Implementation and Application of a Stochastic Agent-based Model for Passenger Flow Distribution Predicting in Planned Metro Station Service Disruption Scenario

Abstract

This paper discusses a method for predicting and evaluating the effects of planned station service disruptions. A three-layer model is used to describe the travel behavior of passengers in terms of the origin and destination station selection process, the path planning process, and the train selection process, and the model is built for both normal situation and station closure situations respectively. On this basis, the calibrated agent-based simulation models under normal and station closure situations are established for simulating and counting the distribution and status of passengers and trains on the metro network, which can effectively predict the passenger flow distribution during a planned station service disruption. Shanghai Metro is taken as a case study, and the simulation results were validated against the historical data demonstrating that the proposed model is able to predict the changes in both number and trend of passenger volume. One platform of People's Square Station has a maximum of over 1000 passengers at a time and needs special attention on waiting zone organization. This study offers a new and cost-effective approach to risk management of metro network for operators.