Theoretical And Artificial Neural Network Computation and Analysis Of Global Solar Radiation At Enugu with Atmospheric Parameter

Abstract

Records of measured sunshine hour’s data for a period of seventeen years (1991-2007) at Enugu 7.55⁰N, 6.47⁰E and 141.50m within the south-east zone of Nigeria were used to compute the global solar radiation of the same location with theoretical and artificial neural network models. The first part of the results (ie January to May )  has similar values with the third part (October to December). This indicate that weather conditions of the location  of study is of two periods, rainy and dry seasons. The values of the results also shows that global solar radiation is directly proportional to sunshine hours both for measured, theoretical and artificial neural network computations. Again Correlation of the two models show high performance of neural network over theoretical computation, as the neural network coefficent of determination R² = 0.96 while coefficent of determination of theoretical computation  is R² = 0.91 respectively. The above result shows that the two models has the ability of computing global solar radiation with sunshine hours, but the artificial neural network computation is  more accurate.

Keywords: Atmospheric parameters, theoretical, neural network, Global solar radiation and sunshine hours.