FIELD STUDIES USING A NEURAL-NET-BASED APPROACH FOR FAULT-DIAGNOSIS IN DISTRIBUTION NETWORKS

The paper discusses results of studies performed on a new fault-diagno sis method for distribution systems using acquired field data. The eff ectiveness of the fault-diagnosis method in distinguishing between fau lted conditions and system conditions that appear fault-like is demons trated, for a field-test system, using data recorded at two utility di stribution systems. The new method uses two major components: a signal preprocessor and a novel supervised clustering-based neural network w hich perform fault detection in the presence of arcing, classification of the fault type and preliminary fault location through the identifi cation of the faulted phase. The work represents the first time that a supervised clustering neural network has been used for distribution f ault diagnosis.