Solar Forecasting Methods: Transition from Classical to Artificial Intelligence-based Forecasting

The variability in photovoltaic power generation generates different negative effects on power grid systems in terms of stability, reliability, and operation planning. Therefore, an accurate estimation of PV power generation is crucial for stabilizing and securing grid operation and promoting large-scale PV power integration. Every year, new techniques and approaches emerge worldwide that reduce the uncertainty in these estimates and improve model accuracy. This study presents a comprehensive survey of solar energy prediction models while summarizing the most recent methodologies and strategies employed to enhance the precision of solar energy production forecasting. The energy sector is highly dynamic and integrates new technologies continually; hence, compilation studies should be updated with new developments. Industries are trying to benefit from artificial intelligence, including the field of solar energy prediction, and this study attempts to capture this trend. In addition to presenting a comparison of the solution methods in artificial intelligence, hybrid approaches to overcome problems are discussed. Different classification models and critical analyses of recent studies based on forecast horizons and historical data are also presented.