D-Type ILC Based Dynamic Modeling and Norm Optimal ILC for High-Speed Trains

Abstract

Differential-type (D-type) iterative learning control (ILC) is a typical control method for repetitive operating nonlinear systems and has been used for speed tracking of high-speed train (HST). However, the derived $\lambda $ -norm convergence property of the D-type ILC may lead to unsafe operation of the HST since huge overshoot phenomenon of tracking errors in the iteration axis may occur. To address this problem, this paper presents a novel dynamic modeling and norm optimal iterative learning control (Dynamic modeling based NOILC) approach. By making full use of the valuable information data generated after each repetitive operation of the HST, a modified iterative learning recursive least squares algorithm is proposed to identify the unknown and time-varying even fast time-varying aerodynamical coefficients in the nonlinear train dynamic model. Then, based on this identified nonlinear train model, a norm optimal iterative learning control with consideration of security, punctuality, and traveling comfort will be designed. Through theoretical analysis, reliable 2-norm convergence of both the model identification error and the tracking control error can be guaranteed. Simulation and experimental studies further verify that the proposed approach achieves a significant improvement in tracking control precision and meanwhile obeys safety requirement.