Design of Iterative Learning Control Schemes Based on Model-Free Feedback Linearization for a Plant with Unknown Dynamics

Abstract

This paper presents a new iterative learning control (ILC) scheme for affine nonlinear systems of repetitive nature with unknown dynamics, combining model-free feedback linearization with a two-dimensional (2D) system framework. The method eliminates the need for prior model knowledge by employing model reference adaptive control (MRAC) together with $Q$-learning to achieve feedback linearization. After linearization, the control law is formulated within the 2D system setting, enabling the integrated design of both feedback and learning controllers to ensure stability and tracking error convergence. Numerical experiments on a thermoplastic injection molding process validate the effectiveness of the proposed approach, demonstrating accurate trajectory tracking across trials.