Automated anomaly correction for substation remote monitoring equipment using a bayesian network-fault tree analysis framework

Abstract

This study addresses the persistent challenge of coordinate drift in substation monitoring systems, where prolonged operation leads to progressive deviations from preset configurations. An optimized BN-FTA framework that integrates Bayesian Networks (BNs) with Fault Tree Analysis (FTA) is proposed to improve anomaly detection in electrical equipment. The BN is incorporated into the FTA-based feature matching process to filter out abnormal data points during feature point matching. Experimental results demonstrate significant performance improvements, including a 25.16% reduction in feature mismatch errors and consistent correction times below 0.71 s. The system achieves a 35.1% increase in image rectification accuracy while maintaining computational efficiency. These advances provide a reliable solution for automated inspection systems that require long-term operational stability. The framework’s lightweight architecture and enhanced precision make it well-suited for substation environments where both accuracy and efficiency are critical. A comparative analysis confirms its superiority over conventional single-technique approaches, with measurable gains in both recognition accuracy and processing speed. This research offers practical solutions for automated inspection of power facilities, delivering robust performance under real-world operating conditions. Furthermore, the successful implementation validates the effectiveness of combining BNs and FTA for industrial monitoring applications.