Network-structured fault location algorithm for distribution networks using multi-source information

In response to the issues of insufficient accuracy and poor adaptability of traditional fault location methods caused by the expansion of distribution network scale, the complexity of topology, and the widespread integration of distributed power sources, this paper proposes a network-based fault location algorithm based on multi-source information. By utilizing dual-clock sources from GPS/BeiDou and the Piecewise Cubic Hermite Interpolation (PCHIP) technique, nanosecond-level spatiotemporal synchronization of multi-source data such as traveling waves, electrical quantities, and communication signals is achieved. A dynamic topology model based on Graph Neural Networks (GNN) is constructed, combined with an improved Dijkstra algorithm, to realize adaptive fault path searching in complex ring networks and multi-source scenarios. Additionally, the introduction of Kalman Filtering (KF) and a dynamic weight allocation strategy helps suppress measurement errors and enhance robustness. The simulation results on the IEEE 33-node distribution network model show that the average positioning error of the proposed algorithm is only 0.48%, and the accuracy of the ring network scene is 98.4%. The comprehensive performance is significantly better than the traditional method, which provides an effective solution for intelligent fault location of distribution network.