MODWT-XGBoost based smart energy solution for fault detection and classification in a smart microgrid

Electrical power being the key driver for any technology driven development, an intelligent technology enabled smart grid which ensures reliable, environment-friendly and power quality certainly provides the necessary fillip to the urban intelligence. This study introduces a novel differential approach of microgrid fault detection and classification as a smart grid enabler. The proposed microgrid protection scheme (MPS) involves an initial phase of pre-processing through anti-aliasing and filtering out of noise of the retrieved system parameters. This is followed by feature extraction process using Maximal Overlap Discrete Wavelet Transform (MODWT) with an abstract wavelet family of mother wavelet 'FejerKorovkin' and three level of decomposition. The differential energy calculated for both three-phase current and its zero-sequence current component at each of the decomposition level of MODWT finally serves as input to an Extreme Gradient Boost (XGBoost) based machine learning model to achieve incipient fault detection and classification. The combination of MODWT and XGBoost as an intelligent MPS working upon a pre-processed de-noised system signals, hitherto untried as per the knowledge of the authors, is tested using standard IEC microgrid test model under varied topological configurations, operational modes, fault conditions, etc. The simulation results, so extensively obtained, prove the effectiveness and robustness of the proposed approach of MPS. The MPS is additionally verified on an IEEE 13 bus microgrid model to reinforce the clam of efficiency.