Kurtosis analysis of the wind speed time series in wind farms

Wind speed forecasting is essential to wind power generation and operation. To improve the performance of wind speed forecasting, analyzing the characteristics of the wind speed time series is necessary, comprehensively, since wind speed change is irregular and may fluctuate violently. In this study, using the technique of the GARCH model kurtosis analysis, the leptokurtosis of wind speed is discussed, and the KA-GARCH theorem is induced and proved, based on the kurtosis definition commonly used in power systems literature. In the case study, GARCH models with three different conditional distributions are presented to simulate the overall kurtosis of wind speed data and a feasible scheme for choosing the best conditional distribution in wind speed forecasting GARCH model is proposed.