HIF detection in distribution networks based on Kullback–Leibler divergence

Abstract

A high-impedance fault (HIF) is mostly associated with the arcing phenomenon, and its main features are low current, randomness, non-linearity, and asymmetry. The low fault current of an HIF is a major challenge for overcurrent-based protection system of distribution networks. Power loss, potential fire hazard, and electric shock are consequences of an undetected HIF. This study presents a time-domain HIF detection algorithm based on monitoring the substation current waveform. Using the Kullback–Leibler divergence similarity measure, the non-linearity, and asymmetry features of two subsequent half cycles of the current waveform are quantified as the HIF detection criterion. A time duration-based criterion is also used to distinguish HIFs from the load, capacitor, feeder, and distributed energy resource switchings and the voltage sag and swell events. The proposed scheme satisfactorily works in the presence of non-linear loads. Also, it does not require the training data set, transformation, and calculation of harmonic/sequence components. Extensive time-domain simulation case studies using the IEEE 13 and 34 node test feeders demonstrate the merits of the proposed HIF detection algorithm.