基于分层神经网络的卫星故障诊断技术应用分析-application analysis of satellite fault diagnosis technology based on hierarchical neural network.docxVIP

AbstractWiththecontinuousdevelopmentofChinasaerospaceindustry,thedeteriorationofthespaceenvironmentandthecomplexityoftheaircraftstructure,theincreaseofspacejunk,vehiclereliabilityrequirementshavebecomemoresophisticated,thelimitationsoffaultdiagnosismethodcannotmeettheaircraftsecurityrequires.Forthecomplexsatellitestructure,modelingdifficultiesandthelargeamountofmonitoringdata,theuseofneuralnetworkhierarchicalfaultdiagnosismethodwillbeverybeneficialtothesatellitefaultdiagnosis.Thispaperanalyzestherespectiveshortcomingsofthesinglediagnosticmethodanddescribesintegrationdiagnosismethodbasedonthedataoftheneuralnetworkandknowledge-basedexpertsystem.Itcancomplementtheiradvantagesanddisadvantages.Theusingofprincipalcomponentanalysistopreprocessthedatasimplifiesthedatamodelandimprovestheefficiencyofdiagnosis.Thecreationofhierarchicalneuralnetworkdiagnosticmodelachievesthediagnosisofcomplexsystems.Workofthispapercontainsthefollowingmaincontents:First,bystudyingtheGRNNnetworkmodelandtheoperatingprinciple,thenatureoffunctionapproximationofGRNNnetworkachievesfaultdiagnosisforsatellitepowersubsystem.Second,throughtheresearchtheSOMneuralnetworksmodelandoperatingprincipleandtheSOMclusteringeffect,theSOMrealizesthefaultdiagnosisofthesatellitepowersubsystem.What’smore,bystudyingthehierarchicaltheory,thepaperdesignsthestructureandtheoperationalgorithmofthehierarchicaldiagnosisforcomplexsystems.Theuseofprincipalcomponentanalysismethodtosimplifythedatamodelimprovestheefficiencyofdiagnosis.Thehierarchicalneuralnetworkachievesthefaultdiagnosisofthesatellitemulti-system.Last,byprogrammingtoachievetheintegrationdiagnosisoftheneuralnetworksandexpertsystems.Theintegrationstructureofembeddedneuralnetworkexpertsystemcompletesthecomprehensivediagnosisandapplicationofthesatellite.Experimentsshowthathierarchicalneuralnetworkfaultdiagnosistechnologynotonlysolvesthelimitationsoftheexistingfaultdiagnosismethods,butalsoimprovetheefficiencyandreliabilityofthediagnosis.Keywords：Satellite，Faultdiagnosis，