PV Array Fault Detection Based on Deep Neural Network

Abstract

This paper develops an artificial intelligence based method to detect different types of faults associated with photovoltaic (PV) arrays. This method is integrated with the powerful deep neural network including multilayer perceptron (MLP) and one-dimension convolutional neural network (1-D CNN). To test and validate the proposed method, a PV system which can simulate typical line-line, line-ground, open-circuit faults is first modeled via Matlab/Simulink and large amounts of normal and fault data are simulated. Then, extensive simulation data are fed into MLP and 1-D CNN to learn the characteristics of different types of faults, and thus detect and distinguish those faults. Finally, the results have shown the high accuracy and effectiveness of the neural network based PV array fault detection method.