A Comparative Analysis of Different Methods for Predicting Solar Radiation Using Deep Neural Networks: An Innovative Approach to Sustainable Community Building

Abstract: In order to deal with the electrical crisis in an efficient manner, it is imperative to promote the use of renewable energy sources, with a specific emphasis on solar energy. Nevertheless, the challenge lies in the variable patterns of solar irradiance, which are influenced by seasonal weather variations, making it a complex factor to predict. The primary aim of this study is to predict the solar radiation on inclined surfaces, while considering the impact of meteorological variables like temperature, wind speed, humidity, and air pressure. The research used the Artificial Neural Network (ANN) methodology to examine the Douala metropolitan area. Consequently, the model may be used to estimate solar irradiance not only inside the specified study area but also across locations with similar climatic conditions, by using different combinations of input data. The model exhibited its proficiency in appropriately evaluating sun ray intensities by generating a noteworthy outcome via its application using (50 concealed-layer neural network networks along the logistic Sigmoid function. Keywords: solar radiation, neural networks, feed-forward neuron networks, and multilayer perceptron’s.