基于历史数据的轨道交通站内客流预测模型  

Passenger Flow Prediction Model of in Rail Transit Station Based on Historical Data

作　　者：沈永发 郑煜 刘辉冉 房志明 张俊[2] SHEN Yongfa;ZHENG Yu;LIU Huiran;FANG Zhiming;ZHANG Jun(School of Management,University of Shanghai for Science and Technology,Shanghai 200093,China;不详)

机构地区：[1]上海理工大学管理学院,上海200093 [2]中国科学技术大学火灾科学国家重点实验室,安徽合肥230026

出　　处：《武汉理工大学学报（信息与管理工程版）》2022年第6期887-893,共7页Journal of Wuhan University of Technology：Information & Management Engineering

基　　金：国家自然科学基金项目(52102414);上海市扬帆计划资金项目(22YF1430000);应急管理部消防救援局科技计划资金项目(2022XFLR30).

摘　　要：轨道交通站点内客流实时状况是进行客流管控与应急管理的基础,为实时监测轨道交通网络客流变化,对大客流进行预警,构建了一种基于历史数据的轨道交通站内客流预测模型。首先利用历史轨道交通每个站点的进出站数据,挖掘不同站点分时段客流OD分布估计及到达时间估计,再基于实时进站数据,预测乘客的终点站选择与行程时间,反推演乘客时空轨迹,最终根据乘客时空轨迹分析各个站点的人群动态变化,从而实现轨道交通站内客流预测。以上海市2015年4月份轨道交通客流数据验证了模型的有效性,并采用MAPE、MRD和MPC 3个参量评价模型优劣。结果表明:与实际客流相比,模型对客流变化规律性较强站点(如南京东路站、人民广场站)的预测误差较小,对数据量较少、客流变化随机性较大的偏远站点(如复兴岛站)的预测误差较大。因此,该模型能够实时预测轨道交通的站内客流变化,从而预警轨道交通大客流,以便采取预防性管控措施。The real-time situation of passenger flow in railway stations is the basis for passenger flow control and emergency management.In order to detect changes in passenger flow in the rail network in real time and to provide early warning of large passenger flows,this paper constructs a rail station passenger flow prediction model based on historical data.Firstly,using the historical rail transit inbound and outbound data of each station,we mine the OD distribution estimation and arrival time estimation of passenger flow at different stations in different time periods,then predict the end station selection and travel time of passengers based on the real-time inbound data,invert the spatio-temporal trajectory of passengers,and finally analyse the dynamic change of crowd at each station based on the spatio-temporal trajectory of passengers,so as to achieve the in-station passenger flow prediction of rail transit.The validity of the model is verified by using the rail passenger flow data in Shanghai in April 2015,and three parameters,MAPE,MRD and MPC,are used to evaluate the model's strengths and weaknesses.The results show that compared with the actual passenger flow,the model has less prediction error for stations with strong regularity of passenger flow changes(e.g.East Nanjing Road Station and People's Square Station)and more prediction error for remote stations with less data and greater randomness of passenger flow changes(e.g.Fuxing Island Station).Therefore,the model is able to predict the change of passenger flow in the station in real time,so as to warn the large passenger flow of the rail transit and take preventive control measures.

关 键 词：交通工程 轨道交通 实时预测 大数据挖掘 动态演化 

分 类 号：U491[交通运输工程—交通运输规划与管理]