基于人工智能的高压电缆接地故障诊断方法研究

打开文本图片集

中图分类号：TM75 文献标志码：A 文章编号：2095-2945（2025）19-0044-04

Abstract：Thispaperproposesa faultdiagnosismethod basedondeeplearming tosolve theproblemsoflowpostioning accuracyandslowdiagnosiseficiencyinhigh-voltagecablegroundingfaultdiagnosis.Themethodfirstpreprocessesandextracts featuresonthecolectedcablefaultsignalsthroughwavelettransform，andbuildsafeaturedatasetcontainingmultipleground faulttypes;thendesignsanimprovedconvolutionalneuralnetworkmodelthatintegratestheatentionmechanismandresidual connectionstructure，realizingadaptivelearingandclassficationoffaultcharacteristics；Finall，areal-timefaultdiagnosis systemisdevelopedtoachieverapidfaultlocationandidentification.Experimentalresultsshowthatinthefaultdiagnosisof10＼～ 35kVhigh-voltage cables，the fault location accuracy of this method reaches 98.5% ，whichis 15% higher than the traditional method，andtheaveragediagnosistimeisshortenedto2.3seconds，anditstillmaintainsstablediagnosisperforancein complexnoiseenvironments.Thismethodhasbeeapliedintransmisson linefaultdiagnosisofaprovincialpowercompany， providing an effective guarantee for the safe and stable operation of the power system.

Keywords：high-voltagecable;groundfaultdiagnosis;deeplearning;faultfeatureextraction;intellgentdiagnosissystem

近年来，随着电力系统规模的不断扩大和供电可靠性要求的持续提高，高压电缆在电力传输中的应用日益广泛。（剩余5841字）