Fault Detection of Power Grid Using Graph Convolutional Networks

Power grids, the linchpin of modern electrical infrastructure, necessitate advanced monitoring systems to ensure operational stability and safety. This paper presents an in-depth investigation into the application of Graph Convolutional Networks (GCN) for fault detection within power grids. Utilizing authentic data collected over two years from a real-world power grid, the research benchmarks the performance of GCN against established algorithms: CNN, LSTM, and ANN. Preliminary findings highlight the unmatched accuracy of GCN, surpassing 91%, emphasizing their proficiency in processing graph-structured data. While CNN and LSTM showcase respectable results, their inherent design indicates certain limitations for grid fault detection. The overarching conclusion suggests a promising avenue for GCN in enhancing power grid monitoring, potentially revolutionizing the methods by which we maintain and secure critical electrical infrastructures.