A separative high-order framework for monotonic convergent iterative learning controller design

This paper proposes a separative high-order framework, both in the iteration axis and in the time axis, for monotonic convergent iterative learning controller (ILC) design. When there exist uncertainties which may be variant from iteration to iteration, i.e., iteration-dependent, the existing ILC design methods cannot be used to achieve monotonic convergence with small error. In this situation, an ILC updating law of high-order in both time-axis and iteration-axis is necessary. It is found that the high-order in time-axis is to condition the system dynamics so that a monotonic convergence can be achieved and the high-order in iteration-axis is to reject the iteration-dependent disturbance by virtue of the internal model principle (IMP). As illustrated in this paper, these two high-order schemes can be designed separately. A detailed design example is presented to illustrate the new design framework proposed in this paper.