Performance-oriented investigation of adaptive robust control in iterative learning control framework with comparative experiments

Abstract

To meet the high-performance motion accuracy requirement in current precision industry, an investigation on the combination of adaptive robust control (ARC) and iterative learning control (ILC) is conducted, which is inspired by our previous research on these two methods. It was theoretically and experimentally illustrated that ARC achieved good tracking performance and guaranteed transient performance even under parametric uncertainties and uncertain disturbances. However, ARC relies on system modeling and the effect of unmodelled part is suppressed by the robust control term, which leads to certain conservativeness of steady tracking performance. On the other hand, without accurate plant model, ILC can achieve excellent steady tracking performance through iteration learning process under repetitive tasks. Nevertheless, ILC needs several iteration learning trials to construct the utmost optimal control input, and is quite sensitive to non-repetitive disturbances and noise. In this paper, noting the merit and deficiency of ARC and ILC, we try to integrate ARC and ILC and aim to synthesize an ARC-ILC control technology for a preliminary investigation. The ARC part in the framework can effectively provide robustness and parametric adaptation, while the ILC part can compensate unmodelled repetitive uncertainties inherent in ARC remarkably. The proposed framework is briefly analyzed and simulated, and comparative experiments are conducted on a linear motor stage. Experimental results demonstrate that the ARC-ILC scheme possesses excellent transient/steady tracking performance even under the existence of parametric uncertainties and external disturbances.