Abstract
With the increasing significance of the power control in wind farms for modern power system, the precise ultra-short term wind power prediction are necessary for achieving the integrated active power control of practical wind farms. This paper focus on the wind power online prediction using the relationship between the wind speed and the wind power generation by ultra-term wind speed forecasting. An improved GM (Grey Model) rolling model is developed in the paper for the ultra-short term wind speed prediction. The simulation results have verified that the developed GM rolling model with the weaken buffer operator for the variable wind data preparation has better prediction performance over the traditional GM rolling model and adaptive alpha-based GM rolling model. This result will obviously optimize the wind power control in the wind farm by predicting the precise wind power generation in real time.