摘要
为提升预测的准确度,提出一种互补集合经验模态分解(CEEMD)、样本熵(SE)、相空间重构(PSR)以及神经网络(BP)的短期风速预测新模型。首先运用CEEMD技术对风速时间序列进行拆解,化繁为简,分离出多个子序列。随后,计算每个子序列的SE,从SE的特征中重组风速序列。继而,将各子序列的预测结果进行相空间重构,获取神经网络预测的输入输出样本。最后运用神经网络预测每个样本,并将所有预测结果累加。此外,还对风电场的实际运行数据进行试验,并将模型的预测结果与其他预测方法进行对比,实验结果显示出此模型在提高风速预测精度方面的显著优势。
Accurate and reliable wind speed forecasting is vital for maintaining the stability of power systems.To improve prediction accuracy,this study introduces a novel short-term wind speed prediction model that integrates complementary ensemble empirical mode decomposition(CEEMD),sample entropy(SE),phase space reconstruction(PSR),and a backpropagation neural network(BP).Initially,CEEMD is employed to decompose the wind speed time series into multiple intrinsic mode functions(IMFs),thereby simplifying the data structure.Following this,the SE of each IMF is calculated,and the wind speed sequence is reconstructed based on SE characteristics.Subsequently,the prediction results of each IMF undergo phase space reconstruction,yielding input-output samples for neural network prediction.Finally,the BP neural network is utilized to forecast each sample,and the predicted values are aggregated.The proposed model is evaluated using real-world data from a wind farm,and its performance is compared with other prediction methods.Experimental results demonstrate that this model significantly enhances wind speed prediction accuracy.
作者
高晟扬
李法社
Gao Shengyang;Li Fashe(Faculty of Mechanical and Electrical Engineering,Yunnan Agricultural University,Kunming 650201,China;School of Metallurgical and Energy Engineering,Kunming University of Science and Technology,Kunming 650093,China)
出处
《太阳能学报》
北大核心
2025年第4期415-422,共8页
Acta Energiae Solaris Sinica
基金
云南省科技厅基础研究项目(202401BD070001-072)
云南省教育厅科学研究项目(2024J0453)
云南农业大学大学生创新项目(XJ2023240)。
关键词
风速预测
样本熵
互补集合经验模态分解
相空间重构
神经网络
时间序列
wind speed forecasting
sample entropy
complementary ensemble empirical mode decomposition
phase space reconstruction
neural networks
time series
