Wind speed forecasting based on autoregressive moving average- exponential generalized autoregressive conditional heteroscedasticity-generalized error distribution (ARMA-EGARCH-GED) model

With the increase of wind power as a renewable energy source in many countries, wind speed forecasting has become more and more important to the planning of wind speed plants, the scheduling of dispatchable generation and tariffs in the day-ahead electricity market, and the operation of power systems. However, the uncertainty of wind speed makes troubles in them. For this reason, a wind speed forecasting method based on time-series is proposed in this paper. We adopt exponential GARCH (EGARCH) models as asymmetric specifications and GARCH-GED for distribution assumptions. The wind speed series are forecasted by using the autoregressive moving average (ARMA)-GARCH model, ARMA-GARCH-M model and ARMA-GARCH-GED model, respectively, after which the forecasting precision of ARMA-GARCH, ARMA-GARCH-M and ARMA-EGARCH-GED models are compared. The results show that ARMA-EGARCH-GED model possesses higher accuracy than ARMA-GARCH-M model (Lalarukh and Yasmin, 1997), and is of certain practical value. However, this study confirms that the conditional generalized error distribution (GED) can better describe the possibility of fat-tailed, non-normal conditional distribution of returns.