基于分解集成框架的铁路货运量预测方法研究  

Research on Railway Freight Volume Forecasting Method Based on Decomposition Integration Framework

作　　者：曹慧 秦江涛[1] CAO Hui;QIN Jiang-tao(School of Management,University of Shanghai for Science&Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学管理学院,上海200093

出　　处：《计算机技术与发展》2023年第8期192-198,共7页Computer Technology and Development

基　　金：国家自然科学基金资助项目(72174121,71774111)。

摘　　要：铁路货运量时间序列受到多种因素影响,数据具有波动性以及随机性的特征,导致预测精度低下。为了提高铁路货运量的预测精度,提出一种基于分解集成框架的铁路货运量预测方法。首先筛选相关影响因素,并使用主成分分析(Principal Component Analysis,PCA)进行降维,得到主成分之后使用互补集合经验模态分解(Complementary Ensemble Empirical Mode Decomposition,CEEMD)将铁路货运量历史数据分解成较为平稳的分量,用样本熵(Sample Entropy,SE)评估分量复杂度并重组分量,将重组分量与主成分构成新的数据集,最后将新数据集通过秃鹰搜索算法(Bald Eagle Search,BES)优化的极限学习机(Extreme Learning Machine,ELM)的模型中预测,得到重组分量的预测结果,叠加预测分量得到最终预测结果。与其他算法对比分析得出,提出的CEEMD-BES-ELM分解集成方法在铁路货运量预测中具有优越性,能够有效提高铁路货运量预测的准确性。The time series of railway freight volume is affected by many factors,and the data has volatility and randomness,which leads to the low prediction accuracy.In order to improve the prediction accuracy of railway freight volume,a railway freight volume prediction method based on decomposition and integration framework is proposed.Firstly,relevant influencing factors were screened,and Principal Component Analysis(PCA)was used for dimension reduction to obtain the principal components.Then the Complementary Ensemble Empirical Mode Decomposition(CEEMD)was used to decompose the historical data of railway freight volume into a relatively stable component.Sample Entropy(SE)was used to evaluate component complexity and reassemble components,which were combined with principal components to form a new data set.Finally,the new data set was predicted in the Extreme Learning Machine(ELM)model optimized by Bald Eagle Search(BES)algorithm to obtain the prediction result of the reconstituted component,and the final prediction result was obtained by superimposing the prediction component.Compared with other algorithms,the proposed CEEMD-BES-ELM decomposition integration method has advantages in railway freight volume prediction and can effectively improve the accuracy of railway freight volume prediction.

关 键 词：互补集合经验模态 秃鹰搜索算法 极限学习机 主成分分析 样本熵 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]