Real-time train regulation with passenger flow control in urban rail systems

Abstract

Passenger flow control is a critical aspect of urban rail system operations. In congested conditions, passenger demand can exceed the available train capacity, leaving some passengers at stations unable to board. This paper aims to enhance rail system performance under such disturbed scenarios. First, a model for train traffic and passenger dynamics is presented using a “discrete-event” approach. Discrete-event traffic models assess train departures from stations as events, without considering the time span between these events. A coupled model is formulated that integrates the relationship between train dwell times at stations and passenger accumulation at platforms, proposing control actions for both train traffic and station facilities.

A discrete-event model predictive control (DE-MPC) strategy is developed with a regulatory objective function and constraints related to safety, feasibility, and the limited capacities of trains and platforms. The objective function also accounts for the number of passengers at platforms to ensure effective flow management during the control period. However, implementing real-time control requires application on a “time-based” system. To bridge this gap, a new method using a virtual discrete-event system is introduced, allowing real-time measurements in a time-based system to be used for evaluating discrete-event states. This enables the application of virtual discrete-event model predictive control (VDE-MPC) on the time-based system, facilitating real-time regulation of passenger flow and train traffic.

Numerical examples demonstrate the effectiveness of the proposed control methods in both discrete-event and time-based frameworks.