A Novel Distribution Network Topology Identification Method Based on Graph Convolutional Network

With the proposal of China’s major strategic goals of "carbon neutrality and carbon peak", a large number of distributed new energy will gradually be integrated into the distribution network. The large-scale integration of distributed energy has greatly increased the uncertainty of the operation of the distribution network itself, posing higher requirements for the planning, scheduling, and operation and maintenance management of the distribution network. Accurate distribution network topology as the foundation for analysis functions such as power flow calculation, fault location, and state estimation in distribution networks, and are also a prerequisite for carrying out power grid planning, operation, and scheduling work. Therefore, studying the topology identification methods of distribution networks is of great significance. In response to the problems of low identification accuracy and poor real-time performance in existing power grid topology identification methods, this paper proposes a distribution network topology identification method based on graph convolutional networks. Firstly, convert the topology of the distribution network into a point line diagram. Then, based on the node features of the distribution network topology, edge features of the distribution network lines are extracted as node features of the line graph. Furthermore, by utilizing the adjacency matrix of the abstract line graph and node features, a graph convolutional network is trained to obtain a distribution network topology identification model. Finally, the effectiveness of the proposed topology identification method was verified through experimental data.