Wavelet-Gaussian process regression model for forecasting daily solar radiation in the Saharan climate

Forecasting solar radiation is fundamental to several domains related to renewable energy where several methods have been used to predict daily solar radiation, such as artificial intelligence and hybrid models. Recently, the Gaussian process regression (GPR) algorithm has been used successfully in remote sensing and Earth sciences. In this paper, a wavelet-coupled Gaussian process regression (W-GPR) model was proposed to predict the daily solar radiation received on a horizontal surface in Ghardaia (Algeria). For this purpose, 3 years of data (2013-15) have been used in model training while the data of 2016 were used to validate the model. In this work, different types of mother wavelets and different combinations of input data were evaluated based on the minimum air temperature, relative humidity and extraterrestrial solar radiation on a horizontal surface. The results demonstrated the effectiveness of the new hybrid W-GPR model compared with the classical GPR model in terms of root mean square error (RMSE), relative root mean square error (rRMSE), mean absolute error (MAE) and determination coefficient (R-2). Design of PV systems requires knowing the solar radiation at a specific location. If monitoring stations don't exist, predictive models can be used. A Wavelet-coupled Gaussian Process Regression (W-GPR) model is compared with other models to predict the daily solar radiation in Ghardaia, Algeria.