Abstract
Abstract
The integration of distributed generation (DG) changes the magnitude and direction of fault currents in distribution networks, which increases the difficulty of fault section location, especially when feeder terminal unit (FTU) information is distorted. To address this problem, this paper proposes a fault section location method based on an improved binary Fire Hawk Optimizer (IBFHO) for active distribution networks with DG. The overall fault section location procedure is as follows. The network topology, DG switching states and ternary FTU reports are first input to construct full-topology binary candidate sets; Sobol and opposition-based learning then initialize candidate fault-section combinations; during each iteration, switching-function fitness evaluation, elite-guided updating, adaptive mutation and binary repair are performed; when the best fitness stagnates, part of the lower-fitness population is restarted; finally, the highest-fitness section set is output as the location result. These mechanisms are designed specifically for fault location: diversified initialization covers alternative section combinations, binary updating and mutation explore and refine fault hypotheses, and restart prevents continued search around a stagnant combination. The method is evaluated through parameter, ablation, statistical and search-behaviour analyses on the IEEE 33-node and IEEE 69-node systems under varied fault, FTU-distortion and DG conditions. The results show that IBFHO transfers to the larger feeder without retuning and retains reliable or competitive location performance across the investigated conditions. In the representative IEEE 33-node comparisons, it combines complete exact-location reliability with the lowest pooled average first hitting generation under the common-iteration setting. Ablation identifies adaptive mutation as the main source of average improvement and stagnation escape as a residual safeguard, while severe FTU distortion can make the true fault set unidentifiable. These results demonstrate that the proposed method provides reliable fault section location under the investigated active-distribution-network conditions.