Design techniques for multivariable ILC:application to an industrial flatbed printer

Iterative Learning Control (ILC) can significantly improve performance of systems that perform repeating tasks. Although in practice many systems are multivariable, frequency-domain ILC design procedures often involve multi-loop single-input single-output (SISO) filters which do not explicitly address interaction in dynamics. The aim of this paper is to i) analyze multi-loop SISO ILC designs, ii) point out the importance of multivariable ILC design, and iii) develop the required multivariable design algorithms. Benefits of the proposed approaches over multi-loop SISO ILC are demonstrated on an industrial printer model example.Iterative Learning Control (ILC) can signi cantly improve performance of systems that perform repeating tasks. Although in practice many systems are multivariable, frequency-domain ILC design procedures often involve multi-loop single-input single-output (SISO) fi lters which do not explicitly address interaction in dynamics. The aim of this paper is to i) analyze multi-loop SISO ILC designs, ii) point out the importance of multivariable ILC design, and iii) develop the required multivariable design algorithms. Bene ts of the proposed approaches over multi-loop SISO ILC are demonstrated on an industrial printer model example