Voltage prediction of a photovoltaic module using artificial neural networks

Appropriate performance prediction of photovoltaic (PV) systems plays an important role in simulation, optimization, design, and control. Mathematical models are usually used to predict PV system behavior, but the main disadvantage of mathematical modeling is the dependency of the models parameters to the operating conditions so that a given set of operating conditions needs a corresponding set of parameters. This drawback greatly restricts the models application. This paper investigates the voltage prediction of a PV module as a function of current, temperature, and solar irradiance by using artificial neural networks (ANNs). For this aim, two ANNs including back propagation (BP) and radial basis function networks are constructed, tested, and compared for modeling of an amorphous silicon PV module. The performance of the BP network is investigated by varying number of neurons, and number of hidden layers and training algorithms. Simulation results indicate that the BP network with one hidden layer produces more accurate results than the other studied networks. Copyright (c) 2013 John Wiley & Sons, Ltd.