基于文本挖掘的高铁信号系统车载设备故障诊断

《基于文本挖掘的高铁信号系统车载设备故障诊断》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、万方数据(2)提出了一种结合先验知识和现场数据的贝叶斯网络结构学习算法。算法首先根据专家知识来确定贝叶斯网络的拓扑结构以及节点间的约束关系，以此来降低搜索空间的复杂度；在此基础上利用搜索打分算法，得到车载设备的诊断贝叶斯网络的结构，提高了诊断准确性。图35幅，表11个，参考文献59篇关键词：高速铁路；车载设备；故障诊断；贝叶斯网络；主题模型；文本挖掘分类号：U284．238万方数据j￡立交通太堂亟±堂焦途塞△旦SI＆△￡!ABSTRACTABSTRACT：Inrailwayindustry,vehic

2、leon-boardequipment(VOBE)ofhighspeedrailwayisoneofthecoreequipmentsofChineseTrainControlSystemandakeyissueaffectingthesafetyandefficiencyofhighspeedrailway．However,statisticsfromWuhan．Guangzhouhighspeedrailwayshowthattherailwaysystemscurrentlysufferfrom

3、ahighrateoffailuresinVOBE．Therefore，itissignificantlyimportanttocarryoutresearchonfaultdiagnosisforVOBE．Inthispaper,300kmVOBEisusedasthestudyobjectwhichiswidelyusedinChinarailwaysystem．BasedonlargeamountoffaultdataandtheanalysisofVOBE，thefaultdiagnosisi

4、sdividedintotwoaspects：1)thefaultdatafeatureextraction；2)faultuncertaintyreasoning．Inrailwayindustry,themaintenancedataisusuallyrecordedinnaturallanguage，whichisabigchallengeforthefaultdiagnosisduetoitsunstmctualfeatureanduncertaintysemantics．Besides，du

5、etothecomplexityofVOBEandarbitrarynatureofthenaturallanguage，thefaultsymptomspaceandfaultspacehavecomplicateduncertaintyrelations，whichisanotherbigchallengeforthefaultdiagnosis．Consideringthearbitrarynatureoffaulttextdata,topicmodelisusedtoanalyzethefau

6、lttextdata,whichCanbeusedtoremovetheinfluencesofsynonymyandpolysemy．Besidesthat，thetextCanbetransformedfromwordspacetotopicspacebyusingtopicmodel，whichcallimprovethediagnosisefficiency．RegardingtotheuncertaintyandcomplexityoffaultdiagnosisforVOBE，aBayes

7、ianNetwork(BN)basedfaultdiagnosissystemforVOBEisproposed．IntheprocessofderivinganappropriateBNstructureforVOBE，anewmethodwhichfullyutilizesdomainexpertknowledgeanddatacalledHDBN—SL(HierarchicalDiagnosticBayesianNetworks—StructureLeaming)isproposed．Atlas

8、t，thecorrectnessandaccuracyoftheproposedmethodhavebeenverifiedbytherealdatafromWuhan—Guangzhouhi曲speedrailwaysignalingsystems．InordertomaketheBNbasedfaultdiagnosissystemfriendlyusefortherailwaystaffs，asoftwaretoolhasbeendeveloped