Key points are not available for this paper at this time.
The precise fault localization holds significant importance in reducing power outage duration and frequency in power systems. The widespread application of synchrophasor measurement technology (PMU) has laid the foundation for achieving accurate fault localization in distribution networks. However, fault localization methods based on PMU often suffer from a significant decrease in accuracy due to topological reconstruction and inaccurate parameters. To address these challenges, this paper proposes a fault location method for distribution networks based on Multi-head Graph Attention Networks (GATs). The proposed method begins by modeling the distribution network as a graph, where nodes represent network components and edges represent the connections between these components. GATs have been employed to learn the underlying relationships between topological structure and electrical characteristics of the distribution network. The results demonstrate that our approach outperforms traditional fault location methods in terms of accuracy and speed. The proposed method achieves high precision which reducing the time required for fault location and enabling faster response times for network maintenance personnel.
Liang et al. (Mon,) studied this question.