Iterative learning control not to interfere with feedback control

Abstract

Abstract This paper discusses about Iterative Learning Control (ILC) which learns an appropriate feedforward input from iterative experiments. In particular, this paper focuses on systems which include integrator, and considers tracking to a desired output which converges to a constant. In this case, the feedforward input should converges to zero, however classical ILC does not take this condition into account, and interfere with the feedback controller. This paper proposes a new design of ILC whose resulting input converges to zero exponentially. The effectiveness of the proposed method is demonstrated by a practical experiment with a flexible arm. Since a flexible arm is known to nonlinear, the proposed method is expected to be effective for nonlinear systems.