第59卷第2期:0190⁃01972023年2月16日HighVoltageApparatusVol.59,No.2:0190⁃0197Feb.16,2023改进朴素贝叶斯模型在电力变压器故障定位中的应用范慧芳1,咸日常1,王涛2,高鸿鹏1,陈蕾1,张冰倩1(1.山东理工大学电气与电子工程学院,山东淄博255049;2.国网山东省电力公司枣庄供电公司,山东枣庄277000)摘要:电力变压器故障能否精准定位一直是制约其状态检修有效开展的技术瓶颈。文中针对目前已有故障定位模型存在的不足,借助变压器故障类型与特征状态量之间的内在关系,将朴素贝叶斯网络模型进行特征属性加权改进,并将其扩展为改进的双层朴素贝叶斯网络模型应用至电力变压器故障定位中。在这一过程中,考虑到特征属性与类别之间和各特征属性之间的依赖关系,采用ReliefF算法和相关系数法分别对特征属性进行加权处理,构造出改进的朴素贝叶斯网络模型,并在MATLAB软件中进行了诊断对比预测,得到了较好的预测结果,文中最后利用实际案例进一步验证了所提模型与分析方法的有效性,可为电力变压器故障诊断提供技术指导。关键词:电力变压器;朴素贝叶斯;属性加权;故障定位;ReliefF算法ApplicationofImprovedNaiveBayesModelinFaultLocationofPowerTransformerFANHuifang1,XIANRichang1,WANGTao2,GAOHongpeng1,CHENLei1,ZHANGBingqian1(1.CollegeofElectricalandElectronicEngineering,ShandongUniversityofTechnology,ShandongZibo255049,China;2.StateGridShandongElectricPowerCompanyZaozhuangPowerSupplyCompany,ShandongZaozhuang277000,China)Abstract:Accuratelocationonfaultofpowertransformerhasalwaysbeenatechnicalbottleneckconstrainingtheef⁃fectivedevelopmentofitsstatemaintenance.Inviewoftheshortcomingofthepresentlyexistedfaultlocationmodel,theinherentrelationshipbetweenthetypeoftransformerfaultandthecharacteristicstateisusedtoimprovethechar⁃acteristicsattributeweightoftheNaiveBayesnetworkmodel,isexpandedintoanimprovedtwo⁃layerNaiveBayesnetworkmodelandappliedtothefaultlocationofpowertransformers.Duringthisprocess,consideringthedependen⁃cybetweenfeatureattributesandcategoriesandbetweeneachfeatureattributes,theReliefFalgorithmandthecorre⁃lationcoefficientmethodareusedtoweightthefeatureattributesrespectivelyandtoconstructanimprovedNaiveBayes...