基于小波包-AR谱的变速器轴承故障特征提取.docx

基于小波包-AR谱的变速器轴承故障特征提取
Abstract
Inthispaper,afaultdiagnosismethodbasedonwaveletpacket-ARspectrumforgearboxbearingsisproposed.Themethodfirstdecomposesthevibrationsignalintodifferentfrequencybandsusingwaveletpacketdecomposition,andthenextractstheautoregressive(AR)spectrogramfromeachbandtoobtainthecharacteristicinformationofthefault.Inordertoverifytheeffectivenessofthemethod,experimentswerecarriedoutonagearboxtestbenchwithbearingfaults.Theresultsshowedthattheproposedmethodcaneffectivelyextractthecharacteristicinformationofthefaultandaccuratelydiagnosethefaulttype.
Introduction
Gearboxbearingsareimportantcomponentsofmechanicalsystemsandplayakeyroleintransmittingtorqueandsupportingtheload.However,duetotheharshworkingenvironmentandlargeworkingload,thebearingsarepronetovariousfaultssuchaswear,pitting,andcracking,whichseriouslyaffectthenormaloperationofthegearboxandevencausecatastrophicaccidents.
Therefore,thefaultdiagnosisofgearboxbearingshasalwaysbeenahotresearchtopicinthefieldofmechanicalfaultdiagnosis.Withthedevelopmentofsignalprocessingtechnology,manyfaultdiagnosismethodsbasedonvibrationsignalshavebeenproposed,suchasfrequencyanalysis,time-frequencyanalysis,andstatisticalanalysis.Amongthem,waveletpacketdecompositionisacommonlyusedmethodintime-frequencyanalysis,whichhastheadvantagesofmulti-resolutionandmulti-scaledecomposition.
Autoregressive(AR)modelisawidelyusedmethodforsignalanalysis,whichcaneffectivelymodelthespectralcharacteristicsofthesignal.ARspectrogramisobtainedbyapplyingARmodeltothepowerspectrumofthesignal,whichcanextractthecharacteristicfrequencyinformationofthesignal.
Inthispaper,afaultdiagnosismethodbasedonwaveletpacket-ARspectrumforgearboxbearingsisproposed.Thevibrationsignalisdecomposedintodifferentfrequencybandsusingwaveletpacketdecomposition,andtheARspectrogramisextractedfromeachbandtoobtainthecharacteristicinformationofthefault.
Methodology
Theproposedfaultdiagnosismethodbasedonwaveletpacket-ARspectrumisshowninFigure1.Thedetailedstepsareasfollows.
Step1: