Comparison of empirical and artificial neural network models for the correlation of monthly average global solar radiation with sunshine hours in Minna, Niger State, Nigeria

Abstract

Monthly average daily values of global solar radiation and sunshine hours over a period of five years (1987-1991) using artificial neural network were developed to predict global solar radiation at Minna which lies on latitude 09.37°N, longitude 06.32° and 265.4 m above sea level. The results were used to compare results from other researchers of different models in the same area. The correlation coefficient of our model was found to be 0.997. These values were found to be higher than the correlation coefficient of other models. The values from our model and other models were tested in terms of mean percentage error (MPE), mean bias error (MBE) and root mean square error (RMSE) and our model was found to have low error values when compared with other models. From the results of these studies, it was found out that all the models have predicting capacity, but our model has better results. This is being recommended for the prediction of global solar radiation for Minna and areas that have similar climate with Minna. Key words: Global solar radiation, artificial neural network, prediction, measured values, models.