Assessment of different combinations of meteorological parameters for predicting daily global solar radiation using artificial neural networks

Abstract

In this study, for determining the best-input scenarios of the used parameters in predicting the Daily Global Solar Radiation (DGSR), a new approach based on Artificial Neural Networks (ANNs) was presented. The proposed approach is based on comparisons between all possible input combinations for determining the best scenarios that can give perfect correlations and approximations with DGSR. Recorded data from 35 stations belonging to different climatic zones (27 in Morocco and 8 in neighboring countries) were reported for training and testing the obtained results. The used input parameters include geographical coordinates, sun declination, day length, day number, clearness index (KI), Top Of Atmosphere (TOA), average ambient temperature (Ta), maximum temperature (Tmax), minimum temperature (Tmin), difference temperature (ΔT), temperature ratio (TR), relative humidity (Rh) and wind speed (Ws). The results revealed 128 best-input scenarios, where the first relevant input combination was found for KI, Ta, ΔT, TR and TOA. This result indicated that the best-input scenario for predicting DGSR is based only on three climatological parameters: KI, function of Ta f(Ta) and TOA. In addition, based on these found best-input scenarios and on the least square regression (LSR) technique, 128 new linear relationships between DGSR and the found best-input combinations were developed. The statistical analysis expressed through statistical criteria indicated perfect correlations and approximations between the predicted and measured values of DGSR.