Probabilistic Solar Generation Forecasting for Rapidly Changing Weather Conditions

Probabilistic solar generation forecasting provides a better means of quantifying generation uncertainties for power grid operations by providing a range of potential power outputs rather than a single-point estimate. The traditional probabilistic models are unreliable under rapidly changing weather conditions due to fluctuating data correlations, necessitating dynamic modeling of spatio-temporal feature correlations under diverse weather scenarios. The correlations represent the interactions across space and time that reflect the impact of weather conditions on solar power output. This paper addresses this critical problem with a novel method by fusing copula theory and machine learning methods to dynamically quantify the spatio-temporal correlations among meteorological data under diverse weather conditions. The meteorological data and the functions employed to estimate spatio-temporal correlations change dynamically based on weather conditions. A data-driven environment-aware model has been developed to produce probabilistic forecasts from this data, effectively quantifying uncertainty in meteorological data. Case studies on real-world datasets demonstrate that the proposed dynamic method exhibits robust performance in solar irradiance and solar power forecasting. Moreover, the model outperforms state-of-the-art models by up to 60% higher accuracy under non-sunny conditions in autumn and winter.