Applications of Graph Computing and Graph Neural Networks in Power Systems: A Survey

With the increased integration of renewable energy sources and power electronic devices, the complexity and uncertainty of power system operations have risen significantly, leading to an evolution in power system computation. Graph computing and Graph Neural Networks (GNNs) offer innovative perspectives and solutions to these challenges. This paper explores the application of graph computing in the analysis of large-scale and complex networks in power systems, as well as the potential of graph neural networks in data-driven power grid operation analysis and decision making. The paper systematically reviews the fundamental theories and key technologies of graph computing and GNNs, along with their specific applications in power systems. These applications include, but are not limited to, load forecasting, state estimation, and fault diagnosis.