欢迎来到天天文库
浏览记录
ID:44051201
大小:767.24 KB
页数:51页
时间:2019-10-18
《基于选择性神经网络的故障诊断分析》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、ABSTRACTSincemodernindustrialproductionsystemshavebeenincreasinglylarger,morecomplexandautomated,onceanaccidenttakesplace,itwillbringhugeeconomiclossesandcasualtiesfortheenterprise.Therefore,howtoimprovethefaultdiagnosistechnologytoensurethenormaloperationofthepr
2、oductionsystemhasbecomeaseriousproblem.Firstofall,thispaperpresentstheoverallframeworkoffaultdiagnosismethodbasedonselectiveneuralnetwork.Thenanalysisthecausesandtheprocessbyusingselectiveneuralnetworkinfaultdiagnosisarea,andpointsouttwokeytechnologiesofthisdiagn
3、osticsystem:oneishowtobuildselectiveneuralnetworkensemble,theotheroneishowtoreducethefaultdatadimensions.Fortheproblemofhowtoconstructselectiveneuralnetworkensemble,thispaperproposesanimprovedbinaryparticleswarmoptimizationmethod(IBPSO),andthismethodisusedtochoos
4、eindividualneuralnetworks,thenestablishesthemodelofselectiveneuralnetworkensemblebasedonIBPSO・Theresultswhichcomparedwithbinaryparticleswarmalgorithm,K-meansclusteringalgorithmandantcolonyalgorithminstandarddatasetsshowthatthegeneralizationperformanceofthepropose
5、dmodelisimproved.Aimingattheproblemofmultiplefaultfeaturesinmodernindustrialproduction,Thekernelprincipalcomponentanalysismethodisintroduced.AnewfaultdiagnosismodelbasedonKPCAandselectiveneuralnetworkensembleisestablishedbybindingthesetwomethods.Theproposedmodeli
6、stestedthroughthesimulationplatformofTennessee・Eastmanprocessandtheactualcaseofsteelworksairsupplysystem.Theoutcomeshowsthatthisalgorithmhasahighercorrectrate,shorterdiagnosistimethanotherneuralnetworksmethods.Finally,aprototypesystemoffaultdiagnosisisdevelopedba
7、sedonselectiveneuralnetworkensemble.Keywords:FaultDiagnosisImprovedBinaryParticleSwarmOptimizationAlgorithmSelectiveNeuralNetworkEnsembleKernelPrincipalComponentAnalysis目录摘要IABSTRACTII1绪论1.1课题来源(1)1.2研究背景和意义(1)1.3故障诊断技术概述(2)1.4神经网络和主元分析方法在故障诊断中的应用概述(6)1.5论文研究内容和组
8、织结构(8)2基于选择性神经网络集成的故障诊断方法的总体框架2.1故障诊断问题描述(10)2.2基于选择性神经网络的故障诊断总体流程(11)2.3故障诊断的测试实例:FH纳西■伊斯曼过程(13)2.4本章小结(15)3基于粒子群算法的选择性神经网络(IBPSOSEN)建模3.1选择性神经网络模型概述(17)3.2改进
此文档下载收益归作者所有