Robust real-time optimisation for train rescheduling and platforming in high-speed railways under a route-locking sectional-release mechanism

The route-locking sectional-release (RLSR) interlocking mechanism has been shown to significantly improve operational efficiency and station throughput in large stations. However, most existing studies rely on simplified interlocking rules, particularly in real-time rescheduling problems. To address this gap and enhance operational efficiency and resilience under dynamic perturbations, including dwell disturbances, running disturbances, and disruptions, this paper proposes a robust real-time optimisation framework for the train rescheduling and platforming problem (TRPP) in high-speed railways, with RLSR explicitly incorporated into the model formulation. Dwell and running disturbances are characterised by uncertainty sets, and the resulting model is reformulated into a tractable robust counterpart, while disruption-related uncertainty is handled through a rolling horizon framework with real-time feedback. To reduce computational complexity and meet real-time requirements, a Lagrangian relaxation-based decomposition algorithm is embedded within the rolling horizon framework. At each rolling horizon stage, the problem is decomposed into two parallelizable subproblems: the train timetable rescheduling problem (TTRP) and the train platforming problem (TPP). Numerical experiments using real-world data validate the effectiveness and computational efficiency of the proposed approach. Furthermore, several managerial insights are derived for practitioners.