Short-term wind speed forecasting combined time series method and arch model

An ARIMA-ARCH model is presented, which considers the heteroscedastic effect between the fluctuation of wind speed and the characteristics of the change of windspeed, to forecast the wind speed.

Abstract

In order to improve the accuracy of the wind speed forecasting in the wind farm, this paper presents an ARIMA-ARCH model, which considers the heteroscedastic effect between the fluctuation of wind speed and the characteristics of the change of wind speed, to forecast the wind speed. First of all, the ARIMA model for the wind speed time series is built by SPSS. After that, the high lag order ARCH effect is found in the residual of the ARIMA model by Lagrange multiplier (LM) test. At last, the GARCH model is built for simulating the residual series and thus to construct the ARIMA-ARCH model. Numerical experiments demonstrate the superiority of the proposed method when comparing with the traditional ARIMA model.