欢迎来到天天文库
浏览记录
ID:23958073
大小:9.45 MB
页数:79页
时间:2018-11-12
《深度自编码网络在滚动轴承故障诊断中的应用研究》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、UNIVERSITYOFELECTRONICSCIENCEANDTECHNOLOGYOFCHINAMASTERTHESIS2015210802111UDC2018.03.282018.05.11201861UDCResearchonFaultDiagnosisofRollingBearingBasedontheDeepAuto-EncoderNetworkAMasterThesisSubmittedtoUniversityofElectronicScienceandTechnologyofChin
2、aDiscipline:PrecisionInstrumentandmachineryAuthor:YangWuSupervisor:A.Prof.KeshengWangSchool:SchoolofMechanicalandElectricalEngineeringDeepAuto-Encoder,DAE1DAEDAEDAEPCAKPCADAE2DAE3DAEBPBackPropagationSVMSupportVectorMachineDAE4DAEDAEDAEDAEIABSTRACTABST
3、RACTRollingbearingisoneofthemostimportantpartsinmechanicalequipment.Itsscontinuousandreliableoperation.Oncetherollingbearingbreaksdown,itwillcauseeconomiclossesorevenadisaster.Therefore,theresearchofrollingbearingfaultdiagnosistechnologyisofgreatsigni
4、ficance.Thefaultdiagnosisprocessisgenerallydividedintofoursteps:faultdatacollection,featureextraction,featureselectionandfaultidentification.Faultfeatureextractionisakeystepinthediagnosisprocess.Thisstepisgenerallysolvedbysignalprocessingmethods,which
5、stronglydependonprofessionalknowledgeandtechnicalpersonnel.TheriseofArtificialIntelligencetechniqueshasopenedupanewworldforrollingbearingfaultdiagnosis.Deeplearningisanemergingforceinthefieldofartificialintelligence.Itcanautomaticallylearntheintrinsic
6、characteristicsfromthedatatoobtainagoodfeatureexpression,whichprovidesnewideasforfaultfeatureextractionanddiagnosisofrollingbearing.Thispaperfocusesontheapplicationofadeepauto-encodernetworkinthefaultdiagnosisofrollingbearings.Themainresearchcontentsa
7、reasfollows:Basedonthedetailedanalysisoftheprincipleofdeepauto-encodernetwork,itsperformanceoffeatureextractionandclassificationabilityisstudiedfromdatareconstructionandclassificationrate.DAEreconstructionnetworkissetupandtheperformanceoffeatureextrac
8、tionisinitiallyverifiedfromtheperspectiveofdatareconstruction.AnexternalstudyofcomparisoninsignalclassificationwithtraditionalfeatureextractionmethodsthatarePCAandKPCAsperformance.DAEdiagnosismodelisconstructed.Astothedecisionofthekeyparameter
此文档下载收益归作者所有