Hierarchical model predictive control for automatic high-speed train operation with multiple traction/braking units

This paper proposes an efficient train operation control scheme for automatic high-speed train operation with multiple traction/braking units by adopting a hierarchical model predictive control framework. The higher-level controller optimizes the dynamic reference speed trajectory for the whole train by tracking off-line speed profiles through a sequential quadratic programming approach with a long sampling time. The lower-level controller focuses on detailed in-train dynamics and distributes control tasks across multiple traction/braking units via a fast alternating direction method of multipliers, operating at a faster rate. This hierarchical structure enables multi-rate execution, allowing each controller to adapt to different time scales and respond effectively to external disturbances. Simulation experiments demonstrate the effectiveness and performance benefits of the proposed method in improving speed tracking accuracy and operational robustness under various scenarios.