AI-Based Forecasting of Hourly Air Temperature in Sub-Saharan Areas of Morocco

Solar energy conversion systems are meticulously engineered to adeptly transform solar energy into electrical energy, thus exemplifying a pivotal focus on sustainable energy initiatives. However, this study focuses on optimizing solar energy conversion by addressing the thermal effect on solar photovoltaic (PV) systems. To predict and improve operating temperature, Deep Learning (DL) techniques are applied, using meteorological data like temperature, dew point temperature, relative humidity, and wind speed. Long Term Memory Network (LSTM), Gated Recurrent Unit (GRU), and the hybrid LSTM&GRU models were evaluated based on Mean Square Error (MSE) and correlation coefficient (R). Each model exhibited noteworthy precision, yet the hybrid model showcased superior accuracy, achieving an R-value of 99.08%. This research proposes an ideal temperature prediction model that can enhance the efficiency of solar PV systems and other solar energy conversion technologies, despite weather fluctuations and environmental conditions.