LAUSR

Abstract With the world-wide deployment of solar energy for a sustainable and renewable future, the stochastic and volatile nature of solar power pose significant challenges to the reliable, economic and secure operation of electrical energy systems. It is therefore imperative to improve the prediction accuracy of solar power to prepare for the unknown conditions in the future. So far, artificial intelligence (AI) algorithms such as machine learning and deep learning have been widely-reported with competitive prediction performance because they can reveal the invariant structure and nonlinear features in solar data. However, these reports have not been fully reviewed. Accordingly, this paper provides a taxonomy research of the existing solar power forecasting models based on AI algorithms. Taxonomy is a process of systematically dividing solar energy prediction methods, optimizers and prediction frameworks into several categories based on their differences and similarities. We also present the challenges and potential future research directions in solar power forecasting based on AI algorithms. This review can help scientists and engineers to theoretically analyze the characteristics of various solar prediction models, thereby helping them to select the most suitable model in any application scenario.