Schedule-Based Rail Transit Path-Choice Estimation using Automatic Fare Collection Data

Abstract

This paper presents a schedule-based passenger’s path-choice estimation model for a multioperator rail transit network, using automatic fare collection (AFC) data from both entry and exit stations. By introducing the train schedule connection network (TSCN), the path-choice estimation is converted into a set generation and weighted assignment problem for feasible TSCN paths (i.e., passenger trajectories). A major factor in path choice, the fail-to-board (FtB) phenomenon because of overcrowding in peak periods, is explicitly modeled. A method for estimating the FtB parameters is described and a weighted assignment function based on FtB parameters is provided. The case of a typical commuting origin-destination pair in the Beijing Subway is analyzed to demonstrate the capability of the proposed model. Results show that: (1) there can be multiple feasible trajectories for one pair of entry and exit records because of the FtB probability, especially in peak periods; and (2) the transfer penalty is more influential during offpeak periods when the in-vehicle times are similar for different paths. The model has potential for locating network capacity bottlenecks at the train run level and for evaluating the effects of adjusting train timetables on the passengers’ path choice.