The increasing complexity of power quality disturbances (PQDs) in modern power grids, presents significant challenges and needs attention on effective detection and classification algorithms. This study employs a time-series-based maximal overlap discrete wavelet transform (MODWT) for feature extraction. This approach facilitates a detailed analysis of 20 distinct types of power quality disturbances that are generated in the MATLAB/Simulink environment and then applied MODWT to get features such as amplitude, phase, energy, and entropy. After feature extraction, principal component analysis (PCA) is applied for feature selection. The classification model is built using extreme gradient boosting (XGBoost) and dataset is separated such that for testing 20% and for training 80% is utilised. This proposed work introduces hybrid technique capable of effectively detecting PQDs, achieving classification accuracy of 99.92%. This proposed topology demonstrates its effectiveness in power quality enhancement for electric vehicle (xEV) charging infrastructures by improving harmonic distortion and voltage stability.
Budumuru et al. (Thu,) studied this question.