基于人工智能的高壓電纜接地故障診斷方法研究

打開文本圖片集

中圖分類號：TM75 文獻標志碼：A 文章編號：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

近年來，隨著電力系統(tǒng)規(guī)模的不斷擴大和供電可靠性要求的持續(xù)提高，高壓電纜在電力傳輸中的應用日益廣泛。（剩余5841字）