基于集成经验模态分解和量子细菌觅食优化的风速预测模型

WIND SPEED FORECASTING METHOD BASED ON EEMD AND QUANTUM BACTERIAL FORAGING OPTIMIZATION

基于风速时间序列内在规律特性,为改善经验模态分解(EMD)模态混叠现象,提出基于集成经验模态分解(EEMD)的风速组合预测模型。同时,针对预测模型参数选择问题,将量子力学的思想引入细菌觅食优化的繁殖算子中,结合量子空间下概率分布模型完成参数寻优。4种算法的参数优化结果表明,改进算法具有更好的全局寻优性能并能提高模型泛化能力。将其应用于组合预测模型中,仿真表明,基于EEMD预测模型能较好地消除EMD的模态混叠现象,具有更高的预测精度。

In view of the intrinsic characteristics of wind speed sequence, a wind speed forecasting method based on empirical mode ensemble empirical mode decomposition （EEMD） was proposed to improve the mode mixing problem of empirical mode decomposition （EMD）. To solve the problem of the uncertain parameters in wind speed forecasting method, a quantum bacterial foraging optimization （QBFO） algorithm was proposed by introducing quantum behavior to the reproduction operator. And then the probability distribution model in quantum space was used to optimize the selection of parameters. The parameter optimizing results of four algorithms show that QBFO has better global search accuracy and generalization performance than other algorithms. By applying the QBFO algorithm to optimize combined forecasting methods, the case study shows the superiority of EEMD in solving the mode mixing problem and the higher accuracy of wind power prediction than EMD model.