×

Predictive optimal iterative learning control. (English) Zbl 0949.93027

An iterative learning control algorithm for tracking systems is proposed. The new idea behind the derivation is the use of predicted errors in the repetitive computation of actual controls. At each iteration the algorithm chooses an input which minimizes the weighted sum of norms of future errors and control increments. It is shown that the rapid convergence with a guaranteed rate is achieved. The convergence is not affected by the details of plant dynamics. The actual implementation of the algorithm has a multimodel structure and uses conventional linear-quadratic design methods. The role of receding-horizon predictive control in the learning process and the properties of the new method are illustrated by the simulation examples.

MSC:

93B51 Design techniques (robust design, computer-aided design, etc.)
PDFBibTeX XMLCite
Full Text: DOI