Comparative Analysis of Empirical and AI-Supported Models in Global Solar Radiation Prediction for İzmir Province

In this study, the performances of different models that can be used to predict global solar radiation for İzmir province were analyzed comparatively. Using ATATEK-Solar software, 14 empirical models commonly used in the literature and a newly developed AI-supported model were tested. Each model was analyzed using three different optimization algorithms (Nelder-Mead Simplex, Pattern Search, Simulated Annealing). Long-term average meteorological data obtained from Turkish State Meteorological Service were used. According to the analysis results, Model 15 performed the most successful predictions with RMSE:0.1451 and R²:0.9995 values. This was followed by Model 5 with RMSE:0.2016 and R²:0.9990 values and Model 6 with RMSE:0.2017 and R²:0.9990 values. When model performances were examined on a monthly basis, it was observed that the lowest prediction errors occurred in spring and summer months. As a result of the study, it is recommended to use Model 15 in evaluating the solar energy potential of İzmir province and system design.