Wind power forecasting based on econometrics theory

Abstract

wind power forecasting is one of the key technical issues for a power system integrated with a large amount of wind farms. Based on analysis of the variation characteristics of wind speed, we applied econometrics theory into the modeling of wind speed, since GARCH has an excellent performance to tracing the variation of those fluctuating sequences. Using the wind power curve, the power output of a wind turbine is easily acquired from the forecasted wind speed. In reference to our study case, its wind data are input into the established model to verify its validity of the approach we proposed. Therefore, the estimated wind power curve for the next day becomes a valuable reference for the dispatch department of a power grid. Compared with the ARIMA and a typical ANN model, GARCH demonstrates its advantage in improving the prediction precision. In addition, in order to understand the applicability of the GARCH model, many numerical simulations have been done and we found that GARCH has better forecasting performances to those sequences with high fluctuation.