Prediction of daily global solar irradiance on horizontal surfaces based on neural-network techniques

In this study, a prediction model of global solar irradiance distribution on horizontal surfaces has been developed. The methodology is based on neural-network techniques and has been applied to the meteorological database of NTUA, Zografou Campus, Athens (37°58′26″N, 23°47′16″E). The investigation of the correlation between weather conditions, duration of daylight and the representative peak value of a Gaussian-type function plays an essential role in the development of the model. The weather conditions are categorized into six different states, whereas the daylight duration is obtained by familiar equations. Thereafter, a correction methodology for the Gaussian-type function—which stands for all six different states—is applied. Finally, the reliability of the developed model is investigated through a suitable validation procedure.