Solar Irradiance Forecasting Based on Deep Learning for Sustainable Electrical Energy in Cameroon

Abstract

Solar energy is indeed a promising source of clean and renewable electricity generation. It is worth noting that there has been a growing interest in the utilization of solar photovoltaic energy as a sustainable energy source. The integration of renewable energy into large-scale power grids poses challenges to grid security and stability due to the intermittent nature of power generation resulting from meteorological parameters. As a result, accurate solar irradiance forecasting is gradually becoming more crucial for system planning as a means of minimizing solar irradiance fluctuations. Increasingly more models are being taken into consideration for forecasting as artificial intelligence technologies, particularly deep learning, advance owing to their greater capacity to handle challenging nonlinear issues. Our team has developed a cutting-edge forecasting model that utilizes Long Short-Term Memory (LSTM) to accurately predict solar radiation. The accuracy of the forecast is being evaluated using meteorological parameters obtained from the city of Douala in Cameroon. Based on the experimental results presented, it appears that the LSTM model may offer superior prediction performance, as evidenced by the reported RMSE of 0.47W/m 2 and MAE of 5.2813W/m 2 .