贾生元博客案例材料109页
Abstract
A model of grounded language acquisition:Sensorimotor features improve lexicaland grammatical learningSteve R.Howell *,Damian Jankowicz,Suzanna BeckerMcMaster University,Hamilton,Ont.,CanadaReceived 1December 2004;revision received 15March 2005Available online 25April 2005AbstractIt is generally accepted that children have sensorimotor mental representations for concepts even before they learn the words for those concepts.We argue that these prelinguistic and embodied concepts direct and ground word learn-ing,such that early concepts provide scaffolding by which later word learning,and even grammar learning,is enabled and facilitated.We gathered numerical ratings of the sensorimotor features of many early words (352nouns,90verbs)using adult human participants.We analyzed the ratings to demonstrate their ability to capture the embodied meaning of the underlying concepts.Then using a simulation experiment we demonstrated that with language corpora of suffi-cient complexity,neural network (SRN)models with sensorimotor features perform significantly better than models without features,as evidenced by their ability to perform word prediction,an aspect of grammar.We also discuss the possibility of indirect acquisition of grounded meaning through ‘‘propagation of grounding’’for novel words in these networks.Ó2005Elsevier Inc.All rights reserved.Keywords:Language acquisition;Features;Semantics;SRN;Neural network;Sensorimotor;Conceptual learningConsiderable evidence suggests that by the time chil-dren first begin to learn words around the age of 10–12months,they have already acquired a fair amount of sensorimotor (sensory/perceptual and motor/physical)knowledge about the environment (e.g.,Lakoff,1987;Lakoff&Johnson,1999;Bloom,2000;Langer,2001),especially about objects and their physical and percep-tual properties.By this age children are generally able to manipulate objects,navigate around their environ-ment,and attend to salient features of the world,includ-ing parental gaze and other cues important for word learning (Bloom,2000).Some have suggested that this pre-linguistic conceptual knowledge has a considerable effect on the processes of language acquisition (Lakoff,1987;Mandler,1992;Smith &Jones,1993)and even on later language processing (e.g.,Glenberg &Kaschak,2002;Barsalou,1999).We also argue that the evidence indicates that this early prelinguistic knowledge has great impact,directly and indirectly,throughout a num-ber of phases of language learning,and we attempt to begin to demonstrate this with a neural networkmodel.Journal of Memory and Language 53(2005)258–276Journal of Memory and Language0749-596X/$-see front matter Ó2005Elsevier Inc.All rights reserved.doi:10.1016/j.jml.2005.03.002*Corresponding author.Present address:Department of Psychology (WJ Brogden Hall),University of Wisconsin-Madison,1202West Johnson Street,Madison,WI 53703,USA.Fax:+16082624029.E-mail address:showell@ (S.R.Howell).To begin with,this prelinguistic conceptual informa-tion helps children to learn theirfirst words,which cor-respond to the most salient and imageable(Gillette, Gleitman,Gleitman,&Lederer,1999)objects and ac-tions in their environment,the ones they have the most experience with physically and perceptually.Generally speaking,the more‘‘concrete’’or‘‘imageable’’a word, the earlier it will be learned.This helps to explain the preponderance of nouns in childrenÕs early vocabularies (see Gentner,1982).The meanings of verbs are simply more difficult to infer from context,as discussed by as demonstrated by Gillette et al.(1999).Only the most clearly observable or‘‘concrete’’verbs make it into chil-drenÕs early vocabularies.However,later verbs are ac-quired through the assistance of earlier-learned nouns. If a language learner hears a simple sentence describing a real-world situation,such as a dog chasing a cat,and already knows the words dog and cat,the only remain-ing word must be describing the event,especially if the learner already has built up a pre-linguistic concept of ‘‘dogs chasing cats’’at the purely observational level. As Bloom(2000)describes,the best evidence for ‘‘fast-mapping’’or one-shot learning of words in chil-dren comes from similar situations in which only one word in an utterance is unknown,and it has a clear, previously unknown,physical referent present.Of course,since the verb chase refers to an event rather than an object,the above example is not an exactfit to the fast-mapping phenomenon as it is usually de-scribed,but it is similar.These veryfirst words that children learn thus help constrain the under-determined associations between the words children hear and the objects and events in their environment,and help children to successfully map new words to their proper referents.This happens through the use of cognitive heuristics such as the idea that a given object has one and only one name(Mark-man&Wachtel,1988),or more basic object-concept primitives(Bloom,2000)such as object constancy.With a critical mass of some50words,children begin to learn how to learn new words,using heuristics such as the count-noun frame,or the adjective frame(Smith, 1999).These frames are consistent sentence formats of-ten used by care-givers that enable accurate inference on the part of the child as to the meaning of the framed word,e.g.,‘‘This is a___.’’These factors combine to produce a large increase in childrenÕs lexical learning at around20months.As they begin to reach another critical mass of words in their lexicon(approaching 300words),they start to put words together with other words—the beginnings of expressive grammar(Bates& Goodman,1999).Around28months of age children en-ter a‘‘grammar burst’’in which they rapidly acquire more knowledge of the syntax and grammar of their lan-guage,and continue to approach mature performance over the next few years.By this account of language acquisition,conceptual development has primacy;it sets the foundation for the language learning that will follow.Words are given meaning quite simply,by their associations to real-world,perceivable events.Words are directly grounded in embodied meaning,at least for the earliest words. Of course,it may not be just simple statistical associa-tions between concepts and words in the environment; the child is an active learner,and processes like joint attention or theory of mind may greatly facilitate the learning of word to meaning mappings(Bloom,2000).Of course,it seems clear that the incredible word-learning rates displayed by older children(Bloom, 2000)indicate that words are also acquired by linguistic context,through their relations to other words.Children simply are learning so many new words each day that it seems impossible that they are being exposed to the ref-erents of each new word directly.The meanings of these later words,and most of the more abstract,less image-able words we learn as adults,must clearly be acquired primarily by their relationships to other known words. It may in fact be true that these meanings can only be ac-quired indirectly,through relationships established to the meanings of other words.Evidence for the indirect acquisition of meaning is not limited to the speed with which children learn words. The work of Landauer and colleagues(e.g.,Landauer and Dumais,1997;Landauer,Laham,&Foltz,1998) provides perhaps the clearest demonstration that word ‘‘meanings’’can be learned solely from word-to-word relationships(although see Burgess&Lund,2000;for a different method called HAL).LandauerÕs Latent Semantic Analysis(LSA)technique takes a large corpus of text,such as a book or encyclopedia,and creates a matrix of co-occurrence statistics for words in relation to the paragraphs in which they occur.Applying singu-lar-value decomposition to this matrix allows one to map the words into a high-dimensional space with dimensions ordered by significance.This high-dimen-sional representation is then reduced to a more manage-able number of dimensions,usually300or so,by discarding the least significant dimensions.The resulting compressed meaning vectors have been used by Landa-uer et.al.in many human language tasks,such as multi-ple choice vocabulary tests,domain knowledge tests,or grading of student exams.In all these cases,the LSA representations demonstrated human-level performance.While models based on these high-dimensional repre-sentations of meaning such as LSA and HAL perform well on real world tasks,using realistically sized vocab-ularies and natural human training corpora,they do have several drawbacks.First,they lack any consider-ation of syntax,since the words are treated as unordered collections(aÔbag of wordsÕ).Second,LSA and HAL meaning vectors lack any of the grounding in reality that comes naturally to a human language learner.Experi-S.R.Howell et al./Journal of Memory and Language53(2005)258–276259ments by Glenberg and Robertson(2000)have shown that the LSA method does poorly at the kinds of reason-ing in novel situations that are simple for human seman-tics to resolve,due largely to the embodied nature of human semantics.So it seems that there are two sources of meaning,di-rect embodied experience,and indirect relations to other words.However,there is an infinite regress in the latter. If words are only ever defined in relation to other words, we can never extract meaning from the system.We would have only a recursive system of self-defined mean-ing,symbols chained to other symbols(similar to SearleÕs Chinese Room argument,Searle,1980).To avoid this dilemma,at least some of the words in our vocabularies must be defined in terms of something external.In children,at least,the earliest words serve this role.They are defined by their mappings to pre-lin-guistic sensory and motor experience,as discussed above.They do not require other words to define their meaning.The most imageable words are thus directly grounded,while the less imageable and more abstract words that are encountered during later learning are more and more indirectly grounded.At some point, we argue,the adult semantic system begins to look much like the LSA or HAL high-dimensional meaning space, with our many abstract words(e.g.,love,loyalty,etc.) defined by relations among words themselves.However, the mature human semantic system is superior to the high-dimensional models,since it can trace its meaning representations back to grounded,embodied meaning, however indirectly for abstract words.Intuitively,this is something like trying to explain an abstract concept like‘‘love’’to a child by using concrete examples of scenes or situations that are associated with love.The abstract concept is never fully grounded in external reality,but it does inherit some meaning from the more concrete concepts to which it is related.Part of the concrete wordsÕembodied,grounded,meaning becomes attached to the abstract words which are often linked with it in usage.By our account,the grounded meaningÔpropagatesÕup through the syntactic links of the co-occurrence meaning network,from the simplest early words to the most abstract.Thus we have chosen to call this the‘‘propagation of grounding’’problem.We argue that this melding of direct,embod-ied,grounded meaning with high-dimensional,word co-occurrence meaning is a vital issue in understanding conceptual development,and hence language develop-ment.We believe it is essential to resolving the disputes between embodied meaning researchers and high-dimen-sional meaning researchers.In previous work(Howell&Becker,2000,2001;Ho-well,Becker,&Jankowicz,2001)we began developing what we consider to be a promising method for model-ing childrenÕs language acquisition processes using neu-ral networks.In this work,we continue this effort,emphasizing the inclusion of pre-linguistic sensorimotor features that will ground in real-world meaning the words that the network will learn.This is a necessary precursor to addressing the‘‘propagation of grounding’’problem itself.Our overall goal is to capture with one model the es-sence of the process by which children learn theirfirst words and theirfirst syntax or grammar.As mentioned above,this is a period stretching from the earliest onset of thefirst true words(10–12months),through the‘‘lex-ical-development burst’’around20months up to the so-called‘‘grammar burst’’around28months.Developing a network that attempts to model the language acquisi-tion that is happening during this period in children is, of course,an ambitious undertaking,and our models are still relatively simple.However,given the discussion on propagation of grounding above,this sort of devel-opmental progression may actually be necessary not just for children learning language,but also for any abstract language learner such as a neural network or other com-putational model.That is,a multi-stage process of con-strained development may be necessary to simplify the problem and make it learnable,with eachÔstageÕprovid-ing the necessary foundation for the next,and ensuring that meaning continues to be incorporated in the pro-cess.As such,we seek to develop and extend a single model that can progress through theseÕstagesÕof lan-guage acquisition,from initial lexical learning,through rapid lexical expansion,to the learning of the earliest syntax of short utterances.Developing a model thatfits developmental behavioral data on child language acqui-sition is one way to ensure that this process is being fol-lowed.For the simulations reported here,we have adopted and extended the Simple-Recurrent Network architecture that has been shown many times to be capa-ble of learning simple aspects of grammar,namely basic syntax(e.g.,Elman,1990,1993;Howell&Becker,2001). Furthermore,SRNÕs have been shown to be able to pro-duce similar results to high-dimensional meaning mod-els.Burgess and Lund(2000)point out that their HAL method using their smallest text window produces simi-lar results in word meaning clustering to an Elman SRN. Also,they state that the SRN is somewhat more sensi-tive to grammatical nuances.SRNÕs may be able to model the acquisition of meaning and grammar,unlike the high-dimensional approaches.The present emphasis of our model is on the inclu-sion of sensorimotor knowledge of concepts or words (for clarity,in what follows we use the term‘‘concept’’to mean the mental representation of a thing or action, and the term‘‘word’’to mean merely the linguistic sym-bol that represents it).This pre-linguistic sensorimotor knowledge(following Lakoff,1987)is represented by a set of features for each word presented to the network, features that attempt to capture perceptual and motor aspects of a concept,such as‘‘size,’’or‘‘hardness,’’or260S.R.Howell et al./Journal of Memory and Language53(2005)258–276‘‘has feathers’’.If a word that the network experiences is accompanied by a set of values or ratings on these fea-ture dimensions,then the network should be able to do more than just manipulate the abstract linguistic symbol of the concept(the word itself).Like a child learning thefirst words,it should then have some access to the meaning of the concept.The networkÕs under-standing would be grounded in embodied meaning,at least at the somewhat abstracted level available to a model without any actual sensory abilities of its own.Unlike most existing language models that employ semantic features(e.g.,Hinton&Shallice,1991;McRae, de Sa,&Seidenberg,1997)our sensorimotor feature set has been designed to be pre-linguistic in nature.That is, features that derive from associative knowledge about which words occur together or other language-related associations are excluded.Only features that a preverbal child could reasonably be expected to experience directly through his or her perceptual and motor interactions with the world are included.As discussed above,while childrenÕsfirst words are obviously learned without any knowledge of language-related word associations, children quickly begin to incorporate linguistic associa-tive information into the semantic meanings of concepts. Certainly,at some point words begin to acquire meaning not only from the sensory properties of the concept,but from the linguistic contexts in which the word has been experienced.We take the conservative stance herein of excluding any linguistic associative influences on senso-rimotor meaning;the sensorimotor feature representa-tions do not change with linguistic experience.This is primarily for practical issues of implementation.The network is capable of learning these associations,but they do not affect the sensorimotor feature representa-tions directly.Whereas most language models employ binary fea-tures,our features are scalar-valued(range0–1),allow-ing a network to makefiner discriminations than merely the binary presence or absence of a feature. For example,two similar items(for example,two cats) may be perceived,but they are not identical;one is lar-ger.Our dimension of size would differentiate the two, with one receiving a rating of0.2,one of0.3.Binary fea-tures cannot easily make suchfine distinctions.Finally, inspired by the work of McRae et al.(1997)on hu-man-generated semantic features,the feature ratings that we use are all derived empirically from human participants.One of the advantages of the neural network model of child language development that we present below is the ability to measure word-learning performance using analogues of lexical comprehension tasks that have been used with children.Since the network learns to associate the sensorimotor features of each concept with a separate phonemic representation of the word, it is possible to examine the strength of the connection in either direction.Thus,given the phonemes of the word,we can measure the degree to which the network produces the appropriate sensorimotor meaning vector. This we refer to as theÔgroundingÕtask,analogous to when a child is asked questions about a concept and must answer with featural information,such as‘‘What kind of noise does a dog make?’’or‘‘Is the dog furry?’’Similarly,we can also ask if,when given the meaning vector alone,the network will produce the proper word. This is an analogue to theÔnamingÕtask in children, where a parent points to an object and asks‘‘What is that?’’In the network,if the completely correct answer is not produced,we can still measure how close the out-put was to the correct answer.For example,we can check whether the answer was a case of‘‘cat’’produced in place ofÔdogÕ,two concepts with a high degree of fea-tural overlap,or whether it was a complete miss.In this paper,we address the grounding task,but not the nam-ing task,although the model can account for both. However,the central aim of this paper is to investigate the contribution of the sensorimotor features to improv-ing the modelÕs lexical and grammatical learning.In Experiments1and2,we describe the empirical collection of feature ratings for nouns and verbs,respec-tively,and describe the results of several analyses per-formed to verify that they are capturing an abstract representation of the wordsÕmeanings.In Experiment 3we describe simulations of a neural network model using these features and trained with a large naturalistic corpus of child-directed speech.We examine the extent to which the inclusion of sensorimotor features improves lexical and grammatical learning over a control condi-tion,in an attempt to demonstrate the utility of feature grounding for language acquisition.However,it is important to note that in referring to‘‘grammatical learning’’we are in fact only considering the simplest as-pects of grammar,namely basic sequence learning. Experiment1—Generation of noun sensorimotor featuresDeveloping a set of sensorimotor dimensions that are plausible for8-to28-month-old infants was an importantfirst stage of this research effort.In our pre-vious models of lexical grounding and acquisition of grammar(Howell&Becker,2001;Howell et al., 2001),we used a more simplistic semantic feature repre-sentation of words(Hinton&Shallice,1991)that was both artificial and confounded wordsÕconceptual semantics with‘‘associative semantics,’’the linguistic relationships between words.We needed a more child-appropriate set of semantic features.Of course, developing these semantic features actually involves two issues:one,collecting the ratings,and two,making sure that the results of the ratings actually reflect realis-tic early semantic relationships.If the collected ratingsS.R.Howell et al./Journal of Memory and Language53(2005)258–276261do not have plausible semantic cohesion,they cannot be very useful in further work.MethodTo avoid having artificial,experimenter-created semantic feature representations,we investigated the McRae et al.(1997)empirically generated feature set. However,of the thousands of features contained in that set,many were non-perceptual(e.g.,linguistically asso-ciative),and few were common across many concepts. To obtain an appropriate set of input features for a neu-ral network model of child language acquisition,we re-quired a more compact,concrete set of features that are perceptual and motor in nature,and could reasonably capture purely pre-linguistic knowledge.Thus,we nar-rowed down the McRae et al.feature list to some200 common and widely represented features.This list was further condensed by converting each set of polar oppo-sites and intermediate points to a single set of19polar-opposite dimensions.For example,‘‘small’’and‘‘large’’became a single continuous dimension of size,ranging from small(0)to large(10),and eliminating the need for‘‘tiny,’’‘‘medium,’’‘‘huge,’’etc.The remaining78 features which could not be unambiguously converted to a set of polar opposites were retained as a condensed list of scalar-valued dimensions,such as color(is_red)or texture(has_feathers),where the numeric value indi-cated the probability of possession of that feature by that concept.We use the termÔfeature dimensionÕorÔdi-mensionÕto refer to all97dimensions,however,since when considered as components of a meaning vector they each represent a spatial dimension in a97-dimen-sional space.This resulting list of features was then reviewed by an independent developmental psychologist,for accessibil-ity to children of the age range in question(8–28 months),and any features that were not considered developmentally appropriate were removed.For exam-ple,‘‘age’’is not reliably perceived by children beyond simply‘‘young’’or‘‘old’’(urel Trainor,private communication,2001)and so was removed.Thefinal list of97sensorimotor feature dimensions (see Appendix A)was small enough to be feasible as in-put for our neural network models,and broad enough to be applicable to many concepts.Given this set of feature dimensions,it was next necessary to obtain ratings of the early concepts along each feature dimension.We used a large sample of human raters to generate the featural ratings for our early words.Our raters were undergrad-uates at McMaster University who participated in this experiment for course credit in an introductory psychol-ogy course.Participants were presented with the concepts and the list of feature dimensions along which to rate them on a computer screen.The display was presented via a web browser,and responses were entered byfilling in re-sponse boxes on the display.Participants were given de-tailed instructions(see Appendix A)as to how to make judgments,and which anchoring points to use in assign-ing numerical values.For example,in rating the size of an object,the smallest item a child might know about might beÔpea,Õwhile for adults it might be something microscopic likeÔvirus.ÕThus,participants were specifi-cally instructed to make judgments taking into account the limited frame of reference that a pre-school child would have,especially relevant for polar-opposite dimensions such as‘‘size.’’Participants entered their data as numbers between1and10,which were later scaled down to the0–1range for easier presentation to neural network models.The rating forms were administered over the Internet as web forms.The data was checked carefully for outli-ers.Three participantsÕdata were excluded due to obvi-ous response patterns(all0Õs,all10Õs,1-2-3Õs,etc.), indicating insufficient attention given to the task.Rat-ings were collected for352noun concepts from the Mac-Arthur Communicative Development Inventory (MCDI,Fenson et al.,2000)in38separate phases with approximately10concepts each during winter,2002. Thefirst two phases had10participants each;the rest had5participants each,for a total of200participants. Participants received course credit for participation so long as the data were not obviously invalid as discussed above The resulting ratings were then averaged across participants yielding a single feature vector of size97 for each concept,352in all.Three forms of analysis were performed on these newly created feature representations,in order to dem-onstrate that they do capture important aspects of the meanings of the words represented:a hierarchical cluster analysis,a Kohonen self-organizing map,and a Euclid-ian-distance-based categorical membership analysis.ResultsWe analyzed the352averaged feature vectors in a hierarchical cluster analysis using SPSS version11.5, to see whether our features captured our intuitive sense of word similarity.The352concepts clearly clustered by meaning,with subcategories merging nicely into super-ordinate categories(see Appendix B).Animals are sepa-rated from people,people and animals are separated from vehicles and inanimate objects,etc.Thus,while the high degree of variability between participantsÕrat-ings was originally a concern,after averaging,the regu-larity inherent in the feature vectors is quite reassuring. To provide another view on the ratings,the ratings vec-tors were fed into a Self-Organizing Map(Kohonen, 1982,1995)neural network,which sought to group the concepts topographically onto a two-dimensional space based on their feature similarity.The resulting topo-262S.R.Howell et al./Journal of Memory and Language53(2005)258–276graphic organization respects the semantic similarities between concepts,showing intuitive groupings based only on the sensorimotor features of concepts(see Fig.1).Note,for example,the grouping of‘‘creatures that fly’’in the top left corner,and the grouping of parts of the body in the middle-left.A more clearly defined measure of success is provided by the categorical analysis.We formed category cent-roids for each of the pre-existing categories of nouns on the MCDI form from which the words were origi-nally drawn.This was done by taking all of the words that belonged to that category and averaging together their feature vector.Then each and every wordÕs feature vector was compared to the centroids of each of the11 categories represented,and the closest match indicated into which category the word should fall.This was done both with the target word included in the centroid gen-eration process,and with it excluded(a more conserva-tive approach).Results are very good,at92.8and88% accuracy,respectively(Chance performance would be 9.1%).See Table1for details.DiscussionWe believe all three analyses indicate the success of the experiment.The hierarchical clusteringanalysis,while vast and somewhat difficult to interpret,shows many clear separations of concepts,and consistent local clusters of meaning.The SOM representation shows clear clustering by meaning,with bothfine-grained and broader similarity structures across the map.Finally, the categorical analysis provides a clear numerical mea-sure of the goodness offit of our features to the preex-isting categorizations of these nouns,with93% accuracy of word to category.The sensorimotor feature ratings thus capture much of the meaning of the con-cepts,definitely enough to be useful as inputs to our lan-guage learning model,and they certainly capture whatÕs important for categorical reasoning.Experiment2—Generation of verb sensorimotor features In this experiment we followed much the same meth-odology as for Experiment1,this time for verb features. However,given that verbs correspond to events in the world rather than to objects,the nature of verb features was expected to be different from that for nouns.Also, there was no pre-existing taxonomy of verb features readily accessible in the literature,as there had been for nouns(what we mean by verb features is different from verbÔclasses,Õthe way verbs are usually grouped). Therefore,our collection of verb features proceeded in two steps.First we conducted a pilot experiment in verb feature generation with human participants,and from that we created a set of verb feature dimensions to be rated in a web-based phase of the experiment exactly as in Experiment1.MethodThe pilot experiment was conducted with12under-graduate participants at McMaster University(see Appendix C for the instructions given to participants). Participants completed a feature generation form for some of the earliest(MCDI,Fenson et al.,2000),and most prototypical(Goldberg,1999)verbs,with the objective being not complete characterization of any gi-ven verb but rather the creative generation of a set of feature dimensions which might be common to many verbs.While fully half of the features generated were unus-able due to contamination by functional relationships with corresponding nouns,associational relationships, etc.,there were sufficiently many perceptual and motor features identified to allow us to create an initial set of feature dimensions.From this beginning,we were able tofill in missing complements of existing dimensions. For example,several participants focused on limb movement to define verbs,which is in line with some existing models of verb definition in computer science (see for example Bailey,Feldman,Narayanan,&Lak-off,1997).From this and considerations of bodily motion and proprioceptive constraints in humans we were able to generate a large primary set of joint-mo-tion dimensions.We also included some other features that had been identified by pilot participants,which brought the total to84feature dimensions(see Appen-dix C for a list)A second group of45participants participated in the rating phase of the verb experiment(see Appendix C for the instructions given to participants).As in Experiment 1,they rated each verb on the list with a value between0 and10on the84feature dimensions.Each participant rated10of the concepts.We then converted these rat-ings to the0–1range,which became the feature repre-sentations for verbs used in the Experiment below.We analyzed the results of the experiment(the feature rat-ings)in the same three ways as in Experiment1:a hierarchical cluster analysis(see Appendix D),a self-Table1Noun category agreement results feature vectors compared to centroids of categories drawn from MCDICategory number Category name Inclusive accuracy Exclusive accuracy 1Animals0.82051280.82051282Vehicles0.91666670.753Toys0.91666670.83333334Food and drink115Clothing0.92857140.89285716Body parts10.8620697Small household items118Furniture and rooms0.84848480.78787889Outside things0.93333330.866666710Places to go0.86363640.636363611People0.84615380.8461538Overall0.92836680.8796562264S.R.Howell et al./Journal of Memory and Language53(2005)258–276。
RTL DFT Techniques to Enhance Defect Coverage
J Electron Test(2010)26:151–164DOI10.1007/s10836-009-5135-1RTL DFT Techniques to Enhance Defect Coverage for Functional Test SequencesHongxia Fang·Krishnendu Chakrabarty·Hideo FujiwaraReceived:30September2009/Accepted:10December2009/Published online:6January2010©Springer Science+Business Media,LLC2009Abstract Functional test sequences are often used in manufacturing testing to target defects that are not de-tected by structural test.However,they suffer from low defect coverage since they are mostly derived in prac-tice from existing design-verification test sequences. Therefore,there is a need to increase their effective-ness using design-for-testability(DFT)techniques.We present a DFT method that uses the register-transfer level(RTL)output deviations metric to select observa-tion points for an RTL design and a given functional test sequence.Simulation results for six ITC 99cir-cuits show that the proposed method outperforms two baseline methods for several gate-level coverage met-rics,including stuck-at,transition,bridging,and gate-Responsible Editor:P.MishraThis research was supported in part by the Semiconductor Research Corporation under Contract no.1588,and by an Invitational Fellowship from the Japan Society for the Promotion of Science.This paper is based on a preliminary version of an invited paper in Proceedings of IEEE International High Level Design Validation and Test Workshop,2009,and a presentation at the IEEE Workshopon RTL and High-Level Testing,2009.H.Fang(B)·K.ChakrabartyDepartment of Electrical and Computer Engineering,Duke University,Box90291,130Hudson Hall,Durham,NC27708,USAe-mail:hf12@K.Chakrabartye-mail:krish@H.FujiwaraGraduate School of Information Science,Nara Institute of Science and Technology,Kansai Science City,Nara630-0192,Japane-mail:fujiwara@is.naist.jp equivalent fault coverage.Moreover,by inserting a small subset of all possible observation points using the proposed method,significant fault coverage increase is obtained for all benchmark circuits.Keywords DFT·Output deviations·RT-level·Test-point insertion·Unmodeled defects1IntroductionVery deep sub-micron(VDSM)process technologies are leading to increasing defect rates for integrated circuits(ICs)[1,24].Since structural test alone is not enough to ensure high defect coverage,functional test is commonly used in industry to target defects that are not detected by structural test[11,28,36].An advantage of functional test is that it avoids overtesting since it is performed in normal functional mode.In con-trast,structural test is accompanied by some degree of yield loss[32].Register transfer(RT)-level fault mod-eling,design-for-testability(DFT),test generation and test evaluation are therefore of considerable interest [5,26,33,35].In RT-level design,a circuit’s behavior is defined in terms of the flow of signals(or transfer of data)between registers and the logical operations performed on those signals.A number of methods have been presented in the literature for test generation at RT-level.In[5],the authors proposed test generation based on a genetic algorithm(GA),targeting statement coverage as the quality metric.In[12,15,23,31],the authors used pre-computed test sets for RT-level modules(adders, shifters,etc.)to derive test vectors for the complete design.In[40],the authors presented a spectral methodfor generating tests using RT-level faults,which has the potential to detect almost the same number of faults as using gate-level test generation.In[19,21],the authors proposed a fault-independent test generation method for state-observable finite state machines(FSMs)to increase the defect coverage.To increase the testability of the complete design and to ease RT-level test generation,various DFT methods at RT-level have also been proposed.The most com-mon methods are based on full-scan or partial scan. However,a scan-based DFT technique leads to long test application time and it is less useful for at-speed testing.On the other hand,non-scan DFT technique [6,9,13,14,30,37]offer low test application time and they facilitate at-speed testing.In[6],non-scan DFT techniques are proposed to increase the testability of RT-level designs.In[30],the authors presented a method called orthogonal scan.It uses functional data-path flow for test data,instead of traditional scan-path flow;therefore,it reduces test application time.In[13], a technique was proposed to improve the hierarchical testability of the data path,which can aid hierarchi-cal test generation.In[14],the authors presented a DFT technique for extracting functional control-and data-flow information from RT-level description and illustrated its use in design for hierarchical testability. This method has low overhead and it leads to shorter test generation time,up to2–4orders of magnitude less than traditional sequential test generation due to the use of symbolic test generation.In[37],the au-thors presented a method based on strong testability, which exploits the inherent characteristic of datapaths to guarantee the existence of test plans(sequences of control signals)for each hardware element in the pared to the full-scan technique,this method can facilitate at-speed testing and reduce test application time.However,it introduces hardware and delay overhead.To reduce overhead,the authors pro-posed a linear-depth time-bounded testability-based DFT method in[9].It ensures the existence of a linear-depth time expansion for any testable fault and exper-iments showed that it offers lower hardware overhead than the method in[37].The goal of the above prior work on RTL DFT was to increase testability and ease RT-level test genera-tion.To enhance the effectiveness of given functional test sequences,the focus in prior work was on func-tional test selection based on various metrics,such as transition input/output(TRIO)[22],toggle coverage [10],validation vector grade(VVG)[34],etc.DFT involving scan-chain insertion at RTL has also been studied[2,20].However,the use of RTL DFT to in-crease the defect coverage of existing functional test se-quences for non-scan designs has largely been ignored. The generation of functional test sequences is a partic-ularly challenging problem,since there is insufficient automated tool support and this task has to be ac-complished manually or at best in a semi-automated manner.Therefore,functional test sequences for manu-facturing test are often derived from design-verification test sequences[16,18,27]in practice.It is impractical to apply all such long verification sequences during time-constrained manufacturing testing.Therefore,shorter subsequences must be used for testing,and this leads to the problem of inadequate defect coverage.Therefore, we focus on RTL DFT to increase the effectiveness of these existing test sequences for non-scan designs.In this paper,we address the problem of improving the defect coverage of given functional test sequences for an RT-level design.The proposed method adopts the RT-level deviation metric from[8]to select the most appropriate observation test points.The devia-tion metric at the gate-level has been used in[38]to select effective test patterns from a large repository of n-detect test patterns.It has also been used in[39] to select appropriate LFSR seeds for reseeding-based test compression.The deviation metric at RT-level has been defined and used in[8]for grading functional test sequences.Simulation results for six ITC 99circuits show that the proposed method outperforms two baseline meth-ods for several gate-level coverage metrics,including stuck-at,transition,bridging,and gate-equivalent fault coverage.Moreover,by inserting a small subset of all possible observation points using the proposed method, significant fault coverage increase is obtained for all benchmark circuits.Since functional test sequences are used to target unmodeled defects,especially when they are used in conjunction with structural testing for mod-eled faults,we evaluate test effectiveness by multiple and different fault models.The remainder of this paper is organized as follows. Section2presents the problem formulation.Section3 describes the RT-level output-deviations metric and its prior application to functional test grading.Section4 presents the proposed observation-point selection method based on RT-level deviations.The design of experiments and experimental results are reported in Section5.Section6concludes the paper and outlines directions for future work.2Problem FormulationWe first formulate the problem being tackled in this paper.Given:–The RT-level description for a design and a func-tional test sequence S;–A practical upper limit n on the number of obser-vation points that can be added.Goal:Determine the best set of n observation points that maximizes the effectiveness of the functional test sequence S.Functional test sequences can be instruction-based for processors or application-based for application specific integrated circuit(ASIC)cores such as an MPEG decoder.They can be in the format of high-level instructions or commands,or in the format of binary bit streams.To increase testability,we can insert an observation point for each register output.We can obtain the high-est defect coverage by inserting the maximum number of observation points.However,it is impractical to do so due to the associated hardware and timing overhead. In fact,the number of observation points that can be added is limited in practice.For a given upper limit n, the challenge is to determine the best set of n obser-vation points such that we can maximizes the defect coverage of the given functional test sequence.Our main premise is that RT-level output deviation can be used as a metric to guide observation-point selection. 3Output Deviations at RT-level:PreliminariesThe RT-level output deviations metric has been defined and used in[8]to grade functional test sequences.The RT-level output deviations metric was used as a surro-gate coverage metric to grade functional test sequences and test sequences with higher deviation values were found to provide higher defect coverage.Before describing RTL output deviations,we intro-duce basic concepts.The first concept is the transition count(TC)of registers.Typically,there is dataflow between registers when an instruction is executed and the dataflow affects the values of registers.For any given bit of a register,if the dataflow causes a change from0to1,it records that there is a0→1transition. Similarly,if the dataflow causes a change from1to0,it records that there is a1→0transition.If the dataflow makes no change to this bit of the register,it records that there is a0→0transition or a1→1transition,as the case may be.After a transition occurs,the value of the bit of a register can be correct or faulty(thus an error may be produced).With any transition of a register bit,a “confidence level”(CL)parameter is associated,which represents the probability that the correct transition occurs.The CL is not associated with the test sequence; rather,it provides a probabilistic measure of the correct operation of instructions at the RT-level.It can be esti-mated from low-level failure data or it can be provided by the designer based on the types of faults of interest. Low CL values can be assigned to registers that are to be especially targeted by functional test sequences for error observation and propagation.Experiments showed that small differences in the CL values have little impact on the effectiveness of grading functional test sequences using RT-level output deviations.Without loss of generality,the CL values for0→0 and1→1are assumed to be higher than that for the transitions0→1and1→0for a register bit.The CL values for0→1and1→0are assumed to be identical. The CL for a register can be described as a4-tuple, e.g.,<0.998,0.995,0.995,0.998>,where the elements in the tuple correspond to the transitions0→0,0→1,1→0,and1→1,respectively.While different CL values can be used for the various registers in a design, without loss of generality,all registers are assumed to have identical CL values.When an instruction is executed,there may be sev-eral transitions for the bits of a register.Therefore, an error may be manifested after the instruction is executed.The CL for instruction I i,C i,is defined as the probability that no error is produced when I i is executed.Similarly,since a functional test sequence is com-posed of several instructions,the CL for a functional test sequence is defined to be the probability that no error is observed when this functional test sequence is executed.For example,suppose a functional test sequence,labeled T1,is composed of instructions I1,I2, ...,I N,and let C i be the CL for I i,as defined above.The CL value for T1,C(T1),is defined as:C(T1)=Ni=1C i.(1)This corresponds to the probability that no error is produced when T1is executed.Finally,the deviation for functional test sequence T1, (T1),is defined as 1−C(T1),i.e.,(T1)=1−Ni=1C i.(2)Based on these definitions,output deviations can be calculated at RT-level for functional test sequences for a given design.Three contributors are considered in the calculation of deviations.The first is the TC of registers. Higher the TC for a functional test sequences,the morelikely is it that this functional test sequences will detect defects.The second contributor is the observability of a regis-ter.The TC of a register will have little impact on defect detection if its observability is so low that transitions cannot be propagated to primary outputs.Therefore,each output of registers is assigned an observability value using a SCOAP-like measure [3].(In SCOAP testability measure,a higher value of the measure for a signal indicates that it might be difficult to control or observe,i.e.,it has lower controllability or lower observability.)The observability vector for a design,composed of the observability values of all registers,is used to model the observability of the the design.We use the Parwan processor [29]as an example to illustrate the calculation of observability vector.The Parwan is an accumulator-based 8-bit processor with a 12-bit address bus.Its architectural block diagram is shown in Fig.1a.Its dataflow diagram,representing all its possible functional paths,is shown in Fig.1b.Each node represents a register.The IN and OUT nodes represent memory.A directed edge between registers represents a possible functional path between registers.For example,there is an edge between the AC node and the OUT node.This edge indicates that there exists a possible functional path from register AC to memory.From the dataflow diagram,we can calculate the observability vector.First,we define the observability value for the OUT node.The primary output OUT has the highest observability since it is directly ing SCOAP-like measure for observability,we de-fine the observability value of OUT to be 0,written as OUT _obs =0.For every other register node,we de-fine its observability parameter as 1plus the minimum of the observability parameters of all its fanout nodes.For example,the fanout nodes of register AC are OUT,SR,and AC itself.Thus the observability parameter of register AC is 1plus the minimal observability para-meter among OUT,SR,AC.That is,the observability parameter of register AC is 1.In the same way,we can obtain the observability parameters for MAR,PC,IR and SR.We define the observability value of a register as the reciprocal of its observability parameter.Fi-nally,we obtain the observability vector for Parwan.Itis simply 1AC _obs ,1I R _obs ,1PC _obs ,1MAR _obs ,1SR _obs,i.e.,(1,0.5,1,1,0.5).The third contributor to RT-level output deviation is the weight vector,which is used to model how much combinational logic a register is connected to.Each register is assigned a weight value,representing the rel-ative sizes of its input cone and fanout cone.Obviously,if a register has a large input cone and a large fanout(b)Fig.1a Architecture of the Parwan processor;b Dataflow graph of the Parwan processor (only the registers are shown)cone,it will affect and be affected by many lines and gates.Thus it is expected to contribute more to defect detection.In order to accurately extract this informa-tion,we need gate-level information to calculate the weight vector.We only need to report the number of stuck-at faults for each component based on the gate-level netlist.This can be easily implemented without gate-level logic simulation by a design analysis tool.Consider each component in Parwan.There are 248stuck-at faults in AC,136in IR,936in PC,202in MAR,96in SR,14690in ALU,and 464in SHU.From this information,we can easily calculate the weight vector (Table 1).Based on the RT-level description of the design,we can determine the fanin cone and fanout cone of a register.For example,the AC register is connected to three components:AC,SHU and ALU.Given a set ofTable1Weight vector for registers(Parwan)AC IR PC MAR SR No.of faults affecting21723389362022020 registerWeight value10.15560.43090.0930.930registers{R i},i=1,2,..,n,let f i be the total number of stuck-at faults in components connected to register R i.Let f max=max{f1,..,f n}.We define the weight of register R i as f i/f max to normalize the size of gate-level logic.Table1shows the numbers of faults affecting registers and weights of registers.We can see that f max=2172for Parwan processor,which is the number of faults in AC.The weight of IR can be calculated as 338/2172,i.e.,0.1556.In this way,weights of other reg-isters can also be obtained.Finally,we get the weight vector(1,0.1556,0.4309,0.093,0.930).In order to calculate output deviations,first the effec-tive TCs for registers should be obtained.Suppose reg-ister Reg makes N10→1transitions for a functional test sequence T S.Then its effective0→1TC equals the product of N1,the observability value of register Reg,and the weight of register Reg.We use Parwan to illustrate how to calculate the effective TCs.For a functional test sequence T S,the TCs corresponding to T S are shown in Table2.In Table2,each row shows the TC for one register,while each column represents the transition type.For exam-ple,the value of third row and second column is206, which implies that the0→0TC for I R is206when functional test sequence T S is executed.By considering the weight vector and the observabil-ity vector,the TC of a register can be transformed to the effective TC.Table3shows the effective TC values for T S for the given observability vector and weight vector.In Table3,each row lists the effective TC for one register.The last row shows the aggregated effective TC for all registers.The columns indicate the various types of transitions.The following example illustrates how to calculate the deviation for a functional test sequences T S.Sup-pose T S is composed of50instructions I1,I2 (I50)Table2TCs for T S0→00→11→01→1 AC67616044 IR206676637 PC7089085205 MAR913251246246 SR67111412Table3Effective TCs for test sequence T SRegister Register0→00→11→01→1 ID nameR1AC67616044R2IR16.035.215.132.88 R3PC305.0838.7836.6388.33 R4MAR84.9123.3422.8822.88 R5SR31.165.1156.515.58 Sum504.18133.45131.19163.67 For each instruction I i,suppose the effective0→0TC for register R k(1≤k≤5)is R ki00,the effective0→1 TC for register R k is R ki01,the effective1→0TC for register R k is R ki10,and the effective1→1TC for reg-ister R k is R ki11.Given the CL vector(1,0.98,0.98,1), the CL value C i for instruction I i can be calculated by considering all possible transitions and the different registers:C i=5k=11R ki00·0.998R ki01·0.998R ki10·1R ki11.(3)Using the property of the exponents,whereby x a·x b= x a+b,Eq.3can be rewritten asC i=15k=1R ki00·0.998 5k=1R ki01·0.9985k=1R ki10·1 5k=1R ki11.(4) Let S i00=5k=1R ki00,S i01=5k=1R ki01,S i10=5k=1 R ki10,and S i11=5k=1R ki11.Equation4can now be written asC i=1S i00·0.998S i01·0.998S i10·1S i11.(5) Based on the deviation definition,the deviation for T S can be calculated as (T S)=1−50i=1C i,i.e.,(T S)=1−50i=11S i00·0.998S i01·0.998S i10·1S i11=1−150i=1S i00·0.998 50i=1S i01·0.998 50i=1S i10·150i=1S i11(6)Let S∗00=50i=1S i00,S∗01=50i=1S i01,S∗10=50i=1S i10and S∗11=50i=1S i11,Eq.6can be rewritten as:(T S)=1−1S∗00·0.998S∗01·0.998S∗10·1S∗11.(7)Note that S∗00is the aggregated effective0→0TC of all registers for all the instructions in T S.The parametersS∗01,S∗10,and S∗11are defined in a similar way.4Observation-point SelectionIn this section,we first define the new concept of RT-level internal deviations.Next,we analyze the fac-tors that determine observation-point selection.Then we present the observation-point selection algorithm based on RT-level deviations.Finally,we introduce recent related work[25],which will be used in Section5 for comparison.4.1RT-level Internal DeviationsThe RT-level output deviation[8]is defined to be a measure of the likelihood that error is manifested at a primary output.Here we define the RT-level internal deviation to be a measure of the likelihood of error being manifested at an internal register node,which means error being manifested at one or more bits of register outputs.In the calculation of RT-level output deviation,a transition in a register is meaningful only when it is propagated to a primary output.On the other hand,in the calculation of RT-level internal deviation, we do not care whether a transition in a register is propagated to a primary output.The method for calcu-lating internal deviations for register can also be used to calculate internal deviations for each bit of a register.The calculation of RT-level internal deviation is sim-ilar to that of RT-level output deviation.In order to calculate RT-level internal deviations for a register, first we need to define the internal effective TCs.Sup-pose a register Reg makes N10→1transitions for a functional test sequence T S.Then its internal effective 0→1TCs equals the product of N1and the weight of register Reg.The observability value of register Reg does not contribute to its effective transition count. Take I R in the Parwan processor as an example.It has the observability value0.5and weight0.1556.Since it makes670→1transitions for functional test se-quence T S,its internal effective0→1transition count is0.1556×67,i.e.,10.43.In the same way,we can calculate the internal effective TCs of all registers for T S,as shown in Table4.Table4Internal effective TCs for test sequence T SRegister Register0→00→11→01→1 ID nameR1AC67616044R2IR32.0610.4310.265.76 R3PC305.0838.7836.6388.33 R4MAR84.9123.3422.8822.88 R5SR62.3210.2313.0211.16After obtaining the internal effective transition counts,we can calculate the RT-level internal deviations.Suppose a functional test sequence T S is composed of m instructions I1,I2...,I m.For each instruction I i,suppose the internal effective 0→0TCs for register R k(1≤k≤t)is I_R ki00, the internal effective0→1TCs for register R k is I_R ki01,the internal effective1→0TCs for register R k is I_R ki10,and the internal effective1→1TCs for register R k is I_R ki11.Let I_S k00=mi=1I_R ki00, I_S k01=mi=1I_R ki01,I_S k10=mi=1I_R ki10,I_S k11= mi=1I_R ki11.The internal deviation of register R k for T S can be calculated for a given CL vector (cl00,cl01,cl10,cl11)as follows:I de v(R k)=1−cl00I_S k00·cl01I_S k01·cl10I_S k10·cl11I_S k11. Note that I_S k00is the aggregated internal effective 0→0TCs of register R k for all the instructions in T S. The parameters I_S k01,I_S k10,and I_S k11are defined in a similar way.In this way,we can calculate the in-ternal deviations of every register for T S.The method for calculating internal deviations for register can also be used to calculate internal deviations for each bit of a register.4.2Analysis of Factors that DetermineObservation-point SelectionThe selection of observation points is determined by three factors:RT-level internal deviations of registers, observability values of registers,and the topological relationship between registers.In this work,we only consider the insertion of observation points at outputs of registers.For a register Reg,we have the following attributes attached with it:I de v(Reg),O de v(Reg),obs(Reg),to represent its internal deviation,output deviation,and observability value,separately.For two registers Reg1and Reg2,when two at-tributes are close in value,we define the following observation-point-selection rules based on the third attribute:Rule1:If I de v(Reg1)>I de v(Reg2),select Reg1; Rule2:If obs(Reg1)<obs(Reg2),select Reg1; Rule3:If Reg1is the logical predecessor of Reg2, select Reg2.For Rule1,the motivation is that if we select a register with higher I de v,its observability will become1. Thus,its O de v will also becomes higher.The higher O de v of this register will contribute more to the cumulative O de v for the circuit.Since we have shown that the cu-mulative O de v is a good surrogate metric for gate-levelfault coverage[8],we expect to obtain better gate-level fault coverage when we select a register with higher I de v.For Rule2,when two registers do not have a prede-cessor/successor relationship with each other,obviously we should select the register with lower observability. For Rule3,if we select Reg1,obs(Reg1)will become1 but this will not contribute to the increase of observabil-ity of Reg2;if we select Reg2,obs(Reg2)will become1 and obs(Reg1)will also be increased due to the prede-cessor relationship between Reg1and Reg2.Therefore, it is possible that the selection of Reg2yields better results than the selection of Reg1,i.e.,the cumulative observability after the insertion of observation point on Reg2is higher than for Reg1.Rule2and Rule3are in conflict with each other on the observability attribute.Rule2selects a register with lower observability while Rule3selects a register with higher observability.In this work,we assume that Rule 3is given higher priority than Rule2.We use RT-level output deviations to guide the se-lection of observation points.We have determined that we should select a register with higher I de v.Since O de v is proportional to I de v and obs,if I de v factor contributes more to O de v,we should select the register with higher O de v.Also,by selecting a register with higher O de v, we are implicitly satisfying the predecessor relationship rule:for two registers Reg1and Reg2whose I de v values are comparable,if Reg1is the predecessor of Reg2, we have obs(Reg1)<obs(Reg2)and O de v(Reg1)< O de v(Reg2).Then we will not select Reg1,which is in accordance with Rule3.4.3RT-level Deviation Based Observation-pointSelectionBased on the RT-level output deviations,we have de-veloped a method for selecting best set of n(where n is a user-specified parameter)observation points for a given RT-level design and a given functional test sequence.In the selecting of observation-points,we target the specific bits of a register.The calculation of I de v,O de v,obs for a register is carried out for each bit of a register.The selection procedure is as following:–Step0:Set the candidate list to be all bits of regis-ters that do not directly drive a primary output.Table5Value of k for each circuitb09b10b12b13b14b15 k482013514730020Table6Gate-level fault coverage(stuck-at and transition)of the design before and after inserting all observation pointsCircuit Original design Design with all observationpointsSFC%TFC%#OP SFC%TFC% b0959.1847.932782.867.86 b1036.8920.191469.0345.67 b1250.2526.6711555.2331.92 b1335.923.334370.8344.02 b1483.9574.616192.3483.32 b159.91 5.3534723.2911.36–Step1:Derive the topology information for the design and save this information in a look-up table.Obtain the weight vector,observability vector,and TCs for each register bit,and calculate RT-level output deviations for each register bit.–Step2:Select a register bit with the highest output deviations as an observation point.Remove this selected register bit from the candidate list.–Step3:If the number of selected observation points reaches n,terminate the selection procedure.–Step4:Update the observability vector using the inserted observation point(selected in Step2)and the topology information.Re-calculate output de-viations for each register bit using the updated observability vector.Go to Step2.In Step1,the topology information of the design can be extracted using a design analysis tool,e.g.,Design Compiler from Synopsys.Topology information ex-tracted here refers to information of the upstream reg-isters for each register.It only needs to be determined once and it can be saved in a look-up table for subse-quent use.In Step4,after selecting and inserting an observation point,we need to update the observability vector because the observability of its upstream nodes will also be enhanced.There is no need to recompute TCs since these depend only on the functional test Table7Gate-level metrics(BCE+and GE score)of the design before and after inserting all observation pointsCircuit Original design Design with all observationpointsBCE+%GE score#OP BCE+%GE score b0945.581212770.13173b1028.041321455.07330b1229.9188911533.521005b1323.112574347.12483b1474.52860116181.238934b15 4.480634710.631987。
MBA经典案例-步鑫生现象
规定：不准迟到早退，违者重罚。有位副厂长从外地出差回来，第二 天上班迟到了 3 分钟，也被按规定扣发工资。以 1983 年计，全厂迟 到者仅 34 人次。步本人开会、办事分秒必争，内 广播一通知开会，两分钟内，全厂 30 名中层以上干部凡是在厂的全 部能到齐。开会的时间一般不超过 15 分钟。
环评爱好者论坛_2011年工程师案例真题分析—贾生元老师
一、2011年环评师考试道路改扩建项目案例题考生发来的案例题，从回忆情况来看，有较多的错别字，不知回忆的是否完整。
本人试着答了一下，也不能保证全部正确，只是与大家交流一下，欢迎提出意见和建议。
一、道路改扩建项目拟对某一现在有省道进行改扩建，其中拓宽路段长16km，新建路段长8km，新建、改建中型桥梁各1座，改造线全线为二级干线公路，设计车速80km/h，路基宽24m，采用沥青路面，改扩建工程需拆迁建筑物6200m2。
该项目沿线两侧分布有大量农田，还有一定数量的果树和路旁绿化带，改建中型桥梁桥址，位于X河集中式饮用水源二级保护区外边缘，其下游4km处为该集中式饮用水源保护区取水口。
新建桥梁跨越的Y河为宽浅型河流，水环境功能类别为II类，桥梁设计中有3个桥墩位于河床，桥址下游0.5km 处为某鱼类自然保护区的边界。
公路沿线分布有村庄、学校等，其中A村庄、B 小学和某城镇规划住宅区的概况及公路营运中期的噪声预测结果见下表：敏感点距红线距离敏感点概况营运中期预测结果路段A村庄 4m 8户超标8dB(A) 拓宽城镇规划住宅区 12m 约200户超标5dB(A) 新建B小学围墙高30m 教学楼120m 生100人、师100人。
夜间无人住宿教学楼昼达标，夜超标2dB(A) 拓宽问题：1、给出A村庄的声环境现状监测时段及评价量。
答：（1）声环境现状监测时段为昼间和夜间。
（2）评价量分别为昼间和夜间的等效声级（Leq,dB(A)）Ld和Ln。
2、针对表中所列敏感点，提出噪声防治措施并说明理由。
答：（1）A村应搬迁。
因为该村超标较高，且处于4a类区，采取声屏障降噪也不一定能取得很好效果，宜搬迁；（2）城镇规划的住宅区，可采取以下措施：a、调整线路方案b、设置声屏障、安装隔声窗以及绿化c、优化规划的建筑物布局或改变前排建筑的功能因为该段为新建路段，可以通过优化线路方案，使线路远离规划的住宅区；也可以设置声屏障并安装隔声窗、建设绿化带的措施达到有效的降噪效果；当然作为规划住宅区，也可以调整或优化规划建筑布局或改变建筑功能。
环评爱好者论坛_贾生元博客整理
2009年固废处理案例某市有一座处理能力600 t/d的生活垃圾填埋场，位于距市区10km处的一条自然冲沟内，场址及防渗措施均符合相关要求。
现有工程组成包括填埋场、填埋气体导排系统、渗滤液收集导排系统，以及敞开式调节池等。
渗滤液产生量约85 m^3/d，直接由密闭罐送距填埋场3km、处理能力为4×10^4 m^3/d的城市二级污水处理厂处理后达标排放。
填埋场产生的少量生活污水直接排入附近的一小河。
随着城市的发展，该市拟新建一座垃圾焚烧发电厂，涉及处理能力为1000 t/d，建设内容包括两座焚烧炉，2×22t/h余热锅炉和2×6MW 发电机组。
设垃圾卸料、输送、分选、储存、焚烧发电，飞灰固化和危险废物暂存等单元，配套建设垃圾渗滤液收集池、处理系统和事故收集池。
垃圾焚烧产生的炉渣、焚烧飞灰固化均送现有的垃圾填埋场。
垃圾焚烧发电厂距现有垃圾填埋场2.5km，不在城市规划区范围内。
厂址及其附近无村庄和其它工矿企业。
问题;1、简要说明现有垃圾填埋场存在的环境问题。
（1）敞开式调节池容易使恶臭气体挥发，影响场外环境空气质量；（2）填埋场产生的生活污水直接排入附近一小河，不符合环保要求（应经处理后回用或达标排放）。
（3）渗滤液未经处理，直接由密闭缺罐送入城市二级污水处理厂不符合规范要求（应在填埋场设置渗滤液处理系统，处理达标后回用、排放，或满足条件后再经污水处理厂处理）。
2、列出垃圾焚烧发电厂主要恶臭因子。
硫化氢（H2S）、甲硫醇（CH3SH）、氨（NH3）、三甲胺（CH3）3N、甲硫醚（CH3）2S、臭气浓度等。
（二甲二硫醚苯乙烯二硫化碳）3、除了垃圾储存池和垃圾输送系统外，本工程产生恶臭的环节还有哪些？垃圾卸料、分选、焚烧、渗滤液收集池及处理系统，甚至事故池等均可产生恶臭。
4、给出垃圾储存池和输送系统控制恶臭的措施。
（1）尽可缩短垃圾在储存池的时间，尽快进入焚烧或填埋系统；（2）垃圾储存池采用封闭式，恶臭气体集中收集，采取活性炭吸附、液体吸收或焚烧等处理方式。
案例
Analysis of clothing supply chain: Integration & Marriage of Lean &AgileBy Mandeep SainiContentsIntroduction 1 Lean and Agile Supply Chain 1 Particular ways of marring lean and agile paradigms 5 The Pareto Curve approach 6 The Decoupling Point 7 Separation of Base and Surge Demand 8 Case: Benetton 8 Case: Hennes & Mauritz (H&M) 10 Case: Zara 12 Conclusion 14List of TablesTable (1): Usage of Lean and Agile 2 Table (2): Difference in Lean and agile 3 Table (3): Market Winner and Qualifier Matrix 4 Table (4): Benefits of Leagile 5List of FiguresFigure (1): The Pareto Curve approach 6 Figure (2): The Decoupling Point 7 Figure (3): Base and Surge Demand 8 Figure (4): Traditional Lean manufacturing process of garments 9 Figure (5): Benetton’s Manufacturing Process 10 Figure (6): H&M’s Supply Chain Model 12 Figure (7): Flow of Information at Zara 13IntroductionModern supply-chains are very complex, with many analogous physical and information flows occurring in order to certify that healthy products are delivered in the right quantities, to the right place in a cost-efficient manner. The current drive towards more efficient supply networks during recent years has resulted in these international networks becoming more vulnerable to disruption. To be precise, there often tend to be very little inventory in the useful professional organisation to buffer interruptions in supply and, therefore, any disruptions can have a rapid impact across the progressive supply networks. This paper contains the significant issues of modern clothing supply chain. Due to globalisation, of rapidly changing markets and vogues of clothing business make it specified in terms of stylish fashion and changing user behaviour. The fashion industries are changing and expending the business while outsourcing; based on shortest lead times. But now, as per the case study “Supplying Fashion Fast” today’s supply chain are not to just serving the market with shortest lead time but it is to react immediately on the demand. . The challenge faced by a supply chain delivering fashion products is to develop a strategy that will improve the match between supply and demand and enable the companies to respond faster to the marketplace”(Naylor, Towill and Christopher, 2000).Lean and Agile Supply chainFor over a decade, companies have been achieving huge cost savings by streamlining their supply chains. While affluent, and thus pleasurable; these trends have also exposed organisations to new sets of paradigms such as Lean, Agile, Integration of Lean and Agile, Relationship driven supply chain etc. The question arise here is, Why there is a need to integrate the lean and agile supply chain? To find the answer the previous pages need to be turned; "Lean" is the name that James Womack gave to the Toyota Production System in the book “The Machine that Changed the World.” Lean was the term that best described Toyota's system versus the rest of the world's automotive manufacturers at the time. Many companies have since applied lean thinking to their organizations withvarying degrees of success. Applying lean to the entire supply chain is not a new concept, but very few have had success doing it. Naylor et. al (1999) defined the lean as, “Leanness means developing a value stream to eliminate all waste including time, and to enable level schedule.” Further the Agility means “using market knowledge and virtual corporation to exploit profitable opportunities in a volatile marketplace.” The leanness is basically to eliminate the waste with in the manufacturing to drive the lowest possible cost and highest quality of the product. Agility is to use the Voice of Customers (VOC) to develop new products to satisfy the demand, this is more flexible and high cost then leanness. “In lean production, the customer buys specific products, whereas in agile production the customer reserves capacity that may additionally need to be made available at very short notice” (Naylor, Towill and Christopher, 2000). Please see Table (1) for the use of lean and agile supply chain and Table (2) for differentiate the lean and agile supply chains. The tables developed by the author to demonstrate the difference, usage and benefits of Lean, Agile and Leagile supply chain paradigms. The table 1, 2 and 4 are influenced by the suggestions by the previous researchers such as Christopher, (2000), Towill, Christopher and Naylor (2000), Crocker & Emmett (2006), Naylor, Naim & Berry (1999) and the other literature found.Table: (1)Usage of Lean and AgileLean•Fluent Manufacturing•Zero inventory•Just in Time (JIT)•Remove waste•Vendor Managed Inventory (VMI)•Total Quality Management (TQM)•Economies of Scale (Low cost)•Commodities•Continuous, Line and High Batch production processAgile •Postponement•Collaborative scheduling•Just In Time (JIT)•Purchasing input capacity (PIC)•Supplier Trade off (Setup Vs Inventory)•House of Quality (HOQ)•Made to Order (High Cost)•Fashion Products•Integration of Micro and Macro environment•Project, Jobbing and low batch processSource: The Present AuthorTable: (2)Difference in Lean and AgileLean •Containing little fat•Product oriented•Reduce stock to minimum•Plan ahead•Satisfy customers by eliminating waste•Measuring output criteria: Quality, Cost and Delivery•Low Cost•Efficiency•Less flexible•Low varietyAgile•Nimble•Customer oriented•Reducing stock in not an issue •Unpredictable demand planning •Satisfy customers by configuring order•Measure output Criteria: Customer satisfaction•High Cost•Effectiveness•High flexible•High varietySource: The Present AuthorAs per the case study “Supply Fashion Fast” the fashion market is volatile and customer driven. Towill and Christopher (2002) suggested the market qualifier and winners in Lean and Agile supply chain (See Table 3). In Agile supply chain the market qualifiers are Quality, cost and lead time and the winner is who produce the high service level. But in Lean supply, the market qualifiers are Quality, Lead time and Service level and the winner is the cost. In addition; Naylor, Towill and Christopher (2000) suggested that agile supply chain is for fashion goods and lean supply chain is for commodities (See Table 3). Now the concept of integration of lean and agile paradigms is originated to capturing the advantage of lean and agile paradigms such as to maximize the efficiency and utilization of the operations and customization of high level of products. Christopher and Towill (2002) pointed that, “the lean concept works well where demand is relatively stable and hence predictable and where variety is low.” Furthermore “Agility is a business wide capability that embraces organisational structure, information systems, logistics process and in particular mind sets.”Table: (3)“Fashion products have a short life cycle and high demand uncertainty, therefore exposing the supply chain to the risks of both stock out and obsolescence. A good example of a fashion product is trendy clothing (Naylor, Towill and Christopher, 2000). To avoid degeneration and to fulfill the high demand uncertainty there is a need to combine the lean and agile to getting the best out of them.This combined approach is known as `Leagility’ and, as it is packed with the best outcomes of lean and agile. Resultant; the integration of lean and agile supply chains can thereby adopt a lean manufacturing approach upstream, enabling a level schedule and opening up an opportunity to drive down costs upstream while simultaneously still ensuring that downstream should have an agile response capable of delivering to an unpredictable marketplace. The need of integration or marring the lean and agile supply chain is to react effectively on a volatile demand while reducing waste and cost and improving quality and service level. Please see table (4) for benefits of ‘Leagile’ supply chain.Table: (4)Benefits of Leagile•Control & view inventory levels across a network•Manage orders between trading partners•Organise collaborative demand plans•Plan replenishment across an internal or external network•Enable Sales and Operation Planning•Monitor and Alert on significant events•Managing JIT approach•Managing Vendor Managed Inventory•Quick response to market•Achieve benefits of postponement•Standardisation of products•Converting voice of customers (VOC) into productsSource: The Present AuthorPractical ways of marring Lean and agile paradigmsThere are particularly three ways of marring lean and agile paradigms suggested by researchers such as, Pareto Curve approach, Decoupling Point and base and surge demand. These three ways of marring lean and agile can be used in any point of time and in any department, such as design, procurement, manufacturing etc. In a particular supply chain these approaches can be used frequently, such as Pareto 80/20 rules and separation of base & surge demand can be used in design, manufacturing, forecasting or while taking the critical decisions such as Standardisation of products, postponement decision etc. These approaches give flexibility to the process and enable to postpone the decisions and lower the inventory and most importantly minimizing the waste while optimizing the performance and quality. De-coupling point approach is the main idea to hold the inventory in shape of incomplete product shape and assemble the products instantly or in a shortest period on customers demand. The Dell computer is a well know example of decoupling approach practice. Practical implication of these approaches gives the benefit of integration of lean and agile supply chain. The practical ways of marring lean and agileprovide available and affordable products, (Christopher & Towill, 2001) instantly to the customers in a volatile demand such as Fashion.Figure (1): The Pareto Curve approachSource: Christopher and Towill (2001)In the late 1940s quality management guru Joseph M. Juran suggested the principle and named it after Italian economist Vilfredo Pareto, who observed that 80% of income in Italy went to 20% of the population. Pareto Analysis is a statistical technique in decision making that is used for the selection of a limited number of tasks that produce significant overall effect; stated Towill, Naylor, Jones (2000), Christopher, Towill (2001) Haughey, (2007). It uses the Pareto Principle; is also know as the 80/20 rule, the idea that by doing 20% of the work you can generate 80% of the benefit of doing the whole job (Haughey, 2007). This rule can be applied on almost anything such as 80% delays arise from 20% of causes, 20% of system defects caused 80% of problems (Towill, Nayloy, Jones, 2000). “The Pareto Principle has many applications in quality control. It is the basis for the Pareto diagram, one of the key tools used in total quality control and Six-Sigma”(Haughey, 2007). In figure (1) Christopher and Towill (2001) suggested that, 20% of theproducts are easily predictable and can be standardised and they lend themselves to lean manufacturing, furthermore the 80% of the products are in agile manufacturing because of less predictability, which require quick response to market”Decoupling pointThe further marring of lean and agile can be achieved by creating decoupling point; in a production process it is common to introduce decoupling points where production lead time is much longer then acceptable order lead time (Christopher and Towill, 2000). The decoupling point takes physical stock to achieve the advantage of different management and control tools to efficiently manage the both side (input & output) of the inventory (Velde and Meijer, 2007). The other side of decoupling point is the natural boundaries of organisations and departments with in the process (Christopher and Towill, 2001, Velde and Meijer, 2007). It is also the hub to meet the need and capability on either side of point. With in a supply chain there can be many numbers of decoupling points (Towill, Naylor and Jones, 2000). “A decoupling point divides the value chain into two distinct parts; one upstream with certain characteristics and one downstream with distinctly different characteristics”(Olhager, Selldin and Wikner, 2006). In figure (3) Christopher and Towill (2001) suggested that, “by utilising the concept of postponement companies may utilise lean method up to decoupling point and agile method beyond that.”Figure (2): The Decoupling PointSource: Christopher and Towill (2000)Separation of Base and Surge DemandSeparating demand patterns into “base” and “surge” elements is an employment of hybrid strategy. “Base demand can be forecast on the basis of past history whereby surge demand typically cannot. Base demand can be met through classic lean procedures to achieve economies of scale whereas surge demand is provided for through more flexible and probably higher cost, processes” stated (Christopher and Towill, 2001). Further Christopher and Towill pointed that; in fashion industry base demand can be sourced in low cost countries and surge demand to be topped up locally”. Base demand can be achieved by classical lean manufacturing with low cost and less flexibility and surge demand by agile with high cost and high flexibility.Figure (3): Responding to combinations of ``base'' and ``surge'' demandsSource: Christopher and Towill (2001)Case: United Colors of BenettonThe Benetton Group exists in 120 countries, with around 5000 stores and produce revenue of around 2 billions. According to the case study the group employees 300 designers and produces 110 million garments a year. The group owns most of the production units in Europe, North Africa, Eastern Europe, and Asia. 90% of the garments are being produced in the Europe and the group invested in highly automated warehouses, near main production centres and stores. Benetton’s stores sell mixed brands, such as the casual wear, fashion oriented products, leisure wear and street wear and the flash collections during the seasons. More then 20% of products are customised to the specific need of each country and reduced by 5-10 percent by standardising the products and strengthening the global brand image and reducing production cost.According to case study Benetton’s goals are to achieve expansion of sales network while minimizing the cost and increase the sales of fashion garments. In order to achieve these goals a higher degree of flexibility is require in the process. But its very hard to achieve flexibility, as the lead times are long; in respect retailers are required to purchase in advance, and the most of the purchase plans are depends upon the generalising the orders. For example; if Benetton needs to wait for a specific number of orders from retailers to buy the fabric in bulk and start manufacturing in order to minimise the cost, but resultant the process will increase the lead time of the finished product in store. See figure (4) for a traditional (lean) manufacturing process of garments.Figure (4): Traditional (Lean) manufacturing process of garmentsSource: The Present AuthorAccording to the case study Benetton the need of fashion industry is the quick response to the market. This requires a higher degree of flexibility in production and decision making. As per the corporate goals of the group, Benetton acquires the strategy of postponement and standardisation of the products. The benefit of the postponement is to enables Benetton to start manufacturing before color choices are made, to react on customer demand and suggestion and to delay the forecast of specific colors. Further more; the product and process standardisation benefits the Benetton with the lower setup cost, manufacturing before dying and give flexibility to produce only a subset of the products.Figure (5): Benetton’s manufacturing processSource: The AuthorIn figure (4) and (5) the manufacturing process is changed due to the dying finished products, in respect of the change in process the setup cost of manufacturing garments parts can be reduced further more the inventory level can also be reduced because the postponement of decision of dying the garments after manufacturing reduced the requirement if keeping much stock of different color of garments. Additionally; postponement is helping the Benetton to produce the fabric under lean manufacturing process while reducing and eliminating cost and waste. It also involves the flexibility to produce variety of colors in a short lead time. This also helped the Benetton to standardise the manufacturing process and further led to gain cost leadership and differentiation strategies. In the context; Dying unit is acting as a decoupling point where the lean manufacturing exists downstream of information flow and agility upstream.As per the case study The Benetton’s 90% of the production is based in the Europe and rest in low cost countries. Here the Pareto 80/20 rule can be applied because 90% of the production is based on to fulfill the surge demand, and the prompt actions can be made on the volatile demand. Reducing the number of customised products by the Benetton is also an attempt to increase the number of standardised products in order to achieve the lowest cost possible and make the product a global brand. The other reason is to gain the benefits of level scheduling of base and surge demand to ensure the usage of capacity.Hennes & Mauritz (H&M)As per the case study and H&M internet media; H&M collections are created and placed centrally in the design and buying department to find the good balance of threecomponents Fashion, Quality and the best Price. H&M is a customer focused company and employees more then 100 designers. A team of 500 people works together to built the range and putting together the colors, fabrics, garment types and theme and provide a feel for new season’s fashion. Furthermore; H&M do not own any manufacturing units, they have more then 700 suppliers in the Asia and Europe, but H&M owns the production offices working closely with the suppliers and ensuring the safety and quality of goods. H&M’s lead time varies 2 weeks to 6 months based on the item. The main transit point of goods is in the Hamburg and company got more then 1500 own stores.As per the company’s business concept Fashion, Price and Quality; H&M produce most of the garments outside Europe to achieve the benefits of leanness. They buy fabric in advance as per the forecast in order to minimise the cost (Li Li, 2007). The production offices situated with in the origin of production act as the second hub of information flow downstream and ensure the quality and the work standard of the suppliers. The other reason of placing production offices is to maximise the efficiency of supplier to achieve the lowest cost and zero defects in the products and minimise the lead time. The transit point in the Hamburg works as a decoupling point, while managing the flow of goods and information upstream and downstream. As H&M is a customer oriented company and learning from customers and serving the surge demand by production in the Europe (Li Li, 2007). The author is tried to develop a model of H&M supply chain to illustrate the particular ways of marriage of lean and agile. To illustrate in easiest way the author had put only one supplier in the Asia and one in Europe, to make it easier the inventory points, are not also explained (see figure 6).Figure (6): H&M SC Model.Source: The AuthorCase: ZaraAs per the case study; under the Zara model, the retail store is the eyes and ears of the company. Instead of relying solely on electronically collected data, Zara utilizes word-of-mouth information to understand more about their customers. Empowered store managers report to headquarters what real customers are saying. Products that are not selling well are quickly pulled and hot items quickly replenished. Their quick turn around on merchandise helps generate cash which eliminates the need for significant debt.Zara hires young designers and trains them to make quick decisions. Decision-making is encouraged and bad decisions are not severely punished. Designers are trained to limit the number of reviews and changes, speeding up the development process and minimizing the number of samples made.Figure: (7) Flow of information at ZaraSource: The Present AuthorAs per the literature available on Zara supply chain and the use of technology the author tried to develop the Figure (7). In the figure it is illustrated that the Zara supply chain starts from the retail stores and customers, the use and flow of information made Zara to convert the high degree of information into opportunity. The agility here is that the stores get feedback from customers and send the feedback to design team. Design team based on the fabric availability design the products by using the “Vanilla Box Design”. Thishelps to make computerised designs instead to waste money and time in making actual samples. Zara is using Pareto 80/20 rule while choosing the designs to send into production. The design team sends the information to cutting department and fabric department to ensure the right pattern is produced, here in production Zara is using the lean manufacturing in specialized factories while standardisation of cutting, stitching and dying process, pointed; Anderson, (2007) Machouca, Lewis and Ferdows, (2005). Un-dyed fabric is produced in advance with the help of long term forecast. Design teams make sure they will only design the garments keeping in mind the availability of specified fabric. The other advantage of integration of all the departments is gaining the benefit of postponement; Zara is dying the finished garments as per the customer’s reaction. Surge demand is managed by producing goods in Europe and base demand in other labor intensive countries (Machouca, Lewis and Ferdows, 2005).ConclusionThe need of supply fashion fast in the volatile demand; led companies such as, Zara, H&M & Benetton to make the changes in lean and agile process and integrate the both to achieve the benefits of lean and agile. The main motive to achieve the leagile is to react fasted on the changing demand. This requires a better control and view of inventory levels across the network, enable sales and replenishment planning across the internal and external network. With the help of IT, Zara achieved the control and monitoring the different event on the market, they are able to act on with the quick response to the market. Zara and Benetton both achieved the benefits of postponement. All there companies achieved the benefits of standardisation. Although; Zara, Benetton and H&M, took the different approach to marring the lean and agile but the overall purpose is the same; “Supply Fashion Fast” with lowest possible price and highest degree of quality.The Figures (4) & (5) Benetton; (6) H&M and (7) (Zara) is developed by the author with the help of the data found on the company website and based on articles and journals of Davanzo, Starr and Lewinski (2004);Machouca, Lewis and Ferdows, (2005); Anderson, (2007); Anderson and Lovejoy (2007); Li Li (2007) and Claburn (2007).ReferencesAnderson K., Lovejoy J.; (2007); The Speeding Bullet; Zara Apparel Supply Chain; March 2007; accessed 06th Dec. 2007; Source:/thelibrary/speeding.htmlAnderson K.; (2007); Fast Fashion Evolves; March 2007; accessed 06th Dec. 2007; Source: /thelibrary/speeding.htmlClaburn T; (2007); Math Whizzes Turbo-Charge an Online Retailer's Sales; 05th Oct. 2007; accessed: 06 Dec. 2007; Source:/info_centers/supply_chain/showArticle.jhtml?articl eID=202300213Christopher, M. and Towell, D.R. (2000): “Supply Chain migration from lean and functional to agile and customized”. Supply Chain Management, Vol. 5 – No. 4 – pp. 206-213.Christopher, M. and Towill, D. (2001), An integrated model for the design of agile supply chains, International Journal of Physical Distribution & Logistics Management , Vol. 31 No.4 , pp.235-246Christopher M. and Towill D; The Supply Chain Strategy Conundrum: To be Lean or Agile or to be Lean and Agile; International Journal of Logistics: Research and applications; Taylor & Francis Ltd; 2002; Vol. 5; No. 3; ISSN 1367-5567Davanzo R. L, Starr C.E and Lewinski H. V;(2004); Supply Chain and the Bottom Line: A Critical Link; Outlook Journal; Feb. 2004; accessed: 05th Dec. 2007; Source:/Global/Research_and_Insights/Outlook/By_Alphabet/Supply Link.htmGilmore D.; (2006); Time for New Supply Chain Icons; 12th Oct. 2006; accessed: 07th Dec. 2007; Source: /assets/FirstThoughts/06-10-12.cfm?cid=771&ctype=contentHaughey D; (2007); Pareto Analysis Step by Step; Accessed 09th Dec 2007; Source: /pareto-analysis-step-by-step.html?gclid=CLy52uKjnp ACFQ2WEgod2zbo7wLi Li: (2007); Fashion Magnates' Supply Chain Contest; 08th May 2007, Accessed 19th Nov. 2007; Source:/index.php?categoryid=Vm10YVlWVXhVbk5SYkVwUlZrUk JPUT09K1M=&p2_articleid=Vm10YWIyUXlTblJWYWs1UlZrUkJPUT09K1M=&p2_p age=2Mason-Jones, R and Towill, D.R. (1997): Information enrichment: designing the supply chain for competitive advantage. Supply Chain Management. Vol. 2, No. 4 – pp. 137-148Mason-Jones, R, Naylor, J.B. and Towill, D.R. (2000): Engineering the leagile supply chain. International Journal of Agile Management Systems. Vol. 2, Iss 1. pp.54Machouca J, Lewis M, Ferdows K; Zara's Secret for Fast Fashion; 21 Feb. 2005; Accessed 18 Nov. 2007; Source /archive/4652.htmlNaylor, J.B, Naim, M.M. and Berry, D. (1999), Leangility: interfacing the lean and agile manufacturing paradigm in the total supply chain, International Journal of Production Economics, Vol. 62, pp.107-18Olhager J, Selldin E, and Wikner J; (2006); Decoupling the value chain; the international journal of value chain management; Vol. 1, No. 1; abstract; Source:/search/index.php?action=record&rec_id=9021&prevQuery =&ps=10&m=orTowill D R., Naylor B. and Jones R M; Lean, agile or leagile? Matching your supply chain to the marketplace;International Journal of Production Research, 2000, VOL. 38, NO. 17, 4061- 4070; ISSN 0020-7543Velde L. N. J and Miejer B. R; (2007); A system approach to supply chain design with a multinational for colorant and coating; accessed 10th Dec. 2007; Source:/mcn/pdf_files/part6_5.pdfWalters D; Demand chain effectiveness – supply chain efficiencies; A role for enterprise information management; Journal of Enterprise Information Management; Volume 19 Number 3 2006 pp. 246-261。
5月10日
5月10日一、夜班1、掘进一区1317（3）材料道，不合格锚杆1根，对责任人谢云苏罚款2元。
2、掘进一区1317（5）材料道，不合格锚杆1根，对责任人汪家勇罚款2元。
早班1、掘进二区22109溜子道，不合格锚杆1根，对责任人盛宏建罚款2元。
中班1、掘进二区22109材料道，不合格锚杆1根，对责任人晁成杭罚款2元。
我们今天所做的一切，说到底都是为了我们的孩子；我们为之献身的事业，最后都要交给我们的孩子。
从这个意义上，我们领悟着胡锦涛同志关于在广大干部群众特别是青少年中进行社会主义荣辱观教育重要论述的深刻含义。
世界上无论大事业还是小事情，都要靠人做。
学会了做人才能学会做事，做好了人才能做好事。
事情做不好，可以从头来。
人做不好，很难重新来。
要让中国特色社会主义事业永葆活力，蓬勃发展，一件根本性的大事，就是要教育青少年一代树立是非分明的社会主义荣辱观，让他们从学会做人做起，学会做事，成就伟业。
青少年树立社会主义荣辱观，需要整个社会“言教”，更需要整个社会“身教”。
教师要为人师表。
“这是老师说的”，“老师要我们这样”，“老师不准我们这样”。
在少年儿童眼里，老师的形象是崇高的，老师的教导是神圣的，老师的话是他们辨是非、分荣辱的标准。
在广大青少年眼里，老师往往是他们做人的榜样，老师的一言一行，往往成为他们模仿的对象，对他们的一生都会产生深刻的影响。
广大教师一定要充分认识自己在培养青少年成长中肩负的光荣责任，要结合社会主义市场经济条件下，社会生活、校园生活、家庭生活的实际，针对当代青少年的特点和思想品德建设的实际，用富有创造性的鲜活形式，将“八荣八耻”引进课堂，作为思想品德教育的重要内容。
特别是广大教师要以身作则，以行育人，以德化人，不能在讲台上滔滔不绝，义愤填膺抨击各种丑恶现象，走下讲台却对各种不正之风熟视无睹甚至迎合追逐。
父母要为子楷模。
父母是孩子第一任老师，孩子往往是父母的影子。
孩子的更多时间是生活在家庭中，父母的人生观、价值观，无时无刻不在潜移默化地影响着孩子的成长。
实施精益生产过程中价值流图析方法的应用
实施精益生产过程中价值流图析办法的应用之邯郸勺丸创作（The Application of Value Stream Mapping inImplementingLean Production）中国汽车技术研究中心杜宏生要实施精益生产却无从下手,这是很多企业都会遇到的问题,所以了解自己价值流真正的状况,对于企业持续改良,实施精益生产来说是十分重要的.本文对价值流和价值流图析通过描绘顾客要求、产品物流、主要供给商和信息流来绘制企业的价值流现状图,基于精益生产中消除浪费思想发明改良的机会和关头过程进行持续改良,进而通过过程的办法实现设定的未来状态图的办法进行介绍,旨在为企业提供一种基于现场的实施精益生产的有效途径和办法.希望能帮忙企业在实施精益生产过程中更好的发明产生浪费的本源并消除之,以提高企业的竞争力.关头词：精益生产价值流价值流图析It is the problem of many enterprises that from where and how to implement Lean Production, so understanding our actualvalue stream is very important for a enterprise to improve continuously and implement Lean Production.This paper introduces the concept of value stream and the method of how to draw current state map of value stream by collect the demands of customer, draw the material flow , main supply and the information flow, find the chance and the key processes to improve continuously based on the thinking of eliminate waste in Lean Production, and realize the future state map of value stream with process method, thus can provide a effective way and method to implement Lean Production for enterprises based on the site. I hope this paper will give some helps for enterprises to find and eliminate roots of wastes better, increase the competitive ability of enterprises.Key words: Lean Production, Value Stream, Value Stream Mapping企业实施精益生产,就是要按照精益思维的原则,在组织、办理、供给链、产品开发和生产运作方面建立有效的生产方法,以消除所有不增加价值的浪费为目标,逐步改良进而最大限度地谋求经济效益和提高竞争力.企业在产品开发、生产制造、办理及办事顾客整个流程中实施精益生产所产生的巨大优势,已通过八十年代的丰田汽车公司、九十年代的戴尔公司以及其他一些企业的巨大成功,为世界企业界所公认.但令人遗憾的是,许多企业在导入精益生产理念和办法后,很少认真地对整个产品的价值流进行阐发,就很快进入了大规模的消除浪费活动,这些改良活动虽然可能改良了产品价值流的一小部分,使之流动得加倍顺畅,但是其它部分的问题仍会导致大量库存,最终的结果是没有降低成本,甚至有所增加.如果仅仅局部实现了精益,那么改良效果的持续性就会受到限制,不克不及实现如大野耐一所说的“在全过程中减少浪费”,这将会导致精益生产的实施无法进行下去.不合行业、不合企业的情况是千差万此外,我们在实施精益的过程中,经常会被企业杂乱无章的布景所迷惑,不知道从哪里、如何实施改良活动,会觉得改良活动无从下手.在这种情况下,就需要有一个有效的东西或办法,能够让我们找出浪费及其原因之所在,然后将其消除,这个东西就是价值流图析技术.价值流图析技术作为一个有效的东西,可以通过做图的办法,帮忙企业考虑整个产品价值流的流动,而不是只考虑孤立的过程,从而使企业能够对其整个价值流进行持续的、系统化的改良,提高企业的效益和在市场中的竞争能力.利用价值流图析技术,不但能够消除浪费,还可以消除产生浪费之本源,使之不至于卷土重来.价值流图析技术已为全球很多企业所接受和采取,并且对实施精益生产起到了良好的效果.一、价值流图析技术与办法在论述价值流图析技术之前,首先介绍什么是价值流.1.1 价值流所谓价值流,是当前产品通过其基本生产过程所要求的全部活动.这些活动包含给产品增加价值和不增加价值两部分,包含了从产品最基本的原资料阶段一直到产品交付顾客的全部过程,如一辆汽车的制造,包含了从顾客要求到概念设计、产品设计、样车制造、试验、定型、投产到交付后的使用、信息反响和回收过程,会经历很多车间、工场、公司,甚至可能经历过多个国家和地区.1.2 价值流图析技术价值流图析技术是帮忙你阐发整个价值流的一个强有力的东西,它可以使整个价值流——通常是缭乱庞杂的,变成可视的一张价值流现状图（如本文图1所市）,使得价值流中的问题显现出来,这样就可以应用各类精益技术将不增值的活动--即浪费消除.这种改良不但能够消除浪费,并且能够消除浪费之源,使之不至于卷土重来,从而提高企业的竞争力.从价值流的定义可以看出,价值流包含整个产品生命周期,地域规模可能包含若干个企业甚至国家和地区,所以做出产品的整个价值流的图析是极为庞杂的任务,但阐发价值流的基本办法是相同的.为了便利起见,这里我们主要讨论工场内的价值流.1.3价值流图析技术在进行价值流图析之前,需要先来明确实施图析的主要步调,如很多技术的实施一样,价值流图析也是一个过程,采取5W1H办法,即确定Why（为什么）,Who（谁做）,What（做什么）,When（何时做）,Where （在哪做）,和How（如何做）,前面我们已经说明了为什么做图析,其它步调具体为以下六方面的事宜：1) Who——确定谁来做需要一位了解产品价值流并且能推进其改良的人,这团体具有领导职责（价值流经理）,由他来领导一个小组进行价值流图析任务.2) What——确定做哪些产品的价值流图析通常我们首先依照80—20原则,对影响企业最大的产品进行图析.3) When——确定何时做应在实施改良之前对价值流进行阐发,以确定应首先改良哪些过程.4) Where——确定在哪里做毫无意义,在现场！只有在现场收集的数据才干真正反应价值流的状况.5) How——进行图析下面将以CAHC公司价值流图析为例,简单说明价值流图析的应用办法.二、价值流图析实际应用2.1 选定产品系列表中所示为CAHC公司2003年供货计划,按照80—20原则—影响大的产品或因素只占全部的20%,可以确定对公司影响较大的产品有8A和8B两种,他们均属于8系列产品,生产过程基底细同,所以确定图析该系列产品.2.2 绘制价值流现状图价值流现状图的绘制应依据以下几个步调进行：·了解并记录顾客的要求如图1中右上角顾客要求框所示,我们主要应了解顾客的需求量、种类、交付频次和要求等.·了解并画出工场内资料流图如图1中下部的线框所示,也是工场内的基本生产过程,我们将能够连续进行的过程列入一个框内,无法连续的,在两框之间用库存三角分隔.·收集并记录每个生产过程的数据（数据框）在每个过程线框下记录主要数据包含生产节拍、换型时间、操纵人数、有效任务时间、设备使用率、废品率等与过程改良有关的数据.注意,这些数据应是现场收集的,而不是某些资料记载的.·了解库存情况小组应对所有库存（包含线上在制品库存）进行盘点,然后记载库存三角下面.注意,是点数而非查帐！·原资料推销和交付的情况如图1种左上角所示,了解主要供给商的供货情况.·了解、记录顾客订货、生产计划、原资料订货过程的信息传递途经及信息,画出信息流如图1上部所示,暗示出顾客订单、资料订单和生产信息传递途径和办法·画出生产时间线和计算相关数据将库存数量依照顾客需求节拍转化为时间,与生产过程时间数据一同画在时间线上,求出生产过程时间占整个时间的百分比.画出的CAHC公司完整价值流现状图如图1所示.2.3 绘制未来价值流图绘制价值流图析现状图的意义就在于通过图析发明工场生产过程中存在的浪费,从图1我们可以看出,188秒的生产时间只占然整个周期时间23.6天的不到1%,大量的时间浪费在库存和等待上了.通过阐发原因,找出关头的浪费及其改良的计划并予以实施——即通过实施精益生产来消除产品价值流中的浪费.阐发现状的目的在于解决价值流中的问题,而这些问题要在未来状态图的制定中通过实施精益价值流来予以解决.2.3.1 发明浪费确定什么是浪费,要按照精益思维的第一个基来源根底则——“从顾客的角度而不是从某个公司、部分或机构的角度确定价值”来确定的.精益生产中把浪费分为两种：即—不增加价值但目前生产、开发等系统要求存在和不增加价值且可以立即消除.2.3.2 设计并实施精益的未来价值流的准则我们画出价值流现状图的目的就是要使当前生产状况所存在的浪用度画图和计算的方法充分显现出来,找出原因,采纳措施逐步完善.价值流图析未来状态图就是使得当前价值流酿成精益的价值流.那么,如何设计并实施精益的未来价值流呢？这里提供7个准则并作简单解释.准则1：按顾客节拍生产使得生产过程的节拍与交付顾客节拍坚持一致,实现准时化生产.准则2：尽可能地实现连续流动尽量消除和减少库存和等待,这样生产过程就可以连续进行.准则3：在无法实现连续流动的地方采取看板拉动办理对与节拍相差悬殊,种类单一的过程,如冲压和焊接这样的无法流动过程,采取看板办理.准则4：努力使得顾客的订单只发到一个过程包管信息的一致性.准则5：在价值流启动过程按时间均匀分派多品种产品的生产实现均衡生产.准则6：在价值流启动过程通过启动一个单位的任务来实现初始拉动这个拉动的“动力源”一定要来自顾客.准则7：在价值流启动过程上游工序形成每天能够制造各类零件的能力多品种、小批量的混流均衡生产,要求上游过程通过减少换型时间和生产批量来提高对下游过程变更的反响速度,这样可以尽可能地减小库存的在制品.图1 CAHC公司的现状图2.3.3 研究现状,找出差别并绘制未来价值流图未来状态图是我们进行精益转化的目标蓝图.依照精益思维和精益价值流的准则来阐发前面做出的现状图,使我们可以发明存在很多方面的浪费,以便我们去消除.阐发研究现状图的关头步调具体来讲分为8步（以CAHC公司为例）；·确定有效任务时间和顾客需求节拍按照顾客需求量18400/月和有效任务时间27600秒/天,可出节拍应为60秒.·确定发运过程是采取顾客拉动还是建立一个成品发运仓库按照公司实际情况,小组确定采取成品发运仓库·确定使用连续流动的过程将焊接、装配这些节拍相近的过程和为一个连续的过程（中间无库存）.·确定采取拉动系统的过程将冲压和后续过程设计成为一个拉动系统,引入看板办理.·确定生产需求传送到价值流的哪一个过程,即价值流启动过程确定发运过程为价值流启动过程.·确定如何在启动过程均衡生产通过需求变更传递到整个过程来实现均衡生产.·确定价值流启动量按照顾客发运的频次和发运方法,确定其动量为20件.·确定设计未来状态图时,为了实现精益价值流,必须改良的哪些过程要实现上述方法,应改良的过程有焊接和装配的整合、冲压与焊接过程的拉动系统、发运过程的拉动系统、信息传递过程的改良等……完成以上阐发任务后,也就可以画出价值流未来状态图.CAHC小组所绘未来状态图如下文图2所示.2.4 计划的制定、实施与效果评估要改良的过程已基本确定,实现过程改良可分为三步：·将未来价值流划分红几个分价值流循环再考虑要实施哪些改良·制定实施计划·评估量划实施的效果改良过程的实施采取PDCA的进程进行,该办法很多文献均有详细介绍,这里就不细说了.2.5 CAHC价值流图析实施中的问题和需注意的事项·首先图析技术是发明价值流过程关头浪费及其本源的办法,以避免不需要的改良活动（这也是一种浪费）,必须通过精益技术消除浪费,才干受到效果.·图析准备：培训十分需要,可以避免标的目的性的错误.·产品选择：在考虑产量同时还应顾及产品产值、利润及其对公司生存与成长等方面的影响.·价值流现状图的绘制：一定要尽可能在较短的时间完成现状图数据的收集.·不但在确定产品适应找出主要因素,在确定顾客要求、过程及其参数、供给商情况时同样要找出影响本公司的主要因素,避免影响图析的效果和后续改良的标的目的和目标.·顾客需求节拍应按照实际供货的历史加以修正.·顾客拉动和成品仓库发货拉动两种方法实际上都是由顾客需求拉动的,关头区别在于是否建立成品库存.在实施精益生产初期,建议采取成品发运仓库的方法,这样能够包管准时交付顾客.随着精益生产实施的深入,可以逐步减少成品库存,最终转酿成顾客拉动.2.6 价值流图析技术实施效果很多企业实施精益生产都遇到过这样的问题,从那里开始做,对全部产品还是某一个？生产过程还是仓库？是单元生产还是一个流？在改良过程中也不成避免的会遇到但某个过程改良了之后,与整条线配合不起来了,导致“还不如不改”的想法.这些问题都是企业在实际生产过程中肯定会遇到的问题.而价值流图析技术是避免这类问题的有效东西,他从企业整个价值流出发来阐发哪些过程需要改良、如何改良和改良到何种状态,这也是本文介绍该技术的底子出发点.从CAHC公司价值流现状图和未来状态图的比较我们可以看出,通过描绘、阐发现状所存在的问题,我们找到了改良点；通过未来状态图的绘制,我们确定了改良的目标,然后通过PDCA办法进行有效实施,达到改良的目的,提高企业的效率和竞争能力.就效果而言,比较现状和未来状态,明显提高了生产的效率和灵活性,更大程度的满足了顾客的要求：·加工时间与生产周期的比例由188秒/23.6天变成169秒/4.5天,降低了生产时间和生产周期,通过生产周期的降低使得对顾客订单反响时间提高了5倍以上,这样可以大量减少库存,而库存量的减少则意味着流动资金、财务成本、产品损耗、仓库办理成本……的降低,也就意味着利润的提高,试想销售额提高5倍谈何容易,而通过改良就可以实现.·人员的减少.焊接装配过程原需要4个作业人员,现在3个足矣.·避免盲目生产.生产信息由原来的多头信息改成由发运过程向前传递,避免由于不了解后道过程需求盲目生产带来的“过量生产”浪费,同时有效的满足顾客不竭变更的要求.·由这些变更带来的研发、试验、生产、办事等方面的改良需求,使得所有改良都能够为提高企业竞争力起到作用,而不是概略文章,还有人的思维变更、顾客满意度的提高……图2 价值流未来状态图三、价值流图析技术的现状和未来价值流图析法一经推出,就得到了汽车行业的一致认同,当年在德国举办的精益生产研讨会上受到了与会专家、学者和企业界人士的一致好评,目前美国、德国、英国等欧美很多企业都把这种办法作为企业价值链阐发和改良的有效东西.成功实施价值流图析的企业业绩证明,有效实施价值流图析技术,可以：·消除50%以上的浪费过程/步调·产品形成周期较当前价值流状况减少1/3以上·需求变更的幅度从30%左右降至5%左右·质量水平得以提高,因为质量问题可以得到及时发明并予以解决·由于实施小批量多品种,增加了运送的次数,这样物流成本会有所提高,但整个价值流的成本得到了大幅度的降低以上我们所介绍的价值流图析,是在工场内的使用的技术.我们知道,几乎所有产品都不克不及在一个工场内完成,其价值流一般都要经过若干个工场、供给商,甚至这些工场在不合的国家,广泛世界五大洲.那么要使产品价值流真正实现精益,仅对一个工场进行价值流图析虽然可以消除这个过程中的浪费,但要完全消除浪费,就要对整个价值流进行阐发.为此丹尼尔?琼斯和詹姆斯?沃麦克教授（?改动世界的机器?和?精益思维?的作者）编著了一本适用于整个价值流图析的书“Seeingthe Whole”.这本书在?价值流图析?（Learning to See）的基础上,将价值流从工场扩展到整个产品实现的价值流.扩展的价值流图析原理上与工场内价值流图析是一致的,区别在于：·每个工场作为一个过程来看待·较工场内价值流图析更加庞杂·物流更加庞杂·信息流也较工场内价值流更加庞杂·考虑的方面更加全面·可以消除更大的浪费但是从扩展的价值流图析我们也可以看出,它是建立在工场内价值流图析的基础上的,因为：·每个工场的过程直接影响到全价值流·扩展价值流图析中实现连续流动、拉动以及均衡生产建立在每个工场的实施基础之上·图析的原理、办法、步调基底细同与所有技术一样,价值流图析技术也是在不竭成长过程中,不但在图析规模方面有所成长,并且在图析内容上也有成长,这里通过图析阐发了物流和信息流,有关专家目前正在考虑资金流动的问题,相信不久会有新的办法出现；此外由于价值流图析是实施精益生产的一种有效东西,那么随着精益技术的成长,在价值流图析过程中也会有新的东西不竭弥补进来,使得价值流图析技术能够加倍有效地反应实际存在的问题,并采纳有力的措施不竭消除浪费,实现全过程精益化.当前精益生产已为企业所广泛接受,从最高办理者到每一位员工都愿努力消除各类浪费,提高企业的竞争力,但苦于无从下手,或实施以后并未达到预期的效果,而价值流图析技术正是帮忙人们有效实施精益生产的一个强有力的东西,它可以把精益思维的基来源根底则融于改良实施过程之中,帮忙企业进行系统化、持续化的改良,而这种改良不但能够消除浪费,并且能够消除产生浪费的本源,使其不至于卷土重来.实施价值流图析技术过程必须充分尊重人的因素,应相信员工能够充分地理解精益思维的理念,在形成精益价值流的过程中充分合作,改动原有的习惯,不竭发明浪费、消除浪费并从中获益.公司是最大的受益者,来自多个方面——竞争力的增加,利润的增加,信誉的提高以及内部员工之间的信任和团结……如精益思维第5项基来源根底则所言,精益生产要不竭追求完美,价值流图析过程中未来状态实现之日,也就是成为此时现状图之时,未来状态转变成现状的循环是没有尽头的,价值流的改良应成为任何企业日常办理活动之一,无论其产品是硬件、软件还是办事,也无论该企业的规模、水平如何……正如我们频频发明的那样,当你在一个循环中消除了一些浪费,在下一个循环还会发明更多的浪费等着你去消除,精益生产的实施就是这些活动的不竭循环.前面曾提到过,精益生产不但仅是汽车行业的生产方法,而是适用于所有的企业.对于任何企业,获得其价值流的方法都是相同的,即从你的公司上游向下游扩展,直到从“原资料的份子到交付用户的成品”的整个过程.那么价值流图析技术作为精益价值流实施的有效东西,将为每个企业提供强有力的支持和办事.哪里有顾客需要的产品,哪里就有价值流,关头是你如何发明它！纵不雅CAHC公司对8系列产品实施价值流图析的全过程,显示了值流图析技术在实施精益生产过程中所起到的巨大威力,同时也证明了价值流图析技术是一种科学有效的东西,它可以帮忙企业不竭发明在价值流中存在的浪费及其产生本源,进而运用精益东西完全消除这些浪费,最终实现企业效益和竞争能力的提高.本文针对CAHC公司价值流图析的研究的另一个目的,希望以此激发我国企业运用科学的办法,认真阐发自己的价值流,并致力于如何实现适合于你自己企业的精益价值流的精益实践之中.。
贾生元博客案例分析
12015年污水处理厂扩建案例解答根据朋友提供的题干，试着解答了一下，仅仅大家讨论。
不足或错误之处，请多批评！1、污泥干化间废气除臭方案是否合理，说明理由。
不合理。
（1）污泥干化产生的臭气温度（约60-65℃）远高于生物滤池除臭设施的适宜温度（22-30℃），而且H2S和NH3的浓度是其他产臭构筑物的8-10倍，不适宜生物滤池除臭。
（2）该车间设置了15m高的排气筒，未说明与其他2个车间以及原污水处理站进水泵房和污泥脱水机房的2个排气筒的位置及其必要性，涉嫌分摊排放量，应首先考虑合并为一根排气筒（或答：应根据其间距、半径200m范围内最高建筑物的高度情况考虑合并，考虑按等效排气筒考核其达标情况）。
（2015年版“案例分析-试题解析”P254页有类似考题及解答）2、本项目是否存在重大危险源，说明理由。
不构成重大危险源。
理由：（1）甲醇储量：16t×6=96（t），小于500t，不构成重大危险源。
（2）沼气量：0.970t×16=15.52（t），小于50t，不构成重大危险源。
（3）沼气罐区与甲醇加药间的距离280m，小于500m，按一个功能单元考虑，其（96/500+15.52/50）<1，也不构成重大危险源。
3、给出本项目大气环境质量现状监测因子。
SO2、NO2、TSP（或答PM10、PM2.5均可）、H2S、NH3、甲硫醇、臭气浓度、甲烷、甲醇。
（特别是臭气浓度、甲烷、甲醇一定要答上。
）4、指出环评机构恶臭影响评价结论存在的问题。
（1）“污泥处理中心3根排气筒对A村庄的恶臭污染物贡献值叠加后满足环境标准限值要求”错误。
（理由：须考虑其是否适用于等效排气筒的计算，而且不只是3根排气筒贡献值的叠加，还需与现状值叠加。
考虑设问情况，理由可以不答）。
（2）“本项目对A村庄的影响可以接受”的结论不可信。
（理由：“满足标准限值”的计算不正确，且与A村庄居民的反映不符。
考虑设问情况，理由可以不答）。
环评师考试--贾生元老师案例分析博客材料补充
贾生元老师案例分析博客材料补充关于2012年环评师案例分析第六题（水利）的一个问题2012年环评师考试结束后，应学员要求我做了部分生态题的答案供大家交流、讨论，只是我个人的答案，并不是标准答案。
但对于本题关于能满足各方面用水的分析比较，一些人认为应该与多年平均径流量相比，而我给出了应该与兴利库容相比。
这一说法引起一部分人的不满。
作为技术交流，有不同意见是十分正常的，我感谢这位网名为cobra提出质疑、叫板拍砖，说明他是经过思考的，不会盲从某个人的意见。
但我个人并不想与其“拍砖”和辩论，因为意义不大。
举个日常生活的例子，我们家里用水来自于水龙头的流出水（相当于径流量），但如果供水并不畅的话，一般家庭会备一个水桶或水缸（相当于水库）储水。
如果水龙头没有水，但水桶或水缸里有水，我们就仍然会喝到水；但如果水龙头没有水，且水桶或水缸里也没水，我们就面临着喝不上水；当然如果水桶或水缸里没有水，但水龙头却有水，那还是可以喝到水的，但你要放出来。
这也就是为何要修水库的最简单的道理，否则就不必修水库了。
多年平均径流量是设计水库库容，包括兴利库容的重要前提条件，但供水量却是与兴利库容有更直接的关系。
2012年电解铜案例题解答2012年考试结束后，没有考生就此题发表过解答要点。
近日本人试着答了一下。
做后发现此题难度不大，而且发现既往工业类考题的难度远比生态类的要小。
因此，建议大家不要轻易放弃做答工业类的考题。
对于本题的答案有错误或不妥之处，希望大家批评，并提出意见和建议。
某公司拟在工业园区建设电解铜箔项目，设计生产能力为8.0×103t/a，电解铜箔生产原料为高纯铜。
生产工序包括硫酸溶铜、电解生箔、表面处理、裁剪收卷。
其中表面处理工序工艺流程见图2-1。
表面处理工序粗化固化工段水平衡见图2-2。
工业园区建筑物高约10m-20m。
图2-1 表面处理工序工艺流程图图2-2 粗化固化工段水平衡图粗化固化工段废气经碱液喷淋洗涤后通过位于车间顶部的排气筒排放，排气筒距地面15m。
贾国宗版主工作日志选集2
贾国宗版主工作日志选集（二）！设置和习练一步功的重要意义！关于设置和习练一步功的重要意义，真学书籍中已经有所论述。
下面我就论坛功友们实践过程中提出的相关问题，作一些强调和补充。
个人认识，谨供大家参考！一、认识、理解和掌握真学的“呼吸形式”。
一步功强调“呼气注意心窝部”的“呼气注意”，从注意力度方面讲，也就是含有仅次于“注意呼气”的意思。
“注意呼气”是真学的精髓，“注意呼气”的实质机理和作用是生产“真气”。
“注意呼气”的基础和前提是自然呼吸，吸气任其自然。
这就是真学的“呼吸形式”，也是设置和习练一步功的重要意义之一。
这种“呼吸形式”在一步功里认识了、理解了、熟悉了、掌握了，才不会在三步功里出现“壮火食气”和“无火无气”的问题。
这种呼吸形式在开始阶段不很习惯，只要要领正确，需要有一个耐心适应的过程。
二、认识、理解和掌握“注意心窝部”的作用。
一步功强调“呼气注意心窝部”，心窝部就是“中丹田”。
“注意心窝部”的作用，就是集中经络里的真气到心窝部。
我们的身体里充满了无数的经络，与各个脏腑紧密相连，已经认知的“十二正经”“奇经八脉”就是真气运行最为明显的通道。
实践证明，真气在贯通任脉上的“中丹田”集中，最为敏感、最为有效、最为合理、最为安全。
为拓展中丹田到下丹田之间的任脉，做好思想、方法、实力准备，打下良好的基础。
同时，注意和“内视”心窝部也是排除杂念最有效的方法。
所谓注意和“内视”就是“全神贯注”，不可分心分神。
三、认识、理解和掌握一步功的“着重点”。
一步功强调“呼气注意心窝部”，就是相对的“有为功法”。
所谓相对的有为功法，就是“有所为、有所不为”。
“有所为”的重点和目的在心窝部，注意和“内视”是方法，着重体会心窝部的热、凉、闷、涨的感觉，有了这些感觉说明真气在中丹田积聚了，一步功目的达到了；至于真气从那里来、要到那里去，就要求大家“有所不为”了，避免“画蛇添足”，不要作其他任何延伸的想象，至于其他地方的某些反应和感觉就更不要管她。
能人贾福建
箍 蕤
：王 智龙 张 军红
贾 建 是 山东汉 予 ，浑 身充满 豪 情 １８ 年 １ ，７岁的他子 承父业 ， ９ １ ０月 １ 来列山 东中条山集 团篦 予沟矿业有限公
司， 到矿 Ｉ 与了 Ｉ ．他凭 首勤 奋好 学 。 ． 人
‘
没有他玩不转的设备
就是电灯泡损坏 ，． 明『题 成为行 车大 ｝ Ⅱ ｛ 《 Ｊ ｛
２ ８日，任 命贾福建为竖 井工段段长兼运 输一 队队长 。 贾福建从大 局出发 ， 愉快接 受了挑战 。
年度荣立运城市“ 二等功”
竖 井工 段 的 工 人们 说— —
隐患。他甩 掉逆 变器， 商接将两个 １７ ２ 伏 的支流电灯泡进行 串联 ，并 利用选矿 厂 废旧皮带做减振垫 ，中『挖 空装上防水 开 ］ 灯头 ， 既不考虑变流变压 ， 义不怕行车震 动 ，把一个复杂 的技术 问题就这么简单 的解决了。 他检查 没备， 眼特尖 。竖井 ３米卷扬
运 输 任 务 ， 誉 为矿 山 生 产 的 “ 动脉 ” 一 个 是 竖 井 被 主 ；
ｉ更 半夜 ｔ － － 岗 ， － 处有 人单 井 经盎 机、 脱岗串 岗的 ， ・ 律从重处 罚。 他到任 的第 一 个月Ｊ Ｔ资 ，仃个 Ｉ ： 职工 一看 自已名 下脯领工 资 为 １０ ０ ０多 元， 有点不敢相信 ｆ己的眼睛 ， ｌ 当他确信
的 ４ ５千 瓦电机 ， ７ 属于全 矿“ 心脏 ” 没 件
只有贾福建来竖井 。 我们才有希 望
２ ０ １ 月问 ， ０ ３ｑ １ 竖井工段的小少 职 工多次 ， 受求 贾福建到呸井 Ｔ段 当队长。
贾福 建说——
既然我来 了。 就要对得起大 家 在第 一次全段职Ｔ见面会 上，贾福 建激动地说 ：弟兄们 ，谢谢大家对我 的 “ 信任。既然 我来 了， 就要对得起大家。不 过 ，今天在这 里我也把话挑 明 ，从今往 后， 谁干活 ， 就给 准钱 ；＃ 诈 干活多 ， 就给谁 钱多； 谁的 活儿 干得好 ， 奖 ， 的活 儿 就 谁
请用右键目标另存为! 教库教育资源交流平台免费课件–教库 (共10张PPT)
贾府人物关系
雍正初年，由于封建统治阶级内部政治斗争的牵连，曹家遭受一系列打击，先后遭遇了革职、家产抄没、下狱、“枷号”等，一年有余。 善于阿谀奉承，深得贾母欢心，独揽贾府大权，成为贾府的实际统治者。 曹 雪 芹（1715—1763） 雍正初年，由于封建统治阶级内部政治斗争的牵连，曹家遭受一系列打击，先后遭遇了革职、家产抄没、下狱、“枷号”等，一年有余。 祖父曹寅做过康熙的伴读和御前侍卫，后任江宁织造，兼任两淮巡盐监察御使，极受康熙宠信，其子也先后继任江宁织造。 他蔑视世俗，卓然独立，反映了他对封建礼教和封建道德的 反抗。 善于阿谀奉承，深得贾母欢心，独揽贾府大权，成为贾府的实际统治者。 介绍贾、史、王、薛四大家族 是一个精明能干、惯于玩弄权术的人。 祖籍辽阳，先世原是汉族，后为满洲正白旗“包衣人”。 第五回：贾宝玉梦游太虚幻境 清代小说家，名 ，字梦沅，号雪芹，又号芹圃、芹溪。 1.
“步步留心，时时在意，不肯轻易多说一句话，多行一步路，惟恐被人耻笑了他去”——寄人篱下
善于阿谀奉承，深得贾母欢心，独揽贾府大权，成为贾府的实际统治者。
介绍贾府环境
“步步留心，时时在意，不
肯轻易多说一句话，多行一步路，
惟恐被人耻笑了他去”——寄人 篱下
王熙凤
是一个精明能干、惯于 玩弄权术的人。为人刁钻狡 黠，明是一盆火，暗是一把 刀。善于阿谀奉承，深得贾 母欢心，独揽贾府大权，成 为贾府的实际统治者。
2. 人ห้องสมุดไป่ตู้的出场，先后适宜，详 略得体，虚实兼用。
3. 人物描写与环境描写交插 进行，配合自然，相得益彰。
环境描写
宏伟的外观
讲究的布局
华贵的设施
人物分析
林黛玉：
为人刁钻狡黠，明是一盆火，暗是一把刀。 是一个精明能干、惯于玩弄权术的人。
贾生元案例重点
贾生元案例重点
一、水利水电
二、公路与铁路
2.1保护耕地的措施：收边坡，降路基；弃土场、服务区，连接线少占地；
造地复垦废弃段，耕作层防污染，临时用地恢复完；耕地保护要记全。
P.61
2.2生态等级：
2.3基本图件：
2.4声环境影响评价等级：1235含人多，一般性评价声二级。
P.63
2.5预测噪声影响时，需要的主要技术资料：工程技术资料、项目运行情况（车流情况）、
环境状况、敏感点参数，
2.6 地形地貌植，水系水源地，土地土壤侵3区，七图生调提。
过p.227
2.7 高速公路运营期突出的环境问题：阻隔景观，敏感点，桥跨水源有风险。
临时占地复重建。
贾p.73,过p.227
2.
贾P.77
2.9。
【最新】引导俞利君
引导：俞利君
1
2006年5月2日凌晨3时23分， “宝华”轮一船舱突然起火，请求 紧急救助！！
2021/2/2
茫茫大海，怎样才能 找到这艘起火的轮船呢？
2
你能向大家介绍你在 教室的位置吗？
2021/2/2
3
5
赵小燕 李娟 孔一 朱丹 王磊 葛小
4
王星 赵林 王小雪 李佳 周一剑 苏叶叶
2021/2/2
13
（2，1）
5
云山村
4
李家村
3
（1 ，2 ）
李家村
2
张家村
心田心农庄（4 ，2） 田
（3，4） （4，3） 心田农庄
1
花鼓村
0
1 2 34
5
警局
2021/2/2
14
山
草地
2021/2/2
出口
湖
15
要求：
1、以小组为单位，将各个游乐项 目贴在你们认为合适的位置上。
2、贴好游乐项目后，设计一条旅 游路线，由小组长填写。
3
余静 方一
王惠
陈月
施丽 张萍
2
（1王,刚2 ） 夏小玉 陈洋
李刚
许阳 王丽
1
李惠 （2李,可1） 王红
林平 叶欣
夏倩
1组 2组
2021/2/2
3组 4组 讲台
5组 6组
4
5
赵小燕 李娟
孔一 朱丹
王磊 葛小
4
王星 赵林 王小雪 李佳 周一剑 苏叶叶
3
余静
方一
王惠 陈月
施丽 张萍
2
王刚 夏小玉 陈洋 李刚 许阳 王丽
3、比一比，哪一组的设计师设计 得又快又合理。
死亡的典范
死亡的典范葛印卡老师Dr Tara Jadhav在1986年参加了首次的内观十日课程之后，就不再到处寻寻觅觅了。
因为她已经找到纯净的正法之道。
她无须再四处寻求任何其它修行方法或技巧了，她全心全意地奉献，开始在这条道上持续前进着。
由于Dr. Tara Jadhav已无其它世俗的牵挂，所以她大部分时间都精勤修持正法。
她已在这条正法之道上修持了好几世，累积了足够的波罗密，因此能够轻易地修习内观。
就如同水中的鱼，不用教它游泳，就能自在地游来游去。
同样地，Tara 也不需接受特别的指导，就能正确地修持。
她不仅有正确的修行技巧与适当的环境，而且全心投入，精进的修持。
由于她已具备了慈悲与无私奉献的特质，因此在1989年被指定为助理老师，1995年成为资深助理老师。
尽管已相当年迈，她依旧全心奉献于法的服务，并且从指导内观学员的过程中，不断增长布施波罗密。
她在八十二岁的高龄，参加了Dhamma Giri-法岗为指导老师所举办的自修课程。
这个课程从1996年12月2日早上的观呼吸法(anapana)开始，她终日专心地禅坐，晚间六至七点的共修时段，她仍在自己的小关房单独禅修，然后回到禅堂听开示。
大约在7：30开示进行不久之后，她双膝着地跪坐着，双手掌触地，俯身向前叩头礼敬。
一次，两次，第三次叩头后，她没有再起身，她在向正法礼敬中，结束了人生的最后一口气息。
她身旁的禅修者起初还纳闷着，因为通常都是开示结束后才会行三跪拜礼的。
她为什么要在开示的前段礼敬呢？原来这是她此生最后一次的礼敬。
而且在每一次礼敬时，她都轻声说着：「无常、无常、无常」(anicca, anicca, anicca)，这就是她最后的话。
所有资深的禅修者都非常了解礼敬不应该变成一种机械式的反应。
只有在对头顶的无常感受保持平等心时，礼敬才有意义，Tara总是以此了知无常的平等心向正法礼敬。
即使在最后一次礼敬也是如此。
Tara 总是对身旁禅修的人说：「我晚年唯一的愿望就是，我要在这块正法之地禅修时，舍离我这个色身。
贾建增元代元日朝会考
贾建增元代元日朝会考元代元日朝会考贾建增◇基本信息摘要：元日朝会是元代非常盛大的朝会活动。
元日朝会的各个环节，百官习仪，帝后升殿，百官待漏、进贺，礼物与贺表的进献，元日大宴等都有严密的制度规范。
地方官府在元日也举行贺正仪式，并遣使入都进献贺表。
元代元日朝会规模宏大，仪式庄严隆重，继承了汉族王朝元日朝会的基本形式，蒙古旧俗也大量融入其中，呈现出明显的蒙古统治王朝的特征。
元代元日朝会从礼仪层面彰显了元代君臣关系，有助于元朝蒙古统治者构建正统性。
作者简介：贾建增，1990年生，山东大学历史文化学院博士后。
文章原刊：《中国史研究》2021年第4期。
元日，又称元旦、元正、正旦，即农历正月初一。
自汉代起，历代王朝都在元日举行朝会仪式。
1唐制，“凡元日，大陈设于含元殿，服衮冕临轩，展宫悬之乐，陈历代宝玉舆辂，备黄麾杖，二王后及百官朝集使、皇亲，并朝服陪位”2。
宋制，“元正一岁之首……圣人重之，制为朝贺之礼焉”3。
辽、金等少数民族建立的政权同样在元日举行朝会仪式。
辽代元日朝会，“臣僚并诸国使昧爽入朝”,执行舞蹈、跪拜、鞠躬等礼节。
4金代元日，皇帝“御大安殿，受皇太子以下百官及外国使贺”5。
元朝是中国历史上第一个由北方游牧民族建立并统一全国的王朝。
大蒙古国时代，“朝会燕飨之礼，多从本俗”6。
忽必烈即位后，推行汉法，逐渐构建起元代的立国规模。
就朝仪而言，忽必烈命刘秉忠、尚文、史杠等主持制定朝仪，礼乐制度逐渐建立起来。
元代朝仪“颇采古礼，杂就金制”7,蒙古旧俗又杂糅其中，因而颇具特色。
关于元代元日朝会，前人研究多有涉及，但未对其细节与意义进行深入挖掘。
8本文拟从《元史·礼乐志》出发，钩稽典章制度与元人诗文材料的记载，以元日朝会的具体过程为重点，对元代元日朝会进行全面考察。
不足之处，请方家批评指正。
一元代元日朝会的举行朝会是蒙元时期诸王、贵戚、百官及藩属各国必须履行的义务。
“国朝凡大朝会，后妃、宗王、亲戚、大臣、将帅、百执事及四方朝附者咸在。
一个可修系统的最优更换模型
一个可修系统的最优更换模型
张元林;贾积身
【期刊名称】《应用数学》
【年(卷),期】1996(9)2
【摘要】本文考虑了单部件、一个修理工组成的可修系统，在故障系统不能“修复如新”的前提下，我们利用几何过程，以系统年龄Ｔ为策略，选择最优的Ｔ使得系统经长期运行单位时间的期望效益达到最大．本文还在一定的条件下证明了最优更换策略Ｔ的唯一存在，且求出了系统经长期运行单位时间的最大期望效益的明显表达式．
【总页数】5页(P180-184)
【关键词】更新过程;更新报酬定理;可修系统;最优更换模型
【作者】张元林;贾积身
【作者单位】东南大学数学力学系,河南机电专科学校
【正文语种】中文
【中图分类】O213.2
【相关文献】
1.服务台可修的排队系统最优更换模型 [J], 贾积身;任仲普
2.冷备可修系统最优更换策略下的数学模型 [J], 贾积身
3.一个可修排队系统的最优更换策略 [J], 张元林
4.两部件冷贮备可修系统的一个最优更换模型 [J], 贾积身;王东升;王秀梅
5.单部件可修系统的一个最优更换策略 [J], 贾积身
因版权原因，仅展示原文概要，查看原文内容请购买。
- 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)。
贾生元老师博客案例材料环评考试总结本人的一点儿小总结，仅供环境影响评价工师职业资格考试的应考朋友们在紧张的学习与复习休闲快乐一下，其中的内涵仅供参考．另外，转载请注明出处。
掌握熟悉和了解，了解透彻易通过；导则标准和方法，出题基础莫轻看；勿言出题偏大纲，其实就在书里边；认真读书三回合，不信当年通不过。
现状调查与评价，方法内容和结果；珍稀种，敏感区，时时刻刻要牢记；影响分析多定向，答题不必扯太多；保护措施是考点，往往要求很具体。
敏感目标要熟记，各个要素均涉及；保护区，要绕避，核心缓冲惹不起；实验区，可特办，不得新增污染源；未分区，虽少见，按照核缓更严厉。
水源地，关生命，风险评价列第一；人在世，需吃饭，没有耕地胃不干；有害物，毒害深，考试主考不正常；改扩建，三本帐，以新代老结旧帐。
验收调查考规范，工况范围和措施；监测布点是关键，影响目标落措施；专题调查敏感点，总量清洁和风险；监测审核未考过，也许以后会出现。
挖煤采矿主考题，生态污染兼俱备；洗选厂，排废水，净化利用莫外排；矿井水，先净化，选电井生还绿化矸石山，废石渣，堆置利用填地下；尾矿库，易溃坝，防尘排洪加绿化；采空区，易塌陷，防护煤柱或搬迁。
油气田，战略源，滚动开发要回顾；三期源项各不同，施工营运和封井；保护区、水源区，一般严禁去开采；实验区，可开采，主管同意调范围；废水废气废泥浆；处理利用有措施；油气都是燃爆品，风险评价要牢记。
水电站，规划先，梯级开发要协调；大坝挡，上下断，都有影响不一般；陆生态，水生态，水质水文又地质；水文情势变化大，影响洄游和三场；库区大，被淹没，移民安置老问题；壤侵蚀，物种移，水土保持硬措施；大坝下，或断流，生态问题数一数；低温水，小流量，过鱼设施切莫忘。
公路铁路与管线，现状调查分区段；如想穿越敏感区，比选方案要考虑；不管能否通过去，环评需先办手续；施工期，影响大，临时工程要弄清；取土场、弃土场，如不恢复土流失；高路堤，深路堑，土石方量会增多；降堤收坡低路基，桥隧代路可考虑；动物行人要通行，留设通道安全过；服务区，有污染，废气污水加油站；时常考个小风险，往往涉及水源地。
平常心，情淡定，仔细看书莫走神；题到手，勿急燥，先看问题析题干；答题时，莫宽泛，原则性的要少谈；简单题，莫得意，一不小心就失意；复杂题，莫慌神，也许玄机在其中；采分点，并不多，一点一条写具体。
第一水电案例附2008年考题一道，参考答案为本人所做，欢迎看过的人提意见：（2008年考题）南方某山区拟建水资源综合利用开发工程，开发目的包括城镇供水，农业灌溉和发电。
工程包括一座库容为2.4×109m3的水库，引水工程和水电站。
水库大坝高54m，水库回水长度27km，水电站装机容量80 MW，水库需淹没耕地230hm2，移民1870人，安置方式拟采用就地后靠，农业灌溉引水主干渠长30km，灌溉面积6×104 hm2，城镇供水范围主要为下游地区的2个县城。
库区及上游地区土地利用类型：林地，耕地，分有自然村落，无城镇、工矿业，河流在水库坝址下游50km河段内有经济鱼类、土著鱼类的索饵场、产卵场。
第一组问题（原题）：1.大坝上游陆域生态环境现状调查应包括那些方面？（1）调查上游陆域大坝蓄水淹没区及影响范围内涉及的物种、种群和生态系统。
（2）重点调查陆域范围内有无受保护的珍稀濒危物种、关键种、土著种和特有种，天然的经济物种及自然保护区等。
如涉及国家级和省级保护生物、珍稀濒危生物和地方特有物种时，应逐个或逐类调查说明其类型、分布、保护级别、保护状况与保护要求等。
（3）植被调查可设置样方，调查植被组成、分层现象、优势种、频率、密度、生物量等指标。
（4）动物调查，应调查动物种类、分布、食源、水源、庇护所、繁殖所及领地范围，生理生殖特性，移动迁徙等活动规律。
（5）调查生态系统的类型、结构、功能和演变过程。
（6）调查相关的非生物因子特征（如气候、土壤、地形地貌、水文及水文地质等）。
（生态现状调查是生态影响评价的基础。
一般从事生态影响评价的技术人员均容易解答该问题，但往往回答不完整。
陆地生态环境现状调查不外乎植物、动物及其生境调查，在具体工作中从物种、种群、生态系统、景观四个层次入手，同时考虑气候、土壤、水系等自然因素。
在复习这样的题时，同时可以联想一下对于下游地区的生态应调查哪些内容？工程建设对上游和下游生态有哪些方面的影响？可注意一下本书的有关内容。
）2.提出水库运行期对下游河段鱼类的主要影响因素。
（1）在工程运行期，由于大坝阻隔，对上下游鱼类物种交流有一定阻碍。
（2）本项目大坝下游有鱼类的索饵、产卵场，被大坝截在上游的鱼类进入索饵和产卵场成为困难，而水库不同方式的放水对下游河道的水生生态影响明显。
（3）水文情势的变化导致饵料生物变化无常，水温、流量、流速、水位、含沙量的变化对鱼类产卵也有不利的影响。
（4）河道中鱼类的种群结构将发生变化、急流性或深水性鱼类将对理游水位变浅不适应。
（水库运行期对上游河段和下游河段的鱼类均有影响，但对下游影响更明显一些。
这类问题提出的方式有多种多样，有问大坝建设对下游鱼类的何影响？对坝上鱼类有何影响？对上游鱼类的三场有何影响？对下游鱼类的三场有何影响？基本大同小异，根据题干提供的信息及提问的形式，结合实际工作来答。
）3、对上游陆生野生动物有哪些影响？（1）施工期大量的施工人员活动和施工机械噪声干扰，以及库区拆迁、清理等，原来两岸活动的动物受到惊扰而发生逃逸。
（2）蓄水营运期，由于水库面积的扩大将造成野生动物生境的淹没、水面扩大影响陆生动物跨河迁移受到阻隔，动物生境有可能缩小。
（以生态影响为主的建设项目，一般可按施工期和营运期两个方面进行影响分析与评价，水力发电站项目施工期和营运期对生态的影响均较为明显。
当然也可以考虑施工前期、营运后期等的影响。
）4.针对工程移民安置，环评需考虑那些环境影响？（1）移民区遗留的环境影响。
遗留的固体废物，建筑残物、生活垃圾，特别是养殖垃圾如不能得到彻底清理，对库区蓄水将造成不利影响；若原来的陡坡开垦的农田不能及地退耕还林还草，其水土流失也将影响库区水质与水位。
（2）移民安置区的环境影响。
对安置区环境容量的影响，增加了安置区的环境与资源压力，移民安置占地及土地开发植被变化对生态的不利影响，农药、化肥面源污染，水土流失等。
（3）移民安置不当可能产生的不利影响。
一是有可能造成移民返迁，加剧库区生态破坏；二是陡坡开垦，使水土流失加剧，甚至诱发地质灾害。
（4）人体健康与民族文化多样性的影响。
（移民安置是水电开发项目关注的重要方面，当然也是考试应该考问的要点。
这个题也可以按环境影响要素来解答。
）5.提出对水库工程需要考虑的环境保护工程措施和环境管理计划。
（1）应主要采取以下环境保护工程措施：①库区移民、库底清理措施，保障库区水质。
②涉及珍稀保护植物的移植、引种栽培、工程防护等措施；涉及保护动物的栖息地保护、建立新的栖息地等措施。
③防治水土流失的护坡、拦挡、导排洪水的排水沟、导水槽等工程措施。
④保障洄游性鱼类的正常洄游的过鱼设施、人工增殖放流，如有竹木流放，则需要设置流放竹木的设施。
解决低温水影响的能够分层取水的措施。
（2）环境管理计划应包括以下内容：①成立环境管理机构、制定管理办法、落实人员与经费，实施施工期的环境监理；②开展库区水质定期监测、生态调查，特别是针对工程对鱼类“三场”的影响及变化情况应进行调查；③严格分层取水、定期放水、保障生态需水量的日常管理；④制定环境风险应急预案，并定期进行演练；⑤建立健全环境管理档案，如实记录工程建设过程及其环境影响与变化过程，环境调查成果及科学研究成果；⑥在工程竣工后落实竣工验收，并在稳定运行后3-5年内进行环境影响后评价。
（本题目也可以单独提出工程措施或环境管理计划来要求回答。
在单独的提问的情况下，要进一步细化答案。
水电项目也应制定环境风险应急预案，本项目有供水功能，而且是向下游两个城镇供水，在机电检修时有可能发生漏油事故；在洪水季节，需注意大坝安全和排洪，也存在安全风险，进而带来环境风险。
本工程兼有供水功能，涉及经济鱼类“三场”，工程规模较大，生态影响应为一级评价，因此，也需要在施工期进行环境监理。
）第二组问题（本人补充）1.针对河流鱼类，工程建设及营运期应采取哪些保护措施？充分调查和研究鱼类的生理生态特性，在大坝设置过鱼通道，根据河道生态及鱼类生长繁殖的要求分层放水，既保证一定的生态流量又尽可能创造较适宜的水文水力条件。
同时，要保证鱼类饵料生物的生活条件。
（在回答这个问题时要注意三个方面：一是鱼类如何通过大坝的问题，二是鱼类生活所需水量、水质及水文水力条件，三是鱼类饵料问题。
）2.大坝建成后，坝下减脱水段将发生哪些变化？（1）先期河床裸露，之后水生生态逐渐向半水生或陆生生态转化；（2）原河道中鱼类及水生生物有可能受到大幅度的消减；（3）农田灌溉用水不足或受冷水灌溉的影响而产量减少。
（引水式电站，其坝下减脱水段生态变化明显，是大家比较关注的。
）3.大坝建成后对鱼类的产卵的影响因素有哪些？（1）大坝建设有可能阻断的通道，使其不能到达产卵场；（2）由于流量、流速、泥沙含量、水温等水文情势的变化，对鱼类产卵将产生不利的影响；（3）生境的变化可能会推迟亲鱼的产卵期、或不能正常孵育、或使产卵后成活率降低。
4.工程建设对上游2个城镇供水有无不利影响？（1）有不利影响。
（2）主要表现为蓄水初期，由于淹没侵入大量的污染物，库区水质较差。
如果营运期不加强流域生态保护，水土流失加剧，库区发生富营养化或流域内开发建设一些污染型或对生态影响较大的项目，也会对城镇供水造成不利影响。
5.30公里农灌水渠施工生态影响主要有哪些？应采取哪些措施？（1）植被破坏、土壤扰动与破坏、水土流失、对野生动物也有不利的干扰影响。
（2）合理选址，避开植被丰富区；控制施工临时用地，尽可能减少对植被的破坏；弃方及时运走，但不能弃置在河道内，在弃土场采取设置挡墙、导排水沟、绿化等有效的水土保持措施；植被临时被毁区在工程结束时及时予以恢复。
第三组问题（本人补充）：1.本工程有为城镇供水的功能，请提出保障城镇供水水质的环保措施？（1）划分水源保护区；（2）控制流域人为开发建设活动，严禁在保护范围内建设污染型项目；（3）大力进行流域绿化，控制水土流失；（4）控制面源污染；防止库区富营养化；（5）加强库区水质监测与日常监督管理；（6）制订环境风险应急预案，一旦发生污染事件，立即启动应急预案。
2.指出水坝对上游农业的影响。
（1）淹没造成农田面积的大量损失，农田及移民区污染物大量进入水库，造成库区水质恶化；（2）农灌用水与城镇供水产生矛盾，需要进一步协调；（3）取水口在库区取深层低温冷水灌溉对农业生产有不利影响。