A symptom-driven fuzzy system for isolating faults

The authors discuss work toward the simplification of fault isolation within distribution networks using a symptom-driven fuzzy system. Based on a symptom description and situation-specific background data, the system draws upon prior experiences with distribution networks to determine which component is most likely to be at fault. The fuzzy system developed narrows the search to a relatively few suspicious components by means of the symptom description and supporting data. At the same time, the system draws upon the prior experiences with networks to determine the membership grade (i.e., the degree to which each suspicious component is indeed faulty) and executes an investigation that focuses on the most likely faulty component. The system has been applied to fault isolation for a distribution network in the west of Taipei which contains four substations and 38 feeders. The results from several trials indicate that the fuzzy system performs well in minimizing the amount of time taken to isolate faults