High Impedance Fault Detection in Distribution Feeder Based on Maximum Overlap Discrete Wavelet Transform

High Impedance Faults (HIFs) represent a common category of faults in distribution networks. These faults occur when power lines come into contact with high-resistance surfaces. Unlike other types of faults, HIFs exhibit unique characteristics such as non-linearity, asymmetry, randomness, and relatively low fault current levels compared to the feeder load current. It is important to note that the low current magnitude of HIFs poses a significant challenge for traditional protection devices in detecting and distinguishing these faults. This study aims to develop an effective approach for detecting HIFs in distribution feeders by extracting features from the current signal at the substation bus. The Maximum Overlap Discrete Wavelet Transform (MODWT) is utilized in this paper as a current signal decomposition technique. This is followed by reconstructing a signal from the appropriate coefficients, which is used to extract features. These features are then employed as input to the Artificial Neural Network (ANN) for distinguishing HIF events from non-HIF events in the distribution system, such as capacitor bank switching, linear and non-linear load switching, and transformer energization. After conducting rigorous testing under various fault and non-fault event scenarios within a radial power feeder system, the proposed technique has proven to be reliable, robust, and effective for HIF detection, with a high accuracy rate of 99.2%.