Investigate the photovoltaic (PV) module performance using Artificial Neural Network (ANN)

Solar energy source is one of renewable source and has been dealing with a research for over a few decades, in order to investigate or develop new method which can increase the performance of solar power. However, there is a limited system used which is able to predict the performance of PV module in term of solar radiation. Then, this study explores the potential of artificial neural network (ANN) to be applied in the system of prediction as fast method used for forecasting output of solar radiation for PV module. The purpose of this study is to predict the solar radiation that receives by PV module. There are five inputs; air temperature, minimum temperature, maximum temperature, relative humidity and wind speed use to predict output solar radiation. The multilayer perceptron (MLP) with back propagation is used as a network topology. The ANN that has an adaptive ability in which the activation function for hidden layer is tansig and the output layer is linear form (purelin). The training algorithm is used Levenberg-Marquardt back propagation which this method is the fastest method and much better learning rate than the other training algorithm. The result of developing a model ANN model, show that network performance best fit of regression, R is 65.4%. The performance MSE for training set is 0.0670, testing set is 0.1169 and validation set is 0.1117. Thus, this network is suitable to do the prediction.