Abstract :
Wind power forecasting falls broadly into two headings: statistical or physical modelling. Statistical modelling makes no real attempt to model the physical processes that drive the wind, but relies on previous wind speed patterns over time to make a prediction. In contrast, physical modelling tries to understand the physical processes that drive the wind, such that with sufficient input data, a model of the physical system can make a prediction some time into the future. In practice, a combination of statistical and physical modelling is used to provide an optimum wind power forecast over a time-horizon that may stretch out to 48 hours or more. To this end, a range of different ´model chains´ can be used to forecast wind power.
Keywords :
atmospheric techniques; electricity supply industry; power system planning; statistical analysis; weather forecasting; wind power; 48 hour; forecast time-horizon; model chains; model input data; optimum wind power forecast; physical modelling; statistical modelling; wind physical processes; wind power forecasting; wind speed patterns;