Assessing the Generalization of Graph Neural Networks for Fault Location Across Increasing Distributed Energy Resource Penetration Levels

2026-07-31

AI Summary

This study addresses the challenges posed by high penetration of distributed energy resources (DERs), which complicate fault characteristics in distribution networks and undermine conventional fault location methods due to bidirectional power flows and intermittent generation. The authors systematically evaluate the generalization capability of spatio-temporal graph neural networks for fault location across varying DER penetration levels, benchmarking STGATv2, GATv2, and GRU on a modified IEEE 123-node system with multiple DERs. They reveal an asymmetric generalization behavior: models trained at 50% penetration generalize robustly to low-penetration scenarios, while those trained at only 10% maintain macro F1 scores of 81–84% at 50% penetration—significantly outperforming baselines. STGATv2 achieves 92–94% macro F1 in-distribution and retains over 85% under measurement noise, demonstrating the critical role of spatio-temporal graph attention in enhancing topological awareness and robustness.

Abstract

Accurate fault location is critical for distribution network reliability. However, increasing distributed energy resource (DER) penetration complicates fault location due to intermittent generation and bidirectional power flows that reshape fault signatures. Spatio-Temporal Graph Neural Networks (STGNNs) have shown promise by jointly modeling spatial and temporal dependencies, but their behavior under increasing DER penetration has not been studied rigorously. In this paper, we (i) systematically benchmark spatio-temporal graph attention network (STGATv2) against purely temporal (gated recurrent unit, GRU), purely spatial (GATv2) and traditional machine learning baselines, and (ii) evaluate how well models generalize across increasing DER penetration levels (10%, 25%, 50%) on a reconfigured IEEE 123-bus feeder with multiple DER injection points and moderate-to-high impedance faults. Results show that STGATv2 consistently outperforms neural baselines, achieving 92-94% macro F1 in-distribution. Notably, generalization across penetration levels is asymmetric: training at 50% penetration retains near in-distribution F1 score at lower levels, whereas training at 10% degrades considerably at 50% - with STGATv2 retaining 81-84% F1 under these drastic shifts, substantially higher than GATv2 and GRU which drop to 69-74% F1 and 73-75% F1 respectively. Under realistic measurement noise, STGATv2 maintains > 85% F1, while GRU drops as low as 33.5% F1, highlighting the critical role of topological awareness for robust fault location in active distribution networks.

Problem

• fault location
• distributed energy resources
• generalization
• graph neural networks
• DER penetration

Innovation

• Spatio-Temporal Graph Neural Networks
• Fault Location
• Distributed Energy Resources
• Generalization
• Graph Attention Network