摘要
中国棉花产量组成的时间序列呈非光滑、非单调的复杂分布形式。运用集成经验模态分解(EEMD)和自回归滑动平均模型(ARMA)相结合对其进行预测。首先利用EEMD方法对原始信号进行分解,得到一组平稳的本征模函数和一个具有趋势性的光滑余波,然后运用ARMA模型分别对本征模函数和余波进行预测,最后将二者的预测值合并,实现对中国棉花产量的精确预测。研究表明,EEMD-ARMA组合模型的平均预测误差仅为0.98936%,比单一ARMA模型的平均预测误差减小了45.60280%。根据EMD-ARMA组合模型,预测得2021年中国棉花产量为610.6475万t,这一预测结果比单一ARMA模型的预测结果更合理。
Combining ensenble empirical mode decomposition(EEMD)decomposition and autoregressive moving average(ARMA)model,the output of cotton in China was predicted.Firstly,EEMD technology was used to decompose the original signal to obtain the single frequency and stable eigenmode function and smooth trend residual wave.Then ARMA model was used to predict the eigenmode function and residual wave respectively.Finally,the predicted values of the two were combined to realize the accurate prediction of the target.The results show that the average prediction error of EEMD-ARMA model is 0.98936%,which is 45.60280%lower than that of ARMA model.The EEMD-ARMA model predicts that the output of cotton in China is 6106475 ton in 2021,which is also more reasonable than the predicted value of ARMA model.
作者
王艳
Wang Yan(School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan/China)
出处
《国际纺织导报》
2022年第2期42-47,共6页
Melliand China
关键词
棉花产量
预测
集成经验模态
自回归滑动平均
output of cotton
prediction
ensenble empirical mode decomposition(EEMD)
autoregressive moving average(ARMA)
