Improved wind forecasting with wavelets

Abstract

An important consequence of a high wind penetration and the difficulty to predict the power generated is the need to entrust the coverage of a large power variations to power spinning reserve. A forecasting system of generation would resolve, at least in part, this problem. The wind variability has a significant negative influence on the forecast models. To avoid the influence of variability, we applied a wavelet filter that can work with softer signals. The results improve significantly when wavelet filtering is applied.