Reducing residual vibrations through iterative learning control, with application to a wafer stage

This paper uses iterative learning control (ILC) to remove terminal residual vibrations. By carefully selecting the ILC observation and actuation windows, ILC brings the system to rest within a finite time interval using feedforward. Experiments conducted on an industrial high-precision set-up show that the vibrations can be removed well within half the period of the dominant vibration. Estimation of the vibration state from different experiments shows that ILC does indeed remove the residual vibrations. Analysis of the experimental results show that the feedforward profile obtained through ILC outperforms feedforward profiles obtained through simulation of LQ-state feedback and does not suffer from model uncertainty.