ON ITERATIVE LEARNING CONTROL FOR SOLVING NEW CONTROL PROBLEMS

Abstract

Iterative learning control (ILC) is an approach for improving the transient performance of uncertain systems that operate repetitively over a fixed time interval. Although ILC has been well established over the past three decades, there are still several open problems to be solved. This thesis aims to apply ILC approach to solve new control problems. In the first part, three different modified ILC schemes are developed to deal with systems with non-repeatable temporal/spatial factors, where a fundamental requirement in traditional ILC that the control system must be strictly repeatable is removed. Then, a new robust ILC algorithm is presented to handle the norm-bounded uncertainties in the second part. The idea behind the proposed controller is to parameterize the bounding functions, and then learn those parametric uncertainties pointwisely in the iteration domain. To further widen the applicability of ILC, the third part extends ILC from ODE systems to PDE systems. Without any simplification or discretization, a design and analysis framework of ILC for linear and nonlinear PDE systems is developed. To the end, ILC approach is applied to a two-link robotic fish in real-time and achieves precise speed tracking performance, which shows that ILC is a powerful motion control method for robotic fish due to its partial model-free property and simplicity in implementation.