High Impedance Fault Detection in Distribution Feeder Based on Frequency Spectrum and ANN

Abstract

High impedance fault (HIF) is among the common faults in distribution networks. This type of fault creates peculiar characteristics in the current signal as a result of the electric arc during the fault, such as irregularity, nonlinearity, and asymmetry, and because this fault occurs when the power line touching with high resistance surfaces, the magnitude of the current drawn during the fault is relatively small compared to the rated load current; therefore, it is difficult for the traditional protection devices to capture it. This paper presents a fault detection method based on the frequency spectrum analysis for the current signal at the substation bus. Fast Fourier Transform (FFT) is proposed in this work as an efficient technique for current signal analysis and extracting harmonics content during HIF and other non-fault events in the distribution system while employing an Artificial Neural Network (ANN) as a features classifier. The ANN is regarded as a vital tool in power system-related applications. It has demonstrated its ability to detect and classify HIF from other normal events such as capacitor bank switching, load switching, and inrush current due to saturation transformer. The results demonstrate that this method has high accuracy (99.34%) for HIF detection with no false positive rate (dependability 100%). MATLAB software (R2021a) is used in this study to perform the simulation.