Study on the adaptability of day-ahead wind power forecast system for on-site use

The precision of wind power forecast system (abbr. WPFS) depends on many field situations, such as geographical conditions, the quality of numerical weather prediction (abbr. NWP), and historical data acquisition. The forecast models and their specifications differ case by case. It is necessary to provide any efficient methodology to design an adaptive WPFS. This paper analyzes, evaluates and then handles the affecting factors, including the space and temporal limitation of NWP, the misplacement error of grid point forecasts and the regression of wind power output model. A systematical solution which is independent on forecasting models is then proposed and verified to improve the adaptability of WPFS that consists of pre-processing of input data, setup of wind power output model and forecast error evaluation.