Research On Passenger Flow Path Selection Of Urban Rail Transit Based On Afc Syetem

Over the last decade,China has entered the peak period of urban rail transit construction.The urban rail transit network with large cities and megacities as the main gathering places has experienced rapid development.With the intensive opening of new lines,the rail transit of each city is accelerating into a network,and the path selection behavior in passenger online networks is increasingly complicated.Traditional passenger flow assignment methods based on equilibrium theory and Logit model have various problems in practical application.Massive passenger travel data recorded by AFC system of urban rail transit provides a new idea for the study of passenger flow assignment.Firstly,this paper reviews the research status of passenger flow distribution in transportation network and intelligent traffic card data mining at home and abroad,and establishes the idea of calculating passenger flow distribution proportion based on ticket data of urban rail transit by means of big data mining and machine learning.Then it introduces the data extraction and processing method of AFC system,and analyses the main factors affecting the passenger path selection behavior.An efficient path set generation algorithm based on maximum transfer times is selected to solve the alternative path set required for passenger flow assignment.This paper investigates the elements of travel time for a trip,and estimates the inbound and outbound time consumption and transfer time consumption,and then estimates the initial value of travel time distribution parameters for a route.Secondly,under the assumption that the single path travel time approximately obeys the normal distribution,the Gauss Mixture Model describing the multi-path OD pair travel time distribution is established,and the unsupervised clustering algorithm-Expectation Maximization algorithm(EM algorithm)is introduced to solve the model.The number of hybrid models is determined by Bayesian Information Criterion,and the algorithm is realized by programming in Python language.Finally,using the real ticket information of Chengdu Metro in a month as the data source,using the passenger flow assignment model established in this paper,the passenger flow assignment ratios of two groups of typical ODs from Fenghuang Street to Chunxi Road and Yipintianxia to Chengdu East Railway Station are solved,and the results of this paper are compared with those of traditional Logit model.The evaluation index of model fitness is proposed,and the interactive verification test is designed to verify the stability and validity of the passenger flow assignment model and its solving algorithm established in this paper.