Iterative Learning-based Train Brake Trajectory Generation for Fixed Stop Position Control with Flexibility to Varying Adhesion Conditions

Motion control of automatic train operations requires accuracy in stopping position and flexibility to environmental changes when applied to conventional railway lines wherein adhesion characteristics may exhibit considerable variation because of disturbances such as rain, snow, and fallen leaves. This study focuses on iterative learning-based control for train motion to achieve precise stopping performance, whereby the motion controller updates the reference control trajectory in every train run on the basis of the results of the previous train runs. The proposed train motion controller iteratively learns the feasible maximum deceleration without excessive wheel slip from the running results and actively adjusts the speed profile for the next train run for compliance with the deceleration limits in accordance with the varying adhesion characteristics. Numerical simulations indicate sufficient stop position accuracy during repeated train runs under varying adhesion characteristics and a capability to return to more efficient speed profiles with higher deceleration and shorter running time as adhesion characteristics recover. The proposed method thus facilitates the automatic train operation while enabling flexibility of control under varying environments.