一种基于EMD的短期风速多步预测方法

A Novel Multi-Step Prediction for Wind Speed Based on EMD

针对风速时间序列的非线性和非平稳性,提出了一种基于经验模式分解(EMD)的短期风速多步预测新方法。该方法首先对风速时间序列进行EMD处理,将其分解为一系列相对平稳的分量,以减少不同特征信息之间的干涉;然后利用游程判定法,将波动程度相近的分量重构为高-中-低频三个分量,使所得分量特征信息集中且预测分量大幅减少;之后针对三分量的特征分别建立相应的多步预测模型;最后将三分量的多步预测结果进行自适应叠加作为最终的预测风速。算例结果表明,运用本文方法使风速多步预测的精度得到了大幅提高,同时在风速波动剧烈时也能保证较好的预测效果。

Aiming at the nonlinearity and nonstationarity of wind speed sequences, a novel multi-step prediction for wind speed is presented. The prediction is primarily based on empirical mode decomposition （EMD）. By means of the EMD technique, the original wind speed sequences are firstly decomposed into a series of functions with more stationary variation. Thus the interferences among the characteristic information embedded in the wind speed can be weakened. Then these functions are reconstructed into three components（high-middle-low frequency components） according to their run-lengths. As a result, not only the characteristics become more centralized, but the predicted components can be greatly reduced. After that, three multi-step prediction models are built on the basis of their respective variation rules. Finally, the prediction values corresponded to the three components are adaptively superposed to obtain the predicted wind speed. A real example is given in this paper. The obtained results show that the proposed approach possesses higher accuracy and the prediction performance is satisfied when the wind speed sharply fluctuates.