Answers are grounded in this paper and cite their sources.
As the penetration of distributed generation (DG) in distribution networks grows, fault currents shift from unidirectional to bidirectional flow, and fault-section location becomes difficult when feeder terminal units (FTUs) are subject to false alarms, missed detections, or false negatives. To address these challenges, this paper proposes a topology-constrained binary fault-location framework for active distribution networks and instantiates it with the topology-constrained adaptive binary Archimedes optimization algorithm (TCAB-AOA). The work corrects three modeling issues common to metaheuristic-based fault-section location: (i) the reverse-current term of the switching function is re-derived so that a downstream DG feeds a reverse current to any network fault rather than only to a fault on its own upstream path; (ii) an unavailable FTU measurement is masked and excluded from the objective rather than equated with a confirmed zero-current reading; and (iii) the FTU observations are generated by an independent nodal-impedance short-circuit model rather than by the switching function itself, so that the reported accuracy reflects generalization to a physical model rather than a self-consistent fit. On this corrected model, the framework integrates hybrid hard-intersection/soft-voting initialization, differential-evolution and Lévy-flight dual perturbation, elite preservation with population restart, and greedy cleanup. On the IEEE 33-node system under nine DG-integrated fault scenarios, TCAB-AOA attains 89.63% average localization accuracy with 0.19 average misjudged sections, significantly outperforming standard AOA (44.81%). A fair-baseline study that equips standard PSO and DE with the same topology-constrained framework yields 88.52% and 92.96%, statistically indistinguishable from TCAB-AOA (p > 0.05, Bonferroni-corrected α = 0.0083), which confirms that the accuracy gain stems primarily from the corrected model and the topology-constrained framework rather than from the specific metaheuristic. The method requires only the existing three-valued FTU upload plus a one-bit availability flag and runs in about 0.5 s per localization, making it directly deployable as a decision-support module on existing feeder-automation systems without additional sensors or communication links.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: