Neuro-adaptive control of high-speed trains under uncertain wheel-rail relationship

Traditional automatic controller designing in train systems is almost based on urban rail transit where the influence of changing wheel-rail relationship caused by the variation of speed and environment is ignored. However, high-speed railway operates in more open environment and higher speed, which leading to a more complex variation of wheel-rail relationship occurring. In this paper, we design an automatic train controller in high-speed railway which can realize the automatic velocity tracking even if the uncertain and nonlinear variation of complex wheel-rail relationship happens. First of all, the train dynamic model is established where the wheel-rail relationship is expressed as an uncertain unknown function and the train operation system is expressed as a third-order nonlinear system. Then, a neural network adaptive controller is designed by using the backstepping method and barrier Lyapunov function. Based on this controller, position and velocity tracking errors are semi-globally uniformly ultimate boundedness. Finally, the effectiveness of the algorithm is verified by simulation experiments.