Research On Passenger Itinerary Inversion Method Of Urban Rail Transit During Peak Hours Based On AFC Data

With the acceleration of urban modernization,the networking characteristics of urban rail transit system have become increasingly prominent.The explosive growth of passenger flow volume on the urban rail transit network has led to endless problems such as congestion and travel safety.It has become an important issue in the operation and management of rail transit to reasonably describe the travel trajectory of passengers in the urban rail transit network according to intelligent means and quantify the congestion effect,to grasp the travel itinerary choice characteristics of passengers,and to timely deduce the passenger flow distribution in network,so as to rationally adjust the train operation plan and formulate the operation control scheme.The AFC data recorded by the AFC system of urban rail transit contains a great deal of real and effective passenger travel information.Based on this,this paper constructs two probability estimation models and puts forward corresponding parameter estimation methods for two behaviors of feasible train choice and effective physical route choice in the urban rail transit passenger travel.Thus,the inversion of passenger itinerary scheme and the reappearance of individual passenger time-space travel trajectory are realized.Firstly,this paper researches the basic characteristics of passenger travel during peak hours of urban rail transit.The generalized cost estimation method for the route is proposed around the travel time factors such as walking time,left behind times,transfer time and on-board time.Based on Bayesian inference,the model of passenger boarding probability and walking time distribution is constructed,and the Expectation Maximization algorithm(EM)is used to iteratively solve the hidden variable of the model,by which the walking time distribution parameters in the station and the left behind times distribution on the platform are obtained simultaneously.According to the distribution characteristics of each component of the travel time,the transfer time is estimated.Therefore,the generalized cost calculation of any route of the rail transit network is realized.Then,the topology matrix of the network is searched to obtain the effective physical route set between OD pairs.The overall distribution characteristics of passenger travel time are counted,and the Gaussian Mixed Model(GMM)is simplified solved by EM algorithm,so as to obtain the actual distribution probability of travel time of each route between OD pairs.Then,a Multinomial Logit Model(MLogit)based on the results of the route generalized cost estimation whose parameters are solved by maximum likelihood estimation according to the fitting results of GMM is constructed,and the sensitivity of the parameters is analyzed.Combined with the estimation results of the physical route choice probability and the boarding scheme choice probability,the passenger itinerary scheme and its choice probability are obtained through passenger segmentation,and the historical time-space travel trajectory of individuals in urban rail transit network is inversed.Finally,taking Beijing Subway network as the research object,and selecting the actual operation data during the morning rush hours of working day,an example is used to verify the feasibility and reliability of the passenger itinerary scheme inversion method proposed in this paper.There are 32 figures,18 tables and 75 references.