Component decomposition to improve precision of short-term wind speed forecasting

Wind speed forecasting is one of the most important technologies to guarantee wind energy integrated into the whole power system smoothly. In this paper, a novel idea is proposed that decomposing effectively different frequency components from wind speed time series contributes to improve short-term wind speed forecasting precision. Based on empirical mode composition, an improved algorithm is presented to reduce the interference between different frequency components. Analysis and simple tests of sinusoidal signals are used to verify that better decomposition can result in more accurate forecast. Case studies of real wind speed data are carried out to investigate the validity of the new algorithm. The results show that the forecasting accuracy of the novel approach can be raised by 4~6% compared to methods without decomposition and increased by 1~2% with conventional decomposed methods.