An iterative learning control algorithm within prescribed input-output subspace

This paper is concerned with an iterative learning control (ILC) method for linear continuous-time systems. With the iteration of experiments, the ILC method yields the desired input for tracking the target trajectory. Most of the former ILC methods use the compensations, such as the time derivative of the error signal or the dual mapping of systems in the learning algorithm. We propose a new ILC algorithm which does not use such compensations contrary to the former methods. In this method, we restrict the input space to the prescribed finite-dimensional subspace, and use the signal sequence which is derived from the projection of the error on this input subspace when the input is updated. The effectiveness of the proposed method is demonstrated by a numerical example and experimental evaluation is performed using a two-mass spring system.