A Method for Fault Location in Distribution Networks Based on Graph Neural Networks and Improved Convolutional Neural Networks
Geng Sun, Xuefeng Liu, Runze Fu, Bin He, Jun Shang · 2024
The reliable operation of distribution networks is crucial for societal stability, and fault location is a key technical challenge for ensuring their reliability. This paper proposes a fault location method for distribution networks based on Graph Neural Networks (GNN), which effectively extracts fault features by constructing a topological graph model and utilizing an improved Graph Convolutional Network. To address the issue of sample imbalance, a strategy adjusting the loss function is employed to enhance the model's performance. Simulation experiments demonstrate that the proposed method outperforms traditional approaches in fault detection and location and exhibits robustness against noise and data incompleteness. The research provides a new perspective for fault location in distribution networks and showcases the potential of GNN applications in power systems.