Column generation-based optimum crew scheduling incorporating network representation for urban rail transit systems

Abstract

Crew scheduling, which usually has its own special requirements, is an essential part of the work undertaken by urban rail transit operators. For example, meal breaks are required during two pre-defined mealtime windows. Shifts are commonly divided into three types that have different restrictions. The ratio of the different types of shifts is determined by a predefined work pattern. Thus, it is crucial to address and solve the crew scheduling problem in urban rail transit systems efficiently after satisfying the above requirements and restrictions. A set of multilayer networks is developed herein to meet the feasibility constraints of shifts. Moreover, the crew scheduling problem is formulated into an extended set covering problem with additional constraints on the ratio of the different types of shifts and solved using the column generation method. The pricing problem of the column generation method is finally modeled as dozens of shortest-path problems based on the proposed network representation. The feasibility of this approach is demonstrated in case studies using real data from three metro lines in Beijing. The experimental results show that the proposed method is well-suited for real-life applications in urban rail transit systems.