A Method for Anomaly Detection in Power Energy Topology Graph Data Based on Domain Knowledge Graph and Graph Neural Networks

With the rapid development of power energy systems, anomaly detection in power energy topology graph data has become increasingly important. However, existing methods often suffer from the lack of domain knowledge and the limited ability to capture complex correlations within the graph data. To address these challenges, this paper proposes a novel method for anomaly detection in power energy topology graph data based on domain knowledge graph and Graph Neural Network (GNN). Firstly, we construct a domain knowledge graph that incorporates expert knowledge and prior information about power energy systems. Then, we utilize the GNN model to learn the representations of nodes and edges in the graph data, capturing their complex relationships. Finally, we apply anomaly detection algorithms on the learned graph representations to identify potential anomalies in power energy topology. Experimental results on realworld power energy datasets demonstrate the effectiveness and efficiency of our proposed method. In conclusion, our method provides a promising approach for more accurate and reliable anomaly detection in power energy topology, contributing to the improvement of power system security and stability.