机械状态监测和故障诊断文献翻译中英文对照.docx
《机械状态监测和故障诊断文献翻译中英文对照.docx》由会员分享,可在线阅读,更多相关《机械状态监测和故障诊断文献翻译中英文对照.docx(13页珍藏版)》请在冰豆网上搜索。
![机械状态监测和故障诊断文献翻译中英文对照.docx](https://file1.bdocx.com/fileroot1/2023-2/13/7d702e12-221b-40de-b454-289adb7bbd14/7d702e12-221b-40de-b454-289adb7bbd141.gif)
机械状态监测和故障诊断文献翻译中英文对照
Abstract
Mechanicalequipmentsarewidelyusedinvariousindustrialapplications.Generallyworkinginsevereconditions,mechanicalequipmentsaresubjectedtoprogressivedeteriorationoftheirstate.Themechanicalfailuresaccountformorethan60%ofbreakdownsofthesystem.Therefore,theidentificationofimpendingmechanicalfaultiscrucialtopreventthesystemfrommalfunction.Thispaperdiscussesthemostrecentprogressinthemechanicalconditionmonitoringandfaultdiagnosis.Excellentworkisintroducedfromtheaspectsofthefaultmechanismresearch,signalprocessingandfeatureextraction,faultreasoningresearchandequipmentdevelopment.Anoverviewofsomeoftheexistingmethodsforsignalprocessingandfeatureextractionispresented.Theadvantagesanddisadvantagesofthesetechniquesarediscussed.Thereviewresultsuggeststhattheintelligentinformationfusionbasedmechanicalfaultdiagnosisexpertsystemwithself-learningandself-updatingabilitiesisthefutureresearchtrendfortheconditionmonitoringfaultdiagnosisofmechanicalequipments.
©2011PublishedbyElsevierLtd.Selectionand/orpeer-reviewunderresponsibilityof[CEIS2011]
Keywords:
Conditionmonitoring;Faultdiagnosis;Vibration;Signalprocessing
1.Introduction
Withthedevelopmentofmodernscienceandtechnology,machineryandequipmentfunctionsarebecomingmoreandmoreperfect,andthemachinerystructurebecomesmorelarge-scale,integrated,intelligentandcomplicated.Asaresult,thecomponentnumberincreasessignificantlyandtheprecisionrequirementforthepartmatingisstricter.Thepossibilityandcategoryoftherelatedcomponentfailuresthereforeincreasegreatly.Malignantaccidentscausedbycomponentfaultsoccurfrequentlyallovertheworld,andevenasmallmechanicalfaultmayleadtoseriousconsequences.Hence,efficientincipientfaultdetectionanddiagnosisarecriticaltomachinerynormalrunning.Althoughoptimizationtechniqueshavebeencarriedoutinthemachinedesignprocedureandthemanufacturingproceduretoimprovethequalityofmechanicalproducts,mechanicalfailuresarestilldifficulttoavoidduetothecomplexityofmodernequipments.Theconditionmonitoringandfaultdiagnosisbasedonadvancedscienceandtechnologyactsasanefficientmeantoforecastpotentialfaultsandreducethecostofmachinemalfunctions.Thisistheso-calledmechanicalequipmentfaultdiagnosistechnologyemergedinthenearlythreedecades[1,2].
Mechanicalequipmentfaultdiagnosistechnologyusesthemeasurementsofthemonitoredmachineryinoperationandstationarytoanalyzeandextractimportantcharacteristicstocalibratethestatesofthekeycomponents.Bycombiningthehistorydata,itcanrecognizethecurrentconditionsofthekeycomponentsquantitatively,predictstheimpendingabnormalitiesandfaults,andprognosestheirfutureconditiontrends.Bydoingso,theoptimizedmaintenancestrategiescanbesettled,andthustheindustrialscanbenefitfromtheconditionmaintenancesignificantly[3,4].
Thecontentsofmechanicalfaultdiagnosiscontainfouraspects,includingfaultmechanismresearch,signalprocessingandfeatureextraction,faultreasoningresearchandequipmentdevelopmentforconditionmonitoringandfaultdiagnosis.Inthepastdecades,therehasbeenconsiderableworkdoneinthisgeneralareabymanyresearchers.Aconcisereviewoftheresearchinthisareahasbeenpresentedby[5,6].Somelandmarksarediscussedinthispaper.Thenovelsignalprocessingtechniquesarepresented.Theadvantagesanddisadvantagesofthesenewsignalprocessingandfeatureextractionmethodsarediscussedinthiswork.Thenthefaultreasoningresearchandthediagnosticequipmentsarebrieflyreviewed.Finally,thefutureresearchtopicsaredescribedinthepointoffuturegenerationintelligentfaultdiagnosisandprognosissystem.
2.FaultMechanismResearch
FaultMechanismresearchisaverydifficultandimportantbasicprojectoffaultdiagnosis,sameasthepathologyresearchofmedical.AmericanscholarJohnSohre,publishedapaperon"Causesandtreatmentofhigh-speedturbomachineryoperatingproblems(failure)",intheUnitedStatesInstituteofMechanicalEngineeringatthePetroleumMechanicalEngineeringin1968,andgaveaclearandconcisedescriptionofthetypicalsymptomsandpossiblecausesofmechanicalfailure.Hesuggestedthattypicalfailurescouldbeclassifiedinto9typesand37kinds[7].Following,Shiraki[8]conducedconsiderableworkonthefaultmechanismresearchinJapanduring60s-70slastcentury,andconcludedabundanton-sitetroubleshootingexperiencetosupportthefaultmechanismtheory.BENTLYNEVADACorporationhasalsocarriedoutaseriesexperimentstostudythefaultmechanismoftherotor-bearingsystem[9].AlargeamountofrelatedworkhasbeendoneinChinaaswell.Gaoetal.[10]researchedthevibrationfaultmechanismofthehigh-speedturbomachinery,investigatedtherelationshipbetweenthevibrationfrequencyandvibrationgeneration,anddrewupthetableofthevibrationfaultreasons,mechanismandrecognitionfeaturesforsubsynchronous,synchronousandsuper-synchronousvibrations.Basedonthetabletheyproposed,theyhaveclassifiedthetypicalfailuresinto10typesand58kinds,andprovidedpreventivetreatmentsduringthemachinedesignandmanufacture,Installationandmaintenance,operation,andmachinedegradation.Xuetal.[11]concludedthecommonfaultsoftherotationalmachines.Chenetal.[12]usedthenonlineardynamicstheorytoanalyzethekeyvibrationproblemsofthegeneratorshaft.Theyestablishedarotornonlineardynamicmodelforthegeneratortocomprehensivelyinvestigatetherotordynamicbehaviorundervariousinfluences,andproposedaneffectivesolutiontopreventrotorfailures.Yangetal.[13]adoptedvibrationanalysistostudythefaultmechanismofaseriesofdieselengines.Otherresearchershavedonealotinthefaultmechanismofmechanicssince1980s,andhavepublishedmanyvaluablepaperstoprovidetheoryandtechnologysupportsintheapplicationoffaultdiagnosissystems[14-18].However,mostofthefaultmechanismresearchisonthequalitativeandnumericalsimulationstage,theengineeringpracticeisdifficulttoimplement.Inaddition,thefaultinformationoftenpresentsstrongnonlinear,nonstationaryandnonGaussiancharacteristics,thesimulationtestscannotreflectthesecharacteristicsveryaccurately.Thefaultdiagnosisresultsandtheapplicationpossibilitymaybeinfluencedsignificantly.Asaresult,thedevelopmentofthefaultdiagnosistechniquestillfacesgreatdifficulties.
3.AdvancedSignalProcessingandFeatureExtractionMethods
Advancedsignalprocessingtechnologyisusedtoextractthefeatureswhicharesensitivetospecificfaultbyusingvarioussignalanalysistechniquestoprocessthemeasuredsignals.Conditioninformationoftheplantsiscontainedinawiderangeofsignals,suchasvibration,noise,temperature,pressure,strain,current,voltage,etc.Thefeatureinformationofacertainfaultcanbeacquiredthroughsignalanalysismethod,andthenfaultdiagnosiscanbedonecorrespondingly.Tomeetthespecificneedsoffaultdiagnosis,faultfeatureextractionandanalysistechnologyisundergoingtheprocessfromtimedomainanalysistoFourieranalysis-basedfrequency-domainanalysis,fromlinearstationarysignalanalysistononlinearandnonstationaryanalysis,fromfrequency-domainanalysistotime-frequencyanalysis.
Earlyresearchonvibrationsignalanalysisismainlyfocusedonclassicalsignalanalysiswhichmadealotofresearchandapplicationprogress.Rotatingmechanicalvibrationisusuallyofstrongharmonic,itsfaultisalsousuallyregisteredaschangesinsomeharmoniccomponents.ClassicalspectrumanalysisbasedonFouriertransform(suchasaveragetime-domaintechniques,spectrumanalysis,cepstrumanalysisanddemodulationtechniques)canextractthefaultcharacteristicinformationeffectively,thusitiswidelyusedinmotivepowermachine,especiallyinrotatingmachineryvibrationmonitoringandfaultdiagnosis.Inamannerofspeaking,classicalsignalanalysisisstillthemainmethodformechanicalvibrationsignalanalysisandfaultfeatureextraction.However,classicalspectrumanalysisalsohasobviousdisadvantages.Fouriertransformreflectstheoverallstatisticalpropertiesofasignal,andissuitableforstationarysignalanalysis.Inreality,thesignalsmeasuredfrommechanicalequipmentareever-changing,non-stationary,non-Gaussiandistributionandnonlinearrandom.Especiallywhentheequipmentbreaksdown,thissituationappearstobemoreprominent.Fornon-stationarysignal,sometime-frequencydetailscannotbereflectedinthespectrumanditsfrequencyresolutionislimitedusingFouriertransform.Newmethodsneedtobeproposedforthosenonlinearityandnon-stationarysignals.Thestrongdemandfromtheengineeringpracticealsocontributestotherapiddevelopmentofsignalanalysis.Newanalyticalmethodsfornon-stationarysignalandnonlinearsignalareemergingconstantly,whicharesoonappliedinthefieldofmachineryfaultdiagnosis.Newmethodsofsignalanalysisaremainincludingtime-frequencyanalysis,waveletanalysis,Hilbert-Huangtransform,independentcomponentanalysis,advancedstatisticalanalysis,nonlinearsignalanalysisandsoon.Theadvantagesanddisadvantagesoftheseapproachesarediscussedbelow.
4.ResearchonFaultReasoning
Atpresent,manymethodsareadoptedintheprocessofdiagnosticreasoning.Accordingtothesubjectsystemswhichtheybelongto,thefaultdiagnosiscanbedividedintothreecategories:
(1)the