Research On Metro Passenger Flowassignment Model Based On Logit

With the increasing pace of subway construction, subway lines change from a single line into a network, Cities metro system operator uses seamless transfer mode, transfer passenger information can not be recorded.Since the passengers’ credit card data extractedfrom subway system database can not determine passengers’ travel route directly,study subway passenger flow distribution, helps to accurately grasp the network traffic distribution subway system, raising the city’s subway system operation and management level.This paper mainly studies the main factors influencing passengers travel route choice, based on the Logit travel route choice model,We built a more accord with the actual improvement of the travel route choice model, the main research content is as follows:(1)the study of characteristics of metro network, mainly to research the composition of the subway network and the influence of different modes of transfer for passenger flow distribution, we consider transfer chain in the subway network, and effective path search algorithm in the subway network is studied.(2)Based on the deep understanding of the travel route choice factors, we established a travel route choice model. For traditional travel route choice model only considers the passenger’s journey time, waiting time, transfer time, traditional factors such as the passenger flow distribution results with obvious deviation problem, We consider alternative factors in the model:angle cost of network topology and line correlation.(3)Mainly studies the parameters in the model calibration method, we first introduce the maximum likelihood estimation method, and then put forward a new statistical method based on bayesian inference. And we research the M-H sampling algorithm and the sampling process, the basic principle of the algorithm is based on markov monte carlo method.(4)According to the shenzhen subway network, We calculate the proportion of passenger flow of different lengths of OD in different times,And analyze the influence of angle cost of network topology and line correlation in passenger flow.