Numerical analysis of application GARCH to short-term wind power forecasting

For a power grid integrated with wind farms, power output predication of wind turbines would influence the accessibility as well as its economy of dispatch schemes. We applied the GARCH algorithm into construct wind speed model, which has excellent characteristics to tracing the variation of the fluctuated time series. With power curve versus wind speed, wind power is easily gotten by putting the forecasted wind speed into the corresponding equation or curve. Using an example, we verified the validity of forecasting model we proposed. The anticipated wind power would offer dispatch operators a valuable reference. Compared with ARIMA, GARCH demonstrated its advantage in improving prediction accuracy. Using a group of sequences with a variety of volatility clustering to do numerical calculation, the results showed that GARCH has better forecasting performance referred to the sequences which highly fluctuate.