Abstract
In this paper, an adaptive model for predicting hourly global, diffuse and direct solar irradiance is described. A dataset of measured air temperature, relative humidity, direct, diffuse and global horizontal irradiance for Jeddah site (Saudi Arabia) were used in this study. Several combinations have been proposed, and the best performance is obtained by using sunshine duration, air temperature and relative humidity as inputs of the developed adaptive
α-model. A good agreement between measured and predicted data is obtained. In fact, the correlation coefficient is more than 97% and the mean bias error is less than 0.8. A comparison between a Feed-Forward Neural Network (FFNN) and the adaptive proposed model is presented in order to demonstrate his performance.