集成智能故障诊断策略及其应用研究

《集成智能故障诊断策略及其应用研究》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、浙江大学博士学位论文集成智能故障诊断策略及其应用研究姓名：武和雷申请学位级别：博士专业：控制理论与控制工程指导教师：朱善安20031201垒星!!坠竺!——AbstractBasedontwofaultdiagnosisinstrumentsystemsforrollingbearingandcar,severaldiagnosisstrategiesusingintegratedintelligencehavebeenbroughtforward．Appliedmethodsthatfaultfeatureisextracte

2、dfromhighfrequencyresonancesignalhavealsobeenpresented．Manyexperimentresultshaveshowntheeffectivenessofthesemethods．Inchapter1，firstly,themeaningoftheresearchisintroducedbriefly,andthedevelopment，keyproblems，andapplicationprospectsonthefaultdiagnosistechnologyarestat

3、edindetail．Secondly,researchprogress，existingproblems，andthedevelopmenttrendonfeatureextractionandintelligentdiagnosisaresummarized．Finally,structureofthedissertationandthemainfruitsoftheresearcharepresented．Inchapter2，energyseparationalgorithm(ESA)isdiscussedindetai

4、l．AMFMmodelofbearingvibrationsignalisconstructedafteranalyzingthevibrationmechanism．FeasibilityofenvelopeanalysisinhighfrequencyresonancesignalwitlltheESAisdemonstrated．TheappliedofdemodulationwitllESAisadvancedwitllwhichdiagnosisprecisionhasbeenimproved，andrealtimed

5、iagnosisandaccuracyratioaresatisfiedonbearingproductionline．AutomaticclassificationmodelusingintegrationofAdaptivewavelettransformnetworkandSelf-OrganizingFeatureM印Networkisintroduced．Thestrongsuitsofwaveletandneuralnetworkaremadeuseofinthismodelandbearingfaultclassi

6、ficationisrealizedautomatically．TheintegratedmethodoftimeandfrequencydomainanalysisandRBFNNisgivenforbearingfaultdiagnosis．Throughselectingrepresentativeparameters，featurespacevectorisconstructedSOtheinputdimensionisreducedandfaultclassificationinformationretained．Th

7、ismethodprovidesareferenceforintellectualizedbearingfaultdiagnosis．Faultdiagnosismethodswimfuzzyinference，rulesinference，case-basedinferenceandcombinationoffuzzyinferenceandrulesinferenceareanalyzedbriefiy．Asthesemethodsexistshortages，advanceddiagnosisstrategyisprese

8、nted，whichcombinesthefuzzyneuralnetworkinsteadoffuzzyinferencetorulesinference．Theproblemofaccuracyratiodeclinebecauseofroughnessof