An Efficient Protection Scheme for High Impedance Fault Diagnosis in Micro Grid Using Machine Learning

In this paper High impedance faults (HIF) in micro grids show reduced fault current amplitudes along with inherent nonlinearity and unpredictable patterns of operation. The fault current during HIF initiation is further reduced by the potential for unpredictable nature of weather in photovoltaic microgrids and the harmonics generated by nonlinear loads and power electronics equipment. The reduced fault current makes it challenging for conventional over current relays to detect HIF. In response to this challenge, For fault detection and categorization under HIF in microgrids, a strong protection method built on a combined architecture of Discrete Wavelet Transform and Tree of Bagged Decision have been developed. The suggested protection technique involves real-time feature extraction from current signals. The effectiveness of the suggested plan was assessed according to its resilience during HIF, and it outperformed Decision Tree (DT) and Support Vector Machine (SVM) approaches.