Efficient data filtering for wind energy assessment

Abstract

The installation of a new wind farm requires the previous acquisition of a great amount of wind speed data in order to perform a correct wind energy assessment. These data are obtained from a meteorological mast installed at each site. Real wind measurements at the meteorological mast include many erroneous data, which must be detected and eliminated before working with them. To localise wrong data, a new method based on the Kalman filter is proposed. This filter is applied to a short-term prediction like the one used in wind power forecasting. The filter parameter tuning is based on general verification metric curves tested with real data from different sites. The filter equations are adapted to different sites depending on the intensity of turbulence and wind direction. In this study all the parameters are explained in detail and the filter is evaluated, showing nice results for wind energy assessment of candidate sites.