Distribution Network Fault Location Based on Graph Attention Network

There have been a lot of research achievements on the power system fault diagnosis technology based on artificial intelligence, but the topology of the distribution network changes frequently, and the traditional artificial intelligence method highly relies on the training data, which brings difficulties to the fault location of the distribution network. This paper proposes a fault location method for the distribution network based on the graph attention network (GAT). The electrical nodes and lines of the distribution network are taken as the vertices and edges of the graph in the graph attention network. The attention coefficient is calculated according to the similarity of the fault features between the adjacent vertices, and the correlation between the vertex features is better integrated into the fault location model, which improves the adaptability of the fault location model to the topology changes. Finally, this paper builds a distribution network fault simulation model to verify that the proposed method has the advantages of high positioning accuracy and good robustness, and it is not affected by fault resistance, fault initial phase angle and fault distance. Under different network topology changes and scenarios, the model has a good application effect in actual comprehensive fault scenarios. © 2021, Power System Technology Press. All right reserved.