適用于深度學習訓練的配電網(wǎng)故障歷史樣本數(shù)據(jù)生成研究

打開文本圖片集

中圖分類號：TP18 文獻標志碼：A 文章編號：2095-2945(2025)14-0063-0

Abstract:Withthedevelopmentofsmartgridanddeep learning technology，theuseof historicalfaultsamples fortraining hasbecomeapowerful meansoffaultdataprocessingindistributionnetworks.Thisstudyadoptsadynamiciterativestrategy: firstly,thedeeplearnigmodelisusedtoidentifythefaulttypesofthedistributionnetwork,andthekeydataissummarized andextractedfromtheidentificationprocessThen，throughacontinuousiterativeprocess，thehistoricalsampledatageerated each time is fed back into the model. Finally,a system model of 10kV line protection detection test is built in Matlab/Simulink, andasimulationtestexampleisbuiltforverification,andtheexperimentalresultsshowthatthemodelisefectiveandfeasible.

Keywords:distributionnetwork failures;faultdetection;deep leaming;dynamic iteration;historical sampledata

在當今信息化、智能化的時代背景下，配電網(wǎng)的自動化水平和智能化[1-2程度不斷提高，對故障診斷技術提出了更高的要求。（剩余8178字）