基于图卷积网络的10 kV线路接地故障定位技术  

Grounding Fault Location Technology for 10 kV Transmission Lines Based on Graph Convolutional Networks

作　　者：王晓园 侯泽东 菅东祥 何皓文 段德毅 WANG Xiaoyuan;HOU Zedong;JIAN Dongxiang;HE Haowen;DUAN Deyi(Turpan Power Supply Company,State Grid Xinjiang Electric Power Co.,Ltd.,Turpan,Xinjiang 838000,China)

机构地区：[1]国网新疆电力有限公司吐鲁番供电公司,新疆吐鲁番838000

出　　处：《自动化应用》2025年第6期138-140,共3页Automation Application

摘　　要：提出了一种基于图卷积网络的10 kV线路接地故障定位技术。该技术通过采集配电自动化系统、气象、历史故障等多源数据,构建了一种配电网图结构模型,进而引入改进的图卷积网络(GCN)模型,捕捉配电网的复杂拓扑关系和运行状态,并进行故障定位。实验结果表明,该技术的故障定位准确率达到95%以上,为配电网运维提供了有力支持。A 10 kV line grounding fault location technique based on graph convolutional network is proposed.This technology constructs a distribution network diagram structure model by collecting multi-source data such as distribution automation systems,meteorology,and historical faults,and then introduces an improved Graph Convolutional Network(GCN)model to capture the complex topological relationships and operating states of the distribution network and perform fault localization.The experimental results show that the fault location accuracy of this technology reaches over 95%,providing strong support for the operation and maintenance of distribution networks.

关 键 词：图卷积网络 多源数据 故障定位 

分 类 号：TM755[电气工程—电力系统及自动化]