面向旋转机械的支持向量机方法及智能故障诊断系统研究

《面向旋转机械的支持向量机方法及智能故障诊断系统研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、浙江大学硕士学位论文面向旋转机械的支持向量机方法及智能故障诊断系统研究’硕士研究生：顾小军学科、专业：机械制造及其自动化指导教师：杨世锡副教授严拱标教授浙江大学研究生院二oO六年二月·奉论文受国家自然科学基金项口(编号：50205025、50505016、50575095)资助浙江大学顾上学付论文AbstractResearchonthelargerotatingmachinefaultdiagnosisisimportantforavoidingmasseconomicIOSSanddisasteraccident．inthi

2、sdissertation．onthebasisofthe“ResearchonNewMethodforRotatingMachineFaultsDiagnosisBasedonIndependentComponentAnalysis”(NationalNatureScienceFundProject，No：50205025)，wehavestudiedthetheoryofSVM(SupportVectorMachine)'faultdiagnosisrecognitionsystemandthemethodforrotati

3、ngmachinefaultsrecognitionbasedonSVM．Thenwehavemadesomesimulationstotesti母ourresearchresult．Intheend，wehaverealizedtheacquiringvibrationalsignalsbasedoftembeddedsystemanddevelopedaprototypeofrotatingmachinefaults’patternrecognitionsystembasedonSVMusingMATLABandLabVIE

4、W．ThedetailsarestudiedasfoIlows：Chapteronediscussestheimportanceofthemonitoringanddiagnosistechnologyforlargerotatingmachineryinmodernmanufacturingandsummarizestheresearchstatusanddevelopmenttrendofmonitoringanddiagnosissystemoflargerotatingmachinery．Wealsodescribeth

5、edevelopmentofthetheoriesandapplicationsofSL丁rStatisticalLearningTheory)andSVM．Atlastthegoalandmaincontentsofthisdissertationarepresented．ChaptertwodescribesthekernelcontentofSLTandanalyzesthestructureoftheSLTandtheproblemsneededtobesolved．ThenSVMandmainideaofitsalgo

6、rithmareintroducedinfaultdiagnosisdomain．WeaddressastruCtureofthepatternrecognitionsystembasedonSVMandintroducethecomponentsoftbesystem．Chaptorthreeaddressestheusualrotatingmachine’svibralfaultmechanismandstudystheclassifiedproblemofrotatingmachinefaultpattern．Ontheb

7、asisoftherotortestbench．wPcomparethreemuti—classSVMclassifiersusedinrotatingmachinefaultspatternrecognitionanddrawtheconclusionthattheDAGclassifiermethodismoresuitableforrotatingmachinefaultdiagnosis．Inchapterfour,accordingtorotatingmachinefaultdiagnosisandpaRernreco

8、gnitionsystem’srequirement,wedesignaembeddedsystemforrotatingmachinevibralsignalacquiringusingdoubledCPUarchitectureofARMandDSPtore