Graph Neural Network-Based Critical Node Classification in Cyber-Physical Power Systems

Cyber-physical power systems (CPPSs), or smart grids, are essential infrastructures whose resilience is critical for maintaining stable electricity supplies, especially during disasters. However, accurately identifying and classifying critical nodes in CPPSs is challenging due to the complexity of their interconnected structures and dynamic behaviors. In this paper, we introduce a novel method for classifying critical nodes using Graph Neural Networks (GNNs). Our approach uniquely integrates centrality metrics from complex network theory with system-level dynamic measures-specifically supply and controllability-as ground truth labels. Simulation results from a real-world CPPS case study in Northeastern Italy validate the effectiveness of our method, demonstrating its potential to significantly enhance CPPS resilience.