基于人工免疫算法的故障診斷問題研究.pdf

1、分類號(hào)：TPl81UDC：Y8584tt0單位代碼：10220學(xué)號(hào)：S0306192大慶石油學(xué)院碩士學(xué)位論文論文題目：基王厶王免疫算法韻故障診斷問題研究碩士生：指導(dǎo)教師：申請(qǐng)學(xué)位級(jí)別：王寨趨一任佳建￡教授)王拳顧專業(yè)、學(xué)科：控壹Ⅱ理論與控制王程提交論文時(shí)間：20o6生3一月一15一旦答辯日期：靼Q車≥月塹目授予學(xué)位日期：2006年3月l5曰ResearchofFaultDiagnosisBasedonArtificialImmuneAl

2、gorithmAbs仃actThreefaultdiagnosismethodsbasedonArtificialImmuneAlgorithra(AlA、areresearchedinthispaperandusethemtodiagnosisthefaultofthepumpjacknledetailworkisasfollows：FaultdiagnosismethodbasedonImmuneneuralnetworkisdes

3、ignedinthispapertheimmunemechanismisintroducedwhenadjusttheweightofneuralnetwork，thismethodmakeuseoftheimmunealgorithm’ScapabilityofkeepingindividualdiversitytosearchtheweightofneuralnetworkglobalyitcanavoidtheBPalgorith

4、mfrombeingtrappedinthelacalvalueefficientlymeanwhileitcanimprovetheconvergencespeedofalgorithm，thefaultdiagnosisresearcharedoneonthepumpjack，andtwomethodsarecomparedandanalysedFaultdiagnosismethodbasedontheinlmunerespons

5、eisproposedTherealcodeisadoptedinthispaperinordertosolvetheproblemtllatusingbinarycodeincreasethecomputingCloneselectionandhypervariationisintroducedbesidesgeneticoperatorbytheauthoritCallimprovethespeedofgeneratingmemor

6、yantibodies，andimproveblindsearchcapabilityofthegeneticalgorithmwhengenarateantibodies，meanwhilethediversityofmemoryantibodiesisensuredInordertOmakethememoryantibodiesexpressstructuresandcharactersformoreantigens，thispap

7、erproposestheideathatusingmulti—antigenstogeneratememoryantibodiesandproposesthestrategythatstimulateandsuppresstheantibodieswhicharegeneratedbythesameantigenbasedonthelivingexpectionandthatstimulateandsuppresstheantibod

8、ieswhicharegeneratedbydifferentantigensbelongtothesamefaultpaRembasedontheconcentration，Thiscanavoid也e曲enomenonofimmaturityconvergenceBesides，knearestneighbor(KNN)methodbasedonthresholdisproposedinthispaperThefaultdiagno

9、sisresearchisdoneonthepumpjack，TheresultshowthatthismethodcangetmemoryantibodieswhichCanidentifyantigenseffectively，andtheeffecttoidentiifyantigenswhicharevariatedispresentedAbnormitydetectionandfaultdiagnosismethodbased

10、onnegativeselectionisdesigned，Theallthorintroducetleideaofalterabledetectorradiusinordertoexertthedetectionfunctionofdetectorsandneedn’ttosetdetectorradiusInordertosolvetheproblemofthedetectors’numberanddetectionarea，thi

11、sPaDeradoptsthemethodthatuseallnealalgorithmtooptimizedetectorsandmakethecoverspacefornonselflagerwithlimiteddetectorsBesidesinordertodetectanddiagnosetheunknownfaultpattern，thispaperputthedetectorswhicharenotactivatedto

12、gethertogetunknownpatternfaultdetectors，andtodetectnewpatternfaultThefaultdiagnosisresearchisdoneonthepumpjack，theresultshowthatthismethodcangetrapidandcorrectresultbasedonalotofnormaldata，andCangetgoodfaultdiagnosiseffe