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

Abstract

Solar energy has been considered a clean and renewable form of energy to generate electricity. As a consequence, the use of solar photovoltaic energy has recently received increasing attention. However, the intermittent power generation resulting from the random nature of meteorological parameters leads to various challenges for the security and stability of power grids when this renewable energy is integrated into large-scale grids. Therefore, accurate forecasting of solar irradiance is gradually increasing its importance in reducing fluctuations in solar irradiance in system planning. With the development of artificial intelligence technologies, especially deep learning, an increasing number of models are being considered for forecasting due to their superior ability to deal with complex nonlinear problems. This paper presents a forecast of solar radiation based on Long Short-Term Memory (LSTM). Samples of the meteorological parameters from the city of Douala in Cameroon are used to assess the accuracy of the proposed forecast. The experimental results demonstrate that the LSTM has a better prediction performance with the RMSE=0.47W/m2 and MAE= 5.2813W/m2.