摘要
铁路客流需求的科学预测是进行运输组织方案决策的重要依据。以高速铁路历史客票数据为基础,结合经验模态分解(empirical mode decomposition,EMD)与机器深度学习中的注意力机制,提出高速铁路日客流量的EMD-Informer组合预测方法。首先采用EMD方法分解高速铁路客流量序列,获得具有周期特征和线路客流内在特征的模态分量,再利用Informer模型分别训练和预测各模态分解分量,并通过多头注意力机制高效挖掘客流数据内在规律和捕捉数据序列中的关键特征,在此基础上重组各分量预测值,从而得到高速铁路日常客流的整体高精度预测值。同时,根据结合问题特征的大量实验,明确可供实际运用参考的超参数设置规则。基于京沪高速铁路全线的实例计算分析表明,相对对比预测方法,EMD-Informer组合预测方法在高速铁路客流的单步预测及超前预测上均具有明显更小的预测误差。
The rational prediction of rail passenger flow is an important basis for the decision-making of the transportation organization scheme.Based on historical high-speed rail ticket data,as well as empirical mode decomposition(EMD)and attention mechanism in machine deep learning,an EMD-Informer combined prediction method for daily passenger flow of high-speed railways was developed.First,the EMD method was used to decompose the passenger flow sequence to obtain railway line modal components with periodic characteristics and intrinsic feature.With the modal components trained and predicted by the Informer model,the internal regulation of passenger flow data and the essential features of flow data sequence were captured by the multi-head attention mechanism.Then the high-accuracy predictive value of high-speed rail passenger flow could be obtained by the reconfiguration of the predictive values of modal components.Meanwhile,according to a large number of experiments and the characteristics of problems,hyperparameter setting rules were formulated which could be used in actual situation.The numerical analysis of Beijing-Shanghai high-speed railway shows that,compared with existing typical prediction methods,the EMD-Informer method has significantly smaller errors in both single-step and multi-step passenger flow prediction.
作者
秦进
胡冉
毛成辉
小虎
徐光明
QIN Jin;HU Ran;MAO Chenghui;XIAO Hu;XU Guangming(School of Traffic and Transportation Engineering,Central South University,Changsha 410075,China)
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2024年第2期1-11,共11页
Journal of the China Railway Society
基金
国家自然科学基金(72171236,U2034208)
湖南省自然科学基金(2022JJ30057,2022JJ30767)。
关键词
高速铁路
客运量预测
经验模态分解
注意力机制
Informer模型
high-speed railway
prediction of passenger flow
empirical mode decomposition
attention mechanism
Informer model
