Combining Physical Modeling with Artificial Intelligence for Solar Power Forecasting

Modern renewable energy forecasting systems blend physical models with artificial intelligence (AI). This paper describes such a system being developed for the Shagaya Renewable Energy Park in Kuwait. This Kuwait Renewable Energy Prediction System uses both physical modeling and machine learning to produce forecasts for short-range forecasting (i.e. the next six hours) as well as for forecasts several days out. These systems work together to implement the best of our physical knowledge of atmospheric flow with actual observations at the site of interest. The system include numerical weather prediction, blending models, statistical learning models for short range prediction, and an analog ensemble to complete the forecast and quantify the uncertainty. This description includes a preliminary assessment of how each of these models performs.