A Model Predictive Control Based Inter-Station Driving Strategy for Virtual Coupling Trains in Railway System.

Virtual Coupling (VC) is a hot research topic in railways about future train operation systems. In VC, trains will drive in a minimum distance as a convoy without physically coupled. The greatest feature of VC over traditional block system is that a following train should maintain a desired distance from its predecessor. Existing research on VC mainly considered the cruise control problem with a constant line speed limitation, which are far from the reality where the speed limitations with respect to track often vary greatly in an inter-station. To solve this problem, we propose a driving strategy that the leader train tracks a desired speed profile and the followers adjust themselves according to their predecessors' state, respectively. A nonlinear model predictive control system is designed to achieve the automatic train control. A nonlinear distance control objective function and continuous variable speed limitations in an inter-station are considered in the constraints. Simulation experiments based on the actual data from a subway system are conducted to show the driving behaviors of trains in an inter-station. The results show that the trains will deviate from the desire state temporarily subjected to the speed limitations. However, all trains will adjust themselves automatically to reach the desired state eventually, which prove the effectiveness of the proposed method.