Comparison of different artificial neural network techniques in prediction of solar radiation for power generation using different combinations of meterological variables

The main objective of present study is to compare ANN model develop with neural network fitting tool (nftool), Radial Basis Function Neural Network (RBFNN) in predicting solar radiation for power generation. The three combinations of input variables are considered for prediction. The RBFNN utilizing input parameters as latitude, longitude, height above sea level and sunshine hours has mean absolute percentage error (MAPE) of 4.94% and absolute fraction of variance (R2) of 96.18% respectively and it give better results than conventional solar radiation prediction models (Angstrom, Akinoglu and Ecevit, Bahel, Almorox and Hontoria). Therefore RBFNN can be used for prediction of solar radiation for solar power generation.