基于AFC数据的城轨站间客流量分布预测

Forecasting of Passenger Flow's Distribution among Urban Rail Transit Stations Based on AFC Data

基于轨道交通自动售检票系统(Automatic Fare Collection,AFC)统计获得的集计型客流数据,依据行为分析理论,提出1种适用于路网结构变化条件下的城轨站间客流量分布预测模型。首先,基于随机效用最大化理论,构建乘客目的地选择模型,选取终点站吸引客流量、列车运行时间、乘客在站换乘时间、乘客换乘次数、起终点站的线位关系和站点属性6个指标构建效用函数,以反映目的地吸引力、城轨服务水平、起终点站之间的线位匹配关系等对乘客目的地选择行为的影响,在此基础上,建立站间客流量分布预测模型;然后,利用代表个人法将AFC数据转化为非集计型数据,基于WESML(Weighted Exogenous Sampling Maximum Likelihood)估计方法,实现对目的地选择的非集计预测模型的参数标定。采用广州地铁6号线开通前后的AFC数据,对该预测模型的预测效果进行检验。结果表明:在新线接入导致地铁线网结构发生变化的条件下,全线网站间客流量分布预测的平均绝对误差仅为36人,因此该预测模型具有较高的预测精度。

Based on the aggregate data from AFC（Automatic Fare Collection）and behavior analysis theory,a model for forecasting passenger flow＇s distribution among stations is proposed which is appropriate for a changed network structure in urban rail transit system.Firstly,a destination choice model is established based on random utility maximization theory,and some properties such as the passenger flow volume attracted by destination,in-vehicle travel time,transfer time,number of transfers,whether origin collinear with destination and land use scale around station,are comprehensively considered in utility function to catch the impact of attractiveness of destination,level of service of urban rail transit system,and line position matching relationship between origin and destination on passenger＇s destination choice behavior.Thereafter,a forecasting model of passenger flow＇s distribution among urban rail transit stations is proposed.Then,based on the disaggregate data transformed from the AFC data by the representative individual method,the proposed model are estimated by WESML（Weighted Exogenous Sampling Maximum Likelihood）method.On the condition of Line 6before and after being put into operation in Guangzhou Metro,based on AFC data,the proposed model is tested,and the results show that the prediction accuracy of proposed method is satisfactory with the mean absolute error of only 36.