控制理论与应用ControlTheoryandApplications《自动化技术与应用》2023年第42卷第6期TechniquesofAutomation&Applications基于小波熵理论的动力发电设备故障自动检测方法叶志晖,石钉科,王柳婧,钱杰（浙江中烟工业有限责任公司，浙江杭州310000）摘要:传统方法检测动力发电设备存在检测效率低、检测灵敏度低且抗噪能力差的问题，基于此提出基于小波熵理论的动力发电设备故障自动检测方法。首先采用小波熵理论中的小波函数，即信号时间熵和信号频率熵，对原始故障数据分别进行故障分类和特征提取，再利用Softmax分类器“数据带”和阶跃函数，对已提取的故障特征样本分别进行内、外故障特征分类，最终根据分类结果实现动力发电设备故障的自动检测。实验结果表明方法效率高、灵敏度高、抗噪能力强且不受故障距离的干扰。关键词:小波熵；特征提取；Softmax分类器；故障检测；故障暂态电流均值中图分类号：TP277;TP206+.3文献标识码:A文章编号:1003-7241(2023)06-0066-04AutomaticFaultDetectionMethodofPowerGenerationEquipmentBasedonWaveletEntropyTheoryYEZhi-hui,SHIDing-ke,WANGLiu-jing,QIANJie(ChinaTobaccoZhejiangIndustrialCo.,Ltd.,Hangzhou310000China)Abstract:Traditionalmethodsfordetectingpowergenerationequipmenthavetheproblemsoflowdetectionefficiency,lowdetectionsensi-tivityandpooranti-noiseability.Basedonthis,anautomaticfaultdetectionmethodforpowergenerationequipmentbasedonwaveletentropytheoryisproposed.Firstitusesthewaveletfunctioninthewaveletentropytheory,namelysignaltimeentropyandsignalfrequencyentropy,toperformfaultclassificationandfeatureextractionontheoriginalfaultdata,andthenusetheSoft-maxclassifier"databand"andstepfunctiontoanalyzetheextractedfaultfeatures.Thesamplesareclassifiedintointernalandex-ternalfaultcharacteristicsrespectively,andtheautomaticdetectionofpowergenerationequipmentfaultsisfinallyrealizedac-cordingtotheclassificationresults.Theexperimentalresultsshowthattheproposedmethodhashighefficiency,highsensitivity,stronganti-noiseabilityandisnotaffectedbythedistanceofthefault.Keywords:waveletentropy;featureextraction;softmaxclassifier;troubleshooting;meanvalueoffaulttransientcurrent收稿日期:2022-01-10DOI:10.20033/j.1003-7241...
