城市轨道交通换乘站点的客流预测方法  

Passenger Flow Prediction Method for Urban Rail Transit Transfer Stations

作　　者：段文杰 朱广宇 DUAN Wenjie;ZHU Guangyu(Chongqing Railway Group Co.,Ltd,Chongqing 401120,China;School of Automation and Software Engineering,Shanxi University,Taiyuan 030031,China)

机构地区：[1]重庆轨道集团有限公司,重庆401120 [2]山西大学自动化与软件学院,太原030031

出　　处：《综合运输》2025年第3期63-68,153,共7页China Transportation Review

基　　金：国家自然科学基金(62433005,62272036)。

摘　　要：换乘站点是城市轨道交通系统中的重要部分,是连接不同线路的关键节点。换乘站点的客流数据中存在多元复杂的时空关系,传统方法无法很好地捕捉这些特征。针对这个问题,本文提出一种基于多尺度时空注意力图卷积(M-STAGCN)的城轨换乘客流预测模型。首先,分析城轨换乘站点的客流的特性。其次,利用时空注意力机制和图卷积网络构建客流预测模型,其中包括四个不同尺度时间分支模块构成:最近的、每日、每周和长期。最终通过实验表明,相较于其他基准模型:ARIMA(Autoregressive Integrated Moving Average model)、LSTM(Long short-term memory),本文提出的客流预测模型在多项评价指标中均更优,证明了本文工作的有效性。The transfer station is an important part of the urban rail transit system and a key node connecting different lines.There are multiple and complex spatiotemporal relationships in the passenger flow data of transfer stations,and traditional methods cannot capture these characteristics well.To solve this problem,this paper proposes a prediction model for urban rail transfer passenger flow based on Multi-scale Spatiotemporal Attention Graph Convolution(M-STAGCN).Firstly,the characteristics of passenger flow at the urban rail transfer station are analyzed.Secondly,the spatiotemporal attention mechanism and graph convolutional network are used to construct the passenger flow prediction model,which includes four time-branching modules at different scales:recent,daily,weekly and long-term.Finally,experiments show that compared with other benchmark models,such as ARIMA(Autoregressive Integrated Moving Average model)and LSTM(Long short-term memory),the passenger flow prediction model proposed in this paper is better than many evaluation indicators,which proves the effectiveness of the work in this paper.

关 键 词：城市轨道交通 换乘站点 客流预测 时空注意力机制 图卷积 

分 类 号：U293.13[交通运输工程—交通运输规划与管理]