Demand-driven flexible-periodicity train timetabling model and algorithm for a rail transit network

Abstract

The optimization models for strict-cyclic and non-cyclic timetables are limited in their ability to be applied to different scenarios. This paper proposes a universal flexible-periodicity timetabling framework for rail transit networks under time-dependent demand, which can provide strict-cyclic, non-cyclic, and partial-cyclic timetables by adjusting the parameters. This approach reduces the complexity of scheduling for operators and allows them to choose the most appropriate timetable for a specific scenario. The demand-driven flexible-periodicity timetabling (DDFPT) model considers passenger waiting time, train operation cost, and timetable periodicity deviation as objectives. To fit the network scenario, the model also needs to take into account the feasibility of rolling stock connections at terminal stations and coordination of timetables on different lines. An adaptive large neighborhood search (ALNS) algorithm with eight destroy operators and a repair operator is proposed to solve the model. A numerical and a real-world network case are used to evaluate the performance of the model and the algorithm. Upon comparing the benefits of various types of timetables, it has been observed that the partial-cyclic timetable strikes a good balance between the passengers’ travel efficiency and the periodicity of the timetable. Additionally, the coordinated optimization of timetables between crossed lines at the network level greatly improves passengers’ travel efficiency at transfer stations.