Short-term Wind Power Forecasting Using Advanced Statistical Methods

This paper describes some of the statistical methods considered in the ANEMOS project for short-termforecasting of wind power. The total procedure typically involves various steps, and all these steps are described in the paper. These steps include downscaling from reference MET forecasts to the actual wind farm, wind farm power curve models, dynamical models for prediction of wind power or wind speed, estimating the uncertainty of the wind power forecast, and finally, methods for upscaling are considered. The upscaling part considers how a total regional production can be estimated using a small number of reference wind farms. Keywords: Forecasting, power curve, wind farmpower curve, upscaling, uncertainty estimation, probabilistic forecasts, adaptation.