An approach for identifying vulnerable nodes in power networks: Neighborhood compactness model based on differential electrical conductivity strength

The extensive scale and intricate connections of power networks have highlighted their vulnerability. The failure of a single vulnerable node can quickly bring down the system due to the networks’ sensitivity to interference. Therefore, it is crucial to efficiently and accurately identify the vulnerable nodes to enhance the safety and management of the power networks. The global information-based vulnerable node identification methods are time-consuming and do not meet the real-time requirements. Meanwhile, the existing local information-based methods lack the exploration of the interaction between neighbors. This paper concentrates on the essential characteristics of the power network topology and aims to mine the vulnerable nodes using broader and richer neighborhood information. A novel vulnerable node identification method is proposed, which combines the neighborhood compactness model and differential electrical conductivity strength (ECS). From the new perspective of the decentered neighborhood, we propose the kth order neighborhood ECS to dynamically quantify multipath connection strength across topological layers, and introduce differential ECS to model incremental connectivity gains from path redundancy and distance decay. Finally, the performance of the neighborhood compactness model is evaluated through average ECS, network efficiency, connectivity, vulnerability, capacity, ranking distribution, and complexity. The proposed algorithm demonstrates superior performance in the six dimensions in comparison experiments on nine power networks.