简介:Forthelargenumberofnonlineardegradationdevicesexistinginaproject,theexistingmethodshavenotsystematicallystudiedtheeffectsofrandomeffectontheremaininglifetime(RL),theaccuracyandefficiencyoftheparametersestimationarenothigh,andthecurrentdegradationstateofthetargetdeviceisnotaccuratelyestimated.Inthispaper,anonlinearWienerdegradationmodelwithrandomeffectisproposedandthecorrespondingprobabilitydensityfunction(PDF)ofthefirsthittingtime(FHT)isdeduced.Aparameterestimationmethodbasedonmodifiedexpectationmaximum(EM)algorithmisproposedtoobtaintheestimatedvalueoffixedcoefficientandthepriorivalueofrandomcoefficientinthemodel.Theposteriorvalueoftherandomcoefficientandthecurrentdegradationstateoftargetdeviceareupdatedsynchronouslybythestatespacemodel(SSM)andtheKalmanfilteralgorithm.ThePDFofRLwithrandomeffectisdeduced.AsimulationexampleisanalyzedtoverifythattheproposedmethodhastheobviousadvantageovertheexistingmethodsinparameterestimationerrorandRLpredictionaccuracy.