资源描述:
《rolling bearing feature extraction based on hht and fault diagnosis based on decision tree》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、研究生课程考核试卷(适用于课程论文、提交报告)科目:SystemsEngineering教师:xx姓名:xx学号:201207xxxxx专业:机械电子工程类别:学术上课时间:2013年3月至2013年5月考生成绩:卷面成绩平时成绩课程综合成绩阅卷评语:阅卷教师(签名)重庆大学研究生院制RollingBearingFeatureExtractionBasedonHHTandFaultDiagnosisBasedonDecisionTreeAbstract:Thelifeofthebearingsisalwaysunpredictable,therefore,itisv
2、eryessentialtoexaminethefaultierdiagnosisandmonitorthedevicetokeepitsafelyrunning.Theprocessoffaultierdiagnosisincludesacquisitionofsignal,featureextraction,failuremodeidentificationetc.FaultierdiagnosisismainlybasedonMechanicalSignalProcessing,SensorTechnologyetc.Inthisregardtheroller
3、bearing6205andnormalbearingwasstudded;themainfailuremodewastheinnerringfailure,theouterringfailureandrollingelementfailure.Featureextractionisthemostimportantstepinfaultierdiagnosis,thefeaturecanbeaccurateandsensitivetorecordthebearingstatus.Basedontherelevantliterature,inthispaper,the
4、featureisappliedbyHHTbothintimedomainandfrequencydomain:ThebearingvibrationsignalisresolvedbyHHTintoHilbertspectrumandpartialHilbertspectrum,thefeatureofthefaultierdiagnosiswastheextracted.Toinvestigatethealgorithminthemulti-faultierdiagnosisproblem,wecanchoosetheperformanceofthemulti-
5、classsupportvectormachineanddecisiontree.Inthispaper,wechoosethealgorithmdecisiontree.ThisalgorithmwastocombinetheHHTanddecisiontreemethod.TheoperationwastofiltertheoriginalsignalusingEMD,theintrinsicmodefunctionswasoperatedinnoiseautocorrelationmethod;themainfrequencyofthefeaturewasex
6、tractedformingtheC4.5decisiontreeofthesamplefeaturesaccordingly,inordertoreducethecomplexityofthealgorithm,wereducetocalculatetheattributionandextractthefeatureinpredictingprocess,theoperationthenodeattributeinformationofthedecisiontreeisfeedbackofthealgorithmisapplied.Therecognizabler
7、ateofthisalgorithmisupto96%.Keywords:rollingbearing;Hilbert-Huangtransform;decisiontree1.IntroductionMechanicalfaultdiagnosis,fromthesystempointofview,isakindofsciencethatusesavarietyofdetectionandidentificationtheorytoidentifymechanicalequipmentrunningstate.Canbetracedbacktothe1960s