基于CEEMD-SE-MM的中长期风速模拟方法

Mid-and long-term wind speed simulation method based on CEEMD-SE-MM

高精度的风速模型对风资源的开发与利用具有重要意义。提出一种基于完全集合经验模态分解-样本熵-马尔可夫模型(CEEMD-SE-MM)的中长期风速模拟方法。利用CEEMD法对风速序列进行特征提取,将风速序列分解成一组固有模态函数和残差;以SE为特征归类固有模态函数合成新模态分量;基于MM对新模态分量片段进行谱聚类;拟合波动片段时长并整合聚类结果得到双层多轨风速模型;在考虑各新模态分量之间相关性的前提下采用双层抽样完成风速模拟。与马尔可夫链蒙特卡洛和改进马尔可夫链蒙特卡洛的结果对比表明,所提风速模型及模拟方法较好地保持了原始风速序列的时序特性和概率特性,且具有更高的精度。

Wind speed model with high precision is of great significance to the development and utilization of wind resources. A mid-and long-term wind speed simulation method based on CEEMD-SE-MM(Complete Ensemble Empirical Mode Decomposition-Sample Entropy-Markov Model) is proposed. The CEEMD method is adopted to extract the characteristics of wind speed sequence and decompose the wind speed sequence into a set of IMFs(Intrinsic Mode Functions) and residual. The SE is taken as characteristic to classify the IMFs and synthesize new IMFs. Spectral clustering of fragments of new IMFs is performed based on MM. The bistratal multi-orbit wind speed model is obtained by fitting the duration of fluctuation fragments and integrating the clustering results. The bistratal sampling is adopted for wind speed simulation under the premise of considering the correlation among new IMFs. Compared with the results of MCMC(Markov Chain Monte Carlo) and improved MCMC,the proposed method well maintains the timing and probability characteristics of the original wind speed sequence with higher precision.