A safety-optimal control method for virtual coupling train control

Virtual Coupling (VC) is an emerging railway signaling concept aimed at increasing line capacity and reducing train headway through Train-to-Train (T2T) communication. Instead of relying on traditional mechanical couplers, trains under VC operate at similar speeds and close spacing by exchanging real-time information via wireless networks. Although VC promises substantial improvements in eciency and flexibility, its practical implementation still faces challenges related to safety assurance, control stability, and communication reliability. This thesis focuses on the longitudinal motion control of VC convoys, addressing the key problem of safety spacing control during operation, under the presence of hazards including communication delays and dynamic uncertainties. A decentralized control framework based on Control Barrier Functions (CBFs) is developed to ensure safe and ecient spacing among trains. The proposed approach integrates safety constraints and reference control into a series of real-time Quadratic Programs (QPs), guaranteeing collision avoidance while maintaining smooth and stable operation. Simulation studies demonstrate that the controller achieves excellent performance in terms of safety, eciency, and computational scalability, showing its potential for real-time implementation in real VC applications.