A Novel Deep Learning Model for Classifying Power Quality Problems in PV-integrated Microgrids using CNN-LSTM

Distributed generation based on photo-voltaic (PV) modules play a vital role to provide continuous power supply for consumers. Power quality (PQ) events caused by integration of distributed energy sources in microgrids (MG) deteriorate the operation of distribution and utility of power supply. In this work, a new CNN-LSTM based deep learning approach is proposed to identify power quality disturbances (PQDs) in PV-integrated microgrid. In a proposed framework, full closed-loop neural architecture containing different layers to facilitate automatic feature extraction and long-time short memory (LSTM) network provide temporal feature for intelligent classifier to improve the training speed. Moreover, comparison with other deep neural networks proves that proposed approach can simplify the procedure of PQ problems with the conventional signal analysis and feature selection methods. A typical PV-integrated microgid is simulated on PSCAD software to prove the robustness of classifier for classification of PQDs in renewable energy integrated distribution networks.