Iterative Learning Based Train Stop Control With Initial State Shifts for Urban Rail Transit

In this paper, a partitioned terminal iterative learning control (TILC) strategy for urban rail transit with initial state shifts is proposed to solve the accurate stop problem. A supervised machine learning method is used to pre-processing the historical operation data (i.e. the initial speed and the position data pairs in the stopping stage). And for each category the TILC algorithm is applied, realizing the update of the control law in a small range. Simulation results show that the proposed algorithm can achieve the stop accuracy of ±15cm even though the initial state varies over a certain range during the iterations. This meets the general requirement of accurate train stop for urban rail transit (±30cm).