On Fast N-1 Contingency Analysis: A Graph Neural Network Approach
Ningkai Tang, Tao Zhang, Junjun Liu, Jixiang Lu, Weiyong Yang · 2024
Rapid N-1 contingency analysis is crucial for ensuring the stability and reliability of power systems, but traditional methods can be computationally prohibitive for large-scale grids. This paper proposes a novel approach leveraging graph neural networks (GNNs) to accelerate this process. By exploiting the inherent graph structure of power systems and the learning capabilities of GNNs, this study train specialized models to predict post-contingency branch flows for each credible N-1 contingency. This eliminates the need for repeated online power flow (PF) calculations, significantly reducing computational burden. Evaluation on the IEEE 118-bus test system demonstrates a remarkable 89.33% reduction in computation time compared to the Newton-Raphson (NR) method, while maintaining high accuracy exceeding 97% and outperforming other GNN architectures. This fast and accurate GNN-based approach promises to enable real-time contingency analysis and enhance situational awareness for improved grid resilience and operational decision-making.