Power Quality Disturbances (PQDs) Classification Analyzed Based on Deep Learning Technique

Abstract

Power Quality (PQ) problems in a distributed generation are mainly appeared due to excess non-linear load in the system. Identification and classification are necessary to ensure the reliability of Power Quality Disturbances (PQDs). This study proposed a signal processing and deep learning approach classify the PQDs by applying Discrete Wavelet Transform (DWT), Multi-Resolution Analysis (MRA) and a one-dimensional Convolutional Neural Network (CNN). For speed up in training, the performance of model a signal processing-based DWT-MRA extracted 54 features and fed it into 1D-CNN. Implementation of 1D-CNN seems more reliable than other machine learning approaches. Simulation results showed good performance and classification of data efficiently. Hence, the proposed approach could open a new era for PQDs in PV/wind smart grid in the near future to obtain more efficient outcomes.