Content area
Assessing solar energy potential is vital for developing solar conversion technologies. Despite Algeria's high solar capacity, the country faces challenges due to a limited number of meteorological stations that measure solar radiation (SR). This paper presents a study investigating the performance of innovative hybrid models (HMs) proposed to improve the SR estimation on inclined surfaces over 5-minute intervals. These HMs are formed by combining five empirical models (EMs) and five transposition models (TMs), resulting in a total of 25 models. The 25 HMs are applied to estimate the SR in two locations in Algeria, Bouzareah and Ghardaia. A comparative study is conducted in MATLAB, evaluating the performance of the suggested HMs and five EMs. The findings demonstrate that the HMs significantly enhance accuracy, particularly under cloudy conditions, reducing the normalized root mean square error (NRMSE) by up to 90% in some cases. For example, on August 16th in Ghardaia, the NRMSE decreased from 25% to 5.32% with our hybrid technique, demonstrating its superiority. Unlike traditional clear-sky models, our method performs well on overcast days. For example, on December 10th in Ghardaia, the Bird and Hulstrom model alone produced a high NRMSE of 35% with a KT value of 0.28; however, combining it with the Temps model reduced the NRMSE to 13.37%. In addition, the HMs based on Bird & Hulstrom-Temps provide the most accurate estimates at both locations, with coefficient of determination (R²) values from 0.9788 to 0.9992 and NRMSE values between 3.33% and 19.64%. In contrast, the Davis and Hay-Hay-based HMs offer the lowest R² values and the highest NRMSE and NMBE values.
