AI-Driven Global Solar Radiation Prediction: Harnessing Machine Learning and Satellite Imagery for Accurate Forecasting
Authors:-Mrs.A.Srujana Jyothi, M.Siri Sathvika, M.Madhur, I.Chathurya, G.Ram Subhash, K.V.K.Varma
Abstract-Accurate prediction of Daily Global Solar Radiation (DGSR) is crucial for applications in renewable energy, agriculture, and climate studies. This paper explores the effectiveness of Machine Learning (ML) algorithms and satellite imagery in enhancing DGSR prediction accuracy. Traditional ML models typically rely on various meteorological parameters (e.g., temperature, wind speed, atmospheric pressure, and sunshine duration) and radiometric parameters (e.g., aerosol optical thickness, water vapour). In this study, we investigate the impact of incorporating normalized reflectance from satellite images across different spectral channels to improve prediction accuracy. We employ two ML-based regression models: Artificial Neural Networks (ANN) and Support Vector Machines (SVM). The results indicate that the selection of input parameters significantly affects the accuracy of daily solar radiation forecasts. Moreover, the ANN model outperforms SVM, demonstrating superior predictive capability.
