Noise tolerant iterative learning control for identification of continuous-time systems

The paper is concerned with both iterative learning control (ILC) and identification of continuous-time systems based on sampled I/O data in the presence of measurement noise. First, we propose a new ILC method which achieves tracking for uncertain plants by iteration of trials. The distinguished feature of this method is that (i) the leaning law works in a finite dimensional parameter space rather than the infinite dimensional input space and (ii) it takes account of noise reduction by using I/O data of all past trials efficiently. Second, it is shown how to estimate parameters of a class of linear continuous-time systems based on the proposed ILC method in noisy circumstances. Its effectiveness is demonstrated through numerical examples.