Anomaly Detection of Electrical Load in Power Systems Based on Graph Neural Networks

In response to the difficulties in feature extraction, poor model generalization ability, and high computational complexity in traditional research on anomaly detection of electrical loads, this article investigated a detection method based on Graph Convolutional Networks (GCNs). By constructing a graph structure of power grid nodes and edges, and using GCN for feature extraction, this method achieved an accuracy of 93.4%, a recall rate of 91.8%, and an F1 score of 92.6% in the experiment, which was significantly better than traditional models. In addition, this article also explored the influence of model parameters, and further verified the robustness of the method by adjusting the learning rate and network layers. The experimental results showed that the GCN-based method can not only effectively detect and locate abnormal electrical load, but also provide effective technical support for intelligent management of the power grid.