Data-Driven Models for Predicting Solar Radiation in Semi-Arid Regions

Solar energy represents one of the most important renewable sources contributing to transition process. Considering that observation daily global solar radiation (GSR) is not affordable in some parts globe, there an imperative need develop alternative ways predict it. Therefore, main objective this study evaluate performance different hybrid data-driven techniques predicting GSR semi-arid regions, such as majority Spanish territory. Here, four ensemble-based models were developed by hybridizing Additive Regression (AR) with Random Forest (RF), Locally Weighted Linear (LWLR), Subspace (RS), and M5P. The base algorithms are scarcely applied previous studies radiation. testing phase outcomes demonstrated AR-RF outperform all other models. provided validated statistical metrics, correlation coefficient (R) root mean square error (RMSE). results proved Scenario #6, utilizing extraterrestrial radiation, relative humidity, wind speed, mean, maximum, minimum ambient air temperatures model inputs, leads accurate predictions among scenarios (R = 0.968–0.988 RMSE 1.274–1.403 MJ/m2⋅d). Also, #3 stood next rank accuracy for both validating stations. AD-RF was best predictive, followed AD-RS AD-LWLR. Hence, recommends new effective methods regions.