Computation-efficient distributed MPC for dynamic coupling of virtually coupled train set

• Published in: Control engineering practice Vol. 145; p. 105846
• Main Authors: Luo, Xiaolin, Tang, Tao, Li, Kaicheng, Liu, Hongjie
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.04.2024
• Subjects: Computational efficiency; Distributed model predictive control; Dynamic coupling; Field test; Virtual coupling; Distributed model predictive control; Dynamic coupling; Field test; Computational efficiency; Virtual coupling
• ISSN: 0967-0661; 1873-6939
• DOI: 10.1016/j.conengprac.2024.105846

Virtual coupling (VC) is an emerging technology to improve the flexibility and capacity of railway services. To adjust the formation of a virtually coupled train set (VCTS) on-the-fly, dynamic coupling control is essential to couple multiple trains (units) stably and efficiently. However, it is still hard to be achieved in real-time, since safety constraints are complex but have to be satisfied for collision avoidance. Thus, this paper proposes a computation-efficient distributed model predictive control (DMPC) approach to solve this problem. First, the movement of VCTS is captured by a three-order dynamics model, while the safety constraints are unspecified and can be defined by arbitrary functions. Then, the DMPC approach is designed which consists of reference planning and tracking. Specifically, we design a prediction-based control scheme to plan reference trajectory for each unit, where the future satisfaction of safety constraints is addressed. Resorting to this design, the reference tracking in DMPC is achieved by solving a computation-efficient quadratic programming problem. The shifting principle is employed in the closed-loop implementation of DMPC to guarantee stability. Finally, experiments are conducted to verify the performance of the proposed approach. Dynamic coupling of a real VCTS is the first time achieved in field tests, where two units are coupled stably and efficiently while satisfying a numerically-evaluated safety constraint. It is a breakthrough in the development of VC technology. •Dynamic coupling of virtually coupled train set in field tests.•Computation-efficient distributed MPC for engineering.•Addressing complex safety constraints in virtual coupling.