Identification of vulnerable nodes in power grids based on graph deep learning algorithm
Xueping Gu, Tong Liu, Shaoyan Li, Xiaodong Yang, Xin Cao
Abstract
Abstract Comprehensive and timely identification of vulnerable nodes is of great significance to ensure the security of the power system. With the development of power grids, traditional identification methods for vulnerable nodes with high operational risks can no longer meet the actual operation needs in terms of model accuracy, and it is prone to ignore some critical nodes. In view of this, a method based on graph deep learning is proposed for grid vulnerable nodes identification to realize fast and accurate identification of vulnerable nodes. Firstly, the GraphSAGE algorithm is used to aggregate the operation states of each node and its neighbours. The node attribute information and topology information are mapped to the output layer at the same time to form a comprehensive representation of the node state through non‐linear transformation. Then according to the similarity between the states, the improved K‐means algorithm is used to cluster the nodes so as to realize the identification of vulnerable nodes. The distances between the nodes and the vulnerability centre are calculated to rank the node vulnerabilities. Finally, the effectiveness and feasibility of the proposed method are verified by the simulation results of the IEEE 30‐bus system and a practical power system.