A Novel QR-based Probabilistic Forecasting Method for Solar power Generation

Solar power is quite intermittent in nature, where it is highly dependent on weather conditions. Therefore, accurate forecasting for solar power generation is essential. Probabilistic forecasts with an uncertainty analysis provide a more comprehensive forecast application. Previous research works have proposed numerous methods for probabilistic forecasts, such as quantile regression, interval, density forecasting, and etc. However, these methods still have many drawbacks. Thus, this paper aims to analyze and control the uncertainty of PV power forecasts, and conducts more in-depth researches on the forecasting methods. The main contribution of this paper is to propose a new probabilistic forecasting process, which can distinguish the data variation of power generation at PV sites and automatically select the optimal method to perform the forecasting. This method can not only obtain accurate results, but also potentially reduce the operation time significantly.