High Impedance Fault Localization Using Discrete Wavelet Transform for Single Line to Ground Fault

Abstract

This paper presents a new approach to determine the high impedance fault location in a distribution network using discrete wavelet transform (DWT). The technique comprises three stages which are fault impedance identification, faulty section localization and fault distance estimation. First, the transient voltage and current waveforms are analyzed using DWT to obtain the energy values of its coefficients. Then artificial neural network is utilized to predict the fault impedance value. Next, the database and trigonometry techniques are used to localize the faulty section and fault distance successively. The proposed method is used to detect single line to ground faults on a 38-node distribution simulated network created using the PSCAD/EMTDC software. The output waveforms are analyzed using MATLAB. The fault impedance and fault distance can be estimated with errors of less than 0.42 and 2.37%, respectively, while the faulty section can be determined within the 6th rank. The encouraging results show that the approach is capable of determining the fault impedance value, localizing the faulty section and estimating the fault distance under various fault inception angles, fault impedances and fault distances.