Enhanced Electrical System Anomaly Detection: The AGNet Model Approach

Abstract

The complexity and integration of renewable energy in modern power systems present significant challenges for traditional anomaly detection methods. These existing techniques struggle to manage the intricate and heterogeneous nature of contemporary electrical systems, necessitating more effective and precise solutions. To address this, we introduce AGNet, a model that leverages the Attentional GRU-GAN architecture. By integrating GRU for time series analysis, GAN for generating synthetic anomaly data, and attention mechanisms for highlighting critical information, AGNet significantly improves anomaly detection performance. Experimental results show that AGNet consistently outperforms both traditional methods and other advanced models across various datasets, demonstrating superior robustness and generalization. This study advances anomaly detection technology in electrical systems, providing a more reliable and efficient monitoring tool. The innovative AGNet model not only resolves current detection challenges but also establishes a framework for future advancements in power system monitoring.