Fault Detection in Photovoltaic Systems Using Optimized Neural Network

Fault detection in photovoltaic (PV) arrays is one of the prime challenges for the operation of solar power plants. This paper proposes an artificial neural network (ANN) based fault detection approach. Partial shading, line-to-line fault, open circuit fault, short circuit fault, and ground fault in a PV array have been investigated, and a data set is synthesized to evaluate the impact on maximum power amplitude and number of power peaks under various exposure of irradiance and temperature. The ANN model has been trained considering irradiance, temperature, maximum power, and the number of power peaks corresponding to the different faulty conditions and non-fault situations. The considered ANN model has been optimized in order to increase the accuracy of fault identification. A particle swarm optimization-based algorithm has been employed to find the optimum number of neurons in the hidden layers to achieve the highest possible prediction accuracy on the test data set. The performance of the optimized neural network has been further cross-validated by an arranged data set containing all the types of faulty conditions. The effectiveness of the proposed technique is verified by comparing the results with existing methods.