Research on anomaly identification method of remote real-time monitoring of power system equipment based on AI
Wei Zhang, Qiong Cao, Shuai Yang, Hao Guo · Australian Journal of Electrical & Electronics Engineering · 2025
The stability of power network operation is seriously threatened by the frequent occurrence of equipment anomalies due to the increasing integration and complexity of power system equipment. This research proposes an AI method-deep convolutional neural network (DCNN) based remote real-time monitoring anomaly diagnosis approach for power system equipment. The technique takes digital pictures of the equipment using a camera, sends them over a network to a remote monitoring centre, and then employs DCNNs to preprocess and extract characteristics from the images of the equipment to identify anomaly situations in the equipment effectively. The method’s operation, including support vector machines, variational modal decomposition, and picture preprocessing, is well explained in this study. Simulation trials in a Beijing-based power firm demonstrate that the approach presented in this study can reliably identify the equipment anomaly in various weather scenarios, with missed and false recognition rates both less than 0.6%. This paper also presents successful cases in real-world applications, confirming the method’s superiority in raising the effectiveness and accuracy of video patrols. The study presented in this paper demonstrates the great practical usefulness and potential applications of the anomaly recognition technique for remote real-time power system equipment monitoring based on DCNNs in challenging situations.