Chapter 2 - Intelligent Data Analytics for PV Fault Diagnosis Using Deep Convolutional Neural Network (ConvNet/CNN)

Abstract

A deep neural network (DNN) using ConvNet/CNN based algorithm has been proposed for automatic fault diagnosis of photovoltaic module (PVM) and localize the anomaly condition in an interconnected photovoltaic (PV) system. The identification of PVM faults have been diagnosed previously by using several intelligent approaches such as nonlinear auto regression neural network with eXogenous inputs (NARX), backpropagation neural network (BPNN), probabilistic neural network (PNN), artificial neural networks (ANN), support vector machine, and fuzzy logic (FL) and its hybrid approaches. However, such types of approaches had not explored the area of PV fault diagnosis with nonlinear load and perturb loading condition. Also, a DNN-based solution for the diagnosis of the type of nonlinear PV faults has not been reported in the literature till date. Hence, to bridge this research gap, real-time experimental image dataset of different PVM faults condition have been used. The colored images of the PV cell of each type of fault in Red-Green-Blue (RGB) index provide the best visual features for PV fault diagnosis. The proposed DNN model using ConvNet/CNN algorithm has been tested multiple times to make its adoptability for online diagnosis of PV module. The obtained results during training and testing phase show its outperformance for PVM failure analysis.