ANewFeatureEvaluationAlgorithmanditsApplicationtoFaultofHighSpeedRailway

ANewFeatureEvaluationAlgorithmanditsApplicationtoFaultofHighSpeedRailway.pdf
Multi-criterionfeaturerankingalgorithmsnotonlyalleviatesthedifficultyonchoosingappropriaterankingcriterioncausedbysingle-rankingalgorithms,butalsoimprovesthereliabilityoffeatmrerankingresults.However,theissueofconflictbetweendifferentsingle-rankingalgorithmsisoftenoverlooked.Bytreatingthistaskasasearchandoptimizationprocess,itispossibletousetheD-Stheoryandevidenceconflicttoreduceconflictsbetweendifferentsingle-criterionsandimprovethestabilityoffeatureevaluation.Thisworkpresentsanewmulti-criterionfeaturerankingalgorithmbasedonD-Stheoryandevidenceconflicttheorycombiningdifferentcriteriaimprovingbothclassificationperformanceandstabilityoffeatureselectionresults.ComparisionbetweenthenewalgorithmandBordaCournt,FuzzyEntropy,Fisher'sRatioandRepresentationEntropymethodsaredoneontrainfaultdataset.Theexperimentialresultsdemonstratemosteffectivenessofthenewalgorithmschemethantheotherfourcriterionsonallcasesconsidered
作者：DUjingJINWeidongCAIZhen-zhen
作者单位：SchoolofElectricalEngineering,SouthwestJiaotongUniversityChengdu610031
母体文献：2015第二届智能交通国际会议论文集
会议名称：2015第二届智能交通国际会议  
会议时间：2015年11月1日
会议地点：成都
主办单位：中国电子学会
语种：chi
分类号：
关键词：featureevaluating  multi-criterionfusion  D-Sevidencetheory  evidenceconflicttheory  faultclassification
在线出版日期：2017年11月24日
基金项目：
相似文献
相关博文

ANewFeatureEvaluationAlgorithmanditsApplicationtoFaultofHighSpeedRailway.pdf

文件大小:

1.74 MB

高速下载