遠(yuǎn)程監(jiān)測診斷及診斷方法的研究.pdf

1、東南大學(xué)博士學(xué)位論文遠(yuǎn)程監(jiān)測診斷及診斷方法的研究姓名：舒賾申請學(xué)位級別：博士專業(yè)：機械制造及其自動化指導(dǎo)教師：黃仁賈民平2000.9.1東南大學(xué)博士學(xué)位論文ABSTRACTAremoteconditionmonitoringandfaultdiagnosissystemisasciencewhichisconnectedwithcomputercommunicationandfaultdiagnosis1’histechniqueisd

2、evelopingathighspeedalongtheinformationsuperhighwayHowever，therearefewremoteconditionmonitoringandfaultdiagnosissystemscomingintooperationwitheconomicbenefitindomesticenterprisesowingtothefactsasfollowsOneistheremotecond

3、itionmonitoringandfaultdiagnosiswhichrequestshighsecuritywhereasthecurrenttechniquedoesn’tbesatisfiedtoneedofproductionTheother；salladvanceddiagnosissystemconnectedwithbasesofmultibranchesofsciencewhichmakesdifficultytot

4、heexistingengineeringInaddition，thediagnosistechniquebasedondataandhavingtheabilityofselflearningandattributereductionaredemandedbutthecurrentdiagnosistechniquecan’tsatisfythesedemandsInthisdissertation，aremoteconditionm

5、onitoringandfaultdiagnosissystemsuitableforproductionisinvestigatedanditsdiagnosistechniqueisalsostudiedFirstlyasystemframeandsecurityarediscussedAthree—levelC／S(Client／Server)framecombinedwiththefileservermodeandCISmode

6、isproposedandastrategyusingmultimeasuresofsecurityisadoptedtoreachthesecuritydemandtotheoperationdataforthekeyequipment11lerequirementstodiagnosistechniqueoftheremotesystemarcanalyzedtooAremotesystemdevelopedbytheauthorg

7、oesintooperationforalargecompressorsystemwhichhasbeenmountedinachemicalenterpriseExperimentalresultsprovethatthesystemissatisfiedtequirementsontherealtime，exactness，securityandreliabilitySecondlycombiningwithrequirements

8、ofremotediagnosis，theauthorsuccessfullyappliesCezaryfuzzydecisiontreetothediagnosisknowledgeacquisitionandreasoningfortherotatingmachineTheinformationentropyofmultilevelsandafuzzydiagnosisreasoningstrategyaredefinedforth

9、efuzzydecisiontreeThirdlysomeproblemoffuzzydiagnosisdecisiontreeisinvestigatedasfollowsTheproblemofsmallsamplesizeistheoreticallystudiedanditisprovedmoreeriectivetotreatproblemsbyusingfuzzyinformationdistributiontheoryTw

10、oproceduresareproposedfortheincompletedataThefirstprocedureistoregulatemembershipfunctionforwithoutreducingattributetreeThesecondprocedareisconstructingmaximumdecisiontreeforreducedattributetreeInvestigatedresultsproveth

11、atthetwoproceduresarefeasibleoftreatingtheproblemofincompletedataAndthentheinfluenceofdifferentToperatoronconstructingandreasoningofdecisiontreeisstudiedTheterminalconditionandpruningmeasurearealsostudiedfortheinfluenceo

12、fnoisedataAtlast，anewreasoningstrategyisproposedtoadapttotheoccasiononwhichthedecisionclassesaredistributedoutoforderinthedecisionspaceFourthlytheparameteroptimizationofattributemembershipfunctionoffuzzydecisiontreeisstu

13、died，andoptimizationmathematicsmodelsofoptimizingattributesarefoundedManytechniqueproblems，suchasseedgroupchoosing，crossing，pruning，andterminatingcriterionetcinoptimizingmultivariableandrestrainedcomplexobjectivefunction