资源描述：
《神经网络优化模型在大坝变形预报中的应用分析》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、AbstractDamisabuildingwhichismainlybuiltincomplexhydrology、engineeringgeologicalconditionsandbearshugeload.Havingareal-timemonitoring,usingalargeamountofdeformationobservationdata,analyzingandaccessingitssafetystatethenmakingaforecasttoitaretheimportantmeanstoensureitssafety.Meanwhile,sin
2、ceaffectedbysomanyfactors,suchaswaterpressure,upliftpressure,temperature,time-dependentandmanyuncertainties,moreover,withstrongrandomnessandcomplexmutualrelationshipofthesefactors,whichmakeitdifficulttodescribetheexactlyquantitativerelationshipbetweenthesefactorsandDamdisplacementwhenusin
3、gthetraditionalmathematicalmodelwhoisfoundontheconditionsofindependentobservationsandzeroexpectedvalue.Especiallywhenhavinglessobservationdataorlargeobservationnoiseerror,thetraditionaldeformationanalysismodelsareusuallylimited.Therefore,studyingthefusionofmultidisciplineknowledgeandtechn
4、ologymethodandestablishingasuitablecombinationofdeformationanalysisandpredictionmodeltoanalyzethedeformationtrendbecomeanimportantsubject.Startedfromthispoint,thepaperintroducesNeuralnetworktheory,Geneticalgorithm，Particleswarmoptimizationalgorithmanditsimprovedalgorithm,thenstudiesthefea
5、sibilityofthesemixedintelligentmodels.Combinedwithconcretedamengineering,thispaperhasanapplicationandcomparisonofthesemixedintelligentmodels.Themainresearchcontentsareasfollows:1)TohaveastudyoftheNeuralnetwork,Geneticalgorithm,PSOalgorithm,basedonthefactthatNeuralnetworkhasastrongrandomne
6、sswheninitialization,slowconvergencerate,easytofallintolocalminimum,thepaperusesGeneticalgorithmandPSOalgorithmtooptimizetheweightsandthresholdsbetweenneuronsandhasanonlinearimprovementoftheweightofthestandardPSOalgorithm.2)UsingvariableselectionmethodbasedonANNtocalculateallfactors’contr
7、ibutionrate,accordingtoit,wemakeachoiceoftheseimpactfactorsanddeterminethefinalfactorsofthedam,thennormalizeallobservationdataandtheseimpactfactors.3)BuildingdamdeformationpredictionmodelbasedonGA,PSOandImprovedPSOalgorithmrespectively.BYprogramingcorrespondingmodel
