基于图卷积神经网络的配电网故障定位方法及算例分析  

Analysis of Fault Location Method and Examples in Distribution Network Based on Graph Convolutional Neural Network

作　　者：彭灵利 PENG Lingli(Guangdong Power Grid Co.,Ltd.,Guangdong 510180,China)

机构地区：[1]广东电网有限责任公司,广东510180

出　　处：《集成电路应用》2024年第8期277-279,共3页Application of IC

摘　　要：阐述为建立配电网故障的自动化定位方法,构建故障定位的算法模型。借助该算法从配电网的拓扑网络结构中提取节点特征和拓扑特征,将其作为模型输入,再经过数据标准化处理、图卷积层运算、全连接层运算,实现故障区段定位。检验结果显示,经过80次迭代后,此算法模型的定位精确度可以达到96%。This paper expounds the establishment of an automated fault location method for distribution networks and the construction of an algorithm model for fault location.It uses this algorithm to extract node and topology features from the topology network structure of the distribution network,takes them as model inputs,and then processes them through data standardization,graph convolutional layer operations,and fully connected layer operations to achieve fault section localization.The test results show that after 80 iterations,the positioning accuracy of this algorithm model can reach 96%.

关 键 词：图卷积神经网络 配电网 故障定位方法 算例分析 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TM73[自动化与计算机技术—控制科学与工程]