Data-based model of metro scheduling for passenger wait-time optimization with constraints: WIP abstract.

This paper presents a data-based model for metro scheduling that aims to minimizes passenger wait time under constraints. In contrast to existing approaches that rely on a statistical model of passenger arrival, we develop a model based on real-world automated fare collection (AFC) data in a metro line of a Korean city for an extended period of time. The model consists of decomposing the travel time for each passenger into wait, ride, and walk times, clustering of passengers by trains they ride and also calculating the number of passengers in each train for any given time. Based on this, for a given train schedule, the total wait time of all the passengers for the entire AFC data period can be calculated. Finally, the minimization problem is formulated using the model under realistic constraints. Refining and validating each component of the model are currently underway before we solve the minimization problem.