Detecting High Impedance Faults Utilizing Combined Phase Voltages with Neutral Line Current

Abstract

This study aims to present a new approach to detecting high impedance faults (HIFs) in the distribution feeder. Discrete wavelet transformations (DWT) and neural networks (NN) have been widely applied in power system research. Consequently, this study developed a novel technique to discriminate effectively between the HIFs and the switch operations by combining DWT with NN. The proposed approach has three distinct features. First, the input signal of this algorithm is neutral line current, rather than the conventional currents based on three individual phases. Second, HIFs identification uses the details at levels 3, 4 and 5 and the approximations at level 5 of the neutral line current are utilized for. Third, the input signals of the three-phase voltages classify the faulty and healthy phases. The results of simulation and field staged fault clearly show that the proposed technique can accurately identify the HIFs in the distribution feeder.