An Artificial Neural Network-Based Fault Detection Technique for PV Array

This paper presents an artificial neural network-based fault detection technique (ANNFD) to detect the short circuit fault (SCF) and hot spot faults (HSF) in PV array. The fault analysis is performed in the MATLAB-Simulink to identify the parameters which provide useful information on the faults. The pattern classification model of artificial neural network (PCANN) is used as a classifier to distinguish the SCF, HSF, and normal operating conditions. Various cases of faults are simulated under different irradiance, temperature, and fault resistance to validate the proposed methods. A sufficient number of training patterns is created through the simulation and used to train the PCANN. In addition, the optimal number of hidden neurons is determined on performance error evaluation. Further, the performance of the ANNFD is evaluated in terms of performance accuracy with the confusion matrix analysis.