Time - frequency adaptive Iterative Learning Control

In many research laboratories, Iterative Learning Control (ILC) has proven itself to be a very effective technique to reduce repetitive control errors that occur in systems that continually perform the same motion or operation. Based on errors from previous operations, the technique iteratively constructs a feedforward signal, with which extremely small tracking errors are obtained [4, 51-However, several characteristics of ILC have prevented it from being widely used as a control technique for industrial applications. First, the learned feed-forward signal depends on the setpoint. If this is altered, the complete learning process has to be repeated, which costs time. For the carrier application of this research, a wafer scanner motion system, this is an important restrain. A second issue when applying standard ILC [5], is the fact that it amplifies non-repetitive disturbances and noise. In the research presented here, both issues have been addressed. Insights from time-frequency analysis of relevant control signals have led t o a time-frequency adaptive ILC that is capable of achieving a tracking performance equivalent to standard ILC, whilst significantly reducing the noise amplification. Further, this has led to one feedforward signal that is suitable for tracking different setpoints. For classes of systems that show behavior comparable to the considered motion system, application of the developed technique-called piecewise ILC-leads to a negligible increase of the tracking errors. This increase is the result of position dependent disturbances and dynamics of the concerning system. de ontwikkelde techniek (piecewise ILC) tot een verwaarloosbare toename van de volgfout. Deze toename is het gevolg van positie-afhankelijke verstoringen en dynamica in de betreffende systemen.