An estimation method for wind power forecast errors based on numerical feature extraction

Estimating wind power forecast errors is very important for the dispatch and control as well as security and defense of wind power systems. Based on numerical feature extraction of historical data and day-ahead forecast data, a method of estimating wind power day-ahead forecast errors is proposed. Firstly, the main factors which influence the wind power forecast errors are analyzed. These include the fluctuation and amplitude of wind power, and the forecast method used. The correlations are calculated by data statistics. Then through analyzing the historical wind power data series, the model of wind power forecast error estimation is established using the multivariate linear regression method. Finally, the performance of the proposed method is verified using the wind power data from the Belgium grid's system operator Elia. This method has also been tested in the dispatch system of a northwest province of China. All of the results demonstrate the feasibility and practicality of this method.