Research On Passenger Flow Assignment Of Urban Rail Transit Based On AFC Data

In recent years,due to the rapid growth of passenger flow some problems such as congestion,insecurity and inefficiency gradually emerged.Thus it has become an urgent problem for operation management in metro to accurately estimate distribution of passenger flow in stations,lines and networks and quantify the congestion in urban rail transit.Automatic Fare Collection(AFC)system could accuretely record each passengers^origin and destination in metro which could provide abundant information for underground company.Therefore,it has become a research hotpot in the aera of passenger flow assignment about how to describe the passenger's micro-behavior characteristics and study the macro-distribution characteristics of passenger flow based on massive AFC data.This paper aims to describe passengers' boarding behavior in detail,estimate the distribution of passengers' access and egress time in stations reasonably,and quantitatively measure the degree of congestion in rail transit.The main work in this paper can be summarized as follows:(1)An estimation method of passengers' boarding probability is put forward.At first,the interaction between passengers and trains on the platform is discussed.Then concept of effective train sets is interpreted and corresponding generation method is also proposed.Moreover,a passenger-to-train assignment model under constraints of train's capacity is contructed to estimate passenger's selection probability for each train in the effective train set.Finally,a calculation method of congestion indicators based on passengers' boarding probability is presented from three aspects:train,station and passenger.(2)An estimation method of passenger travel time distribution is proposed.First of all,this paper presents characteristics of passengers' walking time distribution in stations,and then analyzes limitations of traditional estimation method.What's more,it makes assumptions about distribution of passengers' walking time and classification of passenger.Finally,a probability model of the passengers' walking time distribution with hidden variables is put forward and solved by the Expectation Maximizaiton(EM)algorithm.The method could estimate parameters of passengers' walking time distribution from AFC data and train schedule,which not only solves the time-consuming problem of traditional field investigation,but also overcomes the limitations of the traditional estimation method.It provides a reasonable parameter calibration method for other related research such as passenger flow simulation and passenger flow assignment.(3)A case study is conduted to validate the passenger flow assignment method and parameter estimation method which take Batong Line of Beijing Metro as an example.Firstly,source and preprocessing method of AFC data are introduced.Secondly,parameters of distribution of passengers' egress time in all stations are estimated and compared with results of field survey.Then a simulation experiment is implemented to describe the individual passenger movements and output synthetic AFC data.In order to verify the reliability of the simulation experiment,the synthetic AFC data is compared with actual AFC data.Finally,the effectiveness of passenger flow assignment method is validated using synthetic data.