宁德核电有限公司，宁德352000;2.苏州热工研究院有限公司，苏州215000)摘要：在350 °C 下对17-4P H 不锈钢进行加速热老化后开展力学试验、磁滞回线测试试验与磁多参数测试试验，结果表明，17-4P H 不锈钢热老化后冲击功下降，硬度上升。

磁滞回线表明热老化对17-4P H 不锈钢的宏观磁性无明显影响，且磁参数呈无规律性波动。

应用人工神经网络选取合适的节点数与算法，可建立17-4P H 不锈钢热老化的多参数高精度评估模型，为17-4P H 不锈钢的热老化评估提供新的方法。

关键词：17-4P H 不锈钢；热老化；多参数中图分类号：TG115.28文献标志码： A 文章编号：1000-6656(2021)05-0029-05Nondestructive evaluation of thermal aging of 17-4PH stainless steel bymulti-magnetic parametersCHU Yingjie1, SUN Qi' , LI Qianwu2, SHI Fangjie2, HUANG Fei2, YANG Guangyu2(1. Fujian Ningde Nuclear Power Co., Ltd., Ningde 352000, China ；2. Suzhou Nuclear Power Research Institute Co. ♦ L td., Suzhou 215000, China)Abstract ：The mechanical experiment, hysteresis loop test and multi-magnetic parameters test were carried out after accelerated thermal aging of 17-4PH under 350〇C. Results show that the impact energy of 17-4PH decrease while the hardness increase after thermal aging. The hysteresis loop shows that thermal aging has no impact on macroscopic magnetism of 17-4 PH and the magnetic parameters show no regular change. Through the application of artificial neural network, the proper node and algorithmic was chosen to build the precise module to evaluate the thermal aging of 17-4PH, which provide a new way to evaluate the thermal aging of 17-4PH.Key words ： 17-4PH stainless steel ； thermal aging ； multi-parameter17-4P H (05C r l 7N i 4C u 4N b )马氏体不锈钢因具有较高的强度和优异的耐腐蚀性能，被广泛应用于核电厂的高强度结构部件。

MPI TITAN RF Probe Selection Guide

MPI Probe Selection GuideWith a critical understanding of the numerous measurement challenges associated with today’s RF ap-plications, MPI Corporation has developed TITAN™ RF Probes, a product series specifically optimized for these complex applications centered upon the requirements of advanced RF customers.TITAN™ Probes provide the latest in technology and manufacturing advancements within the field of RF testing. They are derived from the technology transfer that accompanied the acquisition of Allstron, then significantly enhanced by MPI’s highly experienced RF testing team and subsequently produced utilizing MPI’s world class MEMS technology. Precisely manufactured, the TITAN™ Probes include matched 50 Ohm MEMS contact tips with improved probe electrical characteristics which allow the realization of unmat -ched calibration results over a wide frequency range. The patented protrusion tip design enables small passivation window bond pad probing, while significantly reducing probe skate thus providing the out -standing contact repeatability required in today’s extreme measurement environments. TITAN TM Probes with all their features are accompanied by a truly affordable price.The TITAN™ Probe series are available in single-ended and dual tip configurations, with pitch range from 50 micron to 1250 micron and frequencies from 26 GHz to 110 GHz. TITAN™ RF Probes are the ideal choice for on-wafer S-parameter measurements of RF, mm-wave devices and circuits up to 110 GHz as well as for the characterization of RF power devices requiring up to 10 Watts of continuous power. Finally, customers can benefit from both long product life and unbeatable cost of ownership which they have desired foryears.Unique design of the MEMS coplanar contacttip of the TITAN™ probe series.DC-needle-alike visibility of the contact point and the minimal paddamage due to the unique design of the tipAC2-2 Thru S11 Repeatability. Semi-Automated System.-100-80-60-40-200 S 11 E r r o r M a g n i t u d e (d B )Frequency (GHz)Another advantage of the TITAN™ probe is its superior contact repeatability, which is comparable with the entire system trace noise when measured on the semi-automated system and on gold contact pads.CROSSTALKCrosstalk of TITAN™ probes on the short and the bare ceramic open standard of 150 micron spacing compared to conventional 110 GHz probe technologies. Results are corrected by the multiline TRL calibration. All probes are of GSG configuration and 100 micron pitch.-80-60-40-200Crosstalk on Open. Multiline TRL Calibration.M a g (S21) (d B )Frequency (GHz)-80-60-40-200Crosstalk on Short. Multiline TRL Calibration.M a g (S21) (d B )Frequency (GHz)The maximal probe c ontac t repeatability error of the c alibrate S11-parameter of the AC2-2 thru standard by T110 probes. Semi-automated system. Ten contact circles.Cantilever needle material Ni alloy Body materialAl alloy Contact pressure @2 mils overtravel 20 g Lifetime, touchdowns> 1,000,000Ground and signal alignment error [1]± 3 µm [1]Planarity error [1] ± 3 µm [1]Contact footprint width < 30 µm Contact resistance on Au < 3 mΩThermal range-60 to 175 °CMechanical CharacteristicsAC2-2 Thru S21 Repeatability. Manual TS50 System.-100-80-60-40-200S 21 E r r o r M a g n i t u d e (d B )Frequency (GHz)MECHANICAL CHARACTERISTICSThe maximal probe c ontac t repeatability error of the c alibrate S21-parameter of the AC2-2 thru standard by T50 probes. Manual probe system TS50.26 GHZ PROBES FOR WIRELESS APPLICATIONSUnderstanding customer needs to reduce the cost of development and product testing for the high competitive wireless application market, MPI offers low-cost yet high-performance RF probes. The specifically developed SMA connector and its outstanding transmission of electro-magnetic waves through the probe design make these probes suitable for applications frequencies up to 26 GHz. The available pitch range is from 50 micron to 1250 micron with GS/SG and GSG probe tip configurations. TITAN™ 26 GHz probes are the ideal choice for measurement needs when developing components for WiFi, Bluetooth, and 3G/4G commercial wireless applications as well as for student education.Characteristic Impedance 50 ΩFrequency rangeDC to 26 GHz Insertion loss (GSG configuration)1< 0.4 dB Return loss (GSG configuration)1> 16 dB DC current ≤ 1 A DC voltage ≤ 100 V RF power, @10 GHz≤ 5 WTypical Electrical Characteristics26 GHz Probe Model: T26Connector SMAPitch range50 µm to 1250 µm Standard pitch step from 50 µm to 450 µm from 500 µm to 1250 µm25 µm step 50 µm stepAvailable for 90 µm pitch Tip configurations GSG, GS, SG Connector angleV-Style: 90-degree A-Style: 45-degreeMechanical CharacteristicsT26 probe, A-Style of the connectorTypical Electrical Characteristics: 26 GHz GSG probe, 250 micron pitchPROBES FOR DEVICE AND IC CHARACTERIZATION UP TO 110 GHZTITAN™ probes realize a unique combination of the micro-coaxial cable based probe technology and MEMS fabricated probe tip. A perfectly matched characteristic impedance of the coplanar probe tips and optimized signal transmission across the entire probe down to the pads of the device under test (DUT) result in excellent probe electrical characteristics. At the same time, the unique design of the probe tip provides minimal probe forward skate on any type of pad metallization material, therefo -re achieving accurate and repeatable measurement up to 110 GHz. TITAN™ probes are suitable for probing on small pads with long probe lifetime and low cost of ownership.The TITAN™ probe family contains dual probes for engineering and design debug of RF and mm-wave IC’s as well as high-end mm-wave range probes for S-parameter characterization up to 110 GHz for modeling of high-performance microwave devices.Characteristic Impedance 50 ΩFrequency rangeDC to 40 GHz Insertion loss (GSG configuration)1< 0.6 dB Return loss (GSG configuration)1> 18 dB DC current ≤ 1 A DC voltage ≤ 100 V RF power, @10 GHz≤ 5 WTypical Electrical Characteristics40 GHz Probe Model: T40Connector K (2.92 mm)Pitch range50 µm to 500 µmStandard pitch step For GSG configuration:from 50 µm to 450 µm from 500 µm to 800 µmFor GS/SG configuration:from 50 µm to 450 µm 25 µm step 50 µm stepAvailable for 90 µm pitch25 µm stepAvailable for 90/500 µm pitch Tip configurations GSG, GS, SG Connector angleV-Style: 90-degree A-Style: 45-degreeMechanical CharacteristicsTypical Electrical Characteristics: 40 GHz GSG probe, 150 micron pitchT40 probe, A-Style of the connectorCharacteristic Impedance50 ΩFrequency range DC to 50 GHz Insertion loss (GSG configuration)1< 0.6 dB Return loss (GSG configuration)1> 17 dBDC current≤ 1 ADC voltage≤ 100 VRF power, @10 GHz≤ 5 W Typical Electrical Characteristics Connector Q (2.4 mm)Pitch range50 µm to 250 µm Standard pitch stepFor GSG configuration: from 50 µm to 450 µm For GS/SG configuration: from 50 µm to 450 µm 25 µm stepAvailable for 90/500/550 µm pitch 25 µm stepAvailable for 90/500 µm pitchTip configurations GSG, GS, SG Connector angle V-Style: 90-degreeA-Style: 45-degreeMechanical CharacteristicsT50 probe, A-Style of the connectorTypical Electrical Characteristics: 50 GHz GSG probe, 150 micron pitchCharacteristic Impedance50 ΩFrequency range DC to 67 GHz Insertion loss (GSG configuration)1< 0.8 dB Return loss (GSG configuration)1> 16 dBDC current≤ 1 ADC voltage≤ 100 VRF power, @10 GHz≤ 5 W Typical Electrical Characteristics Connector V (1.85 mm)Pitch range50 µm to 250 µm Standard pitch stepFor GSG configuration: from 50 µm to 400 µm For GS/SG configuration: from 50 µm to 250 µm 25 µm step Available for 90 µm pitch25 µm step Available for 90 µm pitchTip configurations GSG Connector angle V-Style: 90-degreeA-Style: 45-degreeMechanical CharacteristicsT67 probe, A-Style of the connectorTypical Electrical Characteristics: 67 GHz GSG probe, 100 micron pitchCharacteristic Impedance 50 ΩFrequency rangeDC to 110 GHz Insertion loss (GSG configuration)1< 1.2 dB Return loss (GSG configuration)1> 14 dB DC current ≤ 1 A DC voltage ≤ 100 V RF power, @10 GHz≤ 5 WTypical Electrical CharacteristicsMechanical CharacteristicsTypical Electrical Characteristics: 110 GHz GSG probe, 100 micron pitchT110 probe, A-Style of the connectorCharacteristic impedance50 ΩFrequency range DC to 220 GHz Insertion loss (GSG configuration)1< 5 dB Connector end return loss(GSG configuration)1> 9 dBTip end return loss(GSG configuration)1> 13 dBDC current≤ 1.5 ADC voltage≤ 50 V Typical Electrical CharacteristicsConnector Broadband interface Pitch range50/75/90/100/125 µm Temperature range -40 ~ 150 ºC Contact width15 µmquadrant compatible(allowing corner pads)Yes recommended pad size20 µm x 20 µm recommended OT (overtravel)15 µmcontact resistance(on Al at 20 ºC using 15 µm OT)< 45 mΩlifetime touchdowns(on Al at 20 ºC using 15 µm OT)> 200,000Mechanical CharacteristicsT220 probe, broadband interface Typical Performance (at 20 ºC for 100 µm pitch)BODY DIMENSIONS PROBES Single-Ended V-StyleT220 GHz Probe1.161.1628.328437.455.6512.5527.73Single-Ended A-StyleCALIBRATION SUBSTRATESAC-series of calibration standard substrates offers up to 26 standard sets for wafer-level SOL T, LRM probe-tip cali -bration for GS/SG and GSG probes. Five coplanar lines provide the broadband reference multiline TRL calibration as well as accurate verification of conventional methods. Right-angled reciprocal elements are added to support the SOLR calibration of the system with the right-angled configuration of RF probes. A calibration substrate for wide-pitch probes is also available.Material Alumina Elements designCoplanarSupported calibration methods SOLT, LRM, SOLR, TRL and multiline TRL Thickness 635 µmSizeAC2-2 : 16.5 x 12.5 mm AC3 : 16.5 x 12.5 mm AC5 : 22.5 x 15 mm Effective velocity factor @20 GHz0.45Nominal line characteristic impedance @20 GHz 50 ΩNominal resistance of the load 50 ΩTypical load trimming accuracy error ± 0.3 %Open standardAu pads on substrate Calibration verification elements Yes Ruler scale 0 to 3 mm Ruler step size100 µmCalibration substrate AC2-2Probe Configuration GSGSupported probe pitch100 to 250 µm Number of SOL T standard groups 26Number of verification and calibration lines5Calibration substrate AC-3Probe Configuration GS/SG Supported probe pitch50 to 250 µm Number of SOL T standard groups 26Number of verification and calibration lines5Calibration substrate AC-5Probe Configuration GSG, GS/SG Supported probe pitch250 to 1250 µm Number of SOL T standard groups GSG : 7GS : 7SG : 7Open standardOn bare ceramic Number of verification and calibration linesGSG : 2GS : 1Typical characteristics of the coplanar line standard of AC2-2 calibration substrate measured using T110-GSG100 probes, and methods recommended by the National Institute of Standard and Technologies [2, 3].2468(d B /c m )F requency (G Hz)α-6-4-202I m a g (Z 0) ()F requency (G Hz)AC2-2 W#006 and T110A-GSG100Ω2.202.222.242.262.282.30 (u n i t l e s s )F requency (G Hz)β/βо4045505560R e a l (Z 0) ()F requency (G Hz)ΩTypical Electrical CharacteristicsMPI QAlibria® RF CALIBRATION SOFTWAREMPI QAlibria® RF calibration software has been designed to simplify complex and tedious RF system calibration tasks. By implementing a progressive disclosure methodology and realizing intuitive touch operation, QAlibria® provides crisp and clear guidance to the RF calibration process, minimizing con-figuration mistakes and helping to obtain accurate calibration results in fastest time. In addition, its concept of multiple GUI’s offers full access to all configuration settings and tweaks for advanced users. QAlibria® offers industry standard and advanced calibration methods. Furthermore, QAlibria® is integrated with the NIST StatistiCal™ calibration packages, ensuring easy access to the NIST mul-tiline TRL metrology-level calibration and uncertainty analysis.MPI Qalibria® supports a multi-language GUI, eliminating any evitable operation risks and inconvenience.SpecificationsRF AND MICROWAVE CABLESMPI offers an excellent selection of flexible cables and acces-sories for RF and mm-wave measurement applications forcomplete RF probe system integration.CablesHigh-quality cable assemblies with SMA and 3.5 mm connectorsprovide the best value for money, completing the entry-level RFsystems for measurement applications up to 26 GHz. Phase stab-le high-end flexible cable assemblies with high-precision 2.92, 2.4, 1.85 and 1 mm connectors guarantee high stability, accuracy and repeatability of the calibration and measurement for DC applications up to 110 GHz.MPI offers these cable assemblies in two standard lengths of 120 and 80 cm, matching the probe system’s footprint and the location of the VNA.Cables Ordering InformationMRC-18SMA-MF-80018 GHz SMA flex cable SMA (male) - SMA (female), 80 cmMRC-18SMA-MF-120018 GHz SMA flex cable SMA (male) - SMA (female), 120 cmMRC-26SMA-MF-80026 GHz SMA flex cable SMA (male) - SMA (female), 80 cmMRC-26SMA-MF-120026 GHz SMA flex cable SMA (male) - SMA (female), 120 cmMRC-40K-MF-80040 GHz flex cable 2.92 mm (K) connector, male-female, 80 cm longMRC-40K-MF-120040 GHz flex cable 2.92 mm (K) connector, male-female, 120 cm longMRC-50Q-MF-80050 GHz flex cable 2.4 mm (Q) connector, male-female , 80 cm longMRC-50Q-MF-120050 GHz flex cable 2.4 mm (Q) connector, male-female , 120 cm longMRC-67V-MF-80067 GHz flex cable 1.85 mm (V) connector, male-female, 80 cm longMRC-67V-MF-120067 GHz flex cable 1.85 mm (V) connector, male-female, 120 cm longMMC-40K-MF-80040 GHz precision flex cable 2.92 mm (K) connector, male-female, 80 cm long MMC-40K-MF-120040 GHz precision flex cable 2.92 mm (K) connector, male-female, 120 cm long MMC-50Q-MF-80050 GHz precision flex cable 2.4 mm (Q) connector, male-female , 80 cm long MMC-50Q-MF-120050 GHz precision flex cable 2.4 mm (Q) connector, male-female , 120 cm long MMC-67V-MF-80067 GHz precision flex cable 1.85 mm (V) connector, male-female, 80 cm long MMC-67V-MF-120067 GHz precision flex cable 1.85 mm (V) connector, male-female, 120 cm long MMC-110A-MF-250110 GHz precision flex cable 1 mm (A) connector, male-female, 25 cm longMPI Global PresenceDirect contact:Asia region: ****************************EMEA region: ******************************America region: ********************************MPI global presence: for your local support, please find the right contact here:/ast/support/local-support-worldwide© 2023 Copyright MPI Corporation. All rights reserved.[1] [2][3] REFERENCESParameter may vary depending upon tip configuration and pitch.R. B. Marks and D. F. Williams, "Characteristic impedance determination using propagation constant measu -rement," IEEE Microwave and Guided Wave Letters, vol. 1, pp. 141-143, June 1991.D. F. Williams and R. B. Marks, "Transmission line capacitance measurement," Microwave and Guided WaveLetters, IEEE, vol. 1, pp. 243-245, 1991.AdaptersHigh-In addition, high-quality RF and high-end mm-wave range adapters are offered to address challenges ofregular system reconfiguration and integration with different type of test instrumentation. MRA-NM-350F RF 11 GHz adapter N(male) - 3.5 (male), straight MRA-NM-350M RF 11 GHz adapter N(male) - 3.5 (female), straightMPA-350M-350F Precision 26 GHz adapter 3.5 mm (male) - 3.5 mm (female), straight MPA-350F-350F Precision 26 GHz adapter 3.5 mm (female) - 3.5 mm (female), straight MPA-350M-350M Precision 26 GHz adapter 3.5 mm (male) - 3.5 mm (male), straight MPA-292M-240F Precision 40 GHz adapter 2.92 mm (male) - 2.4 mm (female), straight MPA-292F-240M Precision 40 GHz adapter 2.92 mm (female) - 2.4 mm (male), straight MPA-292M-292F Precision 40 GHz adapter 2.92 mm (male) - 2.92 mm (female), straight MPA-292F-292F Precision 40 GHz adapter 2.92 mm (female) - 2.92 mm (female), straight MPA-292M-292M Precision 40 GHz adapter 2.92 mm (male) - 2.92 mm (male), straight MPA-240M-240F Precision 50 GHz adapter 2.4 mm (male) - 2.4 mm (female), straight MPA-240F-240F Precision 50 GHz adapter 2.4 mm (female) - 2.4 mm (female), straight MPA-240M-240M Precision 50 GHz adapter 2.4 mm (male) - 2.4 mm (male), straight MPA-185M-185F Precision 67 GHz adapter 1.85 mm (male) -1.85 mm (female), straight MPA-185F-185F Precision 67 GHz adapter 1.85 mm (female) -1.85 mm (female), straight MPA-185M-185M Precision 67 GHz adapter 1.85 mm (male) -1.85 mm (male), straight MPA-185M-100FPrecision 67 GHz adapter 1.85 mm (male) -1.00 mm (female), straightDisclaimer: TITAN Probe, QAlibria are trademarks of MPI Corporation, Taiwan. StatistiCal is a trademark of National Institute of Standards and Technology (NIST), USA. All other trademarks are the property of their respective owners. Data subject to change without notice.。

1_PetroMod_2012.1_InstallGuide

Installation Guide Version 2012.1PetroModPetroMod petroleum systems modeling software PetroModPM*Mark of SchlumbergerCopyright © 2012 Schlumberger. All rights reserved.Copyright © 2012 Schlumberger. All rights reserved.This work contains the confidential and proprietary trade secrets of Schlumberger and may not be copied or stored in an information retrieval system, transferred, used, distributed, translated or retransmitted in any form or by any means, electronic or mechanical, in whole or in part, without the express written permission of the copyright owner.Trademarks & Service MarksSchlumberger, the Schlumberger logotype, and other words or symbols used to identify the products and services described herein are either trademarks, trade names or service marks of Schlumberger and its licensors, or are the property of their respective owners. These marks may not be copied, imitated or used, in whole or in part, without the express prior written permission of Schlumberger. In addition, covers, page headers, custom graphics, icons, and other design elements may be service marks, trademarks, and/or trade dress of Schlumberger, and may not be copied, imitated, or used, in whole or in part, without the express prior written permission of Schlumberger. Other company, product, and service names are the properties of their respective owners.An asterisk (*) is used throughout this document to designate a mark of Schlumberger.iv PetroMod 2012.1 Installation GuideContents1 Information Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-1Schlumberger Product Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-2 About Schlumberger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-2 Documentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-2 Typestyle Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-2 Alert Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-2 Contacting Schlumberger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-3 Technical Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1-32 Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-1Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-2 Audience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-2 System Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-2 Licensing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2-43 Installation (Windows) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-1Downloading the Installation Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-2 Installing PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-3 Installing PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-3 Files Installed During Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-6 Optional: Installing Runtime Environment for Parallel Processing . . . . . . . . . . . . . . . . . . . . . . . .3-7 Installing Runtime Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-7 Activating Parallel Processing in the PetroMod Simulation Interface . . . . . . . . . . . . . . . . . . .3-7 Installing the Licensing Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-9 CodeMeter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3-9 Schlumberger Licensing Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-10 Configuring the PetroMod License . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13 Before you start . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13 Obtaining the license . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-13 Activating a local license . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-14 Activating an external license server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-15 Checking the License Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-16Contents v4 Uninstalling PetroMod (Windows) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-1Uninstalling PetroMod (Windows) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-2 Before You Begin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-2 Uninstalling PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-2 Results of the Uninstallation Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4-25 Installation (Unix) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-1Downloading the Installation Package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-2 Installing PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-3 Installing PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-3 Files Installed During Installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-4 Parallel Processing Set-up (Systems Admin) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-6 Intel MPI runtime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-6 PetroMod Machine Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-6 Configuring users for ssh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-7 Configuring users for rsh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-7 Activating Parallel Processing in PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-8 Load Sharing Facility (LSF) in Conjunction with Parallel PetroMod . . . . . . . . . . . . . . . . . . . . . . .5-9 Editing the MPI location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-9 Intel MPI Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5-9 Running Parallel PetroMod with Queuing Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10 Running PetroMod Software with LSF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10 Configuring the License . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-12 Setting the Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-12 Obtaining a License Key . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-12 Setting up the License Server (Systems Administrator) . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-136 Uninstalling PetroMod (Unix) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6-1Uninstalling PetroMod (Unix) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6-2 Before You Begin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6-2 Uninstalling PetroMod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6-2 Files Removed During Uninstallation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6-2vi PetroMod 2012.1 Installation Guide1Information ResourcesIn This SectionSchlumberger Product Documentation.........................................................1-2About Schlumberger.............................................................................1-2Documentation.....................................................................................1-2Typestyle Conventions..........................................................................1-2Alert Statements..................................................................................1-2 Contacting Schlumberger............................................................................1-3Technical Support.................................................................................1-3Information Resources1-1Schlumberger Product Documentation1-2PetroMod 2012.1 Installation Guide Schlumberger Product DocumentationAbout Schlumberger Schlumberger is the leading oilfield services provider, trusted to deliver superiorresults and improved E&P performance for oil and gas companies around the world.Through our well site operations and in our research and engineering facilities, wedevelop products, services, and solutions that optimize customer performance in asafe and environmentally sound manner.Documentation Documentation is provided in the following electronic formats via the listed location:•PetroMod 2012.1 Installation Guide (Adobe ® Acrobat ® PDF file):https://•PetroMod 2012.1 User Guides (Adobe ® Acrobat ® PDF files):https://•Online help for some applications: PetroMod -> HelpYou must have Adobe ® Reader ® installed to read the PDF files. Adobe Readerinstallation programs for common operating systems are available for a freedownload from the Adobe Web site at .Typestyle Conventions The following conventions are observed throughout this guide:•Bold text is used to designate file and folder names, dialog titles, names ofbuttons, icons, and menus, and terms that are objects of a user selection.•Italic text is used for word emphasis, defined terms, and manual titles.•Monospace text (Courier ) is used to show literal text as you would enter it, or asit would appear onscreen.Alert Statements The alerting statements are Notes, Cautions, and Warnings. These statements areformatted in the following style:• • • • • •Note:Information that is incidental to the main text flow, or to an important pointor tip provided in addition to the previous statement or instruction.• • • • • •Caution:Advises of machine or data error that could occur should the user fail totake or avoid a specified action.• • • • • •Warning:Requires immediate action by the user to prevent actual loss of data orwhere an action is irreversible, or when physical damage to themachine or devices is possible.Contacting SchlumbergerInformation Resources1-3Contacting SchlumbergerTechnical Support Schlumberger has sales and support offices around the world. For information oncontacting Schlumberger, please refer to the information below.For Technical Support for PetroMod software please contact the Customer CareCenter via the Schlumberger Support Portal at https://Internet Postal Mail SchlumbergerAachen Technology Center (AaTC)Ritterstr . 2352072 Aachen - GermanyContacting Schlumberger1-4PetroMod 2012.1 Installation Guide2Getting StartedIn This ChapterOverview...................................................................................................2-2Introduction.........................................................................................2-2System Requirements...........................................................................2-2System Requirements...........................................................................2-2Licensing.............................................................................................2-4Getting Started2-1Overview2-2PetroMod 2012.1 Installation Guide OverviewIntroduction This document describes the steps necessary to install PetroMod* 2012.1. Theinstaller includes a full PetroMod installation.•Installing on a workstation using a local license•Installing on a workstation using a license on a central license serverThis guide also explains the procedures required after installation:•Defining your license environmentThis module has been designed by the Schlumberger Aachen Technology Center(AaTC), Germany.Audience This guide is useful for the following people:•PetroMod users who install PetroMod on their workstations•System Administrator who installs PetroMod on a network shared diskSystem Requirements Before you install PetroMod 2012.1 your machine must meet the followingrequirements:Hardware Requirements Table 2-1Hardware Requirements for workstation (e.g. Dell T7500)Table 2-2Hardware Requirements for laptop (e.g. Dell M4600)Computer CPU 2 x Intel X5667 (quad core) or X5675 (hexa core)Physical Memory12 GB RAM Hard Disk Space 5 GB of free disk space Graphic Card Nvidia Quadro 5000 Network Card 1000 Mbit NicComputerIntel Core i7-2860Q 2.5 GHz Physical Memory16 GB RAM Hard Disk Space5 GB of free disk space Graphic CardNvidia Quadro 2000MOverviewGetting Started2-3Table 2-3Hardware Requirements for Linux Cluster • • • • • •Warning:Linux Users - Due to known issues concerning the instability ofOpenGL graphics, PetroMod only supports local rendering on 3Dgraphic cards with stable graphic drivers. Rendering via a networkcould cause stability issues. In particular, we observed problems withthe Mesa OpenGL package that is delivered with RHEL5 and works asa fall-back when no other driver is installed.Software Requirements Table 2-4Software Requirements • • • • • •Caution:It is possible that graphics do not display correctly when using older ATIgraphics cards and drivers. Thus, we recommend the use of Nvidiagraphics cards. We also recommend to install the latest graphicsdrivers available from Nvidia to avoid OpenGL graphic display errors.The driver version that comes with the OS in most cases is quite old oreven generic drivers are used if the graphics hardware is notrecognized correctly during the installation of the OS. These driversonly support basic functionalities and do not offer the OpenGL featuresrequired by PetroMod. Please be aware that most onboard graphicshardware does not support OpenGL at all. A dedicated graphics card isrequired for PetroMod.• • • • • •Note:Software applications created under older operating system versions willrun under newer operating system versions, but not vice puter CPU 2 x Intel X5667 (quad core) or X5675 (hexa core)Physical Memory48GB RAM Network Card1000 Mbit Nic Hard Disk Space 5 GB of free disk space Microsoft Vista 64-bit Microsoft Windows 7(recommended)64-bit RedHat EnterpriseLinux 5.3(recommended)64 bit Framework 2.0OverviewLicensing Licenses are required to access PetroMod. Certain functionalities or modules willonly be available with the respective licenses. Contact your Schlumberger SISCustomer Support representative to obtain the necessary licensesMaintenance contracts are usually yearly contracts, renewed at any time during theyear. Prior to 2012.1, PetroMod licenses allowed you to step up to a new PetroModversion based on the PetroMod license expiration date without having a validmaintenance contract. Beginning with 2012.1, upgrades are based on yourmaintenance contract expiration date. This is how you read the new licensingformat:FEATURE petrobuilder3D slbsls <yyyy.mm> <dd-mmm-yyyy> <#>Where• <yyyy.mm> is the maintenance expiration year and month• <dd-mmm-yyyy> is the license expiration day, month, year• <#> is the number of licensesMaintenance renewal is required to run any PetroMod version released after yourmaintenance expiration date. You will be automatically contacted by SchlumbergerInformation Systems before your maintenance expires.2-4PetroMod 2012.1 Installation Guide3Installation (Windows)In This ChapterDownloading the Installation Package..........................................................3-2Installing PetroMod.....................................................................................3-3Installing PetroMod...............................................................................3-3Files Installed During Installation...........................................................3-6 Optional: Installing Runtime Environment for Parallel Processing....................3-7Installing Runtime Environment.............................................................3-7Activating Parallel Processing in the PetroMod Simulation Interface..........3-7 Installing the Licensing Tool........................................................................3-9CodeMeter...........................................................................................3-9Schlumberger Licensing Tool................................................................3-10 Configuring the PetroMod License...............................................................3-13Before you start..................................................................................3-13Obtaining the license...........................................................................3-13Activating a local license......................................................................3-14Activating an external license server.....................................................3-15Checking the License Status.................................................................3-16Installation (Windows)3-1Downloading the Installation PackageDownloading the Installation PackageTo install PetroMod, you need the installation package. If you have a DVD, you canuse it. Otherwise, download PetroMod from the Software Download Center.••••••Note:If you are a new user of the Software Download Center, you must registerbefore you can download PetroMod.To download PetroMod 2012 from the Software Download Center1Go to .2Click SIS Software download center.3Log in to the site.4On the Welcome Message page, click Continue.5In the Product Group Name list (in the upper-left corner), click Geology &Geophysics.6In the table on the right, click PetroMod.7In the table of PetroMod downloads, click the Download icon for the PetroMod2012.1 file you need.You are ready to install PetroMod 2012.3-2PetroMod 2012.1 Installation GuideInstalling PetroModInstallation (Windows)3-3Installing PetroModPerform the following tasks prior to beginning the installation:•Ensure that you have admin privileges on the machine on which you are installing PetroMod and/or install the software together with your systems administrator since superuser passwords are required.•Ensure that the “System Requirements” on page 2-2 are met.PetroMod 2012 is a full installation. If you are already using an earlier PetroMod release, copy the new release into a new directory! Do not install the new version ‘over’ the old version to ensure that all programs and files can be updated and will then be compatible.Installing PetroMod The installation ensures that the files required to run PetroMod are installed on yourcomputer.To Install PetroMod 20121Insert the DVD or navigate to the location where you downloaded theinstallation files.2Double-click PetroMod2012.1.exe to start the installation. The folderPetroMod2012.1.msi will be unpacked, then the InstallShield Wizard will open, see Fig. 3-1. Click Next .Fig. 3-1PetroMod InstallShield Wizard3Fill in your User Name and Organization , see Fig. 3-2, then click Next.Installing PetroMod3-4PetroMod 2012.1 Installation GuideFig. 3-2Filling in user name and organization 4Determine the location of the files, see Fig. 3-3. Default is a folder calledSchlumberger in your Program Files folder. If this is not what you want you need to change that manually by clicking the Change button.When you are content with the location, click Next.Fig. 3-3Determining the location of the files 5 A summary of the settings will be displayed, see Fig. 3-4. Click Install .Installing PetroModInstallation (Windows)3-5Fig. 3-4Summary of settings 6You can follow the progress of the installation in the InstallShield Wizard, seeFig. 3-5.Fig. 3-5Installation progress 7Once the installation is complete the InstallShield Wizard will display the finaldialog, see Fig. 3-6. Click Finish .Installing PetroMod3-6PetroMod 2012.1 Installation Guide Fig. 3-6Installation complete 8The PetroMod 2012.1 icon will appear on your desktop . PetroMod 2012.1 willalso be added to the Schlumberger folder in the Progams list of your Start menu.Proceed with the installation of the .NETruntime environment, the MS HPC runtime for parallel processing and / or with installing Flexnet.Files Installed During Installation The following files / folders are installed during the installation of PetroMod:•one folder: PetroMod 2012.1:-client folder incl. sub folders/files-cult folder incl. sub folders/files-def folder incl. sub folders/files-doc folder incl. sub folders/files-geo folder incl. sub folders/files-well folder incl. sub folders/files-WIN64 folder incl. sub folders/files-PetroMod2012.1.batOptional: Installing Runtime Environment for Parallel Processing Optional: Installing Runtime Environment for Parallel ProcessingPetroMod 2012.1 supports parallel processing on Windows platforms usingMicrosoft HPC Pack 2008 R2 SP3. Using parallel processing from the SimulationInterface Microsoft requires the previous installation of Microsoft HPC runtimeenvironment.Installing Runtime Environment You can find the files in the installation package in the RuntimeEnvironment/ Windows folder:•mpi_x64.exe - MS MPI runtime•HpcClient_x64.exe - MS HPC web interface (optional)Once you have installed the files you need to obtain and activate the necessary licenses.••••••Note:If you want to consolidate your existing MS HPC runtime environments (installed with PM 11 or PetroMod 2011) to the latest version that shipswith PetroMod 2012.1, then you have to uninstall the MS HPC Pack 2008SDK on your system and manually set the required environment variable(called CCP_SDK) to C:\Program Files|Microsoft HPC Pack 2008 R2.Otherwise, PetroMod 2011.1 or PetroMod 11 SP4 will complain about themissing MPI runtime environment.Activating Parallel Processing in the PetroMod Simulation Interface 1After the licenses have been activated open the PetroMod Simulation Interface and select Processors for Parallel Run, see Fig. 3-7.Fig. 3-7Activating parallel processing in the PetroMod Simulation InterfaceOptional: Installing Runtime Environment for Parallel Processing2Increase the number of processors in the Processors Selection dialog.••••••Caution:Parallel processing is only supported on your local machine. Youcannot run a simulation on several nodes (as you could on Linuxclusters).Installing the Licensing Tool Installing the Licensing ToolPetroMod 2012 supports the use of the CodeMeter dongle as well as HOSTIDs forlicense authentication.•If you use CodeMeter, you must install CodeMeter software prior to installing theSchlumberger Licensing tool.•If you use HOSTIDs, you can continue by installing the Schlumberger Licensingtool.CodeMeter PetroMod 2012 uses the CodeMeter dongle for license authentication. Before youinsert your CodeMeter dongle into a USB port on your local workstation or on acentral license server, you must install the CodeMeter software. You should useCodeMeter 4.40 in the following circumstances:•If you use a local license (that is, your local workstation is your local licenseserver), install the CodeMeter dongle, appropriate CodeMeter driver, and theSchlumberger Licensing tool on your computer.•If you use a license on a central license server, the Administrator installs theCodeMeter dongle, CodeMeter driver, and the Schlumberger Licensing tool on theserver. Individual users do not need to install any licensing hardware or softwareon their computers.Installing CodeMeter Follow the steps in this section to uninstall previous versions of CodeMeter, andinstall the version required for this PetroMod release.◆To uninstall previous versions of CodeMeterIf you have an older version of CodeMeter installed on your computer, uninstall itand then install the latest version.1If you want to check the version of CodeMeter you are currently using beforeuninstalling it, on the Windows toolbar right-click the CodeMeter icon and selectAbout to open the About CodeMeter window.2Remove your CodeMeter dongle from the USB port.3Select Start > Control Panel > Programs and Features.4On the list of currently installed programs, select CodeMeter Runtime Kit vx.x.5Click Remove.You are ready to install the latest CodeMeter version.◆To install Codemeter••••••Note:CodeMeter automatically installs in your default Program Files directory:%program files%/Codemeter (usually C:/Program Files).1From the Licensing folder in the PetroMod installation package run the correctprogram (.exe) file:Installing the Licensing Tool•For PetroMod 64-bit, run CodeMeterRuntime64.exeAlternatively, go to the CodeMeter website () anddownload CodeMeter 4.40 from their website.The CodeMeter installation wizard opens.2On the Welcome panel, click Next.3On the License Agreement panel, read the agreement, select I accept thelicense agreement, and then click Next.4On the User Information panel, enter your name, your company’s name, andwho will use CodeMeter on your computer, and then click Next.5On the Select Features panel, select the features you want to install and thenclick Next.Schlumberger recommends that you accept the default settings.6On the Ready to Install panel, click Next to begin the installation.7When the installation is complete, on the last panel click Finish.8Restart your computer.The CodeMeter icon appears in your Windows taskbar. When you insert yourCodeMeter dongle into the USB port, the icon changes to show that your computerrecognizes the dongle. If you double-click the icon, you can view information aboutthe dongle.You are ready to install the Schlumberger Licensing tool.Schlumberger Licensing Tool PetroMod uses FlexNet for licensing. The Schlumberger Licensing tool is a simple interface for FlexNet. Using the Schlumberger Licensing 2012 tool to configure and manage your PetroMod license is recommended, but you may use FlexNet tools instead.If you do not already have the Schlumberger licensing tool on your computer, install the Licensing tool as follows:•If you use a local license (that is, your local workstation is your local license server), install the licensing tool on your computer.•If you use a license on a central license server, the Administrator installs the licensing tool on the server. The Schlumberger Licensing tool is not required for the Administrator to install, configure, and manage the PetroMod license. The Administrator may choose to manage PetroMod licenses with FlexNet native tools.The computer that you use to run the Schlumberger licensing tool is the license server for your PetroMod 2012 installaton.Installing the Schlumberger Licensing Tool Follow the steps in this section to install the Schlumberger Licensing tool. If you have a previous version of the licensing tool and want the latest version (2012), uninstall the older version first, and then install the latest version.。

基于自适应带宽核密度估计的载荷外推方法研究

2021年1月农业机械学报第52卷第1期doi:10.6041/j.issn.1000⁃1298.2021.01.042基于自适应带宽核密度估计的载荷外推方法研究牛文铁　才福友　付景静(天津大学机构理论与装备设计教育部重点实验室,天津300350)摘要:为了快速准确地得到玉米收获机车架的载荷谱,针对载荷谱编制过程中传统的载荷外推方法的局限性,提出一种基于四叉树算法的自适应带宽核密度估计(Kernel density estimation)算法,用来进行载荷外推㊂将经过预处理的实测原始载荷数据进行雨流计数统计,得到载荷循环均幅值矩阵,将小于载荷循环最大幅值10%的载荷滤除,其余的载荷循环均幅值数据根据四叉树分割算法进行不同区域的分割,选择高斯核函数,根据拇指法则计算各个区域的局部最优带宽,并根据数据区域内数据点的密集程度对核密度估计的输入进行优化,减少了核密度估计的计算量,最后结合蒙特卡洛模拟算法进行载荷外推㊂采用玉米收获机车架实测载荷数据进行实例验证,结果表明,与传统的固定带宽㊁自适应带宽核密度估计的载荷外推方法相比,本文提出的方法大大提高了计算效率,其概率密度函数图与实际载荷分布更为接近;载荷循环均幅值频次分布相关系数更接近于1,均方根误差更小;载荷循环幅值累积频次曲线的决定系数均大于0.99㊂关键词:玉米收获机;车架;载荷谱;载荷外推;自适应带宽核密度估计;蒙特卡洛模拟中图分类号:S225.5;TU413.4文献标识码:A文章编号:1000⁃1298(2021)01⁃0375⁃10OSID:收稿日期:20200614　修回日期:20200927基金项目:国家重点研发计划项目(2017YFD0700300)作者简介:牛文铁(1971 ),男,副教授,主要从事农业装备试验验证方法与技术和精密机床数字化设计研究,E⁃mail:niuwentie@Load Extrapolation Method Based on Adaptive BandwidthKernel Density EstimationNIU Wentie　CAI Fuyou　FU Jingjing(Key Laboratory of Mechanism Theory and Equipment Design ,Ministry of Education ,Tianjin University ,Tianjin 300350,China )Abstract :For the limitation of the traditional load extrapolation methods in the process of load spectrum compilation,an adaptive bandwidth kernel density estimation algorithm was proposed based on the quad⁃tree algorithm to obtain the load spectrum,which can obtain the corn harvester frame more accurately and quickly.Firstly,the rain flow counting method was applied to count the pretreated measured load data.The load cycles whose amplitudes were less than 10%of the maximum load cycle amplitude value were filtered.The remaining load data were segmented into different regions according to the quad⁃tree segmentation algorithm.The Gaussian kernel function was selected as the kernel function,and the local optimal bandwidth of the data in each region was calculated according to the rule of thumb.In addition,the input of the kernel density estimation was optimized according to the density of data points in the data area,which reduced the calculation consumption of kernel density estimation.The measured load data from the frame of the corn harvester were used for verifying the effectiveness of proposed pared with the traditional load extrapolation methods of fixed and adaptive bandwidth kernel density estimation,the proposed method greatly improved the computational efficiency,the probability density calculated by the proposed method was closer to the actual load distribution.The correlation coefficient of frequency distribution of load cycle mean and amplitude was closer to 1,and the root mean square error was smaller.The determining coefficient of the amplitude cumulative frequency curve was greater than 0.99.The results showed that the research result can provide reference for load extrapolation and load spectrum compilation.Key words :corn harvester;frame;load spectrum;load extrapolation;adaptive bandwidth kerneldensity estimation;Monte Carlo simulation0　引言载荷谱[1-2]是进行零部件疲劳寿命预测和优化设计的重要依据㊂目前,载荷谱已经在航天㊁汽车㊁高铁等领域得到了广泛研究与应用[3-5]㊂而我国大型玉米收获机关键零部件载荷谱的编制方法研究尚处于起步阶段,与国外发达国家相比还有较大的差距㊂国产玉米收获机故障频发,稳定性较差,其主要原因是设计水平不高,在研发设计初期大多以仿制和借鉴为主,很多零部件的结构设计并没有足够的载荷谱数据作为支撑㊂玉米收获机的车架是整车的核心承载部件,承受着割台㊁驾驶室㊁发动机㊁粮仓等传递的力和力矩,在实际工作过程中,经常会出现车架裂痕㊁甚至断裂的现象,严重影响了玉米收获机的整车性能㊂因此,采用合理的载荷外推方法进行多种工况下玉米收获机车架的载荷谱编制,对于玉米收获机车架的结构优化和整车性能提高具有重要意义㊂在实际载荷测试试验中,由于时间㊁天气㊁试验场地和成本等多种原因,只能进行一定时间内载荷时间历程的测量,实测载荷仅反映试验期间的载荷时间历程,不能反映全寿命周期下的载荷分布,因此需要对有限的载荷时间历程进行合理有效的载荷外推[6]㊂载荷外推是影响载荷谱编制准确性的关键环节㊂JOHANNESSON[7]提出了基于POT参数模型的时域外推方法,通过设置峰值的阈值,估计峰值的分布函数,在最大限度保留原始载荷时间序列的基础上,重构得到外推后的载荷时间历程㊂杨子涵等[8]对时域外推过程中的阈值选取方法进行了改进,提出了一种时域外推过程中阈值选取的量化方法,解决了时域外推过程中阈值选取主观性较强的问题㊂张英爽等[9]将经过雨流计数法统计后的载荷循环均值和载荷循环幅值的分布分别用正态分布和威布尔分布进行参数拟合,并以此进行载荷外推㊂翟新婷等[10]㊁GENG等[11]以混合分布函数作为函数拟合模型进行参数估计,该方法的拟合效果优于单分布的参数估计㊂但是,参数外推方法是用纯粹的分布函数来描述载荷的分布规律,对于复杂㊁随机性较大的载荷,则会产生较大的误差,大大降低了外推载荷谱与实际载荷谱的等效性㊂随着研究的不断深入,基于核密度估计法的非参数估计方法被应用于载荷外推中,该方法既可很好地保留载荷数据本身的分布规律,又能实现对任意载荷分布的拟合㊂李凡松等[12]采用基于自适应带宽的核密度估计方法进行载荷外推,每一个数据点都有其自身对应的带宽,该方法相较于固定带宽的核密度估计方法具有更好的拟合效果㊂本文针对核密度估计载荷非参数外推方法中带宽的选择问题,结合改进的四叉树分割算法,对核密度估计的带宽计算进行优化,提出一种改进的自适应带宽核密度估计的载荷外推方法㊂以玉米收获机车架为研究对象,通过与传统的基于核密度估计的固定带宽外推方法和自适应带宽外推方法进行比较,验证该自适应带宽核密度估计的载荷外推方法的准确性和合理性㊂1　核密度估计核密度估计是由ROSENBLATT[13]和PARZEN[14]提出的一种由实测分布函数未知的随机变量来估计其概率密度函数的非参数估计方法㊂该方法不需要知道数据的先验分布,也不必对数据进行任何的假设,只需要确定输入的数据变量㊁核函数以及带宽就可以估计输入数据的概率密度函数㊂目前,核密度估计已经在电力㊁医疗㊁地理等领域得到了非常广泛的应用[15-18]㊂玉米收获机工作时由于其结构复杂,工况多样,导致负载波动大,实测载荷信号具有很强的分散性和随机性,相较于一般的参数估计法来说,核密度估计能更好地描述载荷数据的分布规律,从而使得载荷外推更加准确㊂基于雨流计数矩阵的载荷外推是关于载荷循环均值和幅值的二维问题,二维核密度估计表达式为　^f(x,y)=1nhxh y∑n i=1(K x-x i h x,y-y i h)y(1)其中K(㊃)≥0∬K(㊃)d x={1(2)式中　n 输入数据点的个数h x㊁h y 核密度估计载荷循环幅值㊁均值的带宽x i㊁y i 输入的第i个载荷循环幅值㊁均值K(㊃) 核函数只要选取的核函数和带宽合适,就可以无限制地去逼近任何随机变量真实的概率密度函数㊂大量研究表明,当输入的样本数据量足够大时,核函数的具体形式对概率密度估计的准确性产生的影响相对较小[19],本文选择光滑且连续㊁明显单峰分布的高斯核函数㊂二维高斯核函数表达式为(K x-x i hx,y-y i h)y=12π(exp-(x-x i)22h2x-(y-y i)22h2)y(3)673农　业　机　械　学　报 2021年与核函数相比,带宽h的选择对核密度估计的准确性影响更大[20]㊂带宽h决定了^f(x,y)的光滑程度,若h较大,则有较多的数据点影响此处的概率密度计算,^f(x,y)曲线在此处较光滑,但是其与实际概率密度曲线的偏差较大;若h较小,则有较少的数据点影响此处的概率密度计算,^f(x,y)曲线在此处较陡峭,但是其与实际概率密度曲线的偏差较小㊂因此,为了更加准确地进行核密度估计,带宽h的选择尤为重要㊂为了定量化实现最优带宽计算,提出了平均积分平方误差(MISE), MISE表达式为M ISE(h)=∫E(^f(x)-f(x))2d x(4)式中　^f(x) 估计的概率密度f(x) 实测数据真实的概率密度E(㊃) 求均值函数拇指法则是目前应用最多的固定最优带宽的计算法则,拇指法则的最优带宽计算公式为h(=4d)+21d+4σn-1d+4(5)式中　d 核密度估计的维数,取2σ 输入二维数据样本的标准差此外,还可用无偏交叉验证(Unbiased cross validation)[21]㊁插入法(Plug⁃in)[22]等进行最优带宽的计算㊂上述方法计算的最优带宽是固定的,即每一个数据都有着同样的带宽,不能自动调节和变化,然而,在核密度估计的实际应用中,由于数据的随机性较强,分布不均匀,通过固定带宽核密度估计计算得到的概率密度可能与实际的分布相差较大,所以,希望带宽能够随数据的变化而变化,在数据密集的地方取得小一些,在数据稀疏的地方取得大一些㊂因此,自适应带宽核密度估计[23-24]得到了广泛的应用,该方法的具体计算公式为^f(x,y)=1n∑n i=11(λi h x)(λi h y)(K x-x iλi h x,y-y iλi h)y(6)其中λi=(g-1^f(x i,y i))-α(7)g=n∏n i=1^f(x i,y i)(8)式中　λi 带宽的自适应修订系数α 敏感性参数,取0~1^f(xi,y i) 固定带宽初步估计的概率密度一般情况下当α取0.5时,核密度估计的拟合效果较好[25],相对于固定带宽的核密度估计来说,自适应带宽在一定程度上提高了核密度估计的准确性,但由于数据集中的每一个数据都会影响彼此的带宽,当输入数据集较大时,会大大降低核密度估计的计算效率㊂当一定数量的数据聚集在一个很小的区域时,此区域内带宽的差异对核密度估计的影响十分有限,因此可以根据数据的聚集程度将同一区域内所有数据的带宽设置成相同的,就可以避免对这些点进行不必要的详细的带宽计算,在保证一定核密度估计准确性的前提下,提高了计算效率,所以针对核密度估计最优带宽的选择问题,为了更好兼顾核密度估计的计算效率与准确性,本文在上述方法的基础上进行改进,提出了一种新的自适应带宽计算方法㊂2　基于四叉树算法的自适应带宽核密度估计载荷外推基于四叉树算法的自适应带宽核密度估计载荷外推主要由3个步骤组成,包括:核密度估计的数据输入㊁基于四叉树算法的数据区域分割以及基于四叉树算法的自适应带宽核密度估计㊂具体的载荷外推流程图如图1所示㊂图1　基于四叉树算法的自适应带宽核密度估计载荷外推流程图Fig.1　Adaptive bandwidth kernel density estimation loadextrapolation flow chart based on quad⁃tree algorithm　2.1　核密度估计的数据输入雨流计数法得到的应变与材料的应力应变迟滞回线具有很好的一致性,并且考虑了载荷循环幅值和均值2个变量,符合疲劳载荷本身固有的特性,能够满足疲劳寿命预估以及载荷谱编制的条件,是目前应用最为广泛的实测载荷时间历程的计数统计方法㊂首先将已经预处理的实测载荷利用该方法进行计数统计处理,得到载荷循环均幅值矩阵M(M m㊁M a),其中M m表示载荷循环均值,M a表示载荷循环幅值;在实际工作中,小幅值载荷循环的数量比大幅值载荷循环大,但是其造成的疲劳损伤非常小[26],所以可以将小于最大载荷循环幅值10%的载荷循环只进行简单的线性比例外推,这很大程度上减少了核密度估计的数据输入量,从而降低核密度估计773第1期牛文铁等:基于自适应带宽核密度估计的载荷外推方法研究的计算量,提高了计算效率㊂已经过滤掉小幅值载荷循环的载荷循环均幅值矩阵记为M n (M mn ㊁M an ),并以此作为核密度估计的数据输入㊂2.2　基于四叉树算法的数据区域分割四叉树算法[27]是一种经典的空间分割与索引技术,目前在图像分割与空间索引领域已经得到了广泛的应用[28-29]㊂该算法可以通过对数据空间进行递归四等分割将数据区域分割成密度不等的块,数据越密集的地方,块的数量越多,块相对较小;而数据越稀疏的地方,块的数量越少,块相对较大,分割出的块的数量与大小从一定程度上反映出数据的密集程度㊂数据密集的地方采用较小的带宽,能够更好地反映出该区域分布的细节,与实际分布更加接近,拟合效果较好㊂根据四叉树分割算法分割出不同的数据块,计算各自的局部最优带宽,达到自适应带宽核密度估计的目的㊂将一个区域进行四叉树分割的流程如图2所示㊂图2　四叉树分割流程图Fig.2　Flow chart of quad⁃tree segmentation通过对区域不断执行上述四叉树分割流程,直到所有区域都不满足区域分割的条件,最后可以将原始数据区域分割成大小不尽相同的数据块㊂判断数据区域是否需要分割的阈值,数据区域内数据点个数的最大值N max 和数据区域最小宽度L min 对分割完成后数据块的大小和数量有着重要的影响,不同N max 对概率密度函数的影响如图3所示㊂由图3可知,N max 较大时,图3b 不能很好地反映出该区域内数据点分布的细节信息;N max 较小时,图3c 反映的细节信息和图3a 没有明显的差别,但是N max 越小,四叉树分割的层次就越深,计算量越大㊂四叉树分割相当于根据数据的密集程度进行数据的分类,数据块内的数据点影响着该区域的带宽,而数据块之间不互相影响㊂这与KNN KDE 理论[30]中根据每个数据点由相同个数的相邻点来确定区域有些相似,在每个区域内包含的点数相同的情况下,自适应带宽的大小与数据区域的大小呈正比㊂为了更好地进行数据区域的划分,根据KNN KDE 理论中二维数据的最优相邻个数将数据区域分割的阈值N max 设置成n ,n 为进行四叉树分割数据点的总个数㊂L min 的设置是为了避免四叉树分割进入太深层次㊂在数据特别密集的地方,带宽的变化很小,对最后分布拟合的影响有限,为了减少计算量,不需要进行太深层次的四叉树分割㊂通过雨流计数法得到的载荷均幅值矩阵通常按照64级来统计计数,每一级数据区域的边长定义为Δan =(max M an -min M an )/64(9)Δmn =(max M mn -min M mn )/64(10)式中　Δan 载荷循环幅值每一级的长度Δmn 载荷循环均值每一级的长度因此定义四叉树分割数据块的最小宽度L min =min(Δan ,Δmn ),这既能提高四叉树分割的计算效率,也能最小限度影响核密度估计的准确性㊂2.3　基于四叉树算法的自适应带宽核密度估计利用四叉树分割算法将数据区域分割成不同的数据块后,每个数据块内数据点的个数不相同,数据点带宽的计算流程如图4所示㊂其中,N min 是判断该数据块内数据点带宽计算的阈值条件,N min =N max ,N min 向下取整;h xi ㊁h yi 分别为数据块内载荷循环幅值和载荷循环均值的带宽,L xi ㊁L yi 分别为数据块载荷循环幅值和载荷循环均值的边长㊂不同数据块内数据点的带宽因为数据的差别而有所不同,但同一数据块内数据点的带宽是相同的㊂当一个数据块内的数据点密集程度比较高,彼此相差不大时,每一个数据点贡献的核密度估计分量的差别很微小,数据落入此数据块不同位置的概率近似相同㊂因此,为了进一步提高自适应核密度的计算效率,在数据密集程度比较高的块,不需要每一个数据点都进行核密度估计的计算,只需要将该数据区域数据点的平均值作为核密度估计的数据输入,再乘以该区域数据点的个数即可㊂在此将表示数据块内数据点密集程度的参数定义为γ=L sideH(11)其中L side =min(Δan ,Δmn )/2(12)H =max(h x ,h y )(13)当γ>1时,表示该数据块内的数据点密集程度高,可以对此数据块内的核密度估计计算进行优化㊂依873农　业　机　械　学　报 2021年图3　不同N max 对概率密度函数的影响Fig.3　Effects of different N max on probability densityfunction图4　数据点带宽计算流程图Fig.4　Flow chart of data point bandwidth calculation此改进的自适应核密度估计公式为^f(x ,y )=1N∑N 0i =11hxi h yin (x i ,y i )(K x i -x h xi ,y i -yh )yi(14)式中　N 优化前载荷数据点总个数N 0 优化后载荷数据点总个数当γ>1时,n (x i ,y i )表示此数据块内数据点的个数,x i ㊁y i 分别表示此数据块内数据点的载荷循环幅值以及载荷循环均值的平均值;当γ<1时,n (x i ,y i )等于1,x i ㊁y i 分别表示此数据块内每一个数据点的载荷循环幅值以及载荷循环均值㊂通过上述方法进行自适应带宽核密度估计后,再结合Monte Carlo 模拟算法进行载荷外推㊂3　实例验证3.1　试验方案玉米收获机在实际工作中,工况较多且复杂多变,由于地形㊁车辆载重以及操作行为不规范等原因,车架时常会出现裂纹,大大降低了车架的疲劳强度,影响整车性能,受到很大冲击时,甚至会出现车架整根断裂的问题,严重影响了粮食收获㊂由于试验环境较恶劣,并且测试部位空间有限,需要选用安装方便并且固定牢靠的传感器㊂因此本试验采用中航工业公司BE3504AA 型应变片进行数据测取,并通过有线连接的方式与HBM 公司的SoMat eDAQ 型数据采集仪连接㊂该数据采集仪具有极佳的密封性能和抗震性能,能有效应对水溅㊁扬尘㊁颠簸等测试环境,有线连接的方式能够确保信号传输的稳定性,通过数采的计算机终端软件将采样频率设置为500Hz㊂应变片的安装如图5所示,现场试验如图6所示,数据采集系统组成如图7所示㊂图5　应变片安装图Fig.5　Installation drawing of strain gauge sensor图6　现场试验图Fig.6　Photo of field test图7　数据采集系统组成Fig.7　Composition of data acquisition system　砂石路面工况下行驶较为平稳,载荷变化相对较小,大幅值载荷循环较少,载荷循环的分布相对集973第1期牛文铁等:基于自适应带宽核密度估计的载荷外推方法研究中;田间收获工况下载荷容易受田间地面的软硬程度以及粮仓的使用率等因素的影响,载荷的随机性较高,平稳性较差,大幅值载荷循环相对多一些,载荷循环的分布也更为分散㊂根据通过四叉树分割算法得到的数据块的大小以及密集程度,能够很好地反映出这两种工况下数据的密集程度以及分布情况,并且这两种工况占玉米收获机实际工作工况中很大的比重,具有一定的代表性,为了降低结论的偶然性,本文采用2种工况下各3组样本进行对比验证,经过预处理的试验数据如图8所示㊂图8　预处理后的试验数据Fig.8　Preprocessed experimental data3.2　结果验证通过雨流计数法对上述已经完成预处理的车架实测原始载荷进行统计计数处理,得到的部分载荷循环均幅值频次分布图如图9所示㊂图9　载荷循环均幅值频次分布图Fig.9　Frequency distribution histograms of load cyclemean and range由图9可知,幅值较小的载荷循环占比较大,将小于最大幅值10%的载荷循环过滤掉,过滤完成的数据作为固定带宽和自适应带宽核密度估计的数据输入,本文方法的核密度估计的数据输入在过滤的基础上还需要通过改进的四叉树算法进行优化㊂首图10　载荷循环均幅值散点图Fig.10　Scatter diagrams of load cycle mean and range先将已经筛选好的载荷循环均幅值矩阵中的数据点绘制成如图10所示的散点图,并在此散点图上利用改进的四叉树算法进行数据区域的分割操作,数据区域信息如表1所示㊂083农　业　机　械　学　报 2021年表1　数据区域信息Tab.1　Information of data area工况样本分割后的数据块数分割前核密度估计输入数分割后核密度估计输入数计算减少率/%16464828855.6砂石路面27066234547.936164841236.516879737852.6田间收获26179932059.936475341445.0 经过基于四叉树算法的数据区域分割的计算,数据区域被分割成大小不相同的数据块,每个数据块都有各自的带宽,利用式(14)对数据块的自适应带宽核密度估计进行优化,在之前滤除小幅值载荷循环的基础上又大幅减少了核密度估计的输入,进一步提高了核密度估计的计算效率㊂为了验证本文提出的基于四叉树算法的自适应带宽核密度估计的载荷外推方法的准确性与合理性,与传统的基于核密度估计的固定带宽载荷外推和自适应带宽载荷外推以及实测原始载荷进行对比论证㊂由3种带宽选择方法计算得到的载荷循环均幅值概率密度分布如图11所示㊂将图11与图10进行对比,可以看出传统的固定带宽与自适应带宽核密度估计的曲线较为平滑,而本文提出的自适应带宽核密度估计的曲线较为曲折,能够很好地反映图10数据的密集程度,并且数据密集点的分布情况划分更加具体,细节信息表达更为清楚,能很好地拟合真实概率密度㊂为了进一步评价概率密度函数的拟合程度,将固定带宽㊁自适应带宽以及本文提出的自适应带宽核密度估计的概率密度函数的结果分别结合Monte Carlo 模拟算法进行载荷外推㊂将实测原始载荷数据和上述3种方法进行载荷外推后得到的载荷循环均幅值分别进行分级统计处理,并计算实测原始数据与3种带宽选择外推方法的均方根误差(RMSE),计算结果如表2所示㊂图11　3种不同方法的载荷循环均幅值概率密度分布Fig.11　Mean and range probability density distributions of three different methods of load cycle 表2　载荷循环均幅值的均方根误差Tab.2　RMSE of load cycle mean and range με工况样本序号固定带宽自适应带宽本文方法13.61313.46822.6441砂石路面23.23283.05222.409033.39973.22282.465614.24333.96042.9437田间收获24.51604.20722.656933.98273.70922.6873 由表2可知,相较于传统的固定带宽和自适应带宽核密度估计,本文提出的自适应带宽核密度估计载荷外推方法的均方根误差RMSE 最小,表明通过本文方法计算得到的概率密度更加接近于真实的概率密度㊂将实测原始载荷数据和上述3种方法进行载荷外推后得到的载荷循环均值和幅值数据分别进行分183第1期牛文铁等:基于自适应带宽核密度估计的载荷外推方法研究级统计处理,结果如图12所示㊂为进一步分析频次分布的相关性和拟合程度,计算实测原始数据与3种带宽选择外推方法的相关系数(R )以及均方根误差(RMSE),结果如表3㊁4所示㊂图12　实测原始数据与3种不同方法的载荷循环均值㊁幅值频次分布比较Fig.12　Measured original data compared with mean and range frequency distribution of three different methods表3　载荷循环均值的相关系数及均方根误差Tab.3　R and RMSE of load cycle mean工况样本序号固定带宽自适应带宽本文方法R RMSE /μεR RMSE /μεR RMSE /με10.92853.75550.93023.69470.98361.8156砂石路面20.92213.43580.92753.30410.96542.294230.94022.61370.94272.53280.96911.871110.85945.02570.86234.93350.97302.2493田间收获20.91484.36880.92374.13960.98072.089330.92943.21190.93533.03330.96322.3032表4　载荷循环幅值的相关系数及均方根误差Tab.4　R and RMSE of load cycle range工况样本序号固定带宽自适应带宽本文方法R RMSE /μεR RMSE /μεR RMSE /με10.95828.44330.97336.81720.99552.5550砂石路面20.96338.14230.98395.57530.99791.825830.94749.41920.97217.02230.99402.964410.933612.64630.95989.68200.99503.2939田间收获20.900516.77090.925614.37270.99463.830430.923214.01790.949111.27750.99354.0161 由表3㊁4可知,相较于传统的固定带宽和自适应带宽核密度估计,本文提出的自适应带宽核密度估计载荷外推与实测原始载荷均值和幅值的相关系数R 最大,均方根误差最小,表明通过本文提出的自适应带宽核密度估计进行载荷外推得到的载荷循环均值和幅值分布与实测原始载荷的载283农　业　机　械　学　报 2021年。

RF-EXPLORER 3和RF-EXPLORER 6RF频谱分析仪说明书

RF‑EXPLORER/3: 15-2,700 MHz RF‑EXPLORER/6: 15‑2,700 MHz, 4,850‑6,100 MHz
RF-GENERATOR1
Order No. 25.6000
RF signal generator, 23.4-6,000 MHz • Fully programmable RF generator • Generates carrier and wobble signals from 23.4 MHz
Supplied w/o device.
RF-COVER/BL
Order No. 25.7000
RF-COVER/GE
Order No. 25.7010
RF-COVER/RT
Order No. 25.8000
RF-COVER/SW
Order No. 25.8010
Protective covers, for the RF series.
absolute • 50 Ω SMA connection • Mini USB 2.0 interface for connecting a PC or laptop
as well as for charging the built-in rechargeable lithium polymer battery (1,000 mAh) • Dimensions: 71 x 122 x 25 mm • Weight: 185 g
for setting an individual carrier frequency • Special Wi-Fi analyser with display of 13 WLAN
channels • Max hold function for a reliable detection of ultra-

3GPP协议-36521-1-e40_s00-s05

3GPP TS 36.521-1 V14.4.0 (2017-09)Technical Specification3rd Generation Partnership Project; Technical Specification Group Radio Access Network; Evolved Universal Terrestrial Radio Access (E-UTRA);User Equipment (UE) conformance specification;Radio transmission and reception;Part 1: Conformance Testing(Release 14)The present document has been developed within the 3rd Generation Partnership Project (3GPP TM) and may be further elaborated for the purposes of 3GPP.KeywordsUMTS LTE3GPPPostal address3GPP support office address650 Route des Lucioles - Sophia AntipolisValbonne - FRANCETel.: +33 4 92 94 42 00 Fax: +33 4 93 65 47 16InternetCopyright NotificationNo part may be reproduced except as authorized by written permission.The copyright and the foregoing restriction extend to reproduction in all media.© 2017, 3GPP Organizational Partners (ARIB, ATIS, CCSA, ETSI, TSDSI, TTA, TTC).All rights reserved.UMTS™ is a Trade Mark of ETSI registered for the benefit of its members3GPP™ is a Trade Mark of ETSI registered for the benefit of its Members and of the 3GPP Organizational Partners LTE™ is a Trade Mark of ETSI registered for the benefit of its Members a nd of the 3GPP Organizational Partners GSM® and the GSM logo are registered and owned by the GSM AssociationContentsForeword (92)Introduction (92)1Scope (93)2References (94)3Definitions, symbols and abbreviations (96)3.1Definitions (96)3.2Symbols (98)3.3Abbreviations (100)4General (103)4.1Categorization of test requirements in CA, UL-MIMO, ProSe, Dual Connectivity, UE category 0, UEcategory M1, UE category 1bis, UE category NB1 and V2X Communication (104)4.2RF requirements in later releases (105)5Frequency bands and channel arrangement (106)5.1General (106)5.2Operating bands (106)5.2A Operating bands for CA (108)5.2B Operating bands for UL-MIMO (116)5.2C Operating bands for Dual Connectivity (116)5.2D Operating bands for ProSe (117)5.2E Operating bands for UE category 0 and UE category M1 (118)5.2F Operating bands for UE category NB1 (118)5.2G Operating bands for V2X Communication (118)5.3TX–RX frequency separation (119)5.3A TX–RX frequency separation for CA (120)5.4Channel arrangement (120)5.4.1Channel spacing (120)5.4.1A Channel spacing for CA (121)5.4.1F Channel spacing for UE category NB1 (121)5.4.2Channel bandwidth (121)5.4.2.1Channel bandwidths per operating band (122)5.4.2A Channel bandwidth for CA (124)5.4.2A.1Channel bandwidths per operating band for CA (126)5.4.2B Channel bandwidth for UL-MIMO (171)5.4.2B.1Channel bandwidths per operating band for UL- MIMO (171)5.4.2C Channel bandwidth for Dual Connectivity (171)5.4.2D Channel bandwidth for ProSe (171)5.4.2D.1Channel bandwidths per operating band for ProSe (171)5.4.2F Channel bandwidth for category NB1 (172)5.4.2G Channel bandwidth for V2X Communication (173)5.4.2G.1Channel bandwidths per operating band for V2X Communication (173)5.4.3Channel raster (174)5.4.3A Channel raster for CA (175)5.4.3F Channel raster for UE category NB1 (175)5.4.4Carrier frequency and EARFCN (175)5.4.4F Carrier frequency and EARFCN for category NB1 (177)6Transmitter Characteristics (179)6.1General (179)6.2Transmit power (180)6.2.1Void (180)6.2.2UE Maximum Output Power (180)6.2.2.1Test purpose (180)6.2.2.4Test description (182)6.2.2.4.1Initial condition (182)6.2.2.4.2Test procedure (183)6.2.2.4.3Message contents (183)6.2.2.5Test requirements (183)6.2.2_1Maximum Output Power for HPUE (185)6.2.2_1.1Test purpose (185)6.2.2_1.2Test applicability (185)6.2.2_1.3Minimum conformance requirements (185)6.2.2_1.4Test description (185)6.2.2_1.5Test requirements (186)6.2.2A UE Maximum Output Power for CA (187)6.2.2A.0Minimum conformance requirements (187)6.2.2A.1UE Maximum Output Power for CA (intra-band contiguous DL CA and UL CA) (189)6.2.2A.1.1Test purpose (189)6.2.2A.1.2Test applicability (189)6.2.2A.1.3Minimum conformance requirements (189)6.2.2A.1.4Test description (189)6.2.2A.1.5Test Requirements (191)6.2.2A.2UE Maximum Output Power for CA (inter-band DL CA and UL CA) (192)6.2.2A.2.1Test purpose (192)6.2.2A.2.2Test applicability (192)6.2.2A.2.3Minimum conformance requirements (192)6.2.2A.2.4Test description (192)6.2.2A.2.5Test Requirements (194)6.2.2A.3UE Maximum Output Power for CA (intra-band non-contiguous DL CA and UL CA) (196)6.2.2A.4.1UE Maximum Output Power for CA (intra-band contiguous 3DL CA and 3UL CA) (196)6.2.2A.4.1.1Test purpose (196)6.2.2A.4.1.2Test applicability (196)6.2.2A.4.1.3Minimum conformance requirements (196)6.2.2A.4.1.4Test description (196)6.2.2A.4.1.5Test Requirements (198)6.2.2A.4.2UE Maximum Output Power for CA (inter-band 3DL CA and 3UL CA) (198)6.2.2A.4.2.1Test purpose (199)6.2.2A.4.2.2Test applicability (199)6.2.2A.4.2.3Minimum conformance requirements (199)6.2.2A.4.2.4Test description (199)6.2.2A.4.2.5Test Requirements (201)6.2.2B UE Maximum Output Power for UL-MIMO (201)6.2.2B.1Test purpose (201)6.2.2B.2Test applicability (202)6.2.2B.3Minimum conformance requirements (202)6.2.2B.4Test description (204)6.2.2B.4.1Initial condition (204)6.2.2B.4.2Test procedure (205)6.2.2B.4.3Message contents (205)6.2.2B.5Test requirements (205)6.2.2B_1HPUE Maximum Output Power for UL-MIMO (207)6.2.2B_1.1Test purpose (207)6.2.2B_1.2Test applicability (207)6.2.2B_1.3Minimum conformance requirements (207)6.2.2B_1.4Test description (207)6.2.2B_1.5Test requirements (208)6.2.2C 2096.2.2D UE Maximum Output Power for ProSe (209)6.2.2D.0Minimum conformance requirements (209)6.2.2D.1UE Maximum Output Power for ProSe Discovery (209)6.2.2D.1.1Test purpose (209)6.2.2D.1.2Test applicability (209)6.2.2D.1.3Minimum Conformance requirements (209)6.2.2D.2UE Maximum Output Power for ProSe Direct Communication (211)6.2.2D.2.1Test purpose (211)6.2.2D.2.2Test applicability (211)6.2.2D.2.3Minimum conformance requirements (211)6.2.2D.2.4Test description (211)6.2.2E UE Maximum Output Power for UE category 0 (212)6.2.2E.1Test purpose (212)6.2.2E.2Test applicability (212)6.2.2E.3Minimum conformance requirements (212)6.2.2E.4Test description (212)6.2.2E.4.3Message contents (213)6.2.2E.5Test requirements (213)6.2.2EA UE Maximum Output Power for UE category M1 (215)6.2.2EA.1Test purpose (215)6.2.2EA.2Test applicability (215)6.2.2EA.3Minimum conformance requirements (215)6.2.2EA.4Test description (216)6.2.2EA.4.3Message contents (217)6.2.2EA.5Test requirements (217)6.2.2F UE Maximum Output Power for category NB1 (218)6.2.2F.1Test purpose (218)6.2.2F.2Test applicability (218)6.2.2F.3Minimum conformance requirements (218)6.2.2F.4Test description (219)6.2.2F.4.1Initial condition (219)6.2.2F.4.2Test procedure (220)6.2.2F.4.3Message contents (220)6.2.2F.5Test requirements (220)6.2.2G UE Maximum Output Power for V2X Communication (221)6.2.2G.1UE Maximum Output Power for V2X Communication / Non-concurrent with E-UTRA uplinktransmission (221)6.2.2G.1.1Test purpose (221)6.2.2G.1.2Test applicability (221)6.2.2G.1.3Minimum conformance requirements (221)6.2.2G.1.4Test description (222)6.2.2G.1.4.1Initial conditions (222)6.2.2G.1.4.2Test procedure (222)6.2.2G.1.4.3Message contents (222)6.2.2G.1.5Test requirements (223)6.2.2G.2UE Maximum Output Power for V2X Communication / Simultaneous E-UTRA V2X sidelinkand E-UTRA uplink transmission (223)6.2.2G.2.1Test purpose (223)6.2.2G.2.2Test applicability (223)6.2.2G.2.3Minimum conformance requirements (223)6.2.2G.2.4Test description (224)6.2.2G.2.4.1Initial conditions (224)6.2.2G.2.4.2Test procedure (225)6.2.2G.2.4.3Message contents (226)6.2.2G.2.5Test requirements (226)6.2.3Maximum Power Reduction (MPR) (226)6.2.3.1Test purpose (226)6.2.3.2Test applicability (226)6.2.3.3Minimum conformance requirements (227)6.2.3.4Test description (227)6.2.3.4.1Initial condition (227)6.2.3.4.2Test procedure (228)6.2.3.4.3Message contents (228)6.2.3.5Test requirements (229)6.2.3_1Maximum Power Reduction (MPR) for HPUE (231)6.2.3_1.1Test purpose (231)6.2.3_1.4Test description (232)6.2.3_1.5Test requirements (232)6.2.3_2Maximum Power Reduction (MPR) for Multi-Cluster PUSCH (232)6.2.3_2.1Test purpose (232)6.2.3_2.2Test applicability (232)6.2.3_2.3Minimum conformance requirements (233)6.2.3_2.4Test description (233)6.2.3_2.4.1Initial condition (233)6.2.3_2.4.2Test procedure (234)6.2.3_2.4.3Message contents (234)6.2.3_2.5Test requirements (234)6.2.3_3Maximum Power Reduction (MPR) for UL 64QAM (235)6.2.3_3.1Test purpose (236)6.2.3_3.2Test applicability (236)6.2.3_3.3Minimum conformance requirements (236)6.2.3_3.4Test description (236)6.2.3_3.4.1Initial condition (236)6.2.3_3.4.2Test procedure (237)6.2.3_3.4.3Message contents (237)6.2.3_3.5Test requirements (238)6.2.3_4Maximum Power Reduction (MPR) for Multi-Cluster PUSCH with UL 64QAM (240)6.2.3_4.1Test purpose (240)6.2.3_4.2Test applicability (240)6.2.3_4.3Minimum conformance requirements (240)6.2.3_4.4Test description (241)6.2.3_4.4.1Initial condition (241)6.2.3_4.4.2Test procedure (242)6.2.3_4.4.3Message contents (242)6.2.3_4.5Test requirements (242)6.2.3A Maximum Power Reduction (MPR) for CA (243)6.2.3A.1Maximum Power Reduction (MPR) for CA (intra-band contiguous DL CA and UL CA) (243)6.2.3A.1.1Test purpose (243)6.2.3A.1.2Test applicability (243)6.2.3A.1.3Minimum conformance requirements (244)6.2.3A.1.4Test description (245)6.2.3A.1.5Test Requirements (248)6.2.3A.1_1Maximum Power Reduction (MPR) for CA (intra-band contiguous DL CA and UL CA) for UL64QAM (250)6.2.3A.1_1.1Test purpose (251)6.2.3A.1_1.2Test applicability (251)6.2.3A.1_1.3Minimum conformance requirements (251)6.2.3A.1_1.4Test description (252)6.2.3A.1_1.5Test requirement (254)6.2.3A.2Maximum Power Reduction (MPR) for CA (inter-band DL CA and UL CA) (255)6.2.3A.2.1Test purpose (255)6.2.3A.2.2Test applicability (255)6.2.3A.2.3Minimum conformance requirements (255)6.2.3A.2.4Test description (256)6.2.3A.2.5Test Requirements (260)6.2.3A.2_1Maximum Power Reduction (MPR) for CA (inter-band DL CA and UL CA) for UL 64QAM (263)6.2.3A.2_1.1Test purpose (263)6.2.3A.2_1.2Test applicability (263)6.2.3A.2_1.3Minimum conformance requirements (263)6.2.3A.2_1.4Test description (264)6.2.3A.2_1.5Test Requirements (266)6.2.3A.3Maximum Power Reduction (MPR) for CA (intra-band non-contiguous DL CA and UL CA) (267)6.2.3A.3.1Test purpose (267)6.2.3A.3.2Test applicability (267)6.2.3A.3.3Minimum conformance requirements (268)6.2.3A.3.4Test description (268)6.2.3A.3_1Maximum Power Reduction (MPR) for CA (intra-band non-contiguous DL CA and UL CA) forUL 64QAM (270)6.2.3A.3_1.1Test purpose (270)6.2.3A.3_1.2Test applicability (270)6.2.3A.3_1.3Minimum conformance requirements (270)6.2.3A.3_1.4Test description (271)6.2.3A.3_1.5Test Requirements (272)6.2.3B Maximum Power Reduction (MPR) for UL-MIMO (272)6.2.3B.1Test purpose (272)6.2.3B.2Test applicability (272)6.2.3B.3Minimum conformance requirements (273)6.2.3B.4Test description (273)6.2.3B.4.1Initial condition (273)6.2.3B.4.2Test procedure (274)6.2.3B.4.3Message contents (275)6.2.3B.5Test requirements (275)6.2.3D UE Maximum Output Power for ProSe (277)6.2.3D.0Minimum conformance requirements (277)6.2.3D.1Maximum Power Reduction (MPR) for ProSe Discovery (278)6.2.3D.1.1Test purpose (278)6.2.3D.1.2Test applicability (278)6.2.3D.1.3Minimum conformance requirements (278)6.2.3D.1.4Test description (278)6.2.3D.1.4.1Initial condition (278)6.2.3D.1.4.2Test procedure (279)6.2.3D.1.4.3Message contents (279)6.2.3D.1.5Test requirements (280)6.2.3D.2Maximum Power Reduction (MPR) ProSe Direct Communication (281)6.2.3D.2.1Test purpose (282)6.2.3D.2.2Test applicability (282)6.2.3D.2.3Minimum conformance requirements (282)6.2.3D.2.4Test description (282)6.2.3D.2.4.1Initial conditions (282)6.2.3D.2.4.2Test procedure (282)6.2.3D.2.4.3Message contents (282)6.2.3D.2.5Test requirements (282)6.2.3E Maximum Power Reduction (MPR) for UE category 0 (282)6.2.3E.1Test purpose (282)6.2.3E.2Test applicability (282)6.2.3E.3Minimum conformance requirements (282)6.2.3E.4Test description (282)6.2.3E.4.1Initial condition (282)6.2.3E.4.2Test procedure (283)6.2.3E.4.3Message contents (283)6.2.3E.5Test requirements (283)6.2.3EA Maximum Power Reduction (MPR) for UE category M1 (284)6.2.3EA.1Test purpose (284)6.2.3EA.2Test applicability (284)6.2.3EA.3Minimum conformance requirements (284)6.2.3EA.4Test description (285)6.2.3EA.4.1Initial condition (285)6.2.3EA.4.2Test procedure (287)6.2.3EA.4.3Message contents (287)6.2.3EA.5Test requirements (287)6.2.3F Maximum Power Reduction (MPR) for category NB1 (290)6.2.3F.1Test purpose (290)6.2.3F.2Test applicability (290)6.2.3F.3Minimum conformance requirements (290)6.2.3F.4Test description (291)6.2.3F.4.1Initial condition (291)6.2.3F.5Test requirements (292)6.2.3G Maximum Power Reduction (MPR) for V2X communication (292)6.2.3G.1Maximum Power Reduction (MPR) for V2X Communication / Power class 3 (293)6.2.3G.1.1Maximum Power Reduction (MPR) for V2X Communication / Power class 3 / Contiguousallocation of PSCCH and PSSCH (293)6.2.3G.1.1.1Test purpose (293)6.2.3G.1.1.2Test applicability (293)6.2.3G.1.1.3Minimum conformance requirements (293)6.2.3G.1.1.4Test description (293)6.2.3G.1.1.4.1Initial condition (293)6.2.3G.1.1.4.2Test procedure (294)6.2.3G.1.1.4.3Message contents (294)6.2.3G.1.1.5Test Requirements (294)6.2.3G.1.2 2956.2.3G.1.3Maximum Power Reduction (MPR) for V2X Communication / Power class 3 / SimultaneousE-UTRA V2X sidelink and E-UTRA uplink transmission (295)6.2.3G.1.3.1Test purpose (295)6.2.3G.1.3.2Test applicability (295)6.2.3G.1.3.3Minimum conformance requirements (295)6.2.3G.1.3.4Test description (295)6.2.3G.1.3.4.1Initial conditions (295)6.2.3G.1.3.4.2Test procedure (296)6.2.3G.1.3.4.3Message contents (297)6.2.3G.1.3.5Test requirements (297)6.2.4Additional Maximum Power Reduction (A-MPR) (297)6.2.4.1Test purpose (297)6.2.4.2Test applicability (297)6.2.4.3Minimum conformance requirements (298)6.2.4.4Test description (310)6.2.4.4.1Initial condition (310)6.2.4.4.2Test procedure (339)6.2.4.4.3Message contents (339)6.2.4.5Test requirements (344)6.2.4_1Additional Maximum Power Reduction (A-MPR) for HPUE (373)6.2.4_1.2Test applicability (374)6.2.4_1.3Minimum conformance requirements (374)6.2.4_1.4Test description (375)6.2.4_1.5Test requirements (376)6.2.4_2Additional Maximum Power Reduction (A-MPR) for UL 64QAM (378)6.2.4_2.1Test purpose (378)6.2.4_2.2Test applicability (378)6.2.4_2.3Minimum conformance requirements (378)6.2.4_2.4Test description (378)6.2.4_2.4.1Initial condition (378)6.2.4_2.4.2Test procedure (392)6.2.4_2.4.3Message contents (392)6.2.4_2.5Test requirements (392)6.2.4_3Additional Maximum Power Reduction (A-MPR) with PUSCH frequency hopping (404)6.2.4_3.1Test purpose (404)6.2.4_3.2Test applicability (404)6.2.4_3.3Minimum conformance requirements (405)6.2.4_3.4Test description (405)6.2.4_3.5Test requirements (406)6.2.4A Additional Maximum Power Reduction (A-MPR) for CA (407)6.2.4A.1Additional Maximum Power Reduction (A-MPR) for CA (intra-band contiguous DL CA and ULCA) (407)6.2.4A.1.1Test purpose (407)6.2.4A.1.2Test applicability (407)6.2.4A.1.3Minimum conformance requirements (407)6.2.4A.1.3.5A-MPR for CA_NS_05 for CA_38C (411)6.2.4A.1.4Test description (413)6.2.4A.1.5Test requirements (419)6.2.4A.1_1Additional Maximum Power Reduction (A-MPR) for CA (intra-band contiguous DL CA and ULCA) for UL 64QAM (425)6.2.4A.1_1.1Test purpose (425)6.2.4A.1_1.2Test applicability (425)6.2.4A.1_1.3Minimum conformance requirements (426)6.2.4A.1_1.3.5A-MPR for CA_NS_05 for CA_38C (429)6.2.4A.1_1.3.6A-MPR for CA_NS_06 for CA_7C (430)6.2.4A.1_1.3.7A-MPR for CA_NS_07 for CA_39C (431)6.2.4A.1_1.3.8A-MPR for CA_NS_08 for CA_42C (432)6.2.4A.1_1.4Test description (432)6.2.4A.1_1.5Test requirements (437)6.2.4A.2Additional Maximum Power Reduction (A-MPR) for CA (inter-band DL CA and UL CA) (443)6.2.4A.2.1Test purpose (443)6.2.4A.2.2Test applicability (444)6.2.4A.2.3Minimum conformance requirements (444)6.2.4A.2.4Test description (444)6.2.4A.2.4.1Initial conditions (444)6.2.4A.2.4.2Test procedure (457)6.2.4A.2.4.3Message contents (458)6.2.4A.2.5Test requirements (461)6.2.4A.3Additional Maximum Power Reduction (A-MPR) for CA (intra-band non-contiguous DL CAand UL CA) (466)6.2.4A.3.1Minimum conformance requirements (466)6.2.4A.2_1Additional Maximum Power Reduction (A-MPR) for CA (inter-band DL CA and UL CA) forUL 64QAM (466)6.2.4A.2_1.1Test purpose (466)6.2.4A.2_1.2Test applicability (466)6.2.4A.2_1.3Minimum conformance requirements (467)6.2.4A.2_1.4Test description (467)6.2.4A.2_1.4.1Initial conditions (467)6.2.4A.2_1.4.2Test procedure (479)6.2.4A.2_1.4.3Message contents (480)6.2.4A.2_1.5Test requirements (480)6.2.4B Additional Maximum Power Reduction (A-MPR) for UL-MIMO (484)6.2.4B.1Test purpose (484)6.2.4B.2Test applicability (485)6.2.4B.3Minimum conformance requirements (485)6.2.4B.4Test description (485)6.2.4B.4.1Initial condition (485)6.2.4B.4.2Test procedure (508)6.2.4B.4.3Message contents (508)6.2.4B.5Test requirements (508)6.2.4E Additional Maximum Power Reduction (A-MPR) for UE category 0 (530)6.2.4E.1Test purpose (530)6.2.4E.2Test applicability (531)6.2.4E.3Minimum conformance requirements (531)6.2.4E.4Test description (531)6.2.4E.4.1Initial condition (531)6.2.4E.4.2Test procedure (535)6.2.4E.4.3Message contents (535)6.2.4E.5Test requirements (536)6.2.4EA Additional Maximum Power Reduction (A-MPR) for UE category M1 (542)6.2.4EA.1Test purpose (542)6.2.4EA.2Test applicability (542)6.2.4EA.3Minimum conformance requirements (543)6.2.4EA.4Test description (544)6.2.4EA.4.1Initial condition (544)6.2.4EA.4.2Test procedure (552)6.2.4G Additional Maximum Power Reduction (A-MPR) for V2X Communication (562)6.2.4G.1Additional Maximum Power Reduction (A-MPR) for V2X Communication / Non-concurrentwith E-UTRA uplink transmissions (562)6.2.4G.1.1Test purpose (562)6.2.4G.1.2Test applicability (562)6.2.4G.1.3Minimum conformance requirements (563)6.2.4G.1.4Test description (563)6.2.4G.1.4.1Initial condition (563)6.2.4G.1.4.2Test procedure (564)6.2.4G.1.4.3Message contents (564)6.2.4G.1.5Test Requirements (564)6.2.5Configured UE transmitted Output Power (564)6.2.5.1Test purpose (564)6.2.5.2Test applicability (564)6.2.5.3Minimum conformance requirements (564)6.2.5.4Test description (594)6.2.5.4.1Initial conditions (594)6.2.5.4.2Test procedure (595)6.2.5.4.3Message contents (595)6.2.5.5Test requirement (596)6.2.5_1Configured UE transmitted Output Power for HPUE (596)6.2.5_1.1Test purpose (596)6.2.5_1.2Test applicability (597)6.2.5_1.3Minimum conformance requirements (597)6.2.5_1.4Test description (597)6.2.5_1.4.1Initial conditions (597)6.2.5_1.4.2Test procedure (597)6.2.5_1.4.3Message contents (597)6.2.5_1.5Test requirement (598)6.2.5A Configured transmitted power for CA (599)6.2.5A.1Configured UE transmitted Output Power for CA (intra-band contiguous DL CA and UL CA) (599)6.2.5A.1.1Test purpose (599)6.2.5A.1.2Test applicability (599)6.2.5A.1.3Minimum conformance requirements (599)6.2.5A.1.4Test description (601)6.2.5A.1.5Test requirement (602)6.2.5A.2Void (603)6.2.5A.3Configured UE transmitted Output Power for CA (inter-band DL CA and UL CA) (603)6.2.5A.3.1Test purpose (603)6.2.5A.3.2Test applicability (603)6.2.5A.3.3Minimum conformance requirements (603)6.2.5A.3.4Test description (605)6.2.5A.3.5Test requirement (606)6.2.5A.4Configured UE transmitted Output Power for CA (intra-band non-contiguous DL CA and ULCA) (607)6.2.5A.4.1Test purpose (607)6.2.5A.4.2Test applicability (607)6.2.5A.4.3Minimum conformance requirements (607)6.2.5A.4.4Test description (608)6.2.5A.4.5Test requirement (610)6.2.5B Configured UE transmitted Output Power for UL-MIMO (611)6.2.5B.1Test purpose (611)6.2.5B.2Test applicability (611)6.2.5B.3Minimum conformance requirements (611)6.2.5B.4Test description (612)6.2.5B.4.1Initial conditions (612)6.2.5B.4.2Test procedure (612)6.2.5B.4.3Message contents (613)6.2.5B.5Test requirement (613)6.2.5E Configured UE transmitted Output Power for UE category 0 (614)6.2.5E.4.1Initial conditions (614)6.2.5E.4.2Test procedure (614)6.2.5E.4.3Message contents (614)6.2.5E.5Test requirement (615)6.2.5EA Configured UE transmitted Power for UE category M1 (615)6.2.5EA.1Test purpose (615)6.2.5EA.2Test applicability (615)6.2.5EA.3Minimum conformance requirements (615)6.2.5EA.4Test description (616)6.2.5EA.4.1Initial condition (616)6.2.5EA.4.2Test procedure (617)6.2.5EA.4.3Message contents (617)6.2.5EA.5Test requirements (617)6.2.5F Configured UE transmitted Output Power for UE category NB1 (618)6.2.5F.1Test purpose (618)6.2.5F.2Test applicability (618)6.2.5F.3Minimum conformance requirements (618)6.2.5F.4Test description (619)6.2.5F.4.1Initial conditions (619)6.2.5F.4.2Test procedure (620)6.2.5F.4.3Message contents (620)6.2.5F.5Test requirement (620)6.2.5G Configured UE transmitted Output Power for V2X Communication (620)6.2.5G.1Configured UE transmitted Output Power for V2X Communication / Non-concurrent with E-UTRA uplink transmission (621)6.2.5G.1.1Test purpose (621)6.2.5G.1.2Test applicability (621)6.2.5G.1.3Minimum conformance requirements (621)6.2.5G.1.4Test description (622)6.2.5G.1.4.1Initial conditions (622)6.2.5G.1.4.2Test procedure (622)6.2.5G.1.4.3Message contents (622)6.2.5G.1.5Test requirements (622)6.2.5G.2Configured UE transmitted Output Power for V2X Communication / Simultaneous E-UTRAV2X sidelink and E-UTRA uplink transmission (622)6.2.5G.2.1Test purpose (623)6.2.5G.2.2Test applicability (623)6.2.5G.2.3Minimum conformance requirements (623)6.2.5G.2.4Test description (625)6.2.5G.2.4.1Initial conditions (625)6.2.5G.2.4.2Test procedure (626)6.2.5G.2.4.3Message contents (626)6.2.5G.2.5Test requirements (626)6.3Output Power Dynamics (627)6.3.1Void (627)6.3.2Minimum Output Power (627)6.3.2.1Test purpose (627)6.3.2.2Test applicability (627)6.3.2.3Minimum conformance requirements (627)6.3.2.4Test description (627)6.3.2.4.1Initial conditions (627)6.3.2.4.2Test procedure (628)6.3.2.4.3Message contents (628)6.3.2.5Test requirement (628)6.3.2A Minimum Output Power for CA (629)6.3.2A.0Minimum conformance requirements (629)6.3.2A.1Minimum Output Power for CA (intra-band contiguous DL CA and UL CA) (629)6.3.2A.1.1Test purpose (629)6.3.2A.1.4.2Test procedure (631)6.3.2A.1.4.3Message contents (631)6.3.2A.1.5Test requirements (631)6.3.2A.2Minimum Output Power for CA (inter-band DL CA and UL CA) (631)6.3.2A.2.1Test purpose (631)6.3.2A.2.2Test applicability (632)6.3.2A.2.3Minimum conformance requirements (632)6.3.2A.2.4Test description (632)6.3.2A.2.4.1Initial conditions (632)6.3.2A.2.4.2Test procedure (633)6.3.2A.2.4.3Message contents (633)6.3.2A.2.5Test requirements (633)6.3.2A.3Minimum Output Power for CA (intra-band non-contiguous DL CA and UL CA) (634)6.3.2A.3.1Test purpose (634)6.3.2A.3.2Test applicability (634)6.3.2A.3.3Minimum conformance requirements (634)6.3.2A.3.4Test description (634)6.3.2A.3.4.1Initial conditions (634)6.3.2A.3.4.2Test procedure (635)6.3.2A.3.4.3Message contents (635)6.3.2A.3.5Test requirements (635)6.3.2B Minimum Output Power for UL-MIMO (636)6.3.2B.1Test purpose (636)6.3.2B.2Test applicability (636)6.3.2B.3Minimum conformance requirements (636)6.3.2B.4Test description (636)6.3.2B.4.1Initial conditions (636)6.3.2B.4.2Test procedure (637)6.3.2B.4.3Message contents (637)6.3.2B.5Test requirement (637)6.3.2E Minimum Output Power for UE category 0 (638)6.3.2E.1Test purpose (638)6.3.2E.2Test applicability (638)6.3.2E.3Minimum conformance requirements (638)6.3.2E.4Test description (638)6.3.2E.4.1Initial conditions (638)6.3.2E.4.2Test procedure (639)6.3.2E.4.3Message contents (639)6.3.2E.5Test requirement (639)6.3.2EA Minimum Output Power for UE category M1 (639)6.3.2EA.1Test purpose (639)6.3.2EA.2Test applicability (640)6.3.2EA.3Minimum conformance requirements (640)6.3.2EA.4Test description (640)6.3.2EA.4.1Initial condition (640)6.3.2EA.4.2Test procedure (641)6.3.2EA.4.3Message contents (641)6.3.2EA.5Test requirements (641)6.3.2F Minimum Output Power for category NB1 (641)6.3.2F.1Test purpose (641)6.3.2F.2Test applicability (641)6.3.2F.3Minimum conformance requirements (642)6.3.2F.4Test description (642)6.3.2F.4.1Initial conditions (642)6.3.2F.4.2Test procedure (643)6.3.2F.4.3Message contents (643)6.3.2F.5Test requirements (643)6.3.3Transmit OFF power (643)6.3.3.5Test requirement (644)6.3.3A UE Transmit OFF power for CA (644)6.3.3A.0Minimum conformance requirements (644)6.3.3A.1UE Transmit OFF power for CA (intra-band contiguous DL CA and UL CA) (645)6.3.3A.1.1Test purpose (645)6.3.3A.1.2Test applicability (645)6.3.3A.1.3Minimum conformance requirements (645)6.3.3A.1.4Test description (645)6.3.3A.1.5Test Requirements (645)6.3.3A.2UE Transmit OFF power for CA (inter-band DL CA and UL CA) (646)6.3.3A.2.1Test purpose (646)6.3.3A.2.2Test applicability (646)6.3.3A.2.3Minimum conformance requirements (646)6.3.3A.2.4Test description (646)6.3.3A.2.5Test Requirements (646)6.3.3A.3UE Transmit OFF power for CA (intra-band non-contiguous DL CA and UL CA) (646)6.3.3A.3.1Test purpose (646)6.3.3A.3.2Test applicability (646)6.3.3A.3.3Minimum conformance requirements (647)6.3.3A.3.4Test description (647)6.3.3A.3.5Test Requirements (647)6.3.3B UE Transmit OFF power for UL-MIMO (647)6.3.3B.1Test purpose (647)6.3.3B.2Test applicability (647)6.3.3B.3Minimum conformance requirement (647)6.3.3B.4Test description (647)6.3.3B.5Test requirement (648)6.3.3C 6486.3.3D UE Transmit OFF power for ProSe (648)6.3.3D.0Minimum conformance requirements (648)6.3.3D.1UE Transmit OFF power for ProSe Direct Discovery (648)6.3.3D.1.1Test purpose (649)6.3.3D.1.2Test applicability (649)6.3.3D.1.3Minimum Conformance requirements (649)6.3.3D.1.4Test description (649)6.3.3D.1.5Test requirements (650)6.3.3E UE Transmit OFF power for UE category 0 (650)6.3.3E.1Test purpose (650)6.3.3E.2Test applicability (650)6.3.3E.3Minimum conformance requirement (650)6.3.3E.4Test description (651)6.3.3E.5Test requirement (651)6.3.3EA UE Transmit OFF power for UE category M1 (651)6.3.3EA.1Test purpose (651)6.3.3EA.2Test applicability (651)6.3.3EA.3Minimum conformance requirements (651)6.3.3EA.4Test description (651)6.3.3EA.5Test requirements (652)6.3.3F Transmit OFF power for category NB1 (652)6.3.3F.1Test purpose (652)6.3.3F.2Test applicability (652)6.3.3F.3Minimum conformance requirement (652)6.3.3F.4Test description (652)6.3.3F.5Test requirement (652)6.3.4ON/OFF time mask (652)6.3.4.1General ON/OFF time mask (652)6.3.4.1.1Test purpose (652)6.3.4.1.2Test applicability (653)。

International Journal of System Control and Information Processing.

2. Modeling of Boiler Superheated Steam Temperature Plant
In this section, a coal-fired boiler (#2 furnace 600MW unit) in a certain thermal power plant is taken as an example to introduce the model structure chart of reconstructive boiler steam temperature system (Fig.1). The warm water reduction system of super heater uses two-level four-point cooling structure. First level desuperheater is arranged at two sides between low temperature superheater and dividing panel superheater; Second level desuperheater is arranged at two sides between rear platen superheater and steam lead pipe connected. Except the influence of swinging burner, two-level spray desuperheating adjustment is mainly applied in the main steam temperature control system.
An improved single neuron self-adaptive PID control scheme of superheated steam temperature control system

碳足迹标准PAS 2050

PUBLICLY AVAILABLE SPECIFICATIONPAS 2050:2008Specification for the assessment of the life cycle greenhouse gas emissions of goods and servicesICS code: 13.020.40NO COPYING WITHOUT BSI PERMISSION EXCEPT AS PERMITTED BY COPYRIGHT LAWPAS 2050:2008Publishing and copyright informationThe BSI copyright notice displayed in this document indicates when the documentwas last issued.This Publicly Available Specification comes into effect on 29 October 2008.© BSI October 2008ISBN 978 0 580 50978 0© BSI October 2008PAS 2050:2008ContentsPage Foreword (ii)0 Introduction (iv)1 Scope (1)2 Normative references (1)3 Terms and definitions (2)4 Principles and implementation (6)5 Emission sources, offsetting and unit of analysis (7)6 System boundary (12)7 Data (17)8 Allocation of emissions (22)9 Calculation of the GHG emissions of products (24)10 Claims of conformity (25)AnnexesAnnex A Global warming potential (normative) (26)Annex B Calculation of the weighted average impact of emissionsarising from the use phase and final disposal phase of products (normative) (29)Annex C Calculation of the weighted average impact ofcarbon storage in products (normative) (30)Annex D Calculation of emissions arising from recyclable materialinputs (normative) (31)Annex E Default land use change values for selected countries (normative) (32)Bibliography (34)Further reading (34)© BSI October 2008iPAS 2050:2008© BSI October 2008iiIt has been assumed in the preparation of this PAS that the execution of its provisions will be entrusted to a competent person or persons for whose use it has been produced.Acknowledgement is given to the followingorganizations and individuals who assisted with the development of this specification:Technical author Dr Graham Sinden Organizations •Carbon Trust •Defra•BSI Standards Solutions •E4tech •LEK Consulting •Booz Allen Hamilton •Orion Innovations PAS Steering Group Professor Jim Skea (Chair)UK Energy Research CentreForewordThis Publicly Available Specification (PAS) has been prepared by BSI to specifyrequirements for assessing the life cycle greenhouse gas emissions (GHG) of goods and services. The development of this PAS was co-sponsored by the Carbon Trust and the Department for Environment, Food and Rural Affairs (Defra).Dr Paul JefferissIndependent expert (and Carbon Trust Director)Professor Roland Clift University of Surrey Professor Tim Jackson University of SurreyTerence IlottUK Department for Environment, Food and Rural Affairs Mark KenberThe Climate GroupMichael Roberts/Rhian KellyConfederation of British Industry Stephen ReesonUK Food and Drink Federation Nick Monger-GodfreyJohn Lewis plc.; Park Royal plc.Nigel DickieHeinz UK & IrelandKen Double/Dan StaniaszekEnergy Saving TrustPAS 2050:2008© BSIOctober 2008iiiThis PAS has been prepared and published by BSI which retains its ownership and copyright. BSI reserves the right to withdraw or amend this PAS on receipt ofauthoritative advice that it is appropriate to do so. This PAS will be reviewed at intervals not exceeding two years, and any amendments arising from the review will be published as an amended Publicly Available Specification and publicized in Update Standards. This PAS is not to be regarded as a British Standard,European Standard or International Standard. In the event that this PAS is put forward to form the basis of a full British Standard, European Standard or International Standard, it will be withdrawn.Presentational conventionsThe provisions of this PAS are presented in roman (i.e. upright) type. Its requirements are expressed insentences in which the principal auxiliary verb is “shall”.Its recommendations are expressed in sentences in which the principal auxiliary verb is “should”.Commentary, explanation and general informative material, e.g. Notes, is presented in italic type, and does not constitute a normative element.Contractual and legal considerationsThis publication does not purport to include all thenecessary provisions of a contract. Users are responsible for its correct application.Compliance with this PAS does not in itself confer immunity from legal obligations.PAS 2050:2008© BSI October 2008iv0Introduction0.1 General informationClimate change has been identified as one of the greatest challenges facing nations, governments,business and citizens over future decades (IPCC 2007[1]). Past and current actions, including the release of carbon dioxide (CO 2) and other greenhouse gases through human activities such as the burning of fossil fuels, emissions from chemical processes, and other sources of anthropogenic greenhouse gases, will have an effect on future global climate.While greenhouse gas (GHG) emissions are oftenviewed at global, national, corporate or organizational levels, emissions within these groupings can arise from supply chains within business, between businesses, and between nations. The GHG emissions associated with goods and services reflect the impact of processes,materials and decisions occurring throughout the life cycle of the goods and services.PAS 2050 has been developed in response to broadcommunity and industry desire for a consistent method for assessing the life cycle GHG emissions of goods and services. Life cycle GHG emissions are the emissions that are released as part of the processes of creating,modifying, transporting, storing, using, providing,recycling or disposing of goods and services. PAS 2050recognizes the potential for organizations to use this method to deliver improved understanding of the GHG emissions arising from their supply chains, and to provide a common basis for the comparison and communication of results arising from the use of PAS 2050. Although there is no requirement forcommunication or standardization of communication techniques in this specification, this PAS supports the assessment of life cycle GHG emissions of goods and services that can be later reported and communicated to stakeholders, including consumers. Where an organization implementing this PAS chooses to communicate specific result of the assessment of GHG emissions, it is required to make other information available as specified in this PAS.0.2 Background and benefitsPAS 2050 builds on existing life cycle assessment methods established through BS EN ISO 14040 and BS EN ISO 14044 by specifying requirements for the assessment of the life cycle GHG emissions of goods and services. These requirements further clarify the implementation of the above standards in relation to the assessment of GHG emissions of goods and services,and establish additional principles and techniques thataddress essential aspects of GHG assessment, including:a)business-to-business and business-to-consumer use of partial GHG assessment data in full GHG assessments of goods and services;b)scope of greenhouse gases to be included;c)criteria for global warming potential data;d)treatment of emissions from land use change, and biogenic and fossil carbon sources; e)treatment of the impact of carbon storage in products, and offsetting;f)requirements for the treatment of GHG emissions arising from specific processes;g)data requirements and accounting for emissions from renewable energy generation; and h)claims of conformity.This PAS is intended to benefit organizations,businesses and other stakeholders by providing a clear and consistent method for the assessment of the life cycle GHG emissions associated with goods and services. Specifically, this PAS provides the following benefits:a)For organizations that supply goods and services,this PAS:•allows internal assessment of the existing life cycle GHG emissions of goods and services;•facilitates the evaluation of alternative product configurations, sourcing and manufacturing methods, raw material choices and supplier selection on the basis of the life cycle GHG emissions associated with goods and services;•provides a benchmark for ongoing programmes aimed at reducing GHG emissions;•allows for a comparison of goods or services using a common, recognized and standardized approach to life cycle GHG emissions assessment; and •supports reporting on corporate responsibility.b)For consumers of goods and services, this PAS:•provides a common basis for reporting and communicating the results of life cycle GHG emissions assessments that supports comparison and uniformity of understanding; and•provides an opportunity for greater consumer understanding of life cycle GHG emissions when making purchasing decisions and using goods and services.PAS 2050:2008© BSI October 20081This PAS specifies requirements for the assessment of the life cycle GHG emissions of goods and services (collectively referred to as “products”) based on key life cycle assessment techniques and principles. This PAS is applicable to organizations assessing the GHG emissions of products across their life cycle, and to organizations assessing the cradle-to-gate GHG emissions of products.Requirements are specified for identifying the system boundary, the sources of GHG emissions associated with products that fall inside the system boundary, the data requirements for carrying out the analysis,and the calculation of the results.This PAS addresses the single impact category of global warming, and does not assess other potential social,economic and environmental impacts arising from the provision of products, such as non-greenhouse gas emissions, acidification, eutrophication, toxicity,biodiversity, labour standards or other social, economic and environmental impacts that may be associated with the life cycle of products. The life cycle GHG emissions of products, calculated using this PAS, do not provide an indicator of the overall environmental impact of these products, such as may result from other types of life cycle assessment.This PAS does not include product category-specific rules for goods and services; however, it is intended that selected product category-specific rules for goods and services, developed in accordance with BS ISO 14025, will be adopted where available, as specified in this PAS.It is one of the intentions of this PAS to allow for the comparison of GHG emissions between products, and to enable the communication of this information.However, this PAS does not specify requirements for communication.The following referenced documents are indispensable for the application of this PAS. For dated references,only the edition cited applies. For undated references,the latest edition of the referenced document (including any amendments) applies.BS EN ISO 14021,Environmental labels anddeclarations – Self-declared environmental claims (Type II environmental labelling)BS EN ISO 14044:2006, Environmental management –Life cycle assessment – Requirements and guidelines,Clause 4.3.4.3BS EN ISO/IEC 17050-1,Conformity assessment –Supplier’s declaration of conformity – Part 1: General requirementsISO/TS 14048:2002,Environmental management – Life cycle assessment – Data documentation format,Clause 5.2.2IPCC 2006,Guidelines for National Greenhouse Gas Inventories.National Greenhouse Gas Inventories Programme, Intergovernmental Panel on Climate ChangeNote Subsequent amendments to IPCC 2006 also apply.IPCC 2007,Climate Change 2007: The Physical Science Basis.Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z.Chen, M. Marquis, K.B. Averyt, M. Tignor and ler (eds.)]. Cambridge University Press, Cambridge,United Kingdom and New York, NY , USA, 996 pp.,Chapter 2, Table 2.14Note Subsequent amendments to IPCC 2007 also apply.1Scope2Normative referencesFor the purposes of this PAS the following terms and definitions apply.3.1 allocationpartitioning the input or output flows of a process or a product system between the product system under study and one or more other product systems [BS EN ISO 14044:2006, 3.17]3.2 anticipated life cycle greenhouse gas emissionsinitial estimate of greenhouse gas (see 3.26for adefinition of greenhouse gases) emissions for a product (see 3.37for a definition of product) that is calculated using secondary data (see 3.43for a definition ofsecondary data) or a combination of primary (see 3.36for a definition of primary activity data) and secondary data, for all processes used in the life cycle of the product3.3 biogenicderived from biomass, but not fossilized or from fossil sources (see 3.22for a definition of fossil)3.4 biomassmaterial of biological origin, excluding materialembedded in geological formations or transformed to fossil[Adapted from CEN/TR 14980:2004, 4.3]3.5 business-to-businessprovision of inputs, including products, to another party that is not the end user3.6 business-to-consumerprovision of inputs, including products, to the end user3.7 capital goodsgoods, such as machinery, equipment and buildings,used in the life cycle of products3.8 carbon dioxide equivalent (CO 2e)unit for comparing the radiative forcing of a GHG to carbon dioxide[BS ISO 14064-1:2006, 2.19]Note 1The carbon dioxide equivalent value is calculated by multiplying the mass of a given GHG by its global warming potential (see 3.25for a definition of global warming potential).Note 2 Greenhouse gases, other than CO 2, are converted to their carbon dioxide equivalent value on the basis of their per unit radiative forcing using 100-year global warming potentials defined by the Intergovernmental Panel on Climate Change (IPCC).3.9 carbon sequestrationremoval of carbon from the atmosphere3.10 carbon storageretaining carbon of biogenic or atmospheric origin in a form other than as an atmospheric gas3.11 combined heat and power (CHP)simultaneous generation in one process of useable thermal, electrical and/or mechanical energy3.12 competent personperson with training, experience or knowledge and other qualities, and with access to the required tools,equipment and information, sufficient to enable them to carry out a defined task3.13 consumableancillary input that is necessary for a process to occur but that does not form a tangible part of the product or co-products arising from the processNote 1Consumables include lubricating oil, tools and other rapidly wearing inputs to a process. Consumables differ from capital goods in that they have an expected life of one year or less, or a need to replenish on a one year or less basis.Note 2Fuel and energy inputs to the life cycle of a product are not considered consumables.3.14 consumeruser of goods or services3.15 co-productany of two or more products from the same unit process or product system [BS EN ISO14044:2006, 3.10]Note Where two or more products can be produced from a unit process, they are considered co-products only where one cannot be produced without the other being produced.3.16 data qualitycharacteristics of data that relate to their ability to satisfy stated requirements [BS EN ISO14044:2006, 3.19]PAS 2050:2008© BSI October 200823Terms and definitionsPAS 2050:2008© BSI October 200833.17 downstream emissionsGHG emissions associated with processes that occur in the life cycle of a product subsequent to the processes owned or operated by the organization implementing this PAS3.18 economic valuemarket value of a product, co-product or waste (see 3.50for a definition of waste) at the point of production3.19 emission factoramount of greenhouse gases emitted, expressed as carbon dioxide equivalent and relative to a unit of activityNote For example, kgCO 2e per unit input. Emission factor data would be obtained from secondary data sources.3.20 emissionsrelease to air and discharges to water and land that result in GHGs entering the atmosphere3.21 environmentally extended input–output (EEIO) analysismethod of estimating the GHG emissions (and other environmental impacts) arising from sectors within an economy through the analysis of economic flows Note Alternative terms, such as economic input-output life cycle assessment (EIO-LCA), input output based life cycle assessment (IOLCA) and hybrid life cycle assessment (HLCA)refer to different approaches to implementing EEIO analysis.3.22 fossilderived from fossil fuel or another fossil source,including peat[Adapted from IPCC 2006 Guidelines for National Greenhouse Gas Inventories, Glossary , see Clause 2]3.23 functional unitquantified performance of a product system for use as a reference unit[BS EN ISO 14044:2006, 3.20]3.24 GHG emissionsrelease of GHGs to the atmosphere3.25 global warming potential (GWP)factor describing the radiative forcing impact of one mass-based unit of a given greenhouse gas relative to an equivalent unit of carbon dioxide over a given period of time[BS ISO 14064-1:2006, 2.18]Note Carbon dioxide is assigned a GWP of 1, while the GWP of other gases is expressed relative to the GWP of carbon dioxide from fossil carbon sources. Annex A contains global warming potentials for a 100-year time period produced by the Intergovernmental Panel on Climate Change. Carbon dioxide arising from biogenic sources of carbon is assigned a GWP of zero in specific circumstances specified in this PAS.3.26 greenhouse gases (GHGs)gaseous constituents of the atmosphere, both natural and anthropogenic, that absorb and emit radiation at specific wavelengths within the spectrum of infrared radiation emitted by the Earth's surface, the atmosphere, and cloudsNote The GHGs included in this PAS are specified inAnnex A.3.27 inputproduct, material or energy flow that enters a unit process[BS EN ISO 14040:2006, 3.21]3.28 intermediate productoutput from a unit process that is an input to other unit processes involving further transformation within the system3.29 International Reference Life Cycle Data System (ILCD)series of technical guidance documents with quality, method, nomenclature, documentation and review requirements for quality ensured life cycle data and studies, coordinated for Europe by the European Commission’s Joint Research Centre [2]3.30 life cycleconsecutive and interlinked stages of a product system, from raw material acquisition or generation of natural resources to end of life, inclusive of any recycling or recovery activity[Adapted from BS EN ISO 14040:2006, 3.1]3.31 life cycle assessment (LCA)compilation and evaluation of inputs, outputs and potential environmental impacts of a product system throughout its life cycle[BS EN ISO 14040:2006, 3.2]3.32 life cycle GHG emissionssum of greenhouse gas emissions resulting from all stages of the life cycle of a product and within the specified system boundaries of the productNote This includes all emissions that are released as partof the processes within the boundary of the life cycle of the product, including obtaining, creating, modifying, transporting, storing, operating, using and end of life disposal of the product.3.33 material contributioncontribution from any one source of GHG emissions of more than 1% of the anticipated life cycle GHG emissions associated with a productNote A materiality threshold of 1% has been establishedto ensure that very minor sources of life cycle GHG emissions do not require the same treatment as more significant sources.3.34 offsettingmechanism for claiming a reduction in GHG emissions associated with a process or product through the removal of, or preventing the release of, GHG emissions in a process unrelated to the life cycle ofthe product being assessedNote An example is the purchase of Certified Emission Reductions generated by Clean Development Mechanism projects under the Kyoto Protocol [3].3.35 outputproduct, material or energy that leaves a unit process [Adapted from BS EN ISO 14044:2006, 3.25]Note Materials may include raw materials, intermediate products, co-products, products and emissions.3.36 primary activity dataquantitative measurement of activity from a product’s life cycle that, when multiplied by an emission factor, determines the GHG emissions arising from a process Note 1Examples of primary activity data include the amount of energy used, material produced, service provided or area of land affected.Note 2Primary activity data sources are typically preferable to secondary data sources as the data will reflect the specific nature/efficiency of the process, and the GHG emissions associated with the process.Note 3Primary activity data does not include emission factors.3.37 productany good or serviceNote Services have tangible and intangible elements.The provision of a service can involve, for example, the following:a)an activity performed on a consumer-supplied tangibleproduct (e.g. automobile to be repaired);b)an activity performed on a consumer-supplied intangibleproduct (e.g. the income statement needed to preparea tax return);c)the delivery of an intangible product (e.g. the delivery ofinformation in the context of knowledge transmission); d)the creation of ambience for the consumer (e.g. in hotelsand restaurants);PAS 2050:2008© BSI October 2008 4e)software consists of information and is generallyintangible and can be in the form of approaches,transactions or procedures.[Adapted from BS ISO 14040:2006, 3.9]3.38 product categorygroup of products that can fulfil equivalent functions [BS ISO 14025:2006, 3.12]3.39 product category rules (PCRs)set of specific rules, requirements and guidelines for developing Type III environmental declarations for one or more product categories[BS ISO 14025:2006, 3.5]3.40 product systemcollection of unit processes with elementary and product flows, performing one or more defined functions, that models the life cycle of a product [BS EN ISO 14040:2006, 3.28]3.41 raw materialprimary or secondary material that is used to produce a productNote Secondary material includes recycled material.[BS EN ISO 14040:2006, 3.15]3.42 renewable energyenergy from non-fossil energy sources: wind, solar, geothermal, wave, tidal, hydropower, biomass, landfill gas, sewage treatment plant gas and biogases [Adapted from Directive 2001/77/EC, Article 2[4]]3.43 secondary datadata obtained from sources other than direct measurement of the processes included in the life cycle of the productNote Secondary data is used when primary activity datais not available or it is impractical to obtain primary activity data.3.44 system boundaryset of criteria specifying which unit processes are part of a product system[BS EN ISO 14040:2006, 3.32]3.45 unit processsmallest portion of a life cycle for which data are analysed when performing a life cycle assessment 3.46 upstream emissionsGHG emissions associated with processes that occurin the life cycle of a product prior to the processes owned, operated or controlled by the organization implementing this PAS3.47 use phasethat part of the life cycle of a product that occurs between the transfer of the product to the consumer and the end of life of the productNote For services, the use phase includes the provision of the service.3.48 use profilecriteria against which the GHG emissions arising from the use phase are determined3.49 useful energyenergy that meets a demand by displacing another source of energyNote For example, where heat production from a CHP unit is utilized to meet a demand for heat that was previously met by another form of energy, or meets a new demand for heat that would have required additional energy input, then the heat from the CHP is providing useful energy. Had the heat from the CHP not met a demand, but instead been dissipated (e.g. vented to the atmosphere), the heat would not be considered useful energy (in which case no emissions from the CHP would be assigned to the heat production). 3.50 wastematerials, co-products, products or emissions which the holder discards or intends, or is required to, discard4.1 General requirementsAssessment of the GHG emissions of products shall be carried out using LCA techniques. Unless otherwise indicated, the assessment of the life cycle GHG emissions of products shall be made using the attributional approach, i.e. by describing the inputs and their associated emissions attributed to the delivery of a specified amount of the product functional unit.Note LCA techniques are specified in BS EN ISO 14040 and BS EN ISO 14044. Where the approach described in these standards is incompatible with the requirements of this PAS, the requirements of this PAS take precedence.4.2 PrinciplesOrganizations claiming conformity with this PAS shall ensure that the assessment of the life cycle GHG emissions of a product is complete, and shall be able to demonstrate that the following principles have been taken into consideration when carrying outthe assessment:a)Relevance:select GHG sources, carbon storage,data and methods appropriate to the assessment of the GHG emissions arising from products;b)Completeness:include all specified GHG emissionsand storage that provide a material contributionto the assessment of GHG emissions arising fromproducts;c)Consistency:enable meaningful comparisons inGHG-related information;d)Accuracy:reduce bias and uncertainties as far asis practical;e)Transparency:where the results of life cycle GHGemissions assessment carried out in accordancewith this PAS are communicated to a third party,the organization communicating these resultsshall disclose GHG emissions-related informationsufficient to allow such third parties to makeassociated decisions with confidence.Note The above principles are adapted fromBS ISO14064-1:2006,Clause 3.4.3 Product differentiationThe GHG emissions assessment specified in this PAS shall apply to the product for which the assessment is conducted. Where similar products are being assessed, a separate assessment shall be carried out for each product that is provided to a third party, unless the similar products are provided in such a manner as to make them indistinguishable from each other by the third party.4.4 Supporting dataData supporting the assessment of life cycle GHG emissions, including but not limited to, product and process boundaries, materials, emission factors and emissions, and other data as required in this PAS, shall be documented and a record maintained in a format suitable for analysis and verification for the greater of either five years or the life expectancy of the product. Note Data records should be made available to support claims of conformity no matter what form of verificationis chosen. The basis of support for self verification of conformity is no different from that required for other party verification or independent third party certification (see Clause 10).4.5 Implementation of this PASThe assessment of life cycle GHG emissions for products shall be carried out as either:a)a business-to-consumer assessment, which includesthe emissions arising from the full life cycle of the product; orb)a business-to-business assessment, which includesthe GHG emissions released up to and including the point where the input arrives at a new organization (including all upstream emissions).Note 1The above two approaches are respectively referred to as “cradle-to-grave” approach (see BS EN ISO 14044) and “cradle-to-gate” approach (see BS EN ISO 14040).Note 2See 6.2for the assessment of the emissions arising from part of the life cycle of the product for business-to-business assessment purposes.4Principles and implementation。

关于粒子群优化的英文摘要范文

关于粒子群优化的英文摘要范文Particle Swarm Optimization (PSO) is a powerful and widely-used computational method inspired by the social behavior of birds and fish. Developed by Russell Eberhart and James Kennedy in 1995, it falls under the category of swarm intelligence algorithms, which leverage the collective behavior of decentralized systems to solve optimization problems. PSO is particularly effective for continuous optimization tasks, where it seeks to find optimal or near-optimal solutions from a defined search space.The core concept of PSO involves a group of individual entities, referred to as particles, which traverse the solution space by adjusting their positions based on their own experiences and those of their neighbors. Each particle maintains a record of its best-known position, as well as the best-known position of its group. This dual memory guides theparticles in their movement and allows them to iteratively converge towards the optimal solution. The movement of each particle is influenced by two primary components: cognitive and social. The cognitive component draws from the particle's own history, while the social component draws from thegroup's best-known position.A distinctive feature of PSO is its simplicity and ease of implementation, which makes it suitable for a broad range of applications, including function optimization, neural network training, and engineering design problems. The PSO algorithm can be adjusted with parameters such as the number of particles, their velocities, and the coefficients controlling the cognitive and social influences. These parameters can significantly affect the convergence speed and accuracy of the method.Moreover, PSO has been adapted and enhanced through various strategies, such as employing different topologiesfor particle interactions, incorporating constraints for specific problem domains, and hybridizing with other optimization techniques to improve performance. These modifications allow PSO to effectively tackle more complexand multi-dimensional problems.Despite its advantages, PSO has some limitations, such as the potential for premature convergence and sensitivity to parameter settings. Researchers continue to investigate waysto overcome these challenges, leading to the development of improved variants and hybrid methodologies that combine the strengths of PSO with other optimization strategies.In summary, Particle Swarm Optimization represents a significant advancement in the field of optimization algorithms. Its biologically inspired mechanism, adaptability, and versatility have made it a popular choice for solving a wide array of real-world problems across diverse disciplines. As ongoing research continues to refine and enhance PSO, itis poised to remain an essential tool for optimization in the future.。

节温器布置形式对质子交换膜燃料电池电堆冷却性能的影响

2021年第1期【摘要】运用一维仿真软件建立质子交换膜燃料电池液冷系统模型，研究了不同节温器布置形式对系统的性能影响。

对某额定功率30kW 的燃料电池发动机在4个不同工况点进行液冷系统散热特性仿真：在节温器一进两出的布置形式下仿真结果与试验数据基本一致，电堆出口温度仿真值与实测值相对误差分别为0.5%、1.5%、2.4%、4.9%；节温器两进一出的布置形式下液冷系统中冷却液温度变化平缓而均匀，前10s 和第10~50s 之间的温度变化率之差较一进两出形式低36.85%，更有利于电堆的长期高效运行。

主题词：节温器质子交换膜燃料电池一维仿真液冷系统中图分类号：TM911.42文献标识码：ADOI:10.19620/ki.1000-3703.20200355Effects of Thermostat Layouts on Cooling Performance of ProtonExchange Membrane Fuel Cell StacksCheng Zifeng 1,2,Li Ming 1,2,Guo Qin 3,Ren Yan 4,Qin Guihe 3（1.State Key Laboratory of Automotive Simulation and Control,Jilin University,Changchun 130022;2.College of Automotive Engineering,Jilin University,Changchun 130022;3.College of Computer Science and Technology,Jilin University,Changchun 130022;4.Henan College of Forestry,Luoyang 471000）【Abstract 】In this research,one-dimensional simulation software is used to establish the liquid cooling system model of Proton Exchange Membrane Fuel Cell (PEMFC),and the effects of different thermostat layouts on the performance of the system are studied.Heat dissipation characteristics of a liquid cooling system for a fuel cell engine with rated power of 30kW are simulated at four different operating points,the results are basically consistent with test data under one-in-two-out arrangement of the thermostat.The relative errors between the simulated value and the measured value of the stack outlet temperature are 0.5%,1.5%,2.4%and 4.9%respectively,whereas the change of the coolant temperature in the liquid cooling system is gentle and even under the two-in-one-out arrangement of the thermostat.The difference of temperature change rate between the first 10seconds and the range from the 10th second to the 50th second is 36.85%lower than thatof one-in-two-out arrangement,which is more conducive to the long-term and efficient operation of the stack.Key words:Thermostat,Proton exchange membrane fuel cell,One-dimensional simulation,Liquid cooling system程子枫1,2李明1,2郭勤3任雁4秦贵和3（1.吉林大学，汽车仿真与控制国家重点实验室，长春130022；2.吉林大学，汽车工程学院，长春130022；3.吉林大学，计算机科学与技术学院，长春130022；4.河南林业职业学院，洛阳471000）*基金项目：吉林省科技厅技术攻关项目（20190302120GX ）；汽车仿真与控制国家重点实验室自由探索项目（ascl-zytsxm-202029）。

基于深度多级小波U-Net的车牌雾图去雾算法

第49卷第6期2022年6月Vol.49，No.6Jun.2022湖南大学学报（自然科学版）Journal of Hunan University（Natural Sciences）基于深度多级小波U-Net的车牌雾图去雾算法陈炳权†，朱熙，汪政阳，梁寅聪（吉首大学信息科学与工程学院，湖南吉首416000）摘要：为了解决雾天拍摄的车牌图像边缘模糊、色彩失真的问题，提出了端到端的基于深度多级小波U-Net的车牌雾图去雾算法.以MWCNN为去雾网络的主体框架，利用“SOS”增强策略和编解码器之间的跨层连接整合小波域中的特征信息，采用离散小波变换的像素-通道联合注意力块降低去雾车牌图像中的雾度残留.此外，利用跨尺度聚合增强块补充小波域图像中缺失的空间域图像信息，进一步提高了去雾车牌图像质量.仿真实验表明，该网络在结构相似度和峰值信噪比上具有明显的优势，在处理合成车牌雾图和实际拍摄的车牌雾图上，去雾效果表现良好.关键词：车牌雾图去雾；MWCNN；“SOS”增强策略；跨层连接；注意力；跨尺度聚合中图分类号：TP391.41文献标志码：ADehazing Algorithm of License Plate Fog Image Basedon Deep Multi-level Wavelet U-NetCHEN Bingquan†，ZHU Xi，WANG Zhengyang，LIANG Yincong（College of Information Science and Engineering，Jishou University，Jishou416000，China）Abstract：To solve the problem of edge blurring and color distortion of license plate images taken in foggy weather，an end-to-end depth multilevel wavelet U-Net based algorithm for license plate fog image removal is pre⁃sented.Taking MWCNN as the main frame work of the defogging network，the feature information in the wavelet do⁃main is integrated using the“SOS”enhancement strategy and the cross-layer connection between the codec.The pixel-channel joint attention block of the discrete wavelet transform is used to reduce the fog residue in the defrosted license plate image.In addition，the cross-scale aggregation enhancement blocks are used to supplement the missing spatial domain image information in the wavelet domain image，which further improves the quality of the defogging li⁃cense plate image.The simulation results show that the network has obvious advantages in structural similarity and peak signal-to-noise ratio，and it performs well in dealing with the composite plate fog image and the actual photo⁃graphed plate fog image.Key words：license plate fog image defogging；MWCNN；“SOS”enhancement strategy；cross-layer connection；attention mechanism；cross-scale aggregation∗收稿日期：2021-11-01基金项目：国家自然科学基金资助项目（No.62141601），National Natural Science Foundation of China（No.62141601）；湖南省教育厅重点资助项目（21A0326），The SRF of Hunan Provincial Education Department（No.21A0326）作者简介：陈炳权（1972-），男，湖南常德人，吉首大学副教授，硕士生导师，博士†通信联系人，E-mail：****************文章编号：1674-2974（2022）06-0124-11DOI：10.16339/ki.hdxbzkb.2022293第6期陈炳权等：基于深度多级小波U-Net的车牌雾图去雾算法在大雾天气下使用光学成像器件（如相机、监控摄像头等）对目标场景或物体进行拍摄时，往往会使得图像对比度低，边缘、字符等细节信息模糊.图像去雾是图像处理中典型的不适定问题，旨在从雾天图像中复原出相应的无雾图像，作为一种提升图像质量的方法，已被广泛应用于图像分类、识别、交通监控等领域.近年来，针对不同场景（室内家居场景、室外自然场景、交通道路场景、夜间雾霾场景等）下均匀雾度或非均匀雾度图像的去雾技术引起了广泛关注与研究，但由于实际雾霾对图像影响的复杂多变性，从真实的雾天图像中复原无雾图像仍具有诸多挑战性.图像去雾技术发展至今，主要分为以下三类：基于数学模型的去雾技术，如直方图均衡［1］、小波变换［2］、色彩恒常性理论［3］（Retinex）等；基于大气散射模型（ASM）和相关统计先验的去雾技术，如暗通道先验［4-5］（DCP）、色衰减先验［6］（CAP）、非局部先验［7］（NLP）等；基于深度学习的去雾技术，如Deha⁃zeNet［8］、DCPDN［9］、AODNet［10］等.近年来，深度卷积神经网络在计算机视觉中应用广泛，2019年Liu 等［11］认为传统的卷积神经网络（CNN）在采用池化或空洞滤波器来增大感受野时势必会造成信息的丢失或网格效应，于是将多级小波变换嵌入到CNN中，在感受野大小和计算效率之间取得了良好的折中，因而首次提出了多级小波卷积神经网络（MWCNN）模型，并证实了其在诸如图像去噪、单图像超分辨率、图像分类等任务中的有效性.同年，Yang等［12］也认为离散小波变换及其逆变化可以很好地替代U-Net 中的下采样和上采样操作，因而提出了一种用于单幅图像去雾的小波U-Net网络，该网络与MWCNN结构非常相似.2020年，Yang等［13］将多级小波与通道注意力相结合设计了一种小波通道注意力模块，据此构建了单幅图像去雨网络模型.同年，Peng等［14］则将残差块与MWCNN相结合提出了一种用于图像去噪的多级小波残差网络（MWRN）.2021年，陈书贞等［15］在已有的MWCNN结构上加入多尺度稠密块以提取图像的多尺度信息，并在空间域对重构图像进行进一步细化处理，以弥补小波域和空间域对图像信息表示存在的差异性，从而实现了图像的盲去模糊.为了解决大雾天气下车牌图像对比度低和边缘、字符等信息模糊不清的问题，很多研究人员开始将已有的图像去雾技术应用于车牌识别的预处理中.但大多数只是对已有图像去雾算法进行了简单改进，如对Retinex或DCP等去雾算法进行改进，直接应用于车牌检测识别中.虽然取得了一定的去雾效果，但其并没有很好地复原出车牌图像的特征，且很难应对中等雾和浓雾下的车牌图像.2020年王巧月等［16］则有意识地针对车牌图像的颜色和字符信息进行车牌图像的去雾，提高了车牌图像的质量.受上述研究的启发，本文提出一种基于深度多级小波U-Net的车牌雾图去雾算法，以端到端的方式来实现不同雾度下不同车牌类型的去雾.首先，提出了一种结合DWT、通道注意力（CA）和像素注意力（PA）的三分支结构，该结构可以对编码器每层输出特征的通道和像素进行加权处理，从而让去雾网络聚焦于车牌雾图中的有雾区域；其次，在解码器中引入“SOS”增强模块（“SOS”Block）来对解码器和下层输入的特征进行进一步融合和增强，提高去雾图像的峰值信噪比，并在U-Net编解码结构之间进行了层与层之间的连接，以此来充分利用不同网络层和尺度上的特征信息；最后，为弥补小波域和空间域之间网络表达图像信息的差异，提出了一种结合跨尺度聚合（CSA）的多尺度残差增强模块（CSAE Block），在空间域上进一步丰富网络对于图像细节信息的表达，有效地提高去雾图像的质量.1去雾网络结构本文去雾网络结构如图1所示.该网络主要分为A与B这2大模块，前者在小波域中实现对车牌雾图x LPHaze的去雾，后者在空间域上对模块A输出的无雾图像进行进一步优化，模块A 的网络结构参数见表1.整个网络结构的输出为：y LPDhaze=y B(y A(x LPHaze；θA)；θB)（1）式中：y B(⋅)和y A(⋅)分别为模块A和B的输出，θA 和θB分别表示模块A和B的可学习参数.1.1小波U-Net二维离散小波变换（2D-DWT）可以实现对给定的图像I的小波域分解，分解过程可视为将图像I与4个滤波器进行卷积，即1个低通滤波器f LL和3个高通滤波器（f LH、f HL和f HH），这4个滤波器（即卷积核）分别由低通滤波器f L和高通滤波器f H构成.以Haar小波为例，该卷积核表示为：f L=12[]1，1T，f H=12[]1，-1Tf LL=LL T，f HL=HL T，f LH=LH T，f HH=HH T（2）125湖南大学学报（自然科学版）2022年图1去雾网络结构Fig.1Defogging network structure 表1模块A 网络结构参数Tab.1Network structure parameters of module A网络层层1层2层3层4类型注意力块卷积层残差组注意力块（下、上采样）卷积层残差组注意力块（下、上采样）卷积层残差组注意力块（下、上采样）卷积层残差组卷积层编码器（卷积核大小f ×f ，输出通道c ）éëêêùûúú()2×2，1()1×1，4and ()1×1，16(3×3，16)éëêêùûúú()3×3，16()3×3，16×3éëêêùûúú()2×2，1()1×1，64(3×3，64)éëêêùûúú()3×3，64()3×3，64×3éëêêùûúú()2×2，1()1×1，256(3×3，256)éëêêùûúú()3×3，256()3×3，256×3éëêêùûúú()2×2，1()1×1，1024(3×3，1024)éëêêùûúú()3×3，1024()3×3，1024×3(3×3，1024)输出大小（宽W ，高H ）(64，32)(64，32)(64，32)(32，16)(32，16)(32，16)(16，8)(16，8)(16，8)(8，4)(8，4)(8，4)(8，4)解码器（卷积核大小f ×f ，输出通道c ）—éëêêùûúú()3×3，16()3×3，12éëêêùûúú()3×3，16()3×3，16×3—(3×3，64)éëêêùûúú()3×3，64()3×3，64×3—(3×3，256)éëêêùûúú()3×3，256()3×3，256×3————输出大小（宽W ，高H ）—(64，32)(64，32)—(32，16)(32，16)—(16，8)(16，8)————层2层4层1层3层1层2层3层4离散小波变换DWT 卷积层ConvLayer 注意力块Attention Block “SOS ”增强块“SOS ”Block 残差组ResGroup 离散小波逆变换IDWT 跨尺度聚合增强块CSAE BlockTanh 层模块B模块A 层间多尺度聚合126第6期陈炳权等：基于深度多级小波U-Net 的车牌雾图去雾算法因此，2D-DWT 可以通过将输入图像I 与4个滤波器进行卷积和下采样来实现，从而获得4个子带图像I LL 、I LH 、I HL 和I HH .其操作定义如下：I LL =()f LL ∗I ↓2I LH =()f LH ∗I ↓2I H L=()f HL∗I ↓2I HH=()f HH∗I ↓2（3）其中：∗表示卷积操作；↓2表示尺度因子为2的标准下采样操作.低通滤波器用于捕获图像中光滑的平面和纹理信息，其它3个高通滤波器则提取图像中存在的水平、垂直和对角线方向上的边缘信息.同时，由于2D-DWT 的双正交性质，可以通过二维逆DWT 的精确重建出原始图像.2D-DWT 及其逆变换的分解和重建示意图如图2所示.本文将2D-DWT 及其逆变换嵌入到U-Net 网络结构中，改善原始U-Net 网络的结构.首先，对输入的3通道车牌雾图进行离散小波变换处理，输出图像的通道数变为原来的4倍，图像大小变为原来的12；然后，使用一个单独的卷积层（“3×3卷积+Lea⁃kyReLU ”）将输入图像扩展为16通道的图像；最后，在U-Net 的每层中迭代使用卷积层和离散小波变换用于提取多尺度边缘特征.1.2基于2D-DWT 的通道-像素联合注意力块（DCPA Block ）在去雾网络中平等对待不同的通道和像素特征，对于处理非均匀雾度图像是不利的.为了能灵活处理不同类型的特征信息，Qin 等［17］和Wu 等［18］均采用CA 和PA ，前者主要用于对不同通道特征进行加权，而后者则是对具有不同雾度的图像像素进行加权，从而使网络更关注雾度浓厚的像素和高频区域.引入Hao-Hsiang Yang 等［13］的小波通道注意力块，本文提出了一种基于二维DWT 的通道-像素联合注意力模块，将DWT 、CA 和PA 构建并行的三分支结构，如图3所示.2D-DWT卷积3×3平均池化卷积Leaky ReluLeaky Relu 输入逐元素相乘残差组输出逐元素相加Leaky Relu 1×1卷积2×2Sigmoid 激活函数注意力块图3基于二维DWT 的特征融合注意力模块结构Fig.3Attention module structure of feature fusion basedon two-dimensional DWT其中，两分支结构的注意力块结合了PA （上分支）和CA （下分支）的特点，将具有通道数为C ，高和宽分别为H 、W 的特征图x ∈R C ×H ×W 分别输入到CA 和PA 中，前者通过平均池化将C×H×W 的空间信息转换为C×1×1的通道权重信息，而后者则通过卷积来将C×H×W 的图像转换成1×H×W.CA 由一个平均池化层、一个卷积层和一个LeakyReLU 层构成，其输出为：y CA =LeakyReLU 0.2(Conv 11(AvgPool (x)))（4）式中：y CA ∈R 1×H ×W ；Conv j i (⋅)表示卷积核大小为i ×i 、步长为j 的卷积操作；LeakyReLU 0.2(⋅)表示参数为0.2的LeakyReLU 激活函数；AvgPool (⋅)表示平均池化操作.类似地，PA 有一个卷积层和一个LeakyReLU层，但没有平均池化层，其输出为：y PA =LeakyReLU 0.2(Conv 22(x))（5）式中：y PA ∈R C ×1×1.CA 和PA 通过逐像素相加，并共用一个Sigmoid 激活函数来分别为输入图像通道和像素配置相应的权重参数，其输出为：y A =Sigmoid (y PA ⊕y CA)（6）式中：y A ∈R C ×H ×W ；⊕表示逐元素相加；Sigmoid (⋅)表示Sigmoid 激活函数.最后，和经离散小波变换、卷列行LLLI LII LH I HL HH22222222列I HL I LL I HH222HL H I H2L行HII LH 图22D-DWT 及其逆变换Fig.22D-DWT and its inverse transformation127湖南大学学报（自然科学版）2022年积层和残差组处理后的特征图进行逐元素相乘，从而实现对特征图的加权，其最终输出为：y DCPA =ResGroup (LeakyReLU 0.2(Conv 13(DWT (x))))⊗yA（7）式中：y DCPA ∈R C ×H ×W ；DWT (⋅)为离散小波变换；⊗表示逐元素相乘；ResGroup (⋅)表示残差组函数. 1.3层间多尺度聚合（LMSA ）受到Park 等［19］在图像去雾任务中采用多级连接来复原图像细节信息的启示，将U-Net 编码器中前3层中DCPA Block 的输出进行跨层和跨尺度的特征聚合，使网络能充分利用图像潜在的特征信息，其结构如图4所示.假设编码器的第l 层输出的聚合特征为y lconcat ，输入解码器的第l 层特征为D l in ，其中l =1，2，3，则y lconcat =Cat((Cat i =13F l i (y l DCPA ))，F up (D l +1out ))（8）D l in =LeakyReLU 0.2(Conv 11(LeakyReLU 0.2(y l SEBlock (y l concat))))（9）式中：Cat (⋅)表示级联操作；F l i (⋅)表示从第i 层到第l 层的采样操作；F up (⋅)表示上采样操作；D i +1out为解码器的第i +1层的输出.将每层聚合后的特征图x l concat 输入到SEBlock 中，自适应地调节各个通道特征，其输出为：y lSEBlock (x l concat )=Sigmoid (FC (ReLu (FC (AvgPool (xlconcat)))))（10）式中：FC (⋅)表示全连接层函数；ReLU (⋅)为ReLU 非线性激活函数.SEBlock 的结构如图5所示.平均池化全连接层全连接层ReLuSigmoid 激活函数S图5SEBlock 结构Fig.5SEBlock structure通过“LeakyRelu-Conv-LeakyRelu ”操作减少每层输出特征的通道数，输入到U-Net 解码器中前3层的“SOS ”Block 中，提升重构图像的峰值信噪比.U-Net 网络中的第4层则聚合前2层的DCPABlock 输出特征和第3层的DWT 输出特征，同样经过SEBlock 和“LeakyRelu-Conv-LeakyRelu ”操作后作为第4层的输入特征进行后续操作.其数学表达式为：y 4concat =Cat((Cat i =12F 4i (y 4DCPA ))，E 3out )（11）D 4in =LeakyReLU 0.2(Conv 11(LeakyReLU 0.2(y 4SEBlock (y 4concat))))（12）其中，E 3out 表示解码器第3层的输出.1.4“SOS ”增强模块（“SOS ”Block ）从Romano 等［20］和Dong 等［21］对“Strengthen-Operate-Subtract ”增强策略（“SOS ”）的开发和利用中可知，该增强算法能对输入的图像特征进行细化处理，可以提高输出图像的峰值信噪比.因此，在本文的车牌图像去雾网络解码器结构中，直接将Dong 等［21］采用的“SOS ”增强算法嵌入到U-Net 结构中，提升车牌去雾图像的质量.Dong 等［21］所采用的“SOS ”增强算法的近似数学表达式如下：J n +1=g (I +J n )-J n（13）层1DCPA Block输出层2DCPA Block输出层3DCPA Block输出编码器CCC层1“SOS ”Block层2“SOS ”Block层3“SOS ”Block解码器LeakyReLU-1×1Conv-LeakyReLU SEBlocky 2CSACaty 1coocat y 3CSACat y 2CSAEBlocky 1CSAEBlock y 3CSAEBlockD 2inD 1inD 3in图4层间多尺度聚合结构Fig.4Multi-scale aggregation structure128第6期陈炳权等：基于深度多级小波U-Net 的车牌雾图去雾算法其中：I 为输入雾图像；J n 为第n 层估计的去雾图像；I +J n 表示使用输入雾图进行增强的图像；g (⋅)表示去雾方法或者优化方法.在U-Net 中编解码器之间的嵌入方式如下：将U-Net 编解码器之间的连接方式改为了逐元素相加，即编码器第i 层输出的聚合特征D i in 加上对应的解码器的第i 层输入特征D i +1out（经上采样操作得到与D i in 相同的图片大小）；将逐元素相加后的输出接入到优化单元（即残差组）中进行进一步的特征细化，随后减去解码器的第i 层输入D i +1out .通过上述嵌入方式，模块输出为：y i sos =g i (D i in +(D i +1out )↑2)-(D i +1out )↑2（14）其中：↑2表示尺度因子为2的上采样操作.其与U-Net 相结合的结构示意图如图6所示.解码器的第i +1层输出编码器的第i 层输出逐元素相减优化单元g （·）逐元素相加2D DWT2D IDWT图6“SOS ”深度增强模块Fig.6"SOS"depth enhancement module1.5跨尺度聚合增强模块（CSAE Block ）为了弥补小波域中去雾网络所忽略的精细空间域图像特征信息，本文提出了一种基于残差组的跨尺度聚合增强模块（CSAE Block ），对小波域网络（模块A ）最后输出的重构图像进行空间域上的图像细节特征补充.CSAE Block 结构如图7所示.CSAE Block 主要由卷积层（即“3×3卷积-Lea⁃kyReLU ”）、残差组、平均池化、CSA 和级联操作构成.首先，卷积层和残差组负责对模块A 输出的空间域图像y 模块A 进行特征提取，平均池化将输入特征分解为4个不同尺度（S 1=14，S 2=18，S 3=116和S 4=132）的输出，即：y S 1，y S 2，y S 3，y S 4=AvgPool (ResGroup (Conv 13(y 模块A)))（15）然后，CSA 实现对输入的不同尺度、空间分辨率的特征信息进行聚合，从而达到在所有尺度级别上有用信息的融合，并在每个分辨率级别上生成精细特征；最后，通过“LeakyRelu-Conv-LeakyRelu ”操作来对输入特征的通道数进行调整.该模块可以通过聚合不同尺度之间、不同分辨率之间的特征来使去雾网络获得较强的上下文信息处理能力.该聚合操作可表示如下：y SjCSA =⊕S i∈{}1，2，3，4F S j S i(y Si)（16）式中：y S jCSA 表示第j 个尺度S j 的CSA 输出特征，j =1，2，3，4；F S j Si(⋅)表示将尺度为S i 的特征图采样到尺度为S j 的特征图.同时，在该模块中引入短连接以改y 模块A1/161/321/81/4卷积层残差组平均池化跨尺度聚合（CSA ）y ResGroupy S 2CSAy S 3CSAy S 4CSAy S 1CSA逐元素相加LeakyReLU-1×1Conv-LeakyReLU-1×1Conv-LeakyReLU-UpSample 3×3Conv-y CSACat级联Cy CSAEBlock图7CSAE Block 结构Fig.7CSAE block129湖南大学学报（自然科学版）2022年善其梯度流.综上所述，CSAE Block 总的输出表达式为：ìíîïïïïy CSACat =Cat j =14()F up ()LeakyReLU 0.2()Conv 11()y S jCSAy CSAE Block =Conv 13()Cat ()y CSACat ，y ResGroup （17）其中y CSACat 表示对上采样为统一大小的输出特征进行级联操作.1.6损失函数为了获得更好的去雾效果，本文使用3个损失函数（有图像去雾损失L rh 、边缘损失L edge 以及对比度增强损失L ce ）作为网络训练的总损失L total ，即：L total =αL rh +γL edge -λL ce（18）其中：α，γ和λ均为任意非负常数.1）图像去雾损失函数.本文将L 1损失和L 2损失进行简单结合，构成车牌图像去雾损失函数：L rh =1N ∑i =1N ()I i gt -F Net ()I i haze 1+I i gt -F Net ()I ihaze 2（19）式中：N 表示输入图像像素数；I gt 为干净图像；I haze 为车牌雾图；F Net (⋅)表示车牌去雾网络函数.一方面通过L 1损失函数来获得较高的PSNR 和SSIM 值，另一方面则通过L 2损失函数来尽可能提高去雾图像的保真度.2）边缘损失函数.为了加强输出去雾图像的边缘轮廓信息，本文利用Sobel 边缘检测算法来获得车牌去雾图像和干净图像的边缘轮廓图，分别为E FNet()I haze和E I gt，计算两者的L 1范数，获得边缘损失函数：L edge=E FNet()I haze-E I gt1（20）3）对比度增强损失.为了提高车牌去雾图像的颜色对比度，本文最大限度地提升每个单独颜色通道的方差，即最大化如下表达式：L ce=（21）式中：x 表示图像的像素索引；FˉNet (I haze )为去雾网络输出图像F Net (I haze )的平均像素值.值得注意的是，所期望的输出去雾图像应该增强其对比度，所以L ce 需要最大化，因此在总损失L total 中应减去该项.2训练与测试2.1车牌雾图数据集（LPHaze Dataset ）的制作为了解决车牌去雾网络训练过程中缺失的车牌雾图数据集问题，受RESIDE 数据集制作方法的启示，本文采用成熟的ASM 理论来进行车牌雾图数据集的制作.车牌图像数据主要来源于OpenITS 提供的OpenData V3.1-SYSU 功能性能车牌图像数据库，并以中科大开源数据集CCPD ［22］作为补充，具体制作方法如下：1）预处理.从OpenITS 和CCPD 数据集中随机选取2291张清晰图像，并对这些清晰车牌图像的车牌区域进行截取；2）配置大气散射模型参数值.参照RESIDE 数据集所选取的参数值范围，随机选取如下一组大气光值A =［0.6，0.7，0.8，0.9，1.0］和一组大气散射系数值β=［0.4，0.6，0.8，1.0，1.2，1.4，1.6］，并将场景深度d (x )置为1；3）合成车牌有雾-无雾图像对.采取一张清晰车牌图像对应多张车牌雾图的方法来合成图像对，即根据大气散射模型，结合步骤2中选定的参数值，以一张车牌无雾图像对应35张有雾图像的方法来合成数据集.合成车牌雾图示例如图8所示.（a ）原图（b ）（A =0.6，β=0.4）（c ）（A =0.7，β=0.8）（d ）（A =0.8，β=1.2）（e ）（A =1.0，β=1.6）图8合成车牌雾图示例Fig.8Example of fog map of composite license plate4）划分训练集和验证集.训练集中干净车牌图像共1697张，对应的车牌雾图共59395张；验证集中干净图像共594张，对应车牌雾图共20790张.2.2实验设置本文采用自制的车牌雾图数据集（LPHaze Data⁃130第6期陈炳权等：基于深度多级小波U-Net 的车牌雾图去雾算法set ）作为车牌图像去雾网络的训练和验证数据，其中所有图像像素大小均设置为64×128×3，batch size 为64.并对训练数据进行数据增强操作：随机水平翻转和随机垂直翻转（翻转概率随机取值0或1），以提升网络的鲁棒性.此外，在训练过程中，使用Adam 优化器来优化网络，其参数均采用默认值（β1=0.9和β2=0.999），并通过梯度裁剪策略加速网络收敛，训练800个epoch ，初始学习率为1e -4，总损失函数中α=1、γ=0.1和λ=0.01.采用Pytorch 包进行车牌雾图去雾网络结构代码的编写和训练，整个训练过程均在NVIDIA Tesla T4的显卡上运行.实验主要包括两个部分：其一，测试本文提出的车牌雾图去雾网络模型，其二，对其进行消融实验.上述实验在合成车牌雾图和自然拍摄的车牌雾图上进行测试，所有测试图像均随机选自OpenITS 提供的车牌图像数据库（与LPHaze Dataset 训练集中的数据不重合），并从测试结果中选取如下5种组合类型进行定性和定量分析，分别为（A =0.6，β=0.8）、（A =0.7，β=1.0）、（A =0.8，β=1.2）、（A =0.9，β=1.4）和（A =1.0，β=1.6），同时对其依次编号为组合A 到E.3结果与分析3.1测试实验1）合成车牌雾图去雾结果为了进一步评估算法性能，将本文算法与最近出现的经典去雾算法（基于引导滤波器的暗通道先验算法［4］（GFDCP ）、基于深度学习的端到端的去雾网络［8］（DehazeNet ）、端到端的一体化去雾网络（AODNet ）、端到端门控上下文聚合去雾网络［23］（GCANet ）和端到端特征融合注意力去雾网络［17］（FFANet ））进行比较.以上算法统一在LPHaze Data⁃set 的验证集上进行测试，并选取其中5类合成车牌雾图的测试结果进行实验分析，其结果见表2.由表2可以看出，在5类不同大气光和散射系数的合成车牌雾图上，与GCANet 得到的结果相比，在PSNR 均值上分别提高了5.64、6.74、8.84、10.52、11.88dB ，SSIM 均值上则分别提高了0.0368、0.0599、0.0991、0.1496、0.2225.同时，在图9中的PSNR 和24222018161412108P S N R （d B ）组合A组合B组合C组合D组合EGFDCPDehazeNet AODNet GCANet FFANet 本文算法组合类型（a ）6种算法在5类组合上的PSNR 均值曲线GFDCP DehazeNet AODNet GCANet FFANet 本文算法1.00.90.80.70.60.5S S I M组合A组合B组合C组合D组合E组合类型（b ）6种算法在5类组合上的SSIM 均值曲线图96种算法在5类合成车牌雾图上的PSNR 和SSIM 均值曲线Fig.9PSNR and SSIM mean curves of 6algorithms on5types of composite license plate fog map表2合成车牌雾图去雾图像的PSNR （dB ）/SSIM 均值Tab.2PSNR （dB ）/SSIM mean of defogging image of composite license plate fog image组合类型（A =0.6，β=0.8）（A =0.7，β=1.0）（A =0.8，β=1.2）（A =0.9，β=1.4）（A =1.0，β=1.6）GFDCP20.75/0.946119.23/0.924817.85/0.900715.63/0.861612.70/0.8035DehazeNet19.31/0.895216.92/0.846014.20/0.793611.71/0.74509.57/0.6882AODNet13.79/0.775715.12/0.801314.60/0.745211.64/0.64818.01/0.5349GCANet18.86/0.925516.31/0.890613.82/0.845911.92/0.791310.14/0.7091FFANet18.09/0.894718.65/0.878419.31/0.851212.76/0.71678.61/0.5407本文算法24.50/0.962323.05/0.950522.66/0.945022.44/0.940922.02/0.9316131湖南大学学报（自然科学版）2022年SSIM 均值曲线图中亦可知，在重构图像质量方面，本文算法在处理不同雾度的合成车牌雾图上明显优于上述5类经典算法.最后，从合成车牌雾图的去雾图像中选取部分图片进行效果展示，如图10所示.从去雾效果中可以直观感受到，本文算法相较于其它算法而言，具有较少的雾度残留以及颜色失真.2）自然车牌雾图去雾结果本文还对实际拍摄的自然车牌雾图进行测试，并与上述5种经典算法的去雾效果进行视觉比较.该测试数据选自OpenITS 的车牌图像数据库，共915张实际拍摄的车牌雾图，视觉对比结果如图11所示.从图11可知：在处理常见的蓝底车牌雾图时，本文算法很少出现过度曝光和图像整体偏暗的问题，且雾度残留也很少；对于其它底色的车牌雾图（如图11中的黄底和蓝底车牌），本文算法在去雾效果上相较于上述5种经典算法仍能保持自身优势，并且在颜色、字符等图像信息上也能得到较好的恢复.（a ）（A =0.6，β=0.8）（b ）（A =0.7，β=1.0）（c ）（A =0.8，β=1.2）（d ）（A =0.9，β=1.4）（e ）（A =1.0，β=1.6）合成雾图GFDCPDehazeNetAODNetGCANetFFANet本文算法干净图像图10合成车牌雾图去雾效果展示Fig.10Fog removal effect display of composite license plate图11实际拍摄的车牌雾图去雾效果展示比较Fig.11Comparison of defogging effect of actual license plate fog map自然车牌雾图GFDCP DehazeNet AODNet GCANet FFANet 本文算法132第6期陈炳权等：基于深度多级小波U-Net的车牌雾图去雾算法3.2不同模块对网络性能的影响为了分析其中各个模块的重要性，本文在LP⁃Haze Dataset数据集上进行消融研究分析，以基于ResGroup和“SOS”Block的MWCNN去雾网络作为基准网络模块，对于其他不同的网络模块，其简化形式及说明如下，R1：基于ResGroup和“SOS”Block的MWCNN作为基准网络，该网络不包含DCPA Block、LMSA和CSAE Block；R2：具有DCPA Block的基准网络；R3：具有DCPA Block、LMSA和CSAE Block的基准网络，即最终的车牌雾图去雾网络.上述网络模块均只训练150个epoch，且初始学习率均为1e-4，并在LPHaze Dataset的验证集上进行测试，其测试结果如表3所示.表3不同网络模块在LPHaze Dataset的验证集上测试结果的均值Tab.3Mean value of test results of different networkmodules on the verification set of LPHaze Dataset网络R1 R2 R3“SOS”Block√√√DCPABlock√√LMSA√CSAEBlock√PSNR/dB22.4722.4323.27SSIM0.94210.94320.9513由表3可得，在不加入DCPA Block、LMSA和CSAE Block的情形下，PSNR和SSIM的均值分别可以达到22.47dB和0.9421，而在加入三者后，PSNR 和SSIM均值则分别提升了0.8dB和0.0092，从而使网络能重建出高质量的去雾图像.3.3不同损失函数对网络性能的影响为了分析损失函数的有效性，本文算法分别采用L1、L2、L rh（即L1和L2损失的简单结合）和L total这四类损失函数来进行网络模型的训练，训练150个ep⁃och，且初始学习率为1e-4.分别在LPHaze Dataset的验证集上进行PSNR和SSIM的指标测试，其实验结果如表4所示.从表4可知，只使用L rh损失函数时，表4不同损失函数下车牌去雾网络测试结果Tab.4Test results of license plate defogging networkunder different loss functions损失函数L1L2L rhL total PSNR/dB22.7422.1923.0623.27SSIM0.94170.93710.94710.9513平均PSNR和SSIM可达到23.06dB和0.9471，且相较于单独使用L1或L2时均有着明显提升.而使用总损失函数L total时，平均PSNR和SSIM分别提升了0.21dB和0.0042，从而使网络性能得到较大的改善.4结论本文提出了一种基于深度多级小波U-Net的车牌雾图去雾算法，该算法以MWCNN作为去雾网络主体框架.首先，为了在小波域和空间域中尽可能地整合不同层级和尺度的图像特征，引入了“SOS”增强策略，并在MWCNN中间进行跨层连接，以此对图像特征进行整合、完善和优化；其次，本文将像素注意力、通道注意力和离散小波变换进行有效融合，从而尽可能去除车牌雾图中的雾度；最后，通过跨尺度聚合增强模块来弥补小波域和空间域之间存在的图像信息差异，进一步提高了重构图像质量.自然车牌雾图和LPHaze Dataset的验证集上的实验结果表明，在处理具有不同大气光照和雾度的车牌雾图上，本文算法具有较好的去雾表现，并且在处理具有不同底色的车牌雾图时具有一定的优势.参考文献［1］YADAV G，MAHESHWARI S，AGARWAL A．Foggy image en⁃hancement using contrast limited adaptive histogram equalizationof digitally filtered image：performance improvement［C］//2014In⁃ternational Conference on Advances in Computing，Communica⁃tions and Informatics（ICACCI）．September24-27，2014，Delhi，India．IEEE，2014：2225-2231．［2］RUSSO F．An image enhancement technique combining sharpen⁃ing and noise reduction［J］．IEEE Transactions on Instrumenta⁃tion and Measurement，2002，51（4）：824-828．［3］GALDRAN A，BRIA A，ALVAREZ-GILA A，et al．On the dual⁃ity between retinex and image dehazing［C］//2018IEEE/CVF Con⁃ference on Computer Vision and Pattern Recognition．June18-23，2018，Salt Lake City，UT，USA．IEEE，2018：8212-8221．［4］HE K，SUN J，TANG X.Guided image filtering［C］//DANIILIDISK，MARAGOS P，PARAGIOS puter Vision-ECCV2010，11th European Conference on Computer Vision，Herak⁃lion，Crete，Greece，September5-11，2010，Proceedings，Part I.Springer，2010，6311：1–14.［5］HE K M，SUN J，TANG X O．Single image haze removal usingdark channel prior［C］//IEEE Transactions on Pattern Analysisand Machine Intelligence．IEEE，：2341-2353．［6］ZHU Q S，MAI J M，SHAO L．A fast single image haze removal133。

1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。

But where are the carbon and the methane coming from?
And what other feedback effects can we expect?
:
forbs (e.g. daisies) The UC Berkeley long-term climate-warming experiment in the Colorado Rockies
control plots heated plots
2.5 2 0
5
10 year since 1990
15
20
Saleska et al., Global Biogeochemical Cycles, 2002
Warming on the Tibetan Plateau
mean number of species plot-1
How do we extend insights from local field studies to global scale? The usual approach - Scaling Up with Bioclimate Models based on:
* Agreement that the problem is real and of serious importance; data gaps and some basic science still to be resolved
We can’t just think about the threat of global warming in isolation: The sinister side of synergy
Global Warming Impacts
The China-US Climate Change Forum May 23-24, 2006
John Harte University of California, Berkeley
IMPACTS OF GLOBAL WARMING
*** Greater intensity, frequency & duration of harmful summer heat waves *** Sea level rise of at least ½ m by 2100; loss of some island nations *** Melting of glaciers and sea ice; loss of alpine and arctic habitat; polar bears! *** Reduced snow pack; loss of irrigation water for crops *** Coral reef degradation due to bleaching
This effect is not incorporated in our current climate models. If it were, predicted warming would be considerably larger!
Ice core data tell us how greenhouse gases respond to temperature change:
*** Near unanimity among scientists active in the field; high level of confidence in predictions
MORE IMPACTS OF GLOBAL WARMING
** Increased intensity and possibly frequency of hurricanes and droughts ** Reduced crop yields because of extreme events and persistent drought ** Increased threat of major wildfires
Retreat of N. American Ice Sheet: Evidence for vegetation-mediated feedback Models without spruce trees cannot predict actual rate of retreat of continental ice sheet
Rate of Retreat (km y-1)
Actual rate Rate predicted w/ rock surface
fect f e n tatio e g Ve
Rock, Silt
The spruce trees weren’t following the ice north… they were chasing it!
Feedback # 1: climate-induced change in species composition can alter spring/summer surface albedo
15
Albedo (%)
10
5
Darker plants cause warming
0
Forbs
Sagebrush
(Torn and Harte, Geophys. Res. Lett., 2006)
Temperature anomaly (ºC)
at the lower limit 1.7oC
0.2 o additional warming
at the upper limit 5.8oC
1.9 o additional warming
% Change in plant cover
sagebrush
30 15 0 -15 -30
Harte and Shaw, Science, 1995
Also a 3-fold reduction in flowering success of shallow-rooted forbs in heated plots
A 20% change in regional plant cover will have an effect on local summertime climate that is comparable to that of 2 x CO2
Feedback # 2: warming can alter ecosystem carbon storage, and thus change atmospheric CO2
Ecosystem Functions & Biogeochemical Fluxes
Climate Change
Soil carbon decay rates Drought stress Surface albedo affected by plant cover
Species Composition & Diversity
Spruce Trees
a = 0.1
a = 0.4
Surface Absorption (1-albedo)
The Vostok core implies positive feedback
CO2 release during warming must cause more warming! And CO2 uptake during cooling must cause more cooling.
350 300
CO2 (ppm
250 200 150 100 -6 -4 -2 0 2 4
CO2 (pp m) kya T (ºC )
From such data, the CO2 and CH4 feedback effects on year 2100 global temperature can be calculated
+ +
ozone hole
+ +,-
+,+
global warming
+
+
acid rain
+ +
+
deforestation
+ harmful reinforcement - helpful cancellation
Climate-Ecosystem Feedbacks
Greenhouse gas emissions
** Consensus exists on the underlying science and facts; work needed to sharpen predictions
AND STILL MORE IMPACTS
* Sea level rise up to 40 feet because of Greenland and Antarctica ice melt (“bi-polar disorder”); catastrophic damage to huge numbers of people and to much coastal infrastructure * Extinction episode comparable to K-T boundary; catastrophic loss of ecosystem services * Major spread of infectious tropical and subtropical diseases to the mid-latitudes
Critical pastoral resources in Tibet are vulnerable to warming. Land management (grazing) can partially mediate warming effects.