Discrete-time adaptive iterative learning control with incorporation of a priori process knowledge

Abstract

This work explores utilization of a priori model knowledge to achieve better control performance of iterative learning control systems. Two specific discrete-time adaptive iterative learning control approaches are proposed for different control cases to exactly utilize known information in order to improve control performance. In more detail, a novel time-difference estimator based discrete-time adaptive iterative learning control system is proposed for a time-invariant parametric uncertain system, and a mixed-difference estimator based discrete-time adaptive iterative learning control is proposed for a hybrid parametric uncertain system. Both of these control methods are shown to perform well even though the initial states and the reference trajectories vary randomly along the iteration direction. We show that superior control performance can be achieved compared with the existing iteration-difference estimator based discrete-time adaptive iterative learning control systems by applying the two proposed control methods, which fully utilize known information about the model. Mathematical proof and a simulation study on a permanent magnet linear motor verify the efficiency of the proposed discrete-time adaptive iterative learning control approaches.