Study of the short-term wind power forecasting method for complex terrain wind farm based on the CFD dynamical downscalling

The principal factors for short-term wind farm power forecasting accuracy are terrain complexity which will cause large variability in the wind field in space and time. Based on the combination of mesoscale numerical weather prediction model, and microscale Computational Fluid Dynamics (CFD) model, a short-term wind power dynamical downscaling forecast system is therefore presented. The system is composed of three parts: mesoscale numerical weather prediction model, microscale wind farm characteristic database and wind power forecasting integrated system. This system is able to provide the power forecasting of every turbine in the wind farm in the next 72 hours with 15-minute time step. The influence of turbine maintenance schedule and grid limitation is considered in the prediction result. A test from July 2014 to January 2015 indicates the monthly wind power forecasting error is less than 20%, which conforms to the requirement of the standards. The dynamical downscaling forecast method is not constrained by the terrain complexity or quantity of wind farm histoical data, which means that the system is also applicable on newly-built wind farm.