资源描述：
《Machinery faults detection and forecasting using Hidden Markov Models》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、ProceedingsofESDA20068thBiennialASMEConferenceonEngineeringSystemsDesignandAnalysisJuly4-7,2006,Torino,ItalyESDA2006-95472MACHINERYFAULTSDETECTIONANDFORECASTINGUSINGHIDDENMARKOVMODELSPaoloCalefati1,BiagioAmico1,AntonellaLacasella1,EmanuelMuraca1,MingJ.
2、Zuo21ITIA-CNRInstituteofIndustrialTechnologiesandAutomation,,viadelleMagnolie4-70026Modugno2UniversityofAlberta,MechanicalDepartment,ReliabilityGroupABSTRACTInliterature,severaldiagnostictechniqueshavebeenThepresentworkdescribesanautomaticprocedureforp
3、roposedinthepasttodetectthepresenceoffaultinrotarydiagnosticsandprognosticissues,anditsapplicationtothemachines.Forsuchapplication,aNeuralNetworkclassifierevaluationofgearboxesresiduallifetime.TheHiddenMarkovseemstobeanidealcandidatetocorrelatetheinput
4、datatotheModels-HMM-techniquehasbeenusedtocreatequasi-presenceoffaults,thankstoitscapabilitytolearncomplexandstationaryandstationarymodelsandtotakeadvantagesofthenonlinearmappings.Nevertheless,inmostcases,anexpertmultiplesensordataacquisitionarchitectu
5、re.Atfirst,Markovoperatorisneededtodrawconclusionsaboutthefaultlevelbymodelsfordiagnosticshavebeendefined.Themainadvantagemeansofspectralanalysismethods.Obviously,inordertooftheHMMsapproachisthatallvibrationrawdatameasuredreducecostsandsimplifythediagn
6、osisandprognosticstages,byamultisensorarchitecturecanbeusedwithoutanypre-itwouldbedesirabletomakethefaultdetectionandtheprocessing.AnefforttoadapttheHMMstechniquetotheestimationofresiduallifetimefullyautomatic.prognosticissuehasalsobeencarriedout.Tocre
7、ateMarkovTheHiddenMarkovModelsaresuitabletoperformModelssuitableforprognostics,theViterbiAlgorithmhasbeendetectionandestimationoperationsformachinediagnosticandusedtodefinethebestsequenceofmodelstatesandtoprognosticissues.Inpreviousstudies,thesetechniq
8、ueshaveoptimizeresidualusefullifetimecomputation.Finally,beenappliedtodiagnoseandforecastfaultsofmechanicalexperimentalresultsarediscussed,whichencouragefurthercomponents[1,2].ThemotivationtousetheHMMtheresearcheffortsaccordingtotheprop
