Due to the wind power time-varying and intermittent characteristics

it is difficult to achieve real time energy balance in a power system. Hence

accurate and stable power forecast will make wind energy a schedulable energy. To deal with wind power uncertainties

a multi-scenario power forecasting method based on the optimal fitting wind power curve for single turbine is proposed. Firstly

atmospheric movement model is discretized using computational fluid dynamics (CFD) speedup matrix

which can reflect wind flow distribution accurately. Secondly

a new variable-weight CFD wind forecast combination method for each wind turbine is put forward considering rotor thrust coefficient of free flow field and wakes of actuator disc. Finally

a new site-specific optimal power curve fitting method is developed based on historical data of wind speed and power measured by wind turbine’s anemometers. The contribution rate is the basis of eliminating illegal points by use of iterative algorithm. The engineering applications show that the proposed method can effectively improve the accuracy of wind power curve fitting and forecasting.