A reckoning algorithm for the prediction of arriving passengers for subway station networks

Abstract

Knowing the volume of arriving passengers (APs) is fundamental for optimizing their paths through subway stations and evacuating them under emergency conditions. To predict AP volume online, we first analyze arrival and departure parameters and discuss the relationships among various parameters to determine the train a passenger will most likely take. Interconnecting stations and transfer paths among stations are considered direct connections and direct transfer connections, respectively, to define and construct traveling route sets. Then, travel time chains (TTCs) of transfer and nontransfer passengers are constructed to illustrate the possible routes and time costs between the origin and destination (O/D) stations of passengers. Furthermore, based on TTCs, train capacities and the inbound and outbound times of passengers accessed from an automated fare collection system, we predict the AP volumes at specified stations using a stage-by-stage reckoning algorithm in real time. Finally, to validate the model and the algorithm, we estimate the AP volume for the Beijing Subway network.