Component GARCH-M type models for wind power forecasting

Wind power forecasting is one of the most important aspects for power system with integration of wind power. In this work, Component GARCH-M (CGARCH-M) model is presented for short-term wind power forecasting (STWPF). Moreover, asymmetric and distributional considerations are taken into account to generalize the CGARCH-M type models. The CGARCH-M type models can decompose the volatility structure of wind power series to the permanent component and the transitory component, such as to analyze the relative different influence of components in the volatility of wind power series and improve forecasting performance. By means of the Conditional Maximum Likelihood Estimation (CMLE) method, the parameters are estimated for all the proposed models. Furthermore, with the theoretical derivation, the Augmented News Impact Surface (ANIS) is proposed and the impact of news to the wind power volatility is highlighted. Five-minute ahead short-term wind power forecasting is carried out based on historical coastal wind power data of East China, study results clearly validated that the proposed CGARCH-M type models can provide effective forecasting results and outperform the alternative short-term wind power forecasting models.