Estimating Method of Short-Interval-Traffic Distribution Considering Long-Term-Traffic Dyna
生物力学原理PPT课件
Cartesian coordinate system
Utilizes coordinates for locating a point on a plane by identifying the distance of the point from each of two intersecting lines or ,in space,by the distance from each of three planes or, in space, by the distance from each of three planes intersecting at a point。 2D system,3Dsystem
几个基本概念
机体重量(Body weight) 地面反作用力(Ground reaction force) 空气阻力(Air resistance) 肌力(Muscle force ) 关节反作用力(Joint reaction force )
关节反作用力 (Joint reaction force,JRF )
Relative reference frame describes the position of one limb segment with respect to an adjacent segment。A measurement is made by comparing motion between an anatomic landmark or coordinates of one segment with an anatomic landmark or coordinates of a second segment 。
生物力学原理
为什么要学习生物力学分析
Copula函数的选择及其在金融分析中的若干应用
摘 要Copula理论是一种基于联合分布的建模方法,最大的优点就是把边缘分布和相关结构分离开,它的提出为解决多元联合分布的构建以及变量间的非线性相关性问题提供了一个灵活实用的方法,人们将其广泛的用于金融领域,应用于投资组合、资产定价等方面,对金融数据相关性进行建模、分析有着非常重要的意义和作用。
本文主要讨论了Copula理论在金融领域的应用,分析了基于Copula理论的多金融资产的投资组合优化及风险度量的问题。
主要工作如下:首先介绍Copula函数的相关概念和性质,目前国内外Copula理论研究的进展情况,本文的研究思路、方法及相关应用。
传统的金融数据分析是基于正态分布的假设,但正态假设有其局限性,往往不能满足,本文打破传统的基于正态分布的假设,讨论了Copula理论和Monte Carlo模拟在风险VaR估计中的应用,并选用股票数据实例分析了基于Archimedean Copula的风险VaR估计,结果表明此方法是有效的,而传统的VaR计算方法低估了风险。
并进一步将此方法推广到多维资产的情形,说明与单支股票风险均值相比采用此方法确定的投资组合降低了金融风险。
文章为了进一步提高模型构造的有效性,提出了一种基于Bayes理论的混合Copula构造方法,把多个Copula函数所具有的优点融合到一个混合函数中,通过调整各个函数的权重系数来调整函数尾部相关性强弱,比单个Copula相关结构更为灵活。
另外,将Bootstrap方法引入到Copula中的参数估计中,实例表明采用Bootstrap 方法对参数的估计与实际值比较接近,为我们提供了解决问题的另一种有效的思路。
最后,对本文进行了总结,并对一些本文中可以继续探讨研究的方向进行了进一步的前景展望。
关键词:Copula函数;VaR估计;Bootstrap方法;投资组合Selection of Copulas and its Application on FinanceAbstractCopula functions which based on joint distribution provide a flexible and useful statistic tool to construct multivariate joint distribution and solve the nonlinear correlation problem. One of its advantages is the dependence structure of variables no longer depending on the marginal distributions. Copula theory has gained increasing attention in asset pricing, risk management, portfolio management and other applications. In detail, my research is as follows:We first introduce the ideas of copula theory and several copula functions which belong to Archimedean families. Then we discuss the use of Archimedean Copula in VaR and CVaR calculation without the traditional multidimensional normal distribution assumption in financial risk management. Our empirical analysis which based on stock market data uses Monte Carlo simulation method and the results show that the VaR calculated by copula method is larger than traditional method. It means that traditional method underestimated the risk of stock market, and the Monte Carlo simulation based on Copula is effective for financial Market. Then, this method is extended to the multidimensional case, to show that the VaR of proper portfolio is lower than means of risk with single stock.In order to improve the validation of model construction, we introduce a simple Bayesian method to choose the “best” copula which is a mixture of several different copulas. By estimating parameters of each chosen copula and adjusting their weight coefficients in the mixed copula, the model has all the advantages of the chosen copulas and has more flexibility because different weight coefficient combinations describe different asset correlations. In addition, we introduce Bootstrap method to estimate the parameters of Archimedean Copula. The real analysis also shows the estimated parameter by Bootstrap method gets closer to actual value. So it is another efficient way to solve our problems.At last, we make a summary of the whole paper, and look into the future of some aspects that could be discussed in the coming days.Key Words:Copulas; VaR estimation; Bootstrap method; portfolio目录摘 要 (1)Abstract (III)第一章 绪论 (1)1.1 研究背景与意义 (1)1.2 国内外研究现状 (2)1.3 论文组织结构 (3)第二章 Copula选择及检验 (4)2.1 Copula函数的基本概念 (4)2.1.1 Copula函数定义及性质 (4)2.1.2 阿基米德Copula (5)2.1.3 相关性度量 (6)2.2 常用的二元Archimedean Copula函数与相关性分析 (8)2.2.1 Gumbel Copula函数 (8)2.2.2 Clayton Copula函数 (9)2.2.3 Frank Copula函数 (10)2.3 Copula模型参数估计 (11)2.3.1 Genest and Rivest的非参数估计法 (11)2.3.2 极大似然估计法(The Maximum Likelihood Method) (12)2.3.3 边缘分布函数推断法(The method of Inference of Functionsfor Margins) (13)2.3.4 典型极大似然法(The Canonical Maximum Likelihood Method) (13)2.4 Copula的检验 (13)2.4.1 Klugman-Parsa法则 (13)2.4.2 Copula分布函数检验法则 (14)2.4.3 非参数检验法则 (14)第三章 基于Copula的VaR分析计算 (15)3.1 VaR的基本概念 (15)3.1.1 VaR的定义 (15)3.1.2 VaR的计算要素 (16)3.2 基于Copula的投资组合VaR的计算 (16)3.2.1 Copula-VaR的解析方法 (16)3.2.2 用Copula变换相关系数的VaR分析方法 (17)3.2.3 基于Copula的蒙特卡洛模拟法 (18)3.2.4 实证分析 (19)3.3 基于三维Copula的VaR计算 (25)3.3.1 多元阿基米德Copula的构造 (25)3.3.2 基于Copula的Monte Carlo模拟法 (26)3.3.3 实证分析 (27)第四章 混合Copula的构造与Bootstrap方法的应用 (30)4.1 混合Copula的构造与应用 (30)4.1.1 基于Bayes理论的混合Copula构造 (30)4.1.2 实证分析 (32)4.2 Bootstrap方法的应用 (35)4.2.1 Bootstrap基本原理 (35)4.2.2 Bootstrap估计Copula参数 (36)第五章 结论与展望 (38)5.1 结论 (38)5.2 展望 (38)参考文献 (39)在校期间研究成果 (42)致 谢 (43)附录 数据 (44)附录 程序 (50)第一章 绪论1.1 研究背景与意义当今金融市场的发展达到了空前的规模,国际化、自由化、证券化、金融创新得到了飞速发展,但其不稳定因素也随之增加,脆弱性体现了出来。
基于稀疏表示的双平行线阵二维DOA_估计
第 21 卷 第 7 期2023 年 7 月太赫兹科学与电子信息学报Journal of Terahertz Science and Electronic Information TechnologyVol.21,No.7Jul.,2023基于稀疏表示的双平行线阵二维DOA估计苏龙,李雪,罗德凌(空军航空维修技术学院,湖南长沙410124)摘要:为了对空域目标的方位角和俯仰角进行有效估计,提出一种基于稀疏表示的双平行线阵二维DOA估计方法。
首先需构建包含目标方位角和俯仰角信息的2个空间复合角;然后利用稀疏表示技术求解其中的一个空间复合角,以此作为前提条件,另一个空间复合角就可以解耦为一维波达方向(DOA)估计问题,利用矩阵运算可以求解出来;最后根据已求解的2个空间复合角对方位角和俯仰角进行配对求解。
与现有算法相比较,所提方法受快拍数的影响较小,在信噪比较高、角度间隔较大的情况下,具有良好的性能。
关键词:稀疏表示;双平行线阵;波达方向;空间复合角中图分类号:TN911.7 文献标志码:A doi:10.11805/TKYDA2020710Two-dimensional DOA estimation of double parallel arrays based onsparse representationSU Long,LI Xue,LUO Deling(Air Force Aviation Maintenance Technical College,Changsha Hunan 410124,China)AbstractAbstract::In order to estimate the azimuth angle and pitch angle of airborne target effectively, a two-dimensional Direction Of Arrival(DOA) estimation method based on sparse representation is presented.Firstly, it is necessary to construct two spatial compound angles containing the azimuth and pitchinformation of the target; then, one of the spatial composite angles is solved by sparse representationtechnology, and the other spatial composite angle can be decoupled into one-dimensional DOAestimation problem, which can be solved by matrix operation; finally, the azimuth angle and pitch angleare solved in pairs according to the solved two spatial compound angles. Compared with the existingalgorithms, the proposed method is less affected by the number of snapshots, and has good performanceunder the condition of high signal-to-noise ratio and large angular interval.KeywordsKeywords::sparse representation;double parallel line array;Direction Of Arrival;spatial compound angle波达方向(DOA)的估计在阵列信号处理当中一直是一个热点,引起了许多学者的关注与研究[1-6],在雷达、通信、声呐领域取得了许多成果。
bootstrap置信区间公式
bootstrap置信区间公式## Confidence Intervals for Means Using Bootstrapping.Introduction.Bootstrapping is a technique used to estimatestatistical parameters, such as confidence intervals, by resampling a given dataset with replacement. For example, to calculate a confidence interval for the mean of a population, bootstrapping involves repeatedly sampling from the original dataset, calculating the mean of each sample, and then using the distribution of these sample means to estimate the population mean and the associated confidence interval.Formula.The basic formula for a bootstrap confidence interval for the mean using the percentile method is:CI = (L, U)。
where:L is the lower bound of the confidence interval, which is the _p_th percentile of the sample means, where _p_ is the desired level of confidence.U is the upper bound of the confidence interval, which is the _q_th percentile of the sample means, where _q_ is 1 _p_.For example, if we want a 95% confidence interval, then _p_ = 0.025 and _q_ = 0.975.Steps.The steps for calculating a bootstrap confidence interval for the mean are as follows:1. Resample: Draw B bootstrap samples of size n from the original dataset with replacement.2. Calculate: Compute the mean for each bootstrap sample.3. Percentile: Determine the lower and upper bounds of the confidence interval by finding the _p_th and _q_th percentiles of the sample means.4. Confidence Interval: The interval (L, U) is the bootstrap confidence interval for the mean.Advantages.Bootstrapping has several advantages over traditional methods for estimating confidence intervals, such as:It is non-parametric, so it does not require assumptions about the distribution of the data.It can be used with small sample sizes.It is computationally efficient and easy to implement.Limitations.However, bootstrapping also has some limitations:It can be biased for certain types of data or if the sample size is too small.It can be computationally intensive for large datasets.It may not be accurate if the data is notrepresentative of the population.## 置信区间公式。
心理学专业英语词汇(E2)
心理学专业英语词汇(E2)心理学专业英语词汇(E2)心理学专业英语词汇(E2)eog 动眼电波图eog 眼动图eom 眼外肌运动eonism 男扮女装癖eonism 衣裳倒错症eopsia 暮视症eosophobia 黎明恐怖症ep 诱发电位eparsalgia 过劳病epharmone 适应型epharmony 和谐发育ephebe 男青年ephebiatrics 青春期医学ephebic 青春期的ephebogenesis 青春期身体变化ephebology 青春期学epicene 有异性特征的epicritic 精细觉epicritic sensitivity 后起感觉epidemic 流行的epidemic hysteria 流行性癔病epidemicity 流行性epidemiology 流行病学epidermis 表皮epigamic behavior 吸引异性行为epigenesis 后成论epigenesis 渐成律epigenetic chart 渐成图epigenetic principle 后成原理epilemma 神经梢膜epilemmal nerve ending 梢膜性神经末梢epilepsy 癫痫epileptic attack 癫痫发作epileptic character 癫痫性格epileptic dementia 癫痫性痴呆epileptic fugue 癫痫性神游epileptic furor 癫痫狂怒epileptic psychosis 癫痫性精神病epileptic seizure 癫痫发作epileptic stupor 癫痫僵呆epileptiform 癫痫样的epileptiform seizure 癫痫式发作epileptoid 类癫痫的epileptoid character 类癫痫性格epileptoid convulsion 癫痫状抽搐epileptoid personality 癫痫性人格epileptoid seizure 类癫痫发作epileptology 癫痫学epileptosis 癫痫性精神病epimeletic behavior 护幼行为epinephrine 肾上腺素epineurial canal 神经外管epineurium 神经外膜epinosis 继发性精神病态epiphenomenalism 副现象论epiphenomenon 副现象epiphyseopathy 松果体病epiphysis 松果体epiphysis cerebri 松果体epiplo 网膜episcotister 断续仪episcotister 节光器episode 情节episode analysis 事例分析episodic amnesia 情结性遗忘episodic amnesia 要事失忆症episodic dyscontrol syndrome 发作性控制不良综合症episodic memory 情节记忆episodic movement 意外变动epistemic curiosity 认识性好奇epistemic motivation 求知的动机epistemologist 知识学家epistemology 认识论epistemology 知识论epithalamic 上丘脑的epithalamus 上丘脑epithelium 上皮epitome method 摘要法epochal amnesia 时代性遗忘epochal psychoses 过度期神经失常epsem 等机率选择法epsem sample 等机率选择法样本epsilon movement ε似动epsp 兴奋性突触后电位eq 教育商数equal 相等的equal education 平等教育equal employment opportunity 平等就业机会equal energy spectrum 等能光谱equal interval scale 等距量表equal interval variable 等距变量equal loudness contour 等响线equal loudness curve 等响曲线equal noisiness curve 等噪声曲线equal pitch contour 等高曲线equal probability 均等机率equal sensation function 等感觉函数equal sense distance 均等距离感觉equal sense distance method 感觉等距法equal sign 等号equal temperament 平均律equal weighting 等重量equalinterval variable 等距变量equalitarian 等积的equalitarian family 平等家庭equalitarianism 平等主义equality 平等equalization 均等化equalization phase 均等相equalizing phase 均等相equal appearing interval 等现间距equal appearing method 等现法equal appearing ratio 等现比率equal density contour 等密度线equal interval scale 等级量表equal loudness contour 等响线equal pitch contour 等高线equal probability of selection method 等机率选择法equal status contact hypothesis 等位相交假设equal volume contour 等量线equated score 等价得分equating 等化equating problem 等价问题equation 方程equation of time 时间方程equidistance 等距equidistant 等距离的equilibrant 平衡力equilibrated type 平衡型equilibration 均衡equilibratory sensation 平衡觉equilibrium 平衡equilibrium constant 平衡常数equilibrium model of intimacy 亲密的平衡模型equilibrium of nervous process 神经过程的均衡性equilibrium point 平衡点equilibrium potential 平衡电势equilibrium sense 平衡感equilibrium theory 平衡论equilibrium theory of organization 组织的平衡论equimax rotation 等极限轴转equipment 设备equipment fidelity 设备逼真度equipollence 均等equipollency 均等equipollent 均等的equipotentiality 等势equipotentiality of cortex 大脑皮质等价说equisection method 等分法equity 公平equity of reward 报酬公平性equity theory 公平理论equivalence 等值equivalence belief 等值信念equivalence range 等值范围equivalent anchor items 相等参照测验题equivalent contrast 等效对比equivalent effective temperature 等价有效温度equivalent environment 相同环境equivalent form method 复本法equivalent forms 等值复本equivalent group 相等组equivalent of anxiety attack 焦虑发作等位症equivalent scores 同等分数equivalent sound level 等效声级equivalent temperature 等价温度equivalent test 等价测验equivalent transformational thinking 等值转换思维equivalent group method 等组法equivalent groups procedure 等组法equi distance tendency 等距倾向erben s reflex 埃尔本反射erben s reflex sign 埃尔本反射症erd 诱发反应测定器erect posture 直立姿势erectile dysfunction 勃起功能障碍erection 勃起eremophobia 孤独恐怖症erethism 兴奋增盛erethisophrenia 精神兴奋过度ereuthrophobia 红脸恐怖症erg 视网膜电描记器erg 视网膜电图erg 本能特性erg theory erg理论ergasia 精神活动ergasiatrics 精神病学ergasiomania 工作狂ergasiophobia 手术恐怖症ergasthenia 过劳性衰弱ergastic 有潜能的ergocardiogram 心电动力图ergocardiography 心电动力描记术ergogenic 机能增进的ergograph 肌肉疲劳记录器ergograph 计功器ergography 测力术ergomaniac 工作狂ergometer 测力器ergometer 功能计ergonomic design 工效学设计ergonomic design of work system 工作系统工效学设计ergonomic guiding principles 工效学指导原则ergonomic parameter 工效学参数ergonomic principle 工效学原则ergonomic standard 工效学标准ergonomics 工效学ergonomics of safety 安全工效学ergonomist 工效学家ergophobia 工作恐怖症ergopsychometry 工效心理测量ergotropic mechanism 增进抵抗力机制erikson personality theory 埃里克森个性理论erikson stage 埃里克森阶段划分erikson stage theory of personality 埃里克森人格发展阶段论erikson s personality theory 埃里克森个性理论erogenous 性欲的erogenous zone 性感区eros 性本能erotic 性的erotic delusion 多情妄想erotic instinct 性本能erotic stimulation 色欲刺激erotic type 性爱型erotica 色情事物eroticaphile 好色者eroticaphobe 恶色者eroticism 色情eroticomania 色情狂eroticpsychopathy 变态色情erotism 色欲erotogenesis 性欲发作erotogenic 性感的erotogenic zone 欲源带erotographomania 情书狂erotology 性爱学erotomania 色情狂erotopath 性欲异常者erotopathy 性欲异常erotophobia 性欲恐怖症erotosexual 性欲的erp 早期感受器电位errant 错误的erratic 不稳定的erratic children 乖僻儿童erratic element 不稳定因素erratic fluctuation 不稳定波动error 错误error 误差error analysis 错误分析error band 误差区域error characteristics 误差特征error component 误差因素error control 误差控制error correcting routine 误差校正程序error curve 错误曲线error detection 误差检验error estimation 误差估计error factor theory 错误要因说error free operation 无错操作error in data 数据误差error mean square 误差均方error measure 误差测定error method 误差法error of aggregation 综合误差error of approximation 近似误差error of calculation 计算误差error of central tendency 中心趋势误差error of estimate 估计误差error of measurement 测量误差error of observation 观察误差error of prediction 预测误差error of refraction 折射误差error of sampling 取样误差error of variance 误方差error operating 错误操作error quotient 误差商数error rate 误差率error source 误差源error term 误差项error time 误差时间error variance 误差变量errorless discrimination 无错辨别errorless discrimination learning 无误辨别学习error choice technique 错误选择法erythrochloropia 蓝黄色盲erythrocyte 红血球erythrophobia 红脸恐怖症erythrophose 红幻视erythropsia 红视症erythropsin 视紫红质escalation of commitment 投入的增加escapade 越轨行为escape 逃避escape behavior 逃避行为escape conditioning 逃脱条件反射escape from freedom 逃避自由escape from reality 逃避现实escape hatch 应急出口escape into fantasy 逃入幻想escape learning 逃避学习escape mechanism 逃避作用escape reaction 逃避反应escape rocket 逃逸火箭escape tendency 逃避倾向escape training 逃脱训练escapism 逃避现实eschatology 来世学esodic 传入的esops 员工股份持有制方案esp 超感知觉especial 特别的espial 观察esprit 精神ess 演化的稳定策略essay examination 论文式考试essay examination 作文考试essay test 论文测验essay type question 论文式试题essay type test 作文类型测验essay type test item 论述式题型esse est percipi ence 存在即是被知觉essence 本质essence of beauty 美的本质essential 本质的essential condition 必要条件essential element 要素essential epilepsy 原发性癫痫essentialism 精粹主义essentialism 实在论essentiality 实质性est 电休克疗法establishment 确立establishment stage 建立阶段esteem 尊重esteem need 尊重需要esteem gain hypothesis 自尊增高假设esthematology 感觉学esthesia 触觉esthesia 感觉esthesiodic 感觉传导的esthesiogen 激奋质esthesiogenesis 感觉发生esthesiogenic 发生感觉的esthesiology 感觉学esthesiomania 感觉倒错狂esthesiometer 触觉计esthesiometry 触觉测量法esthesioneure 感觉神经元esthesioneurosis 感觉性神经病esthesiophysiology 感觉生理学esthesis 感觉esthesodic 感觉传导的esthetic 审美esthetic education 美育esthetic feeling 美感esthetic force 美感力esthetic intelligence 审美智力esthetic sentiment 审美情操estheticism 唯美主义esthetics 美学estimate 估计值estimate of error 误差估计量estimate of standard error 标准误差估计量estimated norm 估计常模estimating components of variance 方差分子估计estimation 估计estimation criterion 估计准则estimation difference 估计误差estimation model 估计模式estimation of fixed effects 固定效果估计estimation of parameters 参数估计estimation of population correlation 总体相关系数估计estimation of population mean 总体均数估计estimation of population proportion 总体比例估计estimation of population standard derivation 总体标准差估计estimation of population variance 总体方差估计estimation of strength of relation 相关强度估计estimation of time 时间估计estimation point 估计点estimation range 估计范围estimator 估计量estimator precision 估计量精确度estop 禁止estrangement 疏远estrin 动情素estrinization 动情期变化estrogen 雌激素estrogenicity 动情性estrone 雌酮estronum 雌酮estrous synchronization 发情同步化estruation 动情期estrus 动情期estrus cycle 发情周期eta 相关比etat 状况eternal ear 外耳etg 视丘电图ethchlorvynol 氯乙基戊烯炔醇ether 醚etheromania 乙醚瘾ethical 道德的ethical 伦理的ethical behavior 合伦理的行为ethical characteristics 道德特征ethical code 伦理规章ethical concept 伦理概念ethical conduct 伦理行为ethical dilemma 道德困境ethical discrimination 道德辨别ethical equality 道德平等ethical experience of moral feeling 伦理的道德情感体验ethical issues 伦理问题ethical judgment 道德判断ethical principle 道德原则ethical relativism 伦理相对论ethical sociology 伦理社会学ethical standard 伦理标准ethical value 道德价值ethical risk hypothesis 道德冒险假说ethics 伦理学ethics for modern life 现代生活伦理学ethics of sex 性伦理ethmoid bone 筛骨ethnic cleavage 种族内部分裂ethnic derivation 种族。
大全,MCM备用数学方法中英文对照
MCM备用数学方法中英文对照MCM临近了,专业规范的表达是必不可少的,大家都来热热身吧!各位模友们自己有想到的数学方法或相关方面的专业词汇也可以贴出中英对照的版本![B]计算数学方法:[/B]内插法:interpolation method有限元方法:method of finite element有限差分法:method of finite difference曲线拟合法:method of curve fitting迭代法:method of iteration误差分析法:error analysis样条函数法:method of spline function最小二乘法:method of least square[B]概率论方法:[/B]可靠性分析法:method of reliability analysis时间序列分析法: method of time-series analysis抽样调查法:method of sampling investigation非参数统计法:method of nonparametric statistics实验设计法:method of experiment design故障分析法:method of fault analysis相关分析法:method of correlation analysis测度论方法:method of measure theory统计假设检验法: method of statistical hypothesis testing随机服务法(排队论法):stochastic service system数理统计法:method of mathematical statistics 蒙特卡洛法:Monte Carlo method[B]运筹学方法:[/B]共轭函数法:method of conjugate function动态规划法:method of dynamic programming网络法:method of network优选法:method of optimum seeking图论法:method of graph theory爬山法:method of climbing线性规划法:method of linear programming罚函数法: method of penalty function统筹法:method of overall planning乘子法:method of multiplier最速下降法:method of steepest descent整数规划法:method of integer programming。
Guide to the Expression of Uncertainty in Measurement
Input and Output Quantities
In the generic model Y = f(X1,…,XN), the measurand is denoted by Y
Also called the output quantity
The quantities X1,…,XN are called input quantities
Remember: the correction term or factor itself has uncertainty
A small residual systematic error generally remains after all known corrections have been applied
Benefits
Much flexibility in the guidance Provides a conceptual framework for
evaluating and expressing uncertainty Promotes the use of standard terminology and
Stated Purposes
Promote full information on how uncertainty statements are arrived at
Provide a basis for the international comparison of measurement results
Uncertainty is a more practical concept Evaluating uncertainty allows you to place a
信号完整性基础之十一——三种抖动分解方法
理解力科SDA的三种抖动分解方法张昌骏 美国力科公司深圳代表处在通讯和PC行业,高速串行信号越来越普及,在使用示波器测量和分析这类信号时,通常要求测量总体抖动(Total jitter,简称Tj)和固有抖动(Deterministic jitter,简称Dj),验证是否满足相关规范的要求。
在力科SDA系列示波器中使用了“Normalized Q-Scale method”(简称NQ-Scale方法)来求解Tj。
而Tj分解为固有抖动Dj和随机抖动Rj时,力科SDA提供了三种抖动分解方法,分别为Conventional、effective、MJSQ,如下图所示。
图一:力科SDA的三种抖动分解方法MJSQ方法在Fibre Channel规范已有定义(MJSQ代表Methodologies for jitter and signal quality specification),这种方法在串行数据的抖动分析中被广泛使用。
在MJSQ文档中,Tj 是某一测量样本数量下的TIE抖动的峰峰值,由Rj和Dj组成,Dj是有边界的,而Rj是没有边界的,其概率密度函数满足高斯分布。
Tj的直方图使用dual-Dirac来建模。
Dual-Dirac 模型是由两个满足高斯分布的脉冲组成,左右两个脉冲的均值为μL和μR,两个脉冲的标准偏差都等于σ,Dj = μR - μL,Rj = σ,Tj@BER-12 = 14 * Rj + Dj。
如下图二所示。
图二:Dual-driac模型与MJSQ方法示意图力科SDA中的MJSQ方法直接处理PDF概率密度函数,使用两个高斯分布的曲线分别拟合TIE直方图的左右两边的尾部,调节高斯曲线的标准偏差让曲线能尽量拟合TIE直方图的尾部。
力科SDA的MJSQ分解方法基于传统的MJSQ方法进行了革新,两个高斯分布的均值可以是不以Y轴对称的,标准偏差也可以是不相等的。
拟合的两个高斯曲线的均值之差为Dj,标准偏差的平均值为Rj。
A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection
A Study of Cross-Validation and Bootstrap for Accuracy E stimation and Model SelectionRon KohaviComputer Science DepartmentStanford UniversityStanford, CA 94305ronnykGCS Stanford E D Uh t t p //r o b o t i c s Stanford edu/"ronnykA b s t r a c tWe review accuracy estimation methods andcompare the two most common methods cross-validation and bootstrap Recent experimen-tal results on artificial data and theoretical recults m restricted settings have shown that forselecting a good classifier from a set of classi-fiers (model selection), ten-fold cross-validationmay be better than the more expensive ka\pone-out cross-validation We report on a large-scale experiment—over half a million runs ofC4 5 and aNaive-Bayes algorithm—loestimalethe effects of different parameters on these algonthms on real-world datascts For cross-validation we vary the number of folds andwhether the folds arc stratified or not, for boot-strap, we vary the number of bootstrap sam-ples Our results indicate that for real-worddatasets similar to ours, The best method lo usefor model selection is ten fold stratified crossvalidation even if computation power allowsusing more folds1 I n t r o d u c t i o nIt can not be emphasized enough that no claimwhatsoever 11 being made in this paper that altalgorithms a re equiva lent in practice in the rea l world In pa rticula r no cla im is being ma de tha t ont should not use cross va lida tion in the real world— Wolpcrt (1994a.) Estimating the accuracy of a classifier induced by su-pervised learning algorithms is important not only to predict its future prediction accuracy, but also for choos-ing a classifier from a given set (model selection), or combining classifiers (Wolpert 1992) For estimating the final accuracy of a classifier, we would like an estimation method with low bias and low variance To choose a classifier or to combine classifiers, the absolute accura-cies are less important and we are willing to trade off biasA longer version of the paper can be retrieved by anony mous ftp to starry Htanford edu pub/ronnyk/accEst-long ps for low variance, assuming the bias affects all classifiers similarly (e g esLimates are ")% pessimistic)In this paper we explain some of the assumptions madeby Ihe different estimation methods and present con-crete examples where each method fails While it is known that no accuracy estimation can be corrert allthe time (Wolpert 1994b Schaffer 1994j we are inter ested in identifying a method that ib well suited for the biases and tn rids in typical real world datasetsRecent results both theoretical and experimental, have shown that it is no! alwa>s the case that increas-ing the computational cost is beneficial especiallhy if the relative accuracies are more important than the exact values For example leave-one-out is almost unbiased,but it has high variance leading to unreliable estimates (Efron 1981) l o r linear models using leave-one-out cross-validation for model selection is asymptotically in consistent in the sense that the probability of selectingthe model with the best predictive power does not con-verge to one as the lolal number of observations ap-proaches infinity (Zhang 1992, Shao 1993)This paper \s organized AS follows Section 2 describesthe common accuracy estimation methods and ways of computing confidence bounds that hold under some as-sumptions Section 3 discusses related work comparing cross-validation variants and bootstrap variants Sec lion 4 discusses methodology underlying our experimentThe results of the experiments are given Section 5 with a discussion of important observations We conelude witha summary in Section 62 Methods for Accuracy E s t i m a t i o nA classifier is a function that maps an unlabelled in-stance to a label using internal data structures An i n-ducer or an induction algorithm builds a classifier froma given dataset CART and C 4 5 (Brennan, Friedman Olshen &. Stone 1984, Quinlan 1993) are decision tree in-ducers that build decision tree classifiers In this paperwe are not interested in the specific method for inducing classifiers, but assume access to a dataset and an inducerof interestLet V be the space of unlabelled instances and y theKOHAVI 1137set of possible labels be the space of labelled instances and ,i n ) be a dataset (possibly a multiset) consisting of n labelled instances, where A classifier C maps an unla-beled instance ' 10 a l a b e l a n d an inducer maps a given dataset D into a classifier CThe notationwill denote the label assigned to an unlabelled in-stance v by the classifier built, by inducer X on dataset D tWe assume that there exists adistribution on the set of labelled instances and that our dataset consists of 1 1 d (independently and identically distributed) instances We consider equal misclassifica-lion costs using a 0/1 loss function, but the accuracy estimation methods can easily be extended to other loss functionsThe accuracy of a classifier C is the probability ofcorrectly clasaifying a randoml\ selected instance, i efor a randomly selected instancewhere the probability distribution over theinstance space 15 the same as the distribution that was used to select instances for the inducers training set Given a finite dataset we would like to custimate the fu-ture performance of a classifier induced by the given in-ducer and dataset A single accuracy estimate is usually meaningless without a confidence interval, thus we will consider how to approximate such an interval when pos-sible In order to identify weaknesses, we also attempt o identify cases where the estimates fail2 1 Holdout The holdout method sometimes called test sample esti-mation partitions the data into two mutually exclusivesubsets called a training set and a test set or holdout setIt is Lommon to designate 2/ 3 of the data as the trainingset and the remaining 1/3 as the test set The trainingset is given to the inducer, and the induced classifier istested on the test set Formally, let , the holdout set,be a subset of D of size h, and let Theholdout estimated accuracy is defined aswhere otherwise Assummg that the inducer s accuracy increases as more instances are seen, the holdout method is a pessimistic estimator because only a portion of the data is given to the inducer for training The more instances we leave for the test set, the higher the bias of our estimate however, fewer test set instances means that the confidence interval for the accuracy will be wider as shown belowEach test instance can be viewed as a Bernoulli trialcorrect or incorrect prediction Let S be the numberof correct classifications on the test set, then s is dis-tributed bmomially (sum of Bernoulli trials) For rea-sonably large holdout sets, the distribution of S/h is ap-proximately normal with mean ace (the true accuracy of the classifier) and a variance of ace * (1 — acc)hi Thus, by De Moivre-Laplace limit theorem, we havewhere z is the quanl lie point of the standard normal distribution To get a IOO7 percent confidence interval, one determines 2 and inverts the inequalities Inversion of the inequalities leads to a quadratic equation in ace, the roots of which are the low and high confidence pointsThe above equation is not conditioned on the dataset D , if more information is available about the probability of the given dataset it must be taken into accountThe holdout estimate is a random number that de-pends on the division into a training set and a test set In r a n d o m sub s a m p l i n g the holdout method is re-peated k times and the eslimated accuracy is derived by averaging the runs Th( slandard deviation can be estimated as the standard dewation of the accuracy es-timations from each holdout runThe mam assumption that is violated in random sub-sampling is the independence of instances m the test set from those in the training set If the training and testset are formed by a split of an original dalaset, thenan over-represented class in one subset will be a under represented in the other To demonstrate the issue we simulated a 2/3, 1 /3 split of Fisher's famous ins dataset and used a majority inducer that builds a classifier pre dieting the prevalent class in the training set The iris dataset describes ins plants using four continuous fea-tures, and the task is to classify each instance (an ins) as Ins Setosa Ins Versicolour or Ins Virginica For each class label there are exactly one third of the instances with that label (50 instances of each class from a to-tal of 150 instances) thus we expect 33 3% prediction accuracy However, because the test set will always con-tain less than 1/3 of the instances of the class that wasprevalent in the training set, the accuracy predicted by the holdout method is 21 68% with a standard deviation of 0 13% (estimated by averaging 500 holdouts) In practice, the dataset size is always finite, and usu-ally smaller than we would like it to be The holdout method makes inefficient use of the data a third of dataset is not used for training the inducer 2 2 Cross-Validation, Leave-one-out, and Stratification In fc-fold cross-validation, sometimes called rotation esti-mation, the dataset V is randomly split into k mutuallyexclusive subsets (the folds) , of approx-imately equal size The inducer is trained and tested1138 LEARNINGThe cross-validation estimate is a random number that depends on the division into folds C o m p l e t ec r o s s -v a l id a t i o n is the average of all possibil ities for choosing m/k instances out of m, but it is usually too expensive Exrept for leave-one-one (rc-fold cross-validation), which is always complete, fc-foM cross-validation is estimating complete K-foId cross-validationusing a single split of the data into the folds Repeat-ing cross-validation multiple limes using different spillsinto folds provides a better M onte C arlo estimate to 1 hecomplele cross-validation at an added cost In s t r a t i -fied c r o s s -v a l i d a t i o n the folds are stratified so thaitlicy contain approximately the same proportions of la-bels as the original dataset An inducer is stable for a given dataset and a set of perturbal ions if it induces classifiers thai make the same predictions when it is given the perturbed datasets P r o p o s i t i o n 1 (V a r i a n c e in A>fold C V )Given a dataset and an inducer If the inductr isstable under the pei tur bations caused by deleting theinstances f o r thr folds in k fold cross-validatwn thecross validation < stnnate will be unbiastd and the t a i lance of the estimated accuracy will be approximatelyaccrv (1—)/n when n is the number of instancesin the datasi t Proof If we assume that the k classifiers produced makethe same predictions, then the estimated accuracy has a binomial distribution with n trials and probabihly of success equal to (he accuracy of the classifier | For large enough n a confidence interval may be com-puted using Equation 3 with h equal to n, the number of instancesIn reality a complex inducer is unlikely to be stable for large perturbations unless it has reached its maximal learning capacity We expect the perturbations induced by leave-one-out to be small and therefore the classifier should be very stable As we increase the size of the perturbations, stability is less likely to hold we expect stability to hold more in 20-fold cross-validation than in 10-fold cross-validation and both should be more stable than holdout of 1/3 The proposition does not apply to the resubstitution estimate because it requires the in-ducer to be stable when no instances are given in the datasetThe above proposition helps, understand one possible assumption that is made when using cross-validation if an inducer is unstable for a particular dataset under a set of perturbations introduced by cross-validation, the ac-curacy estimate is likely to be unreliable If the inducer is almost stable on a given dataset, we should expect a reliable estimate The next corollary takes the idea slightly further and shows a result that we have observed empirically there is almost no change in the variance of the cross validation estimate when the number of folds is variedC o r o l l a r y 2 (Variance m cross-validation)Given a dataset and an inductr If the inducer is sta-ble undfi the }>tituibuhoris (aused by deleting the test instances foi the folds in k-fold cross-validation for var-ious valuts of k then tht vartanct of the estimates will be the sameProof The variance of A-fold cross-validation in Propo-sition 1 does not depend on k |While some inducers are liktly to be inherently more stable the following example shows that one must also take into account the dalaset and the actual perturba (ions E x a m p l e 1 (Failure of leave-one-out)lusher s ins dataset contains 50 instances of each class leading one to expect that a majority indu<er should have acruraov about j \% However the eombmation ofthis dataset with a majority inducer is unstable for thesmall perturbations performed by leave-one-out Whenan instance is deleted from the dalaset, its label is a mi-nority in the training set, thus the majority inducer pre-dicts one of the other two classes and always errs in clas-sifying the test instance The leave-one-out estimatedaccuracy for a majont> inducer on the ins dataset istherefore 0% M oreover all folds have this estimated ac-curacy, thus the standard deviation of the folds is again0 %giving the unjustified assurance that 'he estimate is stable | The example shows an inherent problem with cross-validation th-t applies to more than just a majority in-ducer In a no-infornirition dataset where the label val-ues are completely random, the best an induction algo-rithm can do is predict majority Leave-one-out on such a dataset with 50% of the labels for each class and a majontv ind'-cer (the best, possible inducer) would still predict 0% accuracy 2 3 B o o t s t r a pThe bootstrap family was introduced by Efron and is fully described in Efron &. Tibshirani (1993) Given a dataset of size n a b o o t s t r a p s a m p l e is created by sampling n instances uniformly from the data (with re-placement) Since the dataset is sampled with replace-ment, the probability of any given instance not beingchosen after n samples is theKOHAVI 1139expected number of distinct instances from the original dataset appearing in the teat set is thus 0 632n The eO accuracy estimate is derived by using the bootstrap sam-ple for training and the rest of the instances for testing Given a number b, the number of bootstrap samples, let e0, be the accuracy estimate for bootstrap sample i The632 bootstrap estimate is defined as(5)where ace, is the resubstitution accuracy estimate on the full dataset (i e , the accuracy on the training set) The variance of the estimate can be determined by com puting the variance of the estimates for the samples The assumptions made by bootstrap are basically the same as that of cross-validation, i e , stability of the al-gorithm on the dataset the 'bootstrap world" should closely approximate the real world The b32 bootstrap fails (o give the expected result when the classifier is a perfect memonzer (e g an unpruned decision tree or a one nearest neighbor classifier) and the dataset is com-pletely random, say with two classes The resubstitution accuracy is 100%, and the eO accuracy is about 50% Plugging these into the bootstrap formula, one gets an estimated accuracy of about 68 4%, far from the real ac-curacy of 50% Bootstrap can be shown to fail if we add a memonzer module to any given inducer and adjust its predictions If the memonzer remembers the training set and makes the predictions when the test instance was a training instances, adjusting its predictions can make the resubstitution accuracy change from 0% to 100% and can thus bias the overall estimated accuracy in any direction we want3 Related W o r kSome experimental studies comparing different accuracy estimation methods have been previously done but most of them were on artificial or small datasets We now describe some of these effortsEfron (1983) conducted five sampling experiments and compared leave-one-out cross-validation, several variants of bootstrap, and several other methods The purpose of the experiments was to 'investigate some related es-timators, which seem to offer considerably improved es-timation in small samples ' The results indicate that leave-one-out cross-validation gives nearly unbiased esti-mates of the accuracy, but often with unacceptably high variability, particularly for small samples, and that the 632 bootstrap performed bestBreiman et al (1984) conducted experiments using cross-validation for decision tree pruning They chose ten-fold cross-validation for the CART program and claimed it was satisfactory for choosing the correct tree They claimed that "the difference in the cross-validation estimates of the risks of two rules tends to be much more accurate than the two estimates themselves "Jain, Dubes fa Chen (1987) compared the performance of the t0 bootstrap and leave-one-out cross-validation on nearest neighbor classifiers Using artificial data and claimed that the confidence interval of the bootstrap estimator is smaller than that of leave-one-out Weiss (1991) followed similar lines and compared stratified cross-validation and two bootstrap methods with near-est neighbor classifiers His results were that stratified two-fold cross validation is relatively low variance and superior to leave-one-outBreiman fa Spector (1992) conducted a feature sub-set selection experiments for regression, and compared leave-one-out cross-validation, A:-fold cross-validation for various k, stratified K-fold cross-validation, bias-corrected bootstrap, and partial cross-validation (not discussed here) Tests were done on artificial datasets with 60 and 160 instances The behavior observed was (1) the leave-one-out has low bias and RMS (root mean square) error whereas two-fold and five-fold cross-validation have larger bias and RMS error only at models with many features, (2) the pessimistic bias of ten-fold cross-validation at small samples was significantly re-duced for the samples of size 160 (3) for model selection, ten-fold cross-validation is better than leave-one-out Bailey fa E lkan (1993) compared leave-one-out cross-ahdation to 632 bootstrap using the FOIL inducer and four synthetic datasets involving Boolean concepts They observed high variability and little bias in the leave-one-out estimates, and low variability but large bias in the 632 estimatesWeiss and Indurkyha (Weiss fa Indurkhya 1994) con-ducted experiments on real world data Lo determine the applicability of cross-validation to decision tree pruning Their results were that for samples at least of size 200 using stratified ten-fold cross-validation to choose the amount of pruning yields unbiased trees (with respect to their optimal size) 4 M e t h o d o l o g yIn order to conduct a large-scale experiment we decided to use 04 5 and a Naive Bayesian classifier The C4 5 algorithm (Quinlan 1993) is a descendent of ID3 that builds decision trees top-down The Naive-Bayesian clas-sifier (Langley, Iba fa Thompson 1992) used was the one implemented in (Kohavi, John, Long, Manley fa Pfleger 1994) that uses the observed ratios for nominal features and assumes a Gaussian distribution for contin-uous features The exact details are not crucial for this paper because we are interested in the behavior of the accuracy estimation methods more than the internals of the induction algorithms The underlying hypothe-sis spaces—decision trees for C4 5 and summary statis-tics for Naive-Bayes—are different enough that we hope conclusions based on these two induction algorithms will apply to other induction algorithmsBecause the target concept is unknown for real-world1140 LEARNINGconcepts, we used the holdout method to estimate the quality of the cross-validation and bootstrap estimates To choose & set of datasets, we looked at the learning curves for C4 5 and Najve-Bayes for most of the super-vised classification dataaets at the UC Irvine repository (Murphy & Aha 1994) that contained more than 500 instances (about 25 such datasets) We felt that a min-imum of 500 instances were required for testing While the true accuracies of a real dataset cannot be computed because we do not know the target concept, we can esti mate the true accuracies using the holdout method The "true' accuracy estimates in Table 1 were computed by taking a random sample of the given size computing the accuracy using the rest of the dataset as a test set, and repeating 500 timesWe chose six datasets from a wide variety of domains, such that the learning curve for both algorithms did not flatten out too early that is, before one hundred instances We also added a no inform a tion d l stt, rand, with 20 Boolean features and a Boolean random label On one dataset vehicle, the generalization accu-racy of the Naive-Bayes algorithm deteriorated hy morethan 4% as more instances were g;iven A similar phenomenon was observed on the shuttle dataset Such a phenomenon was predicted by Srhaffer and Wolpert (Schaffer 1994, Wolpert 1994), but we were surprised that it was observed on two real world datasetsTo see how well an Accuracy estimation method per forms we sampled instances from the dataset (uniformly without replacement) and created a training set of the desired size We then ran the induction algorihm on the training set and tested the classifier on the rest of the instances L E I the dataset This was repeated 50 times at points where the lea rning curve wa s sloping up The same folds in cross-validation and the same samples m bootstrap were used for both algorithms compared5 Results and DiscussionWe now show the experimental results and discuss their significance We begin with a discussion of the bias in the estimation methods and follow with a discussion of the variance Due to lack of space, we omit some graphs for the Naive-Bayes algorithm when the behavior is ap-proximately the same as that of C 4 5 5 1 T h e B i a sThe bias of a method to estimate a parameter 0 is de-fined as the expected value minus the estimated value An unbiased estimation method is a method that has zero bias Figure 1 shows the bias and variance of k-fold cross-validation on several datasets (the breast cancer dataset is not shown)The diagrams clearly show that k-fold cross-validation is pessimistically biased, especially for two and five folds For the learning curves that have a large derivative at the measurement point the pessimism in k-fold cross-Figure ] C'4 5 The bias of cross-validation with varying folds A negative K folds stands for leave k-out E rror bars are 95% confidence intervals for (he mean The gray regions indicate 95 % confidence intervals for the true ac curaries Note the different ranges for the accuracy axis validation for small k s is apparent Most of the esti-mates are reasonably good at 10 folds and at 20 folds they art almost unbiasedStratified cross validation (not shown) had similar be-havior, except for lower pessimism The estimated accu-racy for soybe an at 2 fold was 7% higher and at five-fold, 1 1% higher for vehicle at 2-fold, the accuracy was 2 8% higher and at five-fold 1 9% higher Thus stratification seems to be a less biased estimation methodFigure 2 shows the bias and variance for the b32 boot-strap accuracy estimation method Although the 632 bootstrap is almost unbiased for chess hypothyroid, and mushroom for both inducers it is highly biased for soy-bean with C'A 5, vehicle with both inducers and rand with both inducers The bias with C4 5 and vehicle is 9 8%5 2 The VarianceWhile a given method may have low bias, its perfor-mance (accuracy estimation in our case) may be poor due to high variance In the experiments above, we have formed confidence intervals by using the standard de-viation of the mea n a ccura cy We now switch to the standard deviation of the population i e , the expected standard deviation of a single accuracy estimation run In practice, if one dots a single cross-validation run the expected accuracy will be the mean reported above, but the standard deviation will be higher by a factor of V50, the number of runs we averaged in the experimentsKOHAVI 1141Table 1 True accuracy estimates for the datasets using C4 5 and Naive-Bayes classifiers at the chosen sample sizesFigure 2 C4 5 The bias of bootstrap with varying sam-ples Estimates are good for mushroom hypothyroid, and chess, but are extremely biased (optimistically) for vehicle and rand, and somewhat biased for soybeanIn what follows, all figures for standard deviation will be drawn with the same range for the standard devi-ation 0 to 7 5% Figure 3 shows the standard devia-tions for C4 5 and Naive Bayes using varying number of folds for cross-validation The results for stratified cross-validation were similar with slightly lower variance Figure 4 shows the same information for 632 bootstrap Cross-validation has high variance at 2-folds on both C4 5 and Naive-Bayes On C4 5, there is high variance at the high-ends too—at leave-one-out and leave-two-out—for three files out of the seven datasets Stratifica-tion reduces the variance slightly, and thus seems to be uniformly better than cross-validation, both for bias and vananceFigure 3 Cross-validation standard deviation of accu-racy (population) Different, line styles are used to help differentiate between curves6 S u m m a r yWe reviewed common accuracy estimation methods in-cluding holdout, cross-validation, and bootstrap, and showed examples where each one fails to produce a good estimate We have compared the latter two approaches on a variety of real-world datasets with differing charac-teristicsProposition 1 shows that if the induction algorithm is stable for a given dataset, the variance of the cross-validation estimates should be approximately the same, independent of the number of folds Although the induc-tion algorithms are not stable, they are approximately stable it fold cross-validation with moderate k values (10-20) reduces the variance while increasing the bias As k decreases (2-5) and the sample sizes get smaller, there is variance due to the instability of the training1142 LEARNING1 igure 4 632 Bootstrap standard deviation in acc-rat y (population)sets themselves leading to an increase in variance this is most apparent for datasets with many categories, such as soybean In these situations) stratification seems to help, but -epeated runs may be a better approach Our results indicate that stratification is generally a better scheme both in terms of bias and variance whencompared to regular cross-validation Bootstrap has low,variance but extremely large bias on some problems We recommend using stratified Len fold cross-validation for model selection A c k n o w l e d g m e n t s We thank David Wolpert for a thorough reading of this paper and many interesting dis-cussions We thank Tom Bylander Brad E fron Jerry Friedman, Rob Holte George John Pat Langley Hob Tibshiram and Sholom Weiss for their helpful com nients and suggestions Dan Sommcrfield implemented Lhe bootstrap method in WLC++ All experiments were conducted using M L C ++ partly partly funded by ONR grant N00014-94-1-0448 and NSF grants IRI 9116399 and IRI-941306ReferencesBailey, T L & E lkan C (1993) stimating the atcuracy of learned concepts, in Proceedings of In ternational Joint Conference on Artificial Intelli-gence , Morgan Kaufmann Publishers, pp 895 900 Breiman, L & Spector, P (1992) Submodel selectionand evaluation in regression the x random case Inttrnational St atistic al Review 60(3), 291-319 Breiman, L , Friedman, J H , Olshen, R A & StoneC J (1984), Cl a ssific ation a nd Regression Trets Wadsworth International GroupEfron, B (1983), 'E stimating the error rate of a pre-diction rule improvement on cross-validation",Journal of the Americ an St atistic al Associ ation 78(382), 316-330 Efron, B & Tibshiram, R (1993) An introduction tothe bootstra p, Chapman & HallJam, A K Dubes R C & Chen, C (1987), "Boot-strap techniques lor error estimation", IEEE tra ns-actions on p a ttern a n a lysis a nd m a chine intelli-gence P A M I -9(5), 628-633 Kohavi, R , John, G , Long, R , Manley, D &Pfleger K (1994), M L C ++ A machine learn-ing library in C ++ in 'Tools with Artifi-cial Intelligence I E E EComputer Society Press, pp 740-743 Available by anonymous ftp from s t a r r y Stanford E DU pub/ronnyk/mlc/ toolsmlc psLangley, P Tba, W & Thompson, K (1992), An anal-ysis of bayesian classifiers in Proceedings of the tenth national conference on artificial intelligence",A A A I Press and M I T Press, pp 223-228Murph' P M & Aha D W (1994), V( I repository of machine learning databases, For information con-tact ml-repository (Ui(,s uci edu Quinlan I R (1993) C4 5 Progra ms for Ma chine Learning Morgan Kaufmann Los Altos CaliforniaSchaffcr C (19941 A conservation law for generalization performance, in Maehinc Learning Proceedings of Lhe E leventh International conference Morgan Kaufmann, pp 259-265Shao, J (1993), Linear model seletion via cross-validation Journ a l of the America n sta tistica l As-sociation 88(422) 486-494 Weiss S M (1991), Small sample error rate estimationfor k nearest neighbor classifiers' I E EE Tr a ns ac tions on Pa ttern An alysis a nd Ma chine Intelligence 13(3), 285-289 Weiss, S M & lndurkhya N (1994) Decision Lreepruning Biased or optimal, in Proceedings of the twelfth national conference on artificial intel-ligence A A A I Press and M I T Press pp 626-632 Wolpert D H (1992), Stacked generalization , Neura lNetworks 5 241-259 Wolpert D H (1994a) Off training set error and a pri-ori distinctions between learning algorithms, tech-mcal Report SFI TR 94-12-121, The Sante Fe ln-stituteWolpert D II {1994b), The relationship between PAC, the statistical physics framework the Bayesian framework, and the VC framework Technical re-port, The Santa Fe Institute Santa Fe, NMZhang, P (1992), 'On the distributional properties of model selection criteria' Journ al of the America nStatistical Associa tion 87(419), 732-737 KOHAVI 1143。
A Neuro-Fuzzy Based Adaptive Set-Point Heat Exchan
Nov. 2012, Volume 6, No. 11 (Serial No. 60), pp. 1584–1588Journal of Civil Engineering and Architecture, ISSN 1934-7359, USAA Neuro-Fuzzy Based Adaptive Set-Point Heat Exchanger Control Scheme in District Heating SystemLiang Huang1, Zaiyi Liao2 and Zhao Lian11. Department of Electrical and Computer Engineering, Ryerson University, Toronto M5B2K3, Canada2. College of Hydraulic and Environmental Engineering, China Three Gorges University, Yichang 443003, ChinaAbstract: The control of heat exchange stations in district heating system is critical for the overall energy efficiency and can be very difficult due to high level of complexity. A conventional method is to control the equipment such that the temperature of hot water supply is maintained at a set-point that may be a fixed value or be compensated against the external temperature. This paper presents a novel scheme that can determine the optimal set-point of hot water supply that maximizes the energy efficiency whilst providing sufficient heating capacity to the load. This scheme is based on Adaptive Neuro-Fuzzy Inferential System. The aim of this study is to improve the overall performance of district heating systems.Key words: District heating system, neuro-fuzzy, inferential sensor, energy efficiency, control.1. IntroductionDistrict heating systems are considered energy efficient and widely used in Canada. Hot water from CHP (combined heat and power) carries heat to the heat exchangers, in which heat is transferred to the water in secondary loops. At individual buildings, appropriate operation of the heat exchangers is essential for harnessing the benefits made possible by district heating systems. The temperature of hot water supply in the secondary loop is conventionally controlled to fluctuate around a set-point, which may be constant for certain period of time or compensated against the external air temperature. Previous studies have shown that these two methods are likely to cause energy waste and/or discomfort [1]. A new approach is to change the set-point according to a measurement of thermal comfort at the buildings using temperature sensors [1]. However, using a lot of temperature sensors in a building can be practically infeasible and unstable. Liao and Dexter [1] proposed a simplifiedCorresponding author: Liang Huang, master, research fields: neuro-fuzzy network, artificial intelligence, building automation system, and control system. E-mail: **********************.physical model for estimating the average indoor air temperature by using measurable variables, such as outdoor temperature, solar radiation, and the power supplied to terminal. This model makes it possible to estimate heating load based on the outputs of simple sensors that are easily available to the controller in practice. In recent years, fuzzy logic [2] and neural networks have been proposed as alternatives to traditional statistical ones in building technology, in terms of improvement of indoor comfort and energy conservation. Researchers extensively applied fuzzy logic to the built environment to improve the performance and to reduce energy consumption [2–6], while neural networks are used for improving performance of built environment [7, 8] and estimate the operative temperature in a building [9, 10] designed an ANFIS based inferential sensor model, which estimates the average air temperature in the buildings that heated by a hydraulic heating system.In this paper, we present a neuro-fuzzy based control scheme that can estimate the heating load and according determine optimal value for the set-point of hot water supply in the secondary loop. When thesystem is operated with such set-point, the energyAll Rights Reserved.A Neuro-Fuzzy Based Adaptive Set-Point Heat Exchanger Control Scheme in District Heating System 1585efficiency can be maximized whilst desired indoor thermal environment is maintained.2. Research MethodsIn the conventional heat exchanger, the heat from the heat source is transferred to the water in secondary loops (Fig. 1) and the required flow rate of hot water from the heat source depends on required heating load, water temperature, and heating transfer rate.sr sssr ( - )*m ( - )*T T T T M η∙∙=(1)where, η is the heating transfer rate of the heatexchanger, m and T s are the water flow rate and temperature at hot-fluid outlet, M and T ss are the hot water flow rate and temperature at hot-fluid inlet, and T sr and T r are the water temperatures at hot fluid inlet and cold-fluid inlet.In this paper, two parameters are used to define the performance of a heating system: overall performance of the heating system and a measure of the thermal comfort in the zone [11]. A comfort range is defined as Φref = [T min , T max ]. The total energy consumption (E) in secondary loop is normalized to the total energy supplied to heat exchanger when the set-point is constant.100%*/o e E E = (2)A measure of the overall performance of the heating system is given by(1)1e e w w γγγγξ-+=+ (3)where, W γ is a weighting constant, which determines the importance of thermal comfort in assessing the overall performance. It should be noted that the largerFig. 1 Shell-and-tube heating exchanger [12]. the value of overall performance, the higher is the overall performance of the heating system [11].The impact of heat exchanger control on the overall performance of heating systems has been studied in simulation. Two types of heat exchanger controllers are studied:• Type I: the constant set-point controller. The supply water temperature set-point is fixed at a constant level specified during commissioning. This is a most commonly used heat exchanger controller because of its simplicity;• Type II: the adaptive set-point controller. The supply water temperature set-point in secondary loop is varied in inverse proportion to a moving average of the external environment in a certain time interval. During the test period, the temperature set-point of Type I is a constant, however the set-point changing of Type II varies based on the required heating load and the capacity of supplied heating load. The adaptive set-point in Type II cannot be varied frequently, since the profile of the control system. The temperature set-point changing time point is decided by estimating the time of instantaneous indoor air temperature equals to the average indoor temperature in one day. To look for a suitable set-point of supplied water temperature in every interval in the test period, indoor temperature comfort is considered firstly, and then, energy efficient. Liao’s simply physical model and Jassar’s [10] model is used in calculating optimal required energy. This optimal set-point need satisfy the system has a lowest energy cost when the indoor temperature in comfortable range during a certain period. The optimal required heating load is⎰1t t d Q Min (4)S.T. 0>sol Qmax min a a a T T T <<max min o o o T T T <<Once the required heating load is decided, the temperature set-point can be calculated byAll Rights Reserved.A Neuro-Fuzzy Based Adaptive Set-Point Heat Exchanger Control Scheme in District Heating System1586)()(t m Q T T dr s =-∙(5)Therefore, the temperature set-pointT QT r ds mt +=)((6)Then, an adaptive neuro-fuzzy inferential heat exchanger control scheme (illustrated in Fig. 2) is proposed and its control process is simulated. The impact of adaptive set-point heat exchanger control scheme on the overall performance of energy efficiency is studied in simulation. The experimental data used to estimate set-point temperature is obtained from a laboratory heating system monitored in an EU CRAFT project [13].3. ResultsIn the proposed control scheme, the temperature set-point estimator estimates the optimal set-point temperature of the hot water in the secondary loop and optimal set-point changing time. The thermalcomfortable range in test period is between 18︒C and 21︒C in our simulation and indoor air temperature is estimated by using adaptive neuro-fuzzy based inferential sensor model [10]. The supplied hot water temperature in the secondary loop is also sensed and the corresponding control signals is generated in heat exchanger operation module, which includes a PID (proportion integration differentiation ) controller, is sent to heat exchanger. In this case, the supplied hot temperature can follow the set-point temperature by controlling the flow rate of the hot water from CHP.In this scheme, the set-point changes twice a day at the 7.58th hour and the 18.67th hour that the indoor air temperature equals to average air temperature of the day.Fig. 3 shows a good performance of adaptive set-point in controlling indoor air temperature in thermal comfortable range. Comparing to constant set-point control heating, the indoor air temperature controlled by adaptive set-point satisfies the desired comfortable temperature range which is between 18︒C and 21︒C.Not only the adaptive has a good performance in keep indoor thermal comfortable, but also it has a good energy cost performance. Fig. 4 shows the adaptive set-point temperature control fulfill the indoor thermal comfort requirement. At the same time, the energy efficiency is also higher than constant set-point control.4. DiscussionThe neuro-fuzzy based adaptive set-point heat exchanger control scheme has a very good performance in maximizing the energy efficiency whilst providing sufficient heating capacity to the load. Jassar’s neuro-fuzzy based inferential sensor model is based on three inputs, power supplied to terminals Q in (derived from temperature difference between hot-fluid inlet and hot-fluid outlet), solar RadiationFig. 2 All Rights Reserved.A Neuro-Fuzzy Based Adaptive Set-Point Heat Exchanger Control Scheme in District Heating System1587Fig. 3 Thermal comfort performances of two types of control.Fig. 4 Impacts of heat exchanger control on the performance of heating systems.Q sol, and external temperature T O. Also, Liao’s simplified physical model for estimation of air temperature is based on the same variables. Therefore, the estimated average air temperature by Jassar’s model is possible to be used in deduction of the optimal set-point of supplied water estimation in secondary loop by using Liao’s model. Although the heating source of the proposed scheme in this paper is heat exchanger not a boiler, they are both hot-water space heating systems.In Fig. 4, the performance of Type I is far below that of the Tpye II, the reasons for the poor performance are as follows:Once commissioned the set-point is fixed for the entire test period.All Rights Reserved.A Neuro-Fuzzy Based Adaptive Set-Point Heat Exchanger Control Scheme in District Heating System 1588•If too high a value of the set-point is selected, more energy will be consumed and the room temperature is more frequently above the upper level of the desired range, resulting in lower overall performance;•If too low a value for the set-point is selected, the benefit of lower energy consumption is at the cost of significant discomfort because the room temperature is more frequently below the lower level of the desired range. Consequently the overall performance remains low.The performance of the Type II controllers is such better than Type I controller. Less energy is consumed and the room temperature is more frequently in the desired comfortable range when the controller is commissioned, such that too high a set-point is used at high external temperatures. As a result the overall performance improved in both energy consumption and comfort ratio.5. ConclusionA neuro-fuzzy based adaptive control scheme is developed to control the heat exchangers in district heating systems for maximize the energy efficiency whilst providing sufficient heating capacity to the load that the indoor temperature is controlled in a thermal comfortable range.In the future, an estimation model which can keep high robustness and high accuracy of the prediction in the indoor temperature estimation will be researched, so that the set-point estimator will have better performance and the robustness of the heat exchanger will be further improved.References[1]Z. Liao and A. L. Dexter, A simplified physical model forestimating the average air temperature in multi-zoneheating systems, Building and Environment 39 (9) (2004)1013–1022.[2]L. Zadeh, Outline of a new approach to the analysis ofcomplex systems and decision processes, in: IEEETransactions on System, Man, and Cybernetics, BrowseJournals & Magazines 3 (1) (1973) 28–44.[3] A. L. Dexter and D. W. Trewhella, Building controlsystems: fuzzy rule-based approach to performanceassessment, Building Services Research and Technology11 (4) (1990) 115–124.[4] A. I. Dounis, M. J. Santamouris and C. C. Lefas, Buildingvisual comfort control with fuzzy reasoning, EnergyConservation and Management 34 (1) (1993) 17–28.[5] A. I. Dounis, M. Bruant, M. Santamouris, G. Guaraccinoand P. Michel, Comparison of conventional and fuzzycontrol of indoor air quality in buildings, Journal ofIntelligent and Fuzzy Systems 4 (1996) 131–140.[6]P. Angelov, A fuzzy approach to building thermalsystems optimization, Vol. 2, in: Proceedings of theeighth IFSA World congress, Taipai, Taiwan, 1999, pp.528–531.[7]J. F. Kreider, Neural networks applied to building energystudies, in: H. Bloem (Ed.), Workshop on ParameterIdentification, Joint Research Center, Ispra, 1995, pp.233–251.[8]S. J. Hepeworth and A. L. Arthur, Adaptive neuralcontrol with stable learning, Mathematics and Computersin Simulation 41 (2000) 39–51.[9]M. S. Moheseni, B. Thomas and P. Fahlen, Estimation ofoperative temperature in buildings using artificial neuralnetworks, Energy and Buildings 38 (2006) 635–640. [10]S. Jassar, Z. Liao and L. Zhao, Adaptive neuro-fuzzybased inferential sensor model for estimating the averageair temperature in space heating systems, Building andEnvironment 44 (8) (2009) 1609–1616.[11]Z. Liao and A. L. Dexter, An inferential control schemefor optimizing the operation of boilers in multi-zoneheating systems, Building Service Engineering Researchand Technology 24 (4) (2003) 245–266.[12]R. K. Shah and D. P. Sekulic, Fundamental of HeatExchanger Design, John Wiley & Sons, Inc., 2003.[13]BRE (Building Research Establishment), ICITE,Controller efficiency improvement for commercial andindustrial gas and oil fired boilers, A CRAFT project,Brittech Controls Europe Ltd., 1999–2001.All Rights Reserved.。
有调节的中介模型检验方法_竞争还是替补_温忠麟解读
724 心理学报 46 卷杂 , 模型中既包括结构方程 , 也包括测量方程 , 因此 , 模型的拟合检验变得很重要 (温忠麟等 , 2012。
无论是显变量还是潜变量 , 都可以利用结构方程建模 , 使用的模型可能比用显变量建立回归模型少 , 如方程 (2和 (3用一个模型就可以了 , 图 4 是其相应的路径图。
本文讨论的有调节的中介模型只涉及一个调节变量 , 如果前半路径和后半路径的调节变量不同, 分别是 U 和 V, 则图 4 和图 7 的中介效应变成 (a1+ 本文 a3U(b1+b2V, 本文提出的检验流程仍然适用。
讨论的模型只涉及一个中介变量 , 更复杂的模型包括有调节的多重中介分析 (Hayes, 2013、多水平数这些复杂据的有调节的中介分析 (刘东等 , 2012等。
模型的检验方法和步骤 , 有待进一步研究。
参考文献 Baron, R. M., & Kenny, D. A. (1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182. Edwards, J. R., & Lambert, L. S. (2007. Methods for integrating moderation and medation: A general analytical framework using moderated path analysis. Psychological Methods, 12, 1–22. Ellis, P. D. (2010. The essential guide to effect sizes: Statistical power, meta-analysis, and the interpretation of research results. Cambridge, NY: Cambridge University Press. Fang, J., & Zhang, M. Q. (2010.Assessing point and interval estimation for the mediating effect: Distribution of the product, nonparametric bootstrap and Markov chain Monte Carlo methods. Acta Psychologica Sinica,44, 1408–1420. [ 方杰 , 张敏强 . (2012. 中介效应的点估计和区间估计 : 乘积分布法、非参数Bootstrap 和 MCMC 法 . 心理学报 , 44, 1408–1420.] Fritz, M. S., & MacKinnon, D. P. (2007. Required sample size to detect the mediated effect. Psychological Science, 18, 233–239. Hayes, A. F. (2009. Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication Monographs, 76, 408–420. Hayes, A. F. (2013. An introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York: Guilford Press. Liu, D., Zhang, Z., & Wang, M.(2012.Moderated mediation and mediated moderation: Theory construction and model checking (Gao, Z. H Trans.. In P. X. Cheng, Y. S. Xu, & L. J. Fan (Eds., Empirical methods in organization and management research (2nd ed, pp. 553–587. Beijing: PekingUniversity Press. [刘东 , 张震 , 汪默 . (2012. 被调节的中介和被中介的调节 : 理论构建与模型检验 ( 高中华译 . 见:陈晓萍 , 徐淑英 , 樊景立 ( 编 . 组织与管理研究的实证方法 ( 第二版 , pp. 553–587. 北京 : 北京大学出版社 .] MacKinnon, D. P. (2008. Introduction to statistical mediation analysis. Mahwah, NJ: Earlbaum. MacKinnon, D. P., Lockwood, C. M., Hoffman, J. M., West, S. G., & Sheets, V. A. (2002. A comparison of methods to test mediation and other intervening variable effects.Psychological Methods, 7, 83–104. MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995. A simulation study of mediated effect measures. Multivariate Behavioral Research, 30, 41–62. Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005. When moderation is mediated and mediation is moderated. Journal of Personality and Social Psychology, 89, 852–863. Preacher, K. J., & Hayes, A. F. (2004.SPSS and SAS procedures for estimating indirect effects in simple mediation models.Behavior Research Methods, Instruments, & Computers,36, 717–731. Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007. Addressing moderated mediation hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral Research, 42, 185−227. Sobel, M. E. (1982. Asymptotic confidence interval for indirect effects in structural equation models.In S. Leinhardt.(Ed., Sociological methodology (pp. 290–312. Washington, DC: American Sociological Association. Wen, Z., Chang, L, & Hau, K. T. (2006. Mediated moderator and moderated mediator. Acta Psychologica Sinica,38, 448–452. [ 温忠麟 , 张雷 , 侯杰泰 . (2006. 有中介的调节变量和有调节的中介变量 . 心理学报 , 38, 448–452.] Wen, Z., Chang, L, Hau, K. T., & Liu, H. (2004. Testing and application of the mediating effects. Acta Psychologica Sinica, 36, 614–620. [ 温忠麟 , 张雷 , 侯杰泰 , 刘红云 . (2004. 中介效应检验程序及其应用 . 心理学报 , 36, 614–620.] Wen, Z., Hau, K. T., & Chang, L. (2005. A comparison of moderator and mediator and their applications. Acta Psychologica Sinica, 37, 268–274. [ 温忠麟 , 侯杰泰 , 张雷 . (2005. 调节效应与中介效应的比较和应用 . 心理学报 , 37, 268–274.] Wen, Z., Liu, H., & Hau, K. T. (2012. Analyses of moderating and mediating effects. Beijing: Educational Science Publishing House. [ 温忠麟 , 刘红云 , 侯杰泰 . (2012. 调节效应和中介效应分析 . 北京 : 教育科学出版社 .] Ye, B., & Wen, Z. (2013. A discussion ontesting methods for mediated moderation models: Discrimination and integration. Acta Psychologica Sinica, 45, 1050–1060. [ 叶宝娟 , 温忠麟 . (2013. 有中介的调节模型检验方法 : 甄别和整合 . 心理学报 ,45, 1050–1060.] Ye, B., Yang, Q., & Hu, Z. (2012. The effect mechanism of parental control, deviant peers and sensation seeking on drug Use among reform school students. Psychological Development and Education, 28, 641–650. 不良同伴和感觉 [叶宝娟 , 杨强 , 胡竹菁 . (2012. 父母控制、寻求对工读生毒品使用的影响机制 , 心理发展与教育 , 28, 641–650.] Ye, B., Yang, Q., & Hu, Z. (2013. Effect of gratitude on adolescents’ academic achievement: Moderated mediating effect.Psychological Development and Education, 29, 192–199. [叶宝娟 , 杨强 , 胡竹菁 . (2013.感恩对青少年学业成就的影响:有调节的中介效应 , 心理发展与教育 , 29, 192–199.] Yuan, Y., & MacKinnon, D. P. (2009. Bayesian mediation analysis. Psychological Methods, 14, 301–322.5期温忠麟等: 有调节的中介模型检验方法:竞争还是替补 725 附录 1 1.0000.262 0.181 –0.019 –0.048 变量间的协方差矩阵文件 (p1.txt 1.000 0.539 0.036 0.0441.000 1.000 1.073 0.608 1.142 0.034 0.082 –0.044 0.036 ANALYSIS: Bootstrap=2000; ! Bootstrap 法抽样 2000 次 MODEL: W on X (a1 U UX (a3; !做 W 对 X,U, UX 的回归 !X 和 UX 的回归系数分别命名为 a1 和 a3 Y on X U W (b1 UW (b2; !做 Y 对 X,U, W, UW 的回归 !W 和 UW 的回归系数分别命名为 b1 和 b2 MODEL CONSTRAINT: new (H1-H7; H1= a1*b2; H2= a3*b1; H3= a3*b2; H4=a1*b1; ! a1b2 的估计 ! a3b1 的估计 ! a3b2 的估计 !当 U 等于 0 时的 (a1+a3U(b1+b2U !的中介效应的值 H5=H4+H1+ H2+ H3; !当 U 等于 1 时的中介效应 (a1+a3U(b1+b2U的值 H6=H4-H1-H2+H3; !当 U 等于 -1 时的中介效应 (a1+a3U(b1+b2U的值 H7=H5-H4; ! U 等于 1 和 0时的 (a1+a3U(b1+b2U之差 OUTPUT: cinterval (bcbootstrap; !输出系数乘积及中介效应之差的偏差校正的百分位 !Bootstrap 计算的中介效应置信区间–0.183 –0.153 –0.180 注释:变量依次为学业成就 (Y 、复原力 (W 、感恩 (X 、感恩与生活事件交互项 (UX及复原力与生生活事件 (U、活事件交互项 (UW。
基于短时PRI变换的多目标分选和识别算法
制导与引信GUIDANCE & FUZE第41卷第4期2020年12月Vol. 4 1 No. 4Dec. 2020文章编号:1671-0576(2020)04-0027-06基于短时PRI 变换的多目标分选和识别算法奚 银,夏新凡,吴迎春,刘永杰,王耀金(上海无线电设备研究所,上海201109)摘要:针对经典脉冲重复间隔(PRI )变换类算法和平面变换类算法无法自主获取复杂信号的实时重频变化特性的问题,在修正PRI 变换算法基础上提出短时PRI 变换方法。
该方法结合对目标脉冲序列短时加窗和PRI 变换的方法,获取了目标信号连续的重频变化信息。
仿真结果表明:这种算法不仅能够高精度地估计目标PRI 中心值,还能获取PRI 的特征信息和调制方式,实现实时有效的多目标分选和截获。
关键词:反辐射雷达;信号分选;短时脉冲重复间隔变换;重频抖动中图分类号:TN971文献标志码:ADOI : 10.3969/j.issn.l 6710576.2020.04.006A Novel Algorithm for Multi-signal Deinterleaving andRecognition Based on Short-time PRI TransformXI Yin , XIA Xin-fan , WU Ying-chun , LIU Yong-jie , WANG Yao-jin(Shanghai Radio Equipment Research Institute, Shanghai 201109, China)Abstract : The classical pulse repetition interval (PRI ) transform algorithm and thetraditionalplanetransform technology can notacquire and describethe PRI modulation characteristics of complex signals automatically. To solve the problem , a novel short.-t.imePRI transform algorithm is suggested based on the improved PRI transform algorithm. It combines the t.wo methods of short-time windowing and PRI transform algorithm , to get the modulation information and changing characteristics of the target signals PRI. Thesimulationresults show thatthe proposed algorithm is successfulin deinterleaving and recognizing mult.i-target.as well as accurately estimating the PRI center value. Moreover , thealgorithmcanobtain morePRIchangingcharacteristics.Inthissense ,theproposed method isarealtimeande f ectivemethodformulti-signaldeinterleavingandinterception.Keywords : ant.i-radiat.ion radar; signal deinterleaving ; short-time PRI transformation ;PRIji t er收稿日期:2020-07-21基金项目:空军预研基金(303020503)作者简介:奚 银(1989—)女,硕士,工程师,主要从事被动导引头雷达信号处理技术研究。
可靠度反分析方法及其在隧道衬砌结构设计中应用
可靠度反分析方法及其在隧道衬砌结构设计中应用原创性声明本人声明,所呈交的学位论文是本人在导师指导下进行的研究工作及取得的研究成果。
尽我所知,除了论文中特别加以标注和致谢的地方外,论文中不包含其他人已经发表或撰写过的研究成果,也不包含为获得南华大学或其他单位的学位或证书而使用过的材料。
与我共同工作的同志对本研究所作的贡献均已在论文中作了明确的说明。
本文的试验数据,是本文作者、导师和生产单位经过巨大努力通过测试获取的。
如有作者需要引用,须经三者书面同意,否则可视为侵权。
作者签名:日期:年月日关于学位论文使用授权说明本人同意南华大学有关保留、使用学位论文的规定,即:学校有权保留学位论文,允许学位论文被查阅和借阅;学校可以公布学位论文的全部或部分内容,可以采用复印、缩印或其它手段保留学位论文;学校可根据国家或湖南省有关部门规定送交学位论文。
作者签名:导师签名:日期:年月日摘要随着基于概率统计基础的结构可靠度理论的发展,经常要根据一个预定或需要的可靠度指标并由其相应的极限状态方程而找到相应的设计参数,该参数须满足相应的结构可靠度水平,传统的正向可靠度理论只能通过试错的方法寻找合适的设计参数,而这种逼近往往是无效的,计算成本也比较大,这就引出了可靠度的反问题分析方法。
本文就可靠度反分析方法及其在隧道衬砌结构设计应用进行了如下研究:(1)首先概要介绍了正向结构可靠度的基本理论,从而引出可靠度的反问题,并系统归纳了可靠度反问题在各种不同情况下的详细计算步骤,并通过算例进行计算比较,证明了该方法的高效性和准确性。
(2)在国内外隧道及地下结构的设计施工中,传统的“荷载-结构”模型仍是设计的主要依据,尤适用于整体式衬砌,而“连续介质”模型可以利用大型有限元软件进行数值模拟,在今后的隧道设计中存在广阔发展空间。
故本文以国内外各种调查统计资料为依据,针对“荷载-结构”模型设计的浅埋隧道衬砌,利用蒙特卡洛-有限元对作用效应进行统计分析,在给定结构目标可靠度的情况下,通过数次迭代计算出各衬砌截面在抗压和抗拉条件下的隧道衬砌截面的厚度(或其均值),并绘制出隧道衬砌轮廓。
generalized method of moments estimator -回复
generalized method of moments estimator -回复什么是广义矩估计法?广义矩估计法(Generalized Method of Moments,简称GMM)是一种用于估计经济模型参数的统计方法。
它是由经济学家Lars Peter Hansen和Thomas J. Sargent于1982年提出的。
GMM的基本思想是通过找到一个或多个矩条件来估计模型参数,其中矩条件是指对于一组给定的数据,模型预测的矩的理论值与实际观测值之间的差异。
GMM的核心是通过最小化模型预测矩与实际观测矩之间的差异来估计模型参数。
为了做到这一点,GMM引入了一个被称为矩条件函数(moment condition function)的函数,用于度量模型预测矩与观测矩之间的距离。
这个函数通常被定义为预测矩与观测矩之差的平方的加权和。
通过最小化矩条件函数,GMM可以找到最优的模型参数估计。
GMM的估计过程可以分为以下几个步骤:1. 确定模型的矩条件:首先,需要确定经济模型的矩条件,即模型的理论预测的矩与实际观测值之间的关系。
这些矩条件通常是经济模型的一些基本性质,如均值、方差、相关系数等。
2. 构建矩条件函数:根据确定的矩条件,构建矩条件函数,用于衡量模型的预测矩与实际观测矩之间的差异。
通常,矩条件函数是预测矩与观测矩之差的平方的加权和。
3. 选择权重矩阵:为了在估计模型参数时,合理地权衡不同矩条件的重要性,需要选择一个权重矩阵。
这个权重矩阵可以反映出不同矩条件的可信度或重要性。
常用的选择方法包括矩方差估计和最小二乘估计。
4. 最小化矩条件函数:使用所选择的权重矩阵和确定的矩条件函数,通过最小化矩条件函数来估计模型参数。
这个最小化过程可以使用各种数值优化算法来实现,如牛顿法、梯度下降法等。
5. 估计参数的标准误差:一旦模型参数估计得到,还需要对估计结果进行统计推断,以获得参数估计的精确度。
通常采用的方法是计算参数估计的标准误差,以衡量估计结果的波动性。
hardy-wilson分级方法
hardy-wilson分级方法English Answer:Hardy-Wilson Confidence Interval for a Proportion.The Hardy-Wilson confidence interval is a method for estimating the true proportion of a population that has a certain characteristic, based on a sample of that population. It is a more accurate method than the simple binomial confidence interval, especially when the sample size is small.The Hardy-Wilson confidence interval is calculated using the following formula:CI = p ± z sqrt((p (1 p)) / n)。
where:p is the sample proportion.z is the z-score corresponding to the desired confidence level.n is the sample size.For example, let's say we have a sample of 100 people, and 60 of them have a certain characteristic. The sample proportion is therefore 0.6. If we want to construct a 95% confidence interval for the true proportion of the population that has this characteristic, we would use the following formula:CI = 0.6 ± 1.96 sqrt((0.6 (1 0.6)) / 100)。
详细议程安排
详细议程安排12月29日(周五)时间日程安排地点08:15-08:30 开幕式经济楼N402 主持人:陈海强,厦门大学致辞:1、洪永淼,康奈尔大学、厦门大学2、汪寿阳,中国科学院3、王维国,东北财经大学4、张永山,《经济研究》杂志社08:30-09:15 主题演讲I 经济楼N402 主持人:方颖,厦门大学嘉宾一:Whitney Newey,麻省理工学院(MIT)题目:TBD09:15-09:45 嘉宾二:汪同三,中国社会科学院题目:数量经济学在中国——对数量经济学的几点认识09:45-10:15 嘉宾三:王维国,东北财经大学题目:婚育间隔收入效应与生育配套政策10:15-10:45 茶歇经济楼N座4楼、5楼10:45-12:25 Parallel Session I-1(English Session)Econ Building N303 Chair:Xingbai Xu, Xiamen UniversityPresenter:Yuying Sun, Chinese Academy of SciencesTitel:Analysis of Trump Election's Impacts on Stock Market with an Interval-Valued Times Series Modelling ApproachDiscussant:Jing Xue,Dongbei University of Finance and EconomicsPresenter: Jing Xue,Dongbei University of Finance and Economics Titel:Profile Likelihood Estimation for Randomly Varying Coefficient Model with EndogeneityDiscussant:Rong Guan,Central University of Finance and Economics时间日程安排地点Presenter: Rong Guan,Central University of Finance and EconomicsTitel: Estimating Probability of Financial Distress:An Interval-data-based MethodDiscussant:Xia Wang,Sun Yat-Sen UniversityPresenter:Xia Wang,Sun Yat-Sen UniversityTitel:Testing for Structural Changes in Factor Models via aNonparametric RegressionDiscussant:Yuying Sun, Chinese Academy of Sciences10:45-12:25 Parallel Session I-2(English Session)Econ Building N301 Chair:Xiaojia Bao, Xiamen UniversityPresenter: Zhonghua Du,Dongbei University of Finance and Economics Titel:The Relationship between Mortality Rates and Income in China: A Nonlinear Panel Data ApproachDiscussant:Mengna Luan,Southwestern University of Finance and EconomicsPresenter:Mengna Luan,Southwestern University of Finance and EconomicsTitel:Air Pollution, Hospital Admissions, and Economic Loss Discussant:Xuebo Wang,Chinese University of Hong KongPresenter:Xuebo Wang, Chinese University of Hong KongTitel:A Tale of Hardships: Gene, Family Education and Children’s Response to Negative ShocksDiscussant:Chee-Ruey Hsieh,The University of Nottingham Ningbo ChinaPresenter:Chee-Ruey Hsieh, The University of Nottingham Ningbo ChinaTitel:The Hidden Costs of Mental Depression: Implications on Social Trust and Life SatisfactionDiscussant:Zhonghua Du,Dongbei University of Finance and Economics10:45-12:25 论文分组报告 I-3 经济楼N501主持人:唐寿宁,《经济研究》编辑部报告人:程磊,武汉大学题目:政治关联丧失与企业的应对策略:基于上市公司官员独董“离职潮”事件的研究评论人:范青亮,厦门大学时间日程安排地点报告人:范青亮,厦门大学题目:举办大型体育赛事对当地经济的影响——以国际马拉松比赛为例评论人:何青,中国人民大学报告人:何青,中国人民大学题目:从政履历是否影响经济增长的同步性?评论人:林建浩,中山大学报告人:林建浩,中山大学题目:威权传统、决策保守性与企业人力资本投资:基于广东三大方言族群的实证研究评论人:程磊,武汉大学10:45-12:25 论文分组报告 I-4 经济楼N401主持人:蒙莉娜,厦门大学报告人:陈鸿,中山大学题目:Transient and Persistent Inefficiency Traps in Chinese Province评论人:梁梓峰,山东大学报告人:梁梓峰,山东大学题目:政府财政支出能否切实改善民生?——基于中国1993—2016年省级面板数据的分析评论人:王瑞瑜,山西财经大学报告人:王瑞瑜,山西财经大学题目:经济增长、居民可支配收入与居民消费的关系研究——基于Panel_VAR 模型的省际面板数据分析评论人:陈鸿,中山大学10:45-12:25 论文分组报告 I-5 经济楼N406主持人:李嘉楠,厦门大学报告人:陈露,东南大学题目:人口结构性变化影响产业区际转移吗?——基于时空影响分析与中国事实检验评论人:李磊,华中科技大学报告人:李磊,华中科技大学题目:人口流动与智能制造:基于资本—技能互补的视角评论人:刘丰,东北财经大学时间日程安排地点报告人:刘丰,东北财经大学题目:城际人口迁移的结构效应——兼论城市最优规模评论人:陈飞,东北财经大学报告人:陈飞,东北财经大学题目:乡城迁移、期望调整与幸福损失评论人:陈露,东南大学10:45-12:25 论文分组报告 I-6 经济楼N118主持人:王利娜,《经济研究》编辑部报告人:付晓琼,中国海洋大学题目:中国GDP数据的真实性:基于291个地级及以上城市2012-2016年NPP-VIIRS夜间灯光的检验评论人:王伟同,东北财经大学报告人:王伟同,东北财经大学题目:国有企业、体制身份与人力资本跨区流动评论人:周末,对外经济贸易大学报告人:周末,对外经济贸易大学题目:成本加成、市场势力与中国工业企业垄断损失——对DLW模型的修正与改进评论人:付晓琼,中国海洋大学10:45-12:25 论文分组报告 I-7 经济楼N308主持人:王学新,厦门大学报告人:程婷婷,南开大学题目:Multi-Step Non- and Semi-Parametric Predictive Regressions for Short and Long Horizon Stock Return Prediction评论人:李欣先,齐鲁工业大学报告人:李欣先,齐鲁工业大学题目:固定效应动态空间面板模型研究评论人:廖小赛,厦门大学报告人:廖小赛,厦门大学题目:A New Approach to Test Predictability in Quantile Regression with Persistent Predictors评论人:王志超,北京航空航天大学时间日程安排地点报告人:王志超,北京航空航天大学题目:纵向成分数据的线性混合效应模型评论人:程婷婷,南开大学10:45-12:25 论文分组报告 I-8 经济楼D235主持人:李培,厦门大学报告人:冯科,北京大学题目:深圳市房地产泡沫检测及其影响因素分析评论人:黄飞鸣,江西财经大学报告人:黄飞鸣,江西财经大学题目:银行信贷扩张与房价上涨的变点分析评论人:田超岳,西南财经大学报告人:田超岳,西南财经大学题目:Termination of Listing Contracts: A Competing Risk Survival Analysis评论人:赵宁如,南京审计大学报告人:赵宁如,南京审计大学题目:House Prices and Home Owner Labor Market Behavior in China 评论人:冯科,北京大学10:45-12:25 论文分组报告 I-9 经济楼D236主持人:谭丽佳,天津大学报告人:林志帆,厦门大学题目:关系文化与企业研发创新评论人:赵玮,南开大学报告人:赵玮,南开大学题目:Beautiful Wife Makes Family Decisions? An Empirical Analysis on Intra‐Household Bargaining Power of Chinese Married Women评论人:肖伟,西南财经大学报告人:肖伟,西南财经大学题目:Power Increases Productivity: Evidence from Chinese Academia 评论人:谭丽佳,天津大学时间日程安排地点报告人:谭丽佳,天津大学题目:Evaluating the Car License Auction Formats of Shanghai,Guangzhou and Singapore: Theory and Experimental Evidence评论人:林志帆,厦门大学10:45-12:25 论文分组报告 I-10 经济楼D336主持人:李迎星,厦门大学报告人:路继业,东北财经大学题目:软钉住还是两极化:汇率制度演进中被淡忘的事实与经验解释评论人:许艺煊,中央财经大学报告人:许艺煊,中央财经大学题目:China’s Trilemma: Monetary Policy Autonomy in an Economy with a Managed Floating Exchange Rate评论人:赵星,华东师范大学报告人:赵星,华东师范大学题目:中国货币政策对美国的溢出效应研究评论人:路继业,东北财经大学12:30-13:30 午餐逸夫楼中餐厅14:00-14:30 主题演讲II 经济楼N402 主持人:金成武,《经济研究》编辑部嘉宾一:刘金全,吉林大学题目:金融发展与经济增长收敛:理论与经验分析14:30-15:00 嘉宾二:汪寿阳,中国科学院题目:计量模型与经济预测的几个问题15:00-15:30 嘉宾三:张晓峒,南开大学题目:季节调整方法在中国的应用15:30-16:00 茶歇经济楼N座4楼、5楼16:00-17:40Parallel Session II-1(English Session)Econ Building N303Chair:Li Chen, Xiamen University时间日程安排地点Presenter:Yingjun Su,Jinan UniversityTitel:A Robust Approach to Estimating ProductionFunctions:Replication of the ACF ProcedureDiscussant:Jilin Wu, Shandong UniversityPresenter:Jilin Wu, Shandong UniversityTitel:Testing for Parameter Constancy in Linear Regressions withAutocorrelation and Changing VarianceDiscussant:Yanfen Zhang, Xiamen UniversityPresenter:Yanfen Zhang, Xiamen UniversityTitle:Random Weighting the Portmanteau Tests for Multivariate WhiteNoise with Unknown Dependent StructureDiscussant:Li Chen, Xiamen UniversityPresenter:Li Chen, Xiamen UniversityTitel:Trending Time Series Models with Endogeneity: A ControlFunction ApproachDiscussant:Yingjun Su,Jinan University16:00-17:40 论文分组报告 II-2 经济楼N501主持人:傅十和,厦门大学报告人:郭萌萌,西南财经大学题目:Does Air Pollution Affect Local Stock Returns in China? 评论人:晋晶,中国人民大学报告人:晋晶,中国人民大学题目:再议“南北供暖”之争——来自断点回归设计的证据评论人:田贤亮,中南财经政法大学报告人:田贤亮,中南财经政法大学题目:Environmental Regulation and Firm Export: Evidence from China’s Environmental Information Disclosure评论人:肖有智,北京大学报告人:肖有智,北京大学题目:碳排放与居民主观幸福感的研究——基于幸福感数据的经验分析评论人:郭萌萌,西南财经大学论文分组报告 II-3 经济楼N401主持人:张庆昭,厦门大学时间日程安排地点16:00-17:40 报告人:曹湛,同济大学题目:Age of Retirement and Consumption评论人:孙榆婷,西南财经大学报告人:孙榆婷,西南财经大学题目:宗教信仰、金融市场参与和家庭资产选择评论人:吴坤,上海财经大学报告人:吴坤,上海财经大学题目:城镇居民家庭储蓄率之谜:基于Age -Period-Cohort分解的再考察评论人:张雅欣,西南财经大学报告人:张雅欣,西南财经大学题目:迁移行为与社会信任——基于CLDS2014数据的实证研究评论人:曹湛,同济大学16:00-17:40 论文分组报告 II-4 经济楼N406主持人:陈小亮,《经济研究》编辑部报告人:高华川,天津财经大学题目:中国GDP的密度预测评论人:姜向荣,山东省科技发展战略研究所报告人:姜向荣,山东省科技发展战略研究所题目:基于季节性的时间序列预测建模及其在短期宏观经济预测中的应用评论人:欧阳志刚,中南财经政法大学报告人:欧阳志刚,中南财经政法大学题目:Economic Policy Uncertainty, Two-Wheel Drive and Economic Growth评论人:田磊,浙江财经大学报告人:田磊,浙江财经大学题目:基建股指收益率、财政支出预期与宏观经济波动评论人:高华川,天津财经大学时间日程安排地点16:00-17:40 论文分组报告 II-5 经济楼N301主持人:范青亮,厦门大学报告人:韩猛,内蒙古财经大学题目:基于累积平方和统计量的动态因子模型结构突变点的探测与检验评论人:柳向东,暨南大学报告人:柳向东,暨南大学题目:非平衡时序数据的动态时间规整过采样方法研究评论人:叶倩婷,华南理工大学报告人:叶倩婷,华南理工大学题目:空间多层次误差移动平均模型的GMM估计与应用——基于三维非平衡面板数据评论人:涂云东,北京大学报告人:涂云东,北京大学题目:函数型核加权最小二乘估计法及其在经济学中的应用评论人:韩猛,内蒙古财经大学16:00-17:40 论文分组报告 II-6 经济楼N308主持人:林娟,厦门大学报告人:李春艳,东北师范大学题目:怎样测量政府R&D资助对产业创新过程的影响?评论人:李世奇,上海社会科学院报告人:李世奇,上海社会科学院题目:地方政府竞争与R&D补贴--基于中国省际面板数据的空间效应研究评论人:王辉,华东师范大学报告人:王辉,华东师范大学题目:政策规制、清洁技术创新与中国环境质量评论人:温焜,南昌大学报告人:温焜,南昌大学题目:提升我国高新技术企业研发效率实证研究评论人:李春艳,东北师范大学16:00-17:40 论文分组报告 II-7 经济楼N118时间日程安排地点主持人:钟威,厦门大学报告人:丁杰,厦门大学题目:“好”的不确定性、“坏”的不确定性与股票市场定价——基于中国A股高频数据的研究评论人:宋䶮娜,北京大学报告人:宋䶮娜,北京大学题目:散户交易、短期轮动和套利非对称性——A股市场特质波动解析评论人:童晨,北京大学报告人:童晨,北京大学题目:Does Aggregate Economic Uncertainty Predict the Volatilityof Financial Assets ?评论人:杨祯奕,中山大学报告人:杨祯奕,中山大学题目:Peer Effect and Corporate Investment between US and China评论人:丁杰,厦门大学16:00-17:40 论文分组报告 II-8 经济楼D235主持人:朱浣君,厦门大学报告人:刘霞,北京师范大学题目:语言沟通能力对企业出口行为的影响——基于国民英语能力的视角评论人:万千,华中科技大学经济学院报告人:万千,华中科技大学经济学院题目:公共教育支出、分配结构与教育代际传递评论人:王泽荣,西南财经大学报告人:王泽荣,西南财经大学题目:本科教育民族间的增加值差距及变化趋势评论人:余海跃,东北财经大学报告人:余海跃,东北财经大学题目:生育成本、代际分工与青年女性就业评论人:刘霞,北京师范大学16:00-17:40论文分组报告 II-9 经济楼D236主持人:谢谦,《经济研究》编辑部时间日程安排地点报告人:黄丽灵,中央财经大学题目:资产价格传染、抛售博弈与中国银行业系统性风险评论人:蒋晓宇,厦门大学报告人:蒋晓宇,厦门大学题目:系统性金融风险测度及跨市场风险传递效应分析——以银行部门和房地产部门为例评论人:倪骁然,厦门大学报告人:倪骁然,厦门大学题目:影子银行、风险传染与股价崩盘:来自同业存单的证据评论人:王俏,首都经济贸易大学报告人:王俏,首都经济贸易大学题目:中国上市公司利润报表真实性的实证分析评论人:黄丽灵,中央财经大学16:00-17:40 论文分组报告 II-10 经济楼D336主持人:孙三百,《经济研究》编辑部报告人:潘宁宁,西南财经大学题目:投资者结构与股价崩盘风险—— 基于金融类机构投资者和一般法人持股的对比分析评论人:奚晓军,闽南师范大学报告人:奚晓军,闽南师范大学题目:我国沪深指数VaR的测度应该考虑自相关性吗?评论人:张传海,中南财经政法大学报告人:张传海,中南财经政法大学题目:市场微观结构噪声真的仅仅是“噪声”么?——基于包含噪声成份收益波动的预测研究评论人:周德才,南昌大学报告人:周德才,南昌大学题目:基于混频损失函数的中国实时金融状况指数另一种构建评论人:潘宁宁,西南财经大学18:00-19:30 晚宴博士生论坛优秀论文颁奖颁奖人:方颖,厦门大学时间日程安排地点08:30-09:00 主题演讲III 经济楼N402 主持人:郑挺国嘉宾一:朱平芳,上海社会科学院题目:研发加计扣除政策对非国有高技术产业的效果评价研究09:00-09:30 嘉宾二:王潼,国家发改委经济体制与管理研究所题目:追踪计量经济学最新进展,深入研究我国经济新常态09:30-10:00 嘉宾三:王少平,华中科技大学题目:Testing for an α0 Moderate Explosiveness process10:00-10:30 茶歇经济楼N座4楼、5楼10:30-12:10 Parallel Session III-1(English Session)Econ Building N303 Chair:Muyi Li,Xiamen UniversityPresenter:Biqing Cai, Huazhong University of Science and TechnologyTitel:Improved inferences in fund performance evaluation using time-varying fund alphas and betas: A nonparametric approach Discussant:Jinyuan Chang,Southwestern University of Finance and EconomicsPresenter:Jinyuan Chang,Southwestern University of Finance and EconomicsTitel:A Frequency Domain Analysis of the Error Distribution from Noisy High-Frequency DataDiscussant:Dong Li, Tsinghua UniversityPresenter:Dong Li, Tsinghua UniversityTitel:Inference for Asymmetric Exponentially Weighted Moving Average ModelDiscussant:Hui Wang, Central University of Finance and EconomicsPresenter:Hui Wang, Central University of Finance and Economics Titel:Inference for Spatial Dynamic Panel Model with Different Spatial Dependence CharacterizationsDiscussant:Biqing Cai, Huazhong University of Science and Technology10:30-12:10 Parallel Session III-2(English Session)Econ Building N301时间日程安排地点Chair:Wei Song, Xiamen UniversityPresenter:Mian Huang, Southwestern University of Finance andEconomicsTitel:Simulation-Based Estimation of Multinomial Discrete ChoiceModel with Fixed EffectsDiscussant:Zhewen Pan,Sun Yat-Sen UniversityPresenter:Zhewen Pan, Sun Yat-Sen UniversityTitel:Semiparametric Estimation of a Censored Regression ModelSubject to Nonparametric Sample SelectionDiscussant:Qiuhua Xu,Southwestern University of Finance andEconomicsPresenter:Qiuhua Xu,Southwestern University of Finance andEconomics Titel:Inferences for Varying-Coefficient Panel DataModels with Cross Sectional DependenceDiscussant:Wei Song, Xiamen UniversityPresenter:Wei Song, Xiamen UniversityTitel:A Semiparametric Estimator for Binary Response Models withEndogenous RegressorsDiscussant:Mian Huang, Southwestern University of Finance andEconomics10:30-12:10 论文分组报告 III-3 经济楼N501主持人:林蔚,首都经济贸易大学报告人:胡晨沛,浙江工商大学题目:农村居民政治参与行为会影响农户人情支出吗?——基于分位数回归模型的实证分析评论人:宁磊,上海财经大学报告人:宁磊,上海财经大学题目:Land Supply, Housing Pricing and Wage Premium: The Case of China评论人:万威,厦门大学报告人:万威,厦门大题目:机票价格市场化损害了消费者的利益吗?评论人:林蔚,首都经济贸易大学时间日程安排地点报告人:林蔚,首都经济贸易大学题目:外资流入的工资效应:基于可变系数模型的估计评论人:胡晨沛,浙江工商大学10:30-12:10 论文分组报告 III-4 经济楼N406主持人:罗楚亮,北京师范大学报告人:徐舒,西南财经大学题目:From Individual Inequality to Household Inequality: the Role of Assortative Mating评论人:丁海燕,浙江财经大学报告人:丁海燕,浙江财经大学题目:Estimate the Income Inequality Using Engel Curve Approach 评论人:文强,西南财经大学报告人:文强,西南财经大学题目:国有经济对要素收入分配的影响评论人:罗楚亮,北京师范大学报告人:罗楚亮,北京师范大学题目:富豪榜、帕累托分布与高收入人群遗漏评论人:徐舒,西南财经大学10:30-12:10 论文分组报告 III-5 经济楼N401主持人:孟磊,厦门大学报告人:王伊攀,山东工商学院题目:环境规制政策与企业空间迁移——基于重污染上市公司的分析评论人:杨欣桐,首都经济贸易大学报告人:杨欣桐,首都经济贸易大学题目:基于空间溢出效应视角的中国城市住房价格影响因素的实证分析评论人:张龙,西北大学报告人:张龙,西北大学题目:劳动收入份额,地区经济增长水平差异与居民消费的关系 --基于分层线性模型的分析评论人:邹薇,武汉大学报告人:邹薇,武汉大学题目:高速铁路、市场准入与经济增长评论人:王伊攀,山东工商学院时间日程安排地点10:30-12:10 论文分组报告 III-6 经济楼N308主持人:马键,广州大学报告人:高艳平,山西财经大学题目:有限样本条件下重复抽样建模方法研究——FSR方法模拟与证明评论人:祁磊,北京师范大学报告人:祁磊,北京师范大学题目:有限混合分布的假设检验及其应用评论人:马键,广州大学报告人:马键,广州大学题目:高维稀疏空间中多值处置效应的双重群组套索估计评论人:钱谊,大连理工大学报告人:钱谊,大连理工大学题目:结构方程、协整关系与SVECM的识别条件——对Gali(1992)数据的再分析评论人:高艳平,山西财经大学10:30-12:10 论文分组报告 III-7 经济楼N118主持人:冯峥晖,厦门大学报告人:梁华杰,华南理工大学题目:监督强度、经济发展水平与控股股东侵占动机转化─金融危机时期的观察评论人:曲彤,北京交通大学报告人:曲彤,北京交通大学题目:Does Support from Government Help Firms Survive? Evidence on Financial and Political Assistance in China,1998-2007评论人:刘方,中国人民大学报告人:刘方,中国人民大学题目:孵化器有效促进了企业创新吗——来自中关村海淀科技园的微观证据评论人:王庆涛,香港城市大学报告人:王庆涛,香港城市大学题目:Getting government subsidy: the roles of executives’ political background and ownership评论人:梁华杰,华南理工大学时间日程安排地点10:30-12:10 论文分组报告 III-8 经济楼D235主持人:林明,厦门大学报告人:陈丽梅,西南财经大学题目:春节期间烟花爆竹禁放政策能否换回城市的蓝天评论人:郝淑双,中南财经政法大学报告人:郝淑双,中南财经政法大学题目:中国区域绿色发展水平影响因素的空间计量评论人:林爱华,华侨大学报告人:林爱华,华侨大学题目:环境污染条件下经济人的“契约式”身份与环境管制评论人:吴雪萍,福州大学报告人:吴雪萍,福州大学题目:基于空间半参模型的空气污染与经济增长关系再检验评论人:陈丽梅,西南财经大学10:30-12:10 论文分组报告 III-9 经济楼D236主持人:周山,厦门大学报告人:钱进,山东财经大学题目:“一带一路”、东道国制度与企业OFDI ——基于动态面板数据GMM 的经验考量评论人:孙照吉,厦门大学报告人:孙照吉,厦门大学题目:企业融入GVC的竞争策略的竞争策略与劳动收入份额评论人:王贺,山西财经大学报告人:王贺,山西财经大学题目:中美贸易差额研究:一个非线性协整检验评论人:吴群锋,北京大学报告人:吴群锋,北京大学题目:企业对外投资与出口产品多元化评论人:钱进,山东财经大学10:30-12:10论文分组报告 III-10 经济楼D336主持人:黎晖晖,厦门大学时间日程安排地点报告人:吕龙,华中科技大学题目:全球股票市场系统性风险溢出研究——基于△CoVaR和社会网络方法的分析评论人:王雪,暨南大学报告人:王雪,暨南大学题目:中国行业指数收益率的信息溢出效应研究——基于有向无环图和网络分析法应用分析评论人:肖毓琨,中国海洋大学报告人:肖毓琨,中国海洋大学题目:国际能源市场对中股的波动溢出效应研究—基于Copula分位数评论人:杨涛,西南财经大学报告人:杨涛,西南财经大学题目:房地产行业与系统性风险——基于动态高维Copula模型评论人:吕龙,华中科技大学12:15-13:30 午餐逸夫楼中餐厅14:00-15:40 论文分组报告 IV-1 经济楼N501主持人:韩晓祎,厦门大学报告人:黄永强,中南财经政法大学题目:交通基础设施投资与经济增长——基于准自然实验的证据评论人:薛飞,西北大学报告人:薛飞,西北大学题目:高铁推动区域经济和就业效应的测度评论人:张梦婷,上海大学报告人:张梦婷,上海大学题目:高铁网络、市场准入与企业生产率评论人:朱州,西南财经大学报告人:朱州,西南财经大学题目:唐山地震及其灾后重建对当地经济增长的长期影响评论人:杨梦,厦门大学报告人:杨梦,厦门大学题目:限购政策对房地产价格影响异质性分析与因地制宜的调控机制设计评论人:黄永强,中南财经政法大学时间日程安排地点14:00-15:40 论文分组报告 IV-2 经济楼N401主持人:尚玉皇,西南财经大学报告人:邓平军,华中科技大学题目:The Effect of Market Quality on the Causality between Returns and Volatilities: Evidences from CSI 300 Index Futures评论人:胡春龙,上海财经大学报告人:胡春龙,上海财经大学题目:随机波动性提高了中国经济增长预测的表现吗?——基于BVAR-SV 和BVAR 模型的比较研究评论人:吕政,贵州财经大学报告人:吕政,贵州财经大学题目:结构突变下国际油价波动对我国石油行业的影响研究评论人:尚玉皇,西南财经大学报告人:尚玉皇,西南财经大学题目:Forecasting Yield Curves with a Mixed-Frequency Affine Model评论人:邓平军,华中科技大学14:00-15:40 论文分组报告 IV-3 经济楼N406主持人:朱蔚宣,厦门大学报告人:方师乐,浙江大学题目:非农就业、农机投资和农机服务利用评论人:林文炼,中山大学报告人:林文炼,中山大学题目:“入学年龄规定”会产生教育不平等吗?——来自1986年《义务教育法》颁布的证据评论人:王剑程,西南财经大学报告人:王剑程,西南财经大学题目:宽带建设、信息关注与农村居民创业评论人:吴艳洁,山西财经大学报告人:吴艳洁,山西财经大学题目:基于金融发展与城市化建设协调配合的城乡居民收入差距研究评论人:方师乐,浙江大学时间日程安排地点14:00-15:40 论文分组报告 IV-4 经济楼N301主持人:倪骁然,厦门大学报告人:蔡俊宇,西安交通大学题目:Structural Break, Stock Prices of Clean Energy Firms and Carbon Market评论人:刘毅男,吉林大学报告人:刘毅男,吉林大学题目:股指期货对冲指数基金的最优套期保值比率研究评论人:孙彦林,吉林大学报告人:孙彦林,吉林大学题目:基于股市和汇市成交量信息视角的股价波动模型研究评论人:田露,厦门大学报告人:田露,厦门大学题目:中国股市和债市“跷跷板”效应是否消失了?--基于金融周期的分析视角评论人:蔡俊宇,西安交通大学14:00-15:40 论文分组报告 IV-5 经济楼N303 主持人:蔡伟贤,厦门大学报告人:韩飞,中央财经大学题目:我国企业社保负担研究:从征缴体制看劳动力需求评论人:胡文骏,厦门大学报告人:胡文骏,厦门大学题目:经济开放的税收效应研究评论人:李丹,浙江大学报告人:李丹,浙江大学题目:中国居民储蓄与政府债务可持续性评论人:刘恩猛,中国计量大学报告人:刘恩猛,中国计量大学题目:市场认可债项信用评级吗?——以中期票据发行为例评论人:王睿霆,日本北海道大学报告人:王睿霆,日本北海道大学题目:中国纳税人的聚束于折点评论人:韩飞,中央财经大学时间日程安排地点14:00-15:40 论文分组报告 IV-6 经济楼N308主持人:肖强,兰州财经大学报告人:刘萍,中山大学题目:The Real Consequences of Financial Stress:Evidence from China评论人:刘晓洁,中南财经政法大学报告人:刘晓洁,中南财经政法大学题目:我国房地产业与经济发展互动关系研究评论人:许可,北京大学报告人:许可,北京大学题目:中国经济潜在增长速度的预测:基于贝叶斯层次模型的实证分析评论人:肖强,兰州财经大学报告人:肖强,兰州财经大学题目:中国金融市场对宏观经济的非对称性影响分析——基于经济政策稳定性视角评论人:刘萍,中山大学14:00-15:40 论文分组报告 IV-7 经济楼N118主持人:柳冠男,厦门大学报告人:李青召,吉林大学题目:基于马尔可夫机制转换动态因子模型对我国经济周期拐点的识别评论人:刘辉,湖南师范大学报告人:刘辉,湖南师范大学题目:中国家庭债务与财政支出效应——基于异质性家庭的DSGE 模型分析评论人:王洪时,吉林大学报告人:王洪时,吉林大学题目:DSGE模型的广义随机模拟方法----改进及应用评论人:向镜洁,华中科技大学报告人:向镜洁,华中科技大学题目:货币传导异质性与实体经济流动性配置的“马太效应”评论人:杨翱,厦门大学报告人:杨翱,厦门大学题目:灾难冲击、技术吸收与中国宏观经济波动评论人:李青召,吉林大学。
非线性最小二乘法Levenberg-Marquardt-method
Levenberg-Marquardt Method(麦夸尔特法)Levenberg-Marquardt is a popular alternative to the Gauss-Newton method of finding the minimum of afunction that is a sum of squares of nonlinear functions,Let the Jacobian of be denoted , then the Levenberg-Marquardt method searches in thedirection given by the solution to the equationswhere are nonnegative scalars and is the identity matrix. The method has the nice property that, forsome scalar related to , the vector is the solution of the constrained subproblem of minimizingsubject to (Gill et al. 1981, p. 136).The method is used by the command FindMinimum[f, x, x0] when given the Method -> Levenberg Marquardt option.SEE A LSO:Minimum, OptimizationREFERENCES:Bates, D. M. and Watts, D. G. N onlinear Regr ession and Its Applications. New York: Wiley, 1988.Gill, P. R.; Murray, W.; and Wright, M. H. "The Levenberg-Marquardt Method." §4.7.3 in Practical Optim ization. London: Academic Press, pp. 136-137, 1981.Levenberg, K. "A Method for the Solution of Certain Problems in Least Squares." Quart. Appl. Math.2, 164-168, 1944. Marquardt, D. "An Algor ithm for Least-Squares Estimation of Nonlinear Parameters." SIAM J. Appl. Math.11, 431-441, 1963.Levenberg–Marquardt algorithmFrom Wikipedia, the free encyclopediaJump to: navigation, searchIn mathematics and computing, the Levenberg–Marquardt algorithm (LMA)[1] provides a numerical solution to the problem of minimizing a function, generally nonlinear, over a space of parameters of the function. These minimization problems arise especially in least squares curve fitting and nonlinear programming.The LMA interpolates between the Gauss–Newton algorithm (GNA) and the method of gradient descent. The LMA is more robust than the GNA, which means that in many cases it finds a solution even if it starts very far off the final minimum. For well-behaved functions and reasonable starting parameters, the LMA tends to be a bit slower than the GNA. LMA can also be viewed as Gauss–Newton using a trust region approach.The LMA is a very popular curve-fitting algorithm used in many software applications for solving generic curve-fitting problems. However, the LMA finds only a local minimum, not a global minimum.Contents[hide]∙ 1 Caveat Emptor∙ 2 The problem∙ 3 The solutiono 3.1 Choice of damping parameter∙ 4 Example∙ 5 Notes∙ 6 See also∙7 References∙8 External linkso8.1 Descriptionso8.2 Implementations[edit] Caveat EmptorOne important limitation that is very often over-looked is that it only optimises for residual errors in the dependant variable (y). It thereby implicitly assumes that any errors in the independent variable are zero or at least ratio of the two is so small as to be negligible. This is not a defect, it is intentional, but it must be taken into account when deciding whether to use this technique to do a fit. While this may be suitable in context of a controlled experiment there are many situations where this assumption cannot be made. In such situations either non-least squares methods should be used or the least-squares fit should be done in proportion to the relative errors in the two variables, not simply the vertical "y" error. Failing to recognise this can lead to a fit which is significantly incorrect and fundamentally wrong. It will usually underestimate the slope. This may or may not be obvious to the eye.MicroSoft Excel's chart offers a trend fit that has this limitation that is undocumented. Users often fall into this trap assuming the fit is correctly calculated for all situations. OpenOffice spreadsheet copied this feature and presents the same problem.[edit] The problemThe primary application of the Levenberg–Marquardt algorithm is in the least squares curve fitting problem: given a set of m empirical datum pairs of independent and dependent variables, (x i, y i), optimize the parameters β of the model curve f(x,β) so that the sum of the squares of the deviationsbecomes minimal.[edit] The solutionLike other numeric minimization algorithms, the Levenberg–Marquardt algorithm is an iterative procedure. To start a minimization, the user has to provide an initial guess for the parameter vector, β. In many cases, an uninformed standard guess like βT=(1,1,...,1) will work fine;in other cases, the algorithm converges only if the initial guess is already somewhat close to the final solution.In each iteration step, the parameter vector, β, is replaced by a new estimate, β + δ. To determine δ, the functions are approximated by their linearizationswhereis the gradient(row-vector in this case) of f with respect to β.At its minimum, the sum of squares, S(β), the gradient of S with respect to δwill be zero. The above first-order approximation of gives.Or in vector notation,.Taking the derivative with respect to δand setting theresult to zero gives:where is the Jacobian matrix whose i th row equals J i,and where and are vectors with i th componentand y i, respectively. This is a set of linear equations which can be solved for δ.Levenberg's contribution is to replace this equation by a "damped version",where I is the identity matrix, giving as the increment, δ, to the estimated parameter vector, β.The (non-negative) damping factor, λ, isadjusted at each iteration. If reduction of S is rapid, a smaller value can be used, bringing the algorithm closer to the Gauss–Newton algorithm, whereas if an iteration gives insufficientreduction in the residual, λ can be increased, giving a step closer to the gradient descentdirection. Note that the gradient of S withrespect to β equals .Therefore, for large values of λ, the step will be taken approximately in the direction of the gradient. If either the length of the calculated step, δ, or the reduction of sum of squares from the latest parameter vector, β + δ, fall below predefined limits, iteration stops and the last parameter vector, β, is considered to be the solution.Levenberg's algorithm has the disadvantage that if the value of damping factor, λ, is large, inverting J T J + λI is not used at all. Marquardt provided the insight that we can scale eachcomponent of the gradient according to thecurvature so that there is larger movement along the directions where the gradient is smaller. This avoids slow convergence in the direction of small gradient. Therefore, Marquardt replaced theidentity matrix, I, with the diagonal matrixconsisting of the diagonal elements of J T J,resulting in the Levenberg–Marquardt algorithm:.A similar damping factor appears in Tikhonov regularization, which is used to solve linear ill-posed problems, as well as in ridge regression, an estimation technique in statistics.[edit] Choice of damping parameterVarious more-or-less heuristic arguments have been put forward for the best choice for the damping parameter λ. Theoretical arguments exist showing why some of these choices guaranteed local convergence of the algorithm; however these choices can make the global convergence of the algorithm suffer from the undesirable properties of steepest-descent, in particular very slow convergence close to the optimum.The absolute values of any choice depends on how well-scaled the initial problem is. Marquardt recommended starting with a value λ0 and a factor ν>1. Initially setting λ=λ0and computing the residual sum of squares S(β) after one step from the starting point with the damping factor of λ=λ0 and secondly withλ0/ν. If both of these are worse than the initial point then the damping is increased by successive multiplication by νuntil a better point is found with a new damping factor of λ0νk for some k.If use of the damping factor λ/ν results in a reduction in squared residual then this is taken as the new value of λ (and the new optimum location is taken as that obtained with this damping factor) and the process continues; if using λ/ν resulted in a worse residual, but using λresulted in a better residual then λ is left unchanged and the new optimum is taken as the value obtained with λas damping factor.[edit] ExamplePoor FitBetter FitBest FitIn this example we try to fit the function y = a cos(bX) + b sin(aX) using theLevenberg–Marquardt algorithm implemented in GNU Octave as the leasqr function. The 3 graphs Fig 1,2,3 show progressively better fitting for the parameters a=100, b=102 used in the initial curve. Only when the parameters in Fig 3 are chosen closest to the original, are thecurves fitting exactly. This equation is an example of very sensitive initial conditions for the Levenberg–Marquardt algorithm. One reason for this sensitivity is the existenceof multiple minima —the function cos(βx)has minima at parameter value and[edit] Notes1.^ The algorithm was first published byKenneth Levenberg, while working at theFrankford Army Arsenal. It was rediscoveredby Donald Marquardt who worked as astatistician at DuPont and independently byGirard, Wynn and Morrison.[edit] See also∙Trust region[edit] References∙Kenneth Levenberg(1944). "A Method for the Solution of Certain Non-Linear Problems in Least Squares". The Quarterly of Applied Mathematics2: 164–168.∙ A. Girard (1958). Rev. Opt37: 225, 397. ∙ C.G. Wynne (1959). "Lens Designing by Electronic Digital Computer: I". Proc.Phys. Soc. London73 (5): 777.doi:10.1088/0370-1328/73/5/310.∙Jorje J. Moré and Daniel C. Sorensen (1983)."Computing a Trust-Region Step". SIAM J.Sci. Stat. Comput. (4): 553–572.∙ D.D. Morrison (1960). Jet Propulsion Laboratory Seminar proceedings.∙Donald Marquardt (1963). "An Algorithm for Least-Squares Estimation of NonlinearParameters". SIAM Journal on AppliedMathematics11 (2): 431–441.doi:10.1137/0111030.∙Philip E. Gill and Walter Murray (1978)."Algorithms for the solution of thenonlinear least-squares problem". SIAMJournal on Numerical Analysis15 (5):977–992. doi:10.1137/0715063.∙Nocedal, Jorge; Wright, Stephen J. (2006).Numerical Optimization, 2nd Edition.Springer. ISBN0-387-30303-0.[edit] External links[edit] Descriptions∙Detailed description of the algorithm can be found in Numerical Recipes in C, Chapter15.5: Nonlinear models∙ C. T. Kelley, Iterative Methods for Optimization, SIAM Frontiers in AppliedMathematics, no 18, 1999, ISBN0-89871-433-8. Online copy∙History of the algorithm in SIAM news∙ A tutorial by Ananth Ranganathan∙Methods for Non-Linear Least Squares Problems by K. Madsen, H.B. Nielsen, O.Tingleff is a tutorial discussingnon-linear least-squares in general andthe Levenberg-Marquardt method inparticular∙T. Strutz: Data Fitting and Uncertainty (A practical introduction to weighted least squares and beyond). Vieweg+Teubner, ISBN 978-3-8348-1022-9.[edit] Implementations∙Levenberg-Marquardt is a built-in algorithm with Mathematica∙Levenberg-Marquardt is a built-in algorithm with Matlab∙The oldest implementation still in use is lmdif, from MINPACK, in Fortran, in thepublic domain. See also:o lmfit, a translation of lmdif into C/C++ with an easy-to-use wrapper for curvefitting, public domain.o The GNU Scientific Library library hasa C interface to MINPACK.o C/C++ Minpack includes theLevenberg–Marquardt algorithm.o Several high-level languages andmathematical packages have wrappers forthe MINPACK routines, among them:▪Python library scipy, modulescipy.optimize.leastsq,▪IDL, add-on MPFIT.▪R (programming language) has theminpack.lm package.∙levmar is an implementation in C/C++ with support for constraints, distributed under the GNU General Public License.o levmar includes a MEX file interface for MATLABo Perl (PDL), python and Haskellinterfaces to levmar are available: seePDL::Fit::Levmar, PyLevmar andHackageDB levmar.∙sparseLM is a C implementation aimed at minimizing functions with large,arbitrarily sparse Jacobians. Includes a MATLAB MEX interface.∙ALGLIB has implementations of improved LMA in C# / C++ / Delphi / Visual Basic.Improved algorithm takes less time toconverge and can use either Jacobian orexact Hessian.∙NMath has an implementation for the .NET Framework.∙gnuplot uses its own implementation .∙Java programming language implementations:1) Javanumerics, 2) LMA-package (a small,user friendly and well documentedimplementation with examples and support),3) Apache Commons Math∙OOoConv implements the L-M algorithm as an Calc spreadsheet.∙SAS, there are multiple ways to access SAS's implementation of the Levenberg-Marquardt algorithm: it can be accessed via NLPLMCall in PROC IML and it can also be accessed through the LSQ statement in PROC NLP, and the METHOD=MARQUARDT option in PROC NLIN.。
HEC-HMS
Converts HEC-1 files into HMS files
HEC-HMS Availability
Available Through HEC Vendors Available at HEC Web Site: “Public Domain” Program
hec-hms hydrologicengineering center's hydrologic modeling system (hms) summary premierhydrologic model today (hec) performsrf-ro calculations basicinput outputoptions precipitationoptions unithydrograph options floodrouting option viewingresults graphsexecution runningactual projects gagedata castrovalley case study keegansexample gis/nexraddata (hec geo-hms) hydrologiccycle legroundwater flow groundwaterdischarge 38 surface discharge 61 evaporation from land 39 moisture over land 385 pre cipitation ocean424 evaporation from ocean surface runoff impervious strata groundwater recharge pre cipitation snow melt uses hecprogram models rainfall-runoffprocess watershedbased watershedphysiographic data modelingoptions computeuh basinareas. floodrouting along streams. estimatingparameters eachbasin based computeddata observeddata hec-1 program history hec-1 modeldevelopment separateprograms: 1967 majorrevision unification:1973 secondmajor revision: 1981 (dam breach, kinematic wave) pcversions: 1984 (partial), 1988 (
estim 词根 -回复
estim 词根-回复"Estim" is a Latin root that means "to estimate" or "to value." This root is used in many words related to estimation, such as "estimate," "estimator," and "estimatee." In this article, we will explore the concept of estimation and its significance in various fields, including mathematics, statistics, and economics. We will also discuss the process of making accurate estimates and the importance of estimations in decision-making.The ability to estimate is a fundamental skill that humans possess. From estimating the time it takes to complete a task to predicting the outcome of an event, estimations play a crucial role in our daily lives. However, in more formal contexts, such as scientific research or business planning, accurate estimations are vital for making informed decisions and setting realistic goals.In the field of mathematics, estimation is often used when the exact value of a quantity is unknown or difficult to determine. For example, in arithmetic, one might estimate the sum of two numbers by rounding them to the nearest ten or hundred. Estimations are also used in geometry to approximate the measurement of angles or lengths of sides in irregular shapes.These estimations provide a close approximation of the actual values and allow mathematicians to work with more manageable numbers.In statistics, estimation plays a crucial role in data analysis and hypothesis testing. The process of estimation involves using sample data to make inferences about a population. For instance, if a researcher wants to estimate the average height of students in a school, they can measure the heights of a sample of students and use this information to estimate the average height of the entire student population. Statistical estimations provide valuable insights into the characteristics of a population and help researchers make generalizations based on limited data.In economics, estimation is vital for various purposes, such as predicting market trends, forecasting future demand, and evaluating investment opportunities. Economists often use historical data and statistical models to estimate future economic indicators, such as GDP growth, inflation rates, or unemployment rates. These estimations are used by policymakers, businesses, and investors to make informed decisions and develop strategies that align with projected economic conditions.The process of making accurate estimates involves several steps. Firstly, it is essential to define the objective or the quantity to be estimated. This requires a clear understanding of the problem and the information available. Secondly, it is crucial to determine the relevant data sources and collect the necessary information. The data should be representative of the population or phenomenon being estimated. Thirdly, one must select an appropriate estimation technique, considering the characteristics of the data and the desired level of accuracy. This may involve using mathematical formulas, statistical models, or expert judgment. Fourthly, the estimation should be carried out using the chosen technique, adjusting for any biases or errors. Finally, the estimated result should be validated and evaluated for its reliability and validity. This may involve comparing it with other estimates, conducting sensitivity analysis, or assessing the confidence interval.In conclusion, estimation is a significant process in various fields, including mathematics, statistics, and economics. It allows us to make informed decisions, predict future outcomes, and understand the characteristics of populations or phenomena. Accurateestimations require clear objectives, relevant data, appropriate estimation techniques, and careful validation. By mastering the skill of estimation, we can enhance our problem-solving abilities and navigate complex situations with more confidence.。
- 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)。
1110IEICE MUN.,VOL.E89–B,NO.4APRIL2006P APER Special Section on Internet Technology VIEstimating Method of Short-Interval-Traffic Distribution Considering Long-Term-Traffic Dynamics for MultimediaQoS Management∗Tadayoshi FUKAMI†,Nonmember,Hiroki NISHIKA W A††,Takuya ASAKA†††a),Members,Tatsuro TAKAHASHI†††,Fellow,and Noriteru SHINAGA W A††,MemberSUMMARY Analyzing short-interval-traffic behaviors is important for network performance management to realize high quality multimedia ap-plications.However,it is difficult to measure short-interval-traffic vol-umes because there are complications in collecting short-interval-traffic data from routers.An example is a heavy load on routers or inaccurate measurement by the short-polling interval;it even demands expensive mea-surement tools.To resolve these disadvantages,an estimating method of short-interval-traffic distribution(EMSIT)has been proposed.This method estimates short-interval-traffic distributions using MIB(Management In-formation Base)data,which collects traffic volumes in cycles of several minutes.In this paper,we propose a new estimation method(EMSIT-LD) based on EMSIT,which applies to the case of long-term-traffic dynamics. We evaluate it using computer simulations and actual traffic data.key words:network traffic,passive measurement,traffic dynamics1.IntroductionThe rapid spread of the Internet has increased the demand for effective network management systems and for provid-ing high-quality network services for its users.However,the current network management systems for the Internet face many difficulties,and we have no clear grasp of the network traffic behaviors in spite of our efforts.It is impossible to understand systematically the network traffic on the Internet because the Internet is an aggregation of many autonomous network systems,and there is an enormous variety of dif-ferent network traffic on it.Therefore,network traffic mea-surement is inseparable from network management;in other words,network traffic measurement is the most important technology for use with network management,and thefirst step in implementing it.Especially,a traffic measurement technique with a large volume of information will become increasingly significant for QoS management of multimedia applications such as live-streaming or V oIP in the future.Today,many traffic measurement tools have been im-Manuscript received July1,2005.Manuscript revised September29,2005.†The author is with NTT DoCoMo Kansai Inc.,Osaka-shi, 530-0001Japan.††The authors are with NTT DoCoMo Inc.,Tokyo,100-6150 Japan.†††The authors are with the Graduates School of Informatics, Kyoto University,Kyoto-shi,606-8501Japan.∗Part of this paper was presented at IEEE Globecom2004,in November-December2004.a)E-mail:asaka@i.kyoto-u.ac.jpDOI:10.1093/ietcom/e89–b.4.1110plemented and a large number of network quality measuringprojects have been launched[1]–[4].Network administra-tors choose them according to the network traffic to be mea-sured.For example,we choose passive measurement toolswhen measuring long-term-traffic dynamics or their corre-lations at a local point on networks.On the other hand,wechoose active measurement tools when measuring an avail-able bandwidth or RTT on an End-to-End path.In case weobserve network traffic continuously,we should use trafficmeasurement tools that are easier and less expensive oper-ate.A network traffic measurement method with MIB(Man-agement Information Base)on a router satisfies these re-quirements.It outputs the throughputs every unit-time witha SNMP(Simple Network Management Protocol)[5].The World Wide Web(WWW)generates an ever-growing proportion of traffic on the Internet,and networkquality intervals of a few seconds,such as8s,is importantfor the Web,because such a short interval corresponds to thetime taken to display an entire Web page on a screen.Thus,measuring traffic dynamics periods of a few seconds is animportant aspect of network performance management.Ifthe traffic distribution in a few seconds were measured,net-work administrators could manage network qualities at the α%point of its distribution(such as the80%point),where we define traffic distribution in a few seconds as traffic distri-bution which is calculated by traffic data measured in cyclesof a few seconds.In other words,the administrators coulddetermine“the worst-case quality”represented by the tail ofthe traffic distribution.Thus,in this paper,we treat an orderof a second as a“Short interval.”The target for network management in this paper is toobserve short-interval traffic distribution of the minute-levelorder interval in the steady state,such as20min,and to op-erate network resources to allocate appropriately in an orderof more than just minutes.Examples of these operations areshown below.(1)Link capacity planning(2)Accommodation planning of access links to back-bone links(3)Planning of L2path routing(4)Network information disclosure to usersMost conventional methods and tools have shortcom-ings in measuring short-interval traffic volumes[9].For ex-ample,using a MIB(Management Information Base)andCopyright c 2006The Institute of Electronics,Information and Communication EngineersFUKAMI et al.:ESTIMATING METHOD OF SHORT-INTERV AL-TRAFFIC DISTRIBUTION1111Fig.1Tra ffic volume to be measured.SNMP (Simple Network Management Protocol)[5],[7],[8]with a short-polling-interval imposes a heavy load on a router,and these measurements may be inaccurate.Fur-thermore,probe hardware devices [6],protocol analyzers,and PCs can also collect many types of tra ffic information,but they are expensive,especially for high-speed links.To make matters worse,simultaneous measurements of many links require many such expensive devices.To resolve these di fficulties,an estimating method of a short-interval-tra ffic distribution (such as tra ffic distribution during a few seconds)has been proposed which we hereafter call EMSIT(an Estimating Method of Short-Interval-Tra ffic distribution)[9].It imposes a lighter load on a router than actual short-interval measurement since EMSIT estimates a short-interval tra ffic distribution only with MIB data,which are measured in cycles of a few minutes (Fig.1)[11].It is easy to operate EMSIT because it requires only MIB and SNMP,which are already implemented on routers,and does not require any particularly expensive devices.Unfortunately,however,EMSIT performance deterio-rates under conditions where the network tra ffic has a long-term-tra ffic dynamics [12].We define long-term-tra ffic dy-namics (such as the tra ffic dynamics during a few minutes)as fluctuations of average tra ffic volumes over a period of a few hours.The change of tra ffic at daytime and night is the one example.Moreover,in the long-term-tra ffic dynamics,there exists average tra ffic volume between “an estimation section †”(term to be estimated such as 20min)and “a mea-surement section”(whole measured term)without the esti-mation section.Moreover,when long-term-tra ffic dynamic exists,the di fference of the average tra ffic volume in the es-timation section to that of the measurement section without the estimation section is greater than the threshold β.To overcome the disadvantage of EMSIT,we propose a suitable tra ffic estimation method,called EMSIT-LD (an Es-timating Method of Short-Interval-Tra ffic distribution con-sidering Long-term-tra ffic Dynamics),for long-term-tra ffic dynamics.EMSIT-LD estimates current short-interval-tra ffic distribution using tra ffic trends measured over the pre-vious couple of hours.We think that the required accuracy depends on both the content of these operations and man-agement policy of operators.Thus,in general,it is di fficult to judge whether it is possible to allow the level of error given by EMSIT or EMSIT-LD.In many cases,we believe that EMSIT-LD is better than EMSIT because the amount of required measurement data and the complexity of theirprocedures are the same.2.EMSIT—Basic MethodIn general,there exist parametric and non-parametric ap-proaches to estimate the probability distribution function [13].The parametric approach can easily be applied to es-timate a distribution,though it requires a priori information of the specification of the distribution.The non-parametric approach can estimate without a priori information,though it may require many parameters and it may become di fficult to handle the model.EMSIT (and also EMSIT-LD proposed in this paper)is a parametric approach and the distribution to be estimated is assumed a normal distribution.Next,we review the basic method of EMSIT.The goal is to estimate the distribution of the average tra ffic volume X (1)(n )(n =1,2,···,m 1m 2k )for short interval T (e.g.8sec.).Using SNMP,we obtain historical data on average tra ffic volumes X (m 1)(n )(n =1,2,···,m 2k )during interval m 1T (m 1>0)from MIB on a router.This estimation method,EMSIT,is applied after the server has gathered the tra ffic information from MIB using SNMP.EMSIT procedureStep 1:Calculate sample average ¯x and sample varianceσ2(m 1)of X (m 1)(n ).Step 2:Calculate variance σ2(m 1m 2)of X(m 1m 2)(n ),which is obtained by averaging the original sample X (m 1)(n )in non-overlapping sub-blocks of size m 2.Step 3:Calculate the variance σ2(1)of X (1)(n )usinglog 10σ2(1)=log 10σ2(m 1)·log 10m 1m 2−log 10σ2(m 1m 2)·log 10m 1log 10m 2.(1)Step 4:Set the distribution of X (1)(n )as the normal distri-bution with average ¯x and variance σ2(1).In Step 3,assuming (2)to be available,(1)can be easily derived from (2).σ2(M )=σ2(1)·M−θ(2)where θis a constant and 0<θ≤1.Equation (2)represents the relationship among di fferent tra ffic statistics X (M )(M =1,2,...),and we assume X (M )(n )to be a stationary random process [15],[16].For intervals shorter than RTT,(2)has not been ob-served [18].When Eq.(2)exists and 0<θ<1,process X (1)(n )is called exactly second-order self-similar with Hurst parame-ter H (the value of H gives the degree of self-similarity and is expressed as H =1−θ/2)[14].The most striking fea-ture of self-similarity is that the correlation structures of the process do not degenerate as M →∞,unlike those of the†“Estimation section”and “measurement section”are de-scribed in detail in Sect.3.1112IEICE MUN.,VOL.E89–B,NO.4APRIL2006traditional Markovian models(θ=1),which all do.Self-similarity has been observed in traffic measurement in ac-tual networks(e.g.,[15]).Those conventional works mainlyfocused on the existence of self-similarity in traffic or on amethod of calculating H.They showed that Internet traffichas self-similarity and that Eq.(2)empirically exists.In Step4,the traffic distribution of X(1)(n)is assumed tobe a normal distribution because the central limited theoremapplies to the case in which manyflows are multiplexed andindependent of each other.This assumption is particularlyreasonable in backbone links[19].3.EMSIT-LD—Our ProposalThe conventional method,EMSIT,has been used only onsteady traffic conditions.If traffic has long-term dynam-ics,we have no idea what the estimation result with EMSITwill be.This is because EMSIT does not always satisfy(2)when estimating traffic distribution in the long-term-trafficdynamics zone.We demonstrate the unavailability of EM-SIT under the condition of long-term-traffic dynamics in thenext section.In this section,we propose the new estimationmethod EMSIT-LD that takes into account long-term-trafficdynamics.In Fig.2,this method estimates the distribution duringtime interval of the“estimation section”(e.g.20min)usingthe“measurement section”(e.g.2hrs).The time intervalof the estimation section should be determined according tonetwork management policy(e.g.30min),and it is deter-mined by positive integer parameter p.The estimation sec-tion shifts every m1T(from j-th estimation section to j+1-thestimation section).We explain the details of the estimationprocedure as follows.EMSIT-LD procedureStep1:Calculate the varianceσ2(m1)andσ2(m1m2)of the traf-fic X(m1)(n)and X(m1m2)(n)inside the measurement sec-tion[0,m1m2kT](Fig.2).Then,calculate the slope s of the regression line l on the Variance Time Plots(Fig.3).s=log10σ2(m1m2)−log10σ2(m1)log10m2.(3)Step2:Divide the measurement section into p sub-sections,and calculate the averageµi(i=1,2,···,p)of the traffic X(m1)(n)in each sub-section.Next,calcu-late the difference between the maximumµmax and theminimumµmin ofµi.If the difference is smaller thanthe thresholdβ(µmax−µmin≤β),substitute the slope s to u;otherwise,do not,where u is a real number.Step3:Calculate the averageˆµand the varianceˆσ2(m1)of the traffic X(m1)(n)inside the estimation section[(1−1 q )m1m2kT,m1m2kT](Fig.2),where the positive inte-ger q is the divisor of ing the variancelog10ˆσ2(m1)and the slope u,describe the regression lineˆl.The y-intercept ofˆl is our estimation variancelog10ˆσ2(1).Fig.2Measurement and estimation sections in EMSIT-LD,and its repetitive calculationroutines.Fig.3Image of EMSIT-LD estimation process.Step4:Estimate the T-interval traffic distribution as the normal distribution of the averageˆµand the variance ˆσ2(1).Step5:Shift the measurement section to the next position, and repeat the same process[Step1–Step4](Fig.2).Step1calculates slope s of the regression line l for the measurement section just like original the EMSIT.In Step 2,EMSIT-LD divides the measurement section into several sections(such as12sections)using parameter p,and com-pares the traffic averages in each section.The parameter p determines the interval time to calculate the traffic average, and p and q can be determined independently.If the dif-ference between the maximum and minimum average traffic is smaller than the thresholdβ,EMSIT-LD judges that the measurement section is in the steady traffic state and uses u calculated in the current routine.On the other hand,if the difference is larger than the thresholdβ,EMSIT-LD judges that the measurement section is in the dynamic traffic state and uses u in its memory that was calculated in the last pe-riod of steady traffic.Step3estimates the variance of the estimated distribution using slope u,which was calculated in Steps1or2,and Step4specifies the traffic distribution during the estimation section.Step5shifts the estimation section.The measurement section requires some time(such as a couple of hours)because EMSIT(-LD)secures enough sam-ples to draw the regression line l on the Variance Time Plots. In the case where the measurement and estimation sections are2hours and20minutes,respectively,and m1m2T=600 (that is,the10-min variance calculation cycle),the measure-FUKAMI et al.:ESTIMATING METHOD OF SHORT-INTERV AL-TRAFFIC DISTRIBUTION1113 ment section provides12samples,whereas the estimationsection provides only two samples to calculate the varianceσ2(m1m2)of the traffic X(m1m2)(n).To overcome this problem,in Step3,assuming that the regression line l in the mea-surement section and the regression lineˆl in the estimation section have the same trend,we parallel-shift the regression line l in order to draw the regression lineˆl.EMSIT cannot estimate correct traffic distribution when estimating during the long-term-traffic dynamics zone because it does not always satisfy Eq.(2)and cannot provide the correct regression line l.On the other hand,our pro-posed method EMSIT-LD can estimate correct traffic dis-tribution regardless of long-term-traffic dynamics because EMSIT-LD estimates traffic distribution using the past re-gression line l,which has been obtained in the past steady traffic states.The disadvantage of EMSIT and the advantage of EMSIT-LD are confirmed numerically in the next section.4.Simulations4.1Simulation ModelFigure4outlines our simulation model,on which network conditions are implemented.Simulation models used in our paper are based on models in conventional studies[20],[21]. There are150clients at the ends of the network,and each client has its own access link.We use ns-2as the network simulator,and evaluate EMSIT and EMSIT-LD in the cases below.Simulation1(S1):Each client requests100-kbytefiles on average,which are represented by an exponential dis-tribution.Traffic volume increases after150min have passed.Figure5illustrates traffic volume per1-minFig.4Simulationmodel.Fig.5Traffic volume(S1).cycle.Simulation2(S2):Each client requests100-kbytefiles on average,which are represented by a Pareto distribu-tion(shape parameter is set to1.5)[22].Traffic vol-ume increases after180min have passed.Figure8il-lustrates traffic volume per1-min cycle.Simulation3(S3):Each client requests100-kbytefiles on average,which are represented by an exponential dis-tribution.Traffic volume increases for a brief period between160min and165min.Figure11illustrates traffic volume per1-min cycle.Simulation4(S4):Each client requests100-kbytefiles on average,which are represented by an exponential dis-tribution.The RTT of half the clients(75clients)is 200ms and the others(75clients)is20ms.Traffic vol-ume increases after160minutes has passed.Figure14 illustrates traffic volume per1-min cycle.Our four models are characterized by distribution of transferredfile size,RTT setting,and tendency for traffic to increase.The two former are used in conventional studies as well,while the latter one is originally introduced in this paper.For the two former ones,these models are set to eval-uate diverse traffic conditions.Thus,individual models have no special meaning for EMSIT and EMSIT-LD.The latter one,however,the tendency for traffic to increase,simulates actual traffic dynamics.In(S1),(S2),(S3),and(S4),tendencies for traffic to in-crease were changed by a change in the average inter-arrival time of the TCP connection for every client.Tendency for traffic to increase in(S1),(S2),and(S4)simulated actual traffic dynamics in the time zone from night to day.The average traffic volume increased every5min,and the aver-age amount of trafficfinally increased by up to six times. Tendency for traffic to increase in(S3)simulated temporal traffic concentration such as a DDoS attack.The average traffic was set to ten times only for a specific5-min interval.The parameters of EMSIT(-LD)are set to T=10,k= 12,m1=6,m2=10,p=12,q=6.The parameterβis set to0.25Mbps(the value was determined in trial experi-ment).These parameters indicate that our estimating target is traffic behavior during a10sec period(T=10),the mea-surement section is2hrs(7,200sec:m1m2kT=7,200),and the estimation section is20min(q=6).The value of short interval T value was determined because second-order qual-ity is significant for many applications.Parameter m1was set based on a general MIB polling interval,and Parameter q was set as a least number to calculate variance for traffic data obtained from MIB during an estimation section.For parameters m2and k,we set these values based on results of the trial experiments,and we confirmed that their values did not strongly influence the performance of EMSIT-LD.Fur-thermore,EMSIT(-LD)obtains the MIB data every1min (m1T=60)and calculates the variance during1-min and 10-min intervals(m1T=60,m1m2T=600).The differ-ence of the average traffic volumes in the12sub-sections (p=12)in the measurement section informs us whether or1114IEICE MUN.,VOL.E89–B,NO.4APRIL2006Fig.6Variance dynamics(S1).Fig.7Comparison of each distribution (S1at 185min).Fig.8Tra ffic volume (S2).not long-term tra ffic dynamics would occur.4.2Simulation ResultWe compare the performances of EMSIT and EMSIT-LD in the (S1)–(S4)situations we mentioned above.Figures 6,9,12and 15illustrate the actual (loga-rithm)variance dynamics calculated by measured data and the those estimated by each estimating method,around the long-term-tra ffic dynamics zone in each simulation.In the conventional method EMSIT,we regard the estimated vari-ances inside the whole measurement section as the one in the estimation section in Fig.2,because EMSIT doesnot Fig.9Variance dynamics(S2).Fig.10Comparison of each distribution (S2at 195min).Fig.11Tra ffic volume (S3).define its estimation section.In those figures,estimated variances by EMSIT-LD fol-low actual variances,though estimated ones by EMSIT do not do so in the long-term-tra ffic dynamics zone.Particu-larly,in Fig.12,estimated variances by EMSIT-LD also fol-low actual variances;however,EMSIT continuously outputs irrelevant variances as long as several-hundred minutes.Figures 6,9,and 15show that accuracies of the esti-mated variances by EMSIT-LD are essentially similar.That is,distribution types of file size and the di fference of RTT do not a ffect the performance of EMSIT(-LD).Moreover,these figures show that estimated variances by EMSIT are smaller than those by EMSIT-LD during period when tra ffic trendsFUKAMI et al.:ESTIMATING METHOD OF SHORT-INTERV AL-TRAFFIC DISTRIBUTION1115Fig.12Variance dynamics(S3).Fig.13Comparison of each distribution (S3at 200min).to increase.This is because the long-term dynamics is rel-atively larger than short-term dynamics and that the slope u of the regression line in EMSIT becomes small.From the viewpoint of network performance management,high traf-fic probability may be underestimated in period when tra ffic trends to increase,thus EMSIT-LD is superior to EMSIT.Figures 7,10,13and 16illustrate the actual tra ffic dis-tribution and distributions calculated by each estimated vari-ance and measured average,where they are calculated at the 185-min (S1),195-min (S2),200-min (S3)and 180-min (S4)marks.In those figures,the estimated tra ffic distribu-tion by EMSIT-LD closely estimates the actual one,particu-larly at higher throughputs.On the other hand,the estimated tra ffic distribution by EMSIT is widely separated from the actual one,particularly in Fig.13.Figure 12shows that estimated variances by EMSIT did not follow actual variances in the case of temporal traf-fic increasing.This is because EMSIT uses all the measured data during a measurement section to estimate variance σ2.Both temporal tra ffic dynamics and the measurement sec-tion’s length a ffect the estimated variance for a long time.On the other hand,EMSIT-LD uses all the measured data during the measurement section only to calculate the slope u of the regression line l .Therefore,temporal tra ffic dy-namics a ffects the estimated variance only for the estimationsection.Fig.14Tra ffic volume(S4).Fig.15Variance dynamics(S4).Fig.16Comparison of each distribution (S4at 180min).5.Numerical ExamplesWe compare performances of EMSIT and EMSIT-LD with actual measured tra ffic data.The measured link was used as the main external connection line (6Mbps)between the LANs in the R&D center of a certain company and the Inter-net.Applications used in the link consist of mainly WWW,e-mail,and FTP.P2P applications were not used.More-over,the LANs comprised more than a thousand PCs.The measurements were conducted for downlink from the Inter-net during the daytime on a weekday.The daytime tra ffic increased in the morning and decreased in the evening.1116IEICE MUN.,VOL.E89–B,NO.4APRIL2006Fig.17Tra fficvolume.Fig.18Variance dynamics.Figure 17illustrates the measured tra ffic volume per 1-min.In Fig.17,the tra ffic has a light load from 0min to 500min,then a heavy load after 500min.We set estimation parameters as follows,T =10,k =12,m 1=6,m 2=10,p =12,q =6.These parameters are same as the simulation ones we evaluated above.Figure 18illustrates estimated and measured (loga-rithm)variance dynamics,where each point is sampled in a 20-min cycle.Estimated variances by EMSIT-LD follow the dynamics of measured variances;however,estimated vari-ances by EMSIT do not follow them.Figures 19and 20illustrate comparisons of estimated and measured distributions at the 515-min and the 920-min marks in Fig.17.The 515-min mark is in the midst of the tra ffic dynamics,while the 920-min mark is in steady traf-fic after the tra ffic dynamics.In Fig.19,only EMSIT pro-duces a dissimilar distribution to the measured one.On the other hand,in Fig.20,both methods produce similar distri-butions to the measured one except for the range between 2and 3Mbps.These results (including the simulation ones)indicate that EMSIT estimates incorrect distributions in the tra ffic dynamics zone and possible distributions after the tra ffic dy-namics zone,whereas EMSIT-LD always estimates almost-correct ones regardless of tra ffic dynamics.Next,we evaluate the performance of the proposed method under various values of threshold parameter βin Step 2of EMSIT-LD,because parameter’s setting mayaf-Fig.19Comparison of each distribution (at 515min).Fig.20Comparison of each distribution (at 920min).Fig.21Threshold βand error ratio.fect the accuracy of the proposed method.Here,we define the error ratio R e as follows,R e ≡ˆσ2(1)−σ2(1) σ2(1),(4)where σ2(1)is an actual tra ffic variance for interval T ,andˆσ2(1)is an estimated variance for T .Figure 21shows error ratios when βwas varied in the case of Fig.17.Each error ratio in this figure is the aver-age error ratio for the whole estimation section.This result shows that the presence of an appropriate βvalue and a smallFUKAMI et al.:ESTIMATING METHOD OF SHORT-INTERV AL-TRAFFIC DISTRIBUTION1117βvalue does not lead to accurate estimation.Figure21shows that the optimal value of the thresh-oldβexists near0.8Mbps for that case.Whenβis smaller than the optimal value,the error ratio greatly deteriorates. In many these cases,“a long-term-traffic dynamics”was de-tected and past traffic measurement data were used to esti-mate the distribution.On the other hand,whenβis larger than the optimal value,the error ratio deteriorates only a lit-tle.We propose an approach to determineβby using actual short-interval measurement data for each link.For example, we measure short-interval measurement data only for one day,and set the parameterβusing these data for after that. We can measure actual short-interval measurement data for many links by equipping a different link with the same de-vice every day.In this approach,probe devices for every link in the wide-area network to measure short-interval data are not necessary,and we can determine value of the parameter βusing only a small number of probe devises.6.ConclusionWe proposed an estimation method of short-interval-traffic distributions considering long-term-traffic dynamics (EMSIT-LD)for QoS network management of multime-dia applications.This method estimates the current short-interval traffic distribution using traffic trends of the pre-vious several hours,which are obtained from the previ-ous steady traffic states.EMSIT-LD enables us to esti-mate the traffic distribution in a long-term-traffic dynamics zone,which was impossible for the conventional estimat-ing method,EMSIT.According to some of the simulation result and numerical examples,the EMSIT-LD successfully demonstrated its performance.Some research problems remain concerning the pro-posed method.The parameter-setting method should be dis-cussed more deeply.EMSIT-LD requires some parameters such as q,βand so on.Furthermore,the value of T and an interval time for the estimation section depend on the network management policy.We should discuss this pol-icy from the perspective of application-level quality.On the other hand,βand q may be not related to the policy,there-fore,we also need to examine a parameter-setting method from the viewpoint of estimation performance.We think investigating using actual traffic data for parameter setting will be useful.Moreover,EMSIT-LD should be evaluated under various network and traffic conditions.To improve the EMSIT-LD,we need to investigate how to integrate it with other statistical methods such as time-series analysis.AcknowledgementThis work was partly supported by a Grant-in-Aid from the 21st Century COE Program(Grant No.14213201)of the Ministry of Education,Culture,Sports,Science and Tech-nology(MEXT),Japan and a Grant-in-Aid for Scientific Research(C)(no.15560327)from the Japan Society for the Promotion of Science(JSPS).References[1]IPPM WG,/,2004.[2]CAIDA,/home/,2004.[3] A.Feldmann,A.Greenberg,C.Lund,N.Reingold,and J.Rexford,“NetScope:Traffic engineering for IP networks,”IEEE Commun.Mag.,vol.14,no.2,pp.11–19,2000.[4]Network Monitoring Tools,/˜cottrell/tcom/nmtf-tools.html,2004.[5]W.Stallings,SNMP,SNMPv2,and RMON practical network man-agement,Second ed.,Addison-Wesley Publishing Company,1996.[6]Agilent Technologies,/,2004.[7]MRTG,/mrtg.html,2005.[8]OpenView,/,2005.[9]T.Asaka,“Method of estimating short-interval-traffic distributionusing MIB,”IEICE mun.,vol.E85-B,no.5,pp.1038–1041,May2002.[10]Y.Breitbart, C.Y.Chan,M.Garofalakis,R.Rastogi,andA.Silberschatz,“Efficiently monitoring bandwidth and la-tency in IP networks,”IEEE INFOCOM2001,/user/minos/Abstracts/infocom01-abs.html,2001.[11]S.Belenki and S.Tafvelin,“Analysis of errors in network load mea-surements,”Computer Communication Review,vol.12,no.1,pp.64–79,2000.[12]T.Fukami,T.Asaka,and T.Takahashi,“Method of estimating short-interval-traffic distribution considered long-term-trafficfluctuation,”IEICE Technical Reports,NS2003–42,June2003.[13] D.S.Moore,The Basic Practice of Statistics,W H Freeman&Co(Sd),2003.[14]Z.Sahinoglu and S.Tekinay,“On multimedia networks:Self-similartraffic and network performance,”IEEE Commun.Mag.,vol.13, no.1,pp.48–52,1999.[15]M.Grossglauser and J.-C.Bolot,“On the relevance of long-rangedependence in network traffic,”IEEE/ACM w.,vol.7, no.5,pp.629–640,2000.[16]M.Roughan,D.Veitch,and P.Abry,“Real-time estimation of theparameters of long-range dependence,”IEEE/ACM w., vol.8,no.4,pp.467–478,2000.[17]H.Furuya,T.Hill,and H.Nakamura,“Influence of transmissiondelay on self-similar scaling behavior of aggregated TCP/IP traffic,”IEICE Technical Reports,SSE2000-10,April2000.[18] A.Feldmann,A.C.Gilbert,and W.Willinger,“Data networks as cas-cades:Investigating the multifractal nature of Internet W AN traffic,”ACM SIGCOMM’98,pp.25–38,Aug.2000.[19]R.Morris and D.Lin,“Variance of aggregated Web traffic,”IEEEINFOCOM2000,pp.360–366,March2000.[20]R.Kawahara,K.Ishibashi,T.Asaka,and K.Ori,“A method of IPtraffic management using the relationship between TCPflow behav-ior and link utilization,”IEICE mun.,vol.E86-B,no.11, pp.3244–3256,Nov.2003.[21]M.Ishizuka,M.Aida,and S.Kuribayashi,“Measurement-basedevaluation of TCP throughput,”IEICE mun.,vol.E87-B, no.12,pp.3637–3649,Dec.2004.[22]M.Nabe,M.Murata,and H.Miyahara,“Analysis and modeling ofWorld Wide Web traffic for cap acity dimensioning of Internet access links,”Performance evaluation,vol.34,no.4,pp.249–271,1998.。