汽車發(fā)動機故障診斷中不確定性問題的貝葉斯網絡解法.pdf

1、東北大學博士學位論文汽車發(fā)動機故障診斷中不確定性問題的貝葉斯網絡解法姓名：張曉丹申請學位級別：博士專業(yè)：計算機應用技術指導教師：趙海20050601東北大學博士學位論文ABSTRACTSolutionofBayesianNetworkforUncertainProblemofMotorFaultDiagnosisAbstractUncertainproblemsarekeyproblemsinfaultdiagnosisfields，w

2、hichareresultedbymanyimportantreasons，includingcomplexdiagnosisobjects，limittestmeans，andtheinexactdiagnosisknowledge，elcMotorislargemechaatronicsequipment，whichhascomplex，correlativerelationexitinginitsunits，madtheunits

3、arefullofuncertainfactorsandinformation，Sothefaultofmotorispossiblemulti—kinds，correlativeIntheall：themostimportantproblemofsolvingmotorfaultdiagnosisistosolvetheuncertainproblemTherearemanycommonmethodsforsolvinguncerta

4、inproblems，suchasBayesianmethod，fuzzytheoryDStheory，certaintheoryetc，Andtherearemanyexpertsresearchingonthesemethods，andhasobtainedthat，BayesiannetworkbasedonBayesiantheoryisthebestmethodforsohinguncertainproblemsnowInth

5、ispaperforsolvinguncerlainproblems，thefaultdiagnosisstructuremodelandfunctionfusionmodelwasproposed，themodelswerebasedontheinfonnationfusiontechnologymadtheresearchofuncertainproblemsinfaultdiagnosisprocess，andthediagnos

6、isobjectwasAudiA618LmotorBayesiannetworkfusionalgorithmwasproposedfortheknowledgeexpressionandconstitutionofthemodels，whichwereresearcheddeeplyConstructofBayesiannetworkincludedstructurelearningandparameterlearningTherea

7、remanymethodsforstructurelearningAccordingtotheinfornaationobtainedbysensorstimelyanimprovedstructurelearningmethodofBayesiannetworkwasproposedinordertogetmoreexactdiagnosisresultTheon—linelearningideawasadoptedforthecon

8、structofstructure，andthelearnedstructureshouldbeclosetotheformeJ。structure，theMDLoflearnedstructureshouldbelessthantheformerstructureThenthedatafromsensorsinputtingintothes)，stemcontinuouslythenewstructurewasobtainedbyle