Adaptive critic design of automatic train regulation of MRT system

In view of the stochastic environment in mass rapid transit systems, the robustness against disturbance and punctuality subject to run time constraint are important issues in designing the automatic train regulation system. This paper addresses the design of automatic train regulation reinforced by adaptive critic mechanism to pursue minimization on schedule deviation and headway deviation. Dual heuristic approximate dynamic programming is used to develop the adaptive critic automatic train regulation system. The results are verified by operation data drawn from blue line of Taipei Mass Rapid Transit system. Comparing with that obtained by linear quadratic optimal design, the adaptive critic approach is superior in complying with disturbance appearing in the environment.