Comparison of solar radiation models using meteorological parameters

The research is an attempt to overcome a paucity of solar radiation data and forecasting models in Batna, Algeria, using three approaches: regression, fuzzy logic, and artificial neural networks. The following input parameters were used in this work: the measured values of the duration of insolation(S), the time between sunset and sunrise (S0), the daily mean of irradiation at the top of the atmosphere G0 (extraterrestrial), temperature (T), humidity (Hu), pressure (Pr), wind speed (Ws), and rainfall (R). For all three techniques’ models. The data utilized are from the Helio-Clim 1 website and span ten years (1996–2005). The statistical evaluation criteria are utilized to determine the effectiveness of the suggested paradigm. The correlation rate for the forty models used is more than 82%. The trials undertaken for this study lead us to the following conclusion: humidity is the most important meteorological parameter for Multiple-Layer Perceptron (MLP) and regression predictors, whereas pressure and precipitation are dominant for Fuzzy 3 and Fuzzy 7, respectively. Furthermore, when compared to empirical and fuzzy models, the proposed Artificial Neural Network (ANN) models have shown enhanced prediction skills.