Anomaly Analysis of Optical Panels Based on TU-AIS Alert System

With the continuous expansion of the power grid scale, the occurrence of faults has become an inevitable phenomenon. Therefore, rapidly and accurately locating faults has become increasingly crucial. To improve the efficiency and accuracy of fault location in power grids, this paper proposes an optical panel anomaly classification method based on the grey wolf optimized random forest (GWO-RF) model, aimed at addressing optical panel anomalies in terminal unit alarm indication signal (TU-AIS) alerts. The algorithm combines the global search capability of the GWO algorithm with the strong classification performance of RF, enabling precise classification of anomalous optical panels and significantly enhancing fault detection accuracy. The simulation results show that the proposed GWO-RF model outperforms single-metric detection, support vector classification, and RF algorithms in multiple indicators such as accuracy, specificity, recall, and $\mathbf{F 1}$ score, demonstrating the potential and advantages of this method in power grid fault location.