Monotonic Convergent Iterative Learning Controller Design Based on Interval Model Conversion

摘要

This note presents a robust iterative learning controller design method for plants subject to interval model uncertainty in the A-matrix of their state-space description. First-order perturbation theory is used to find bounds on the eigenvalues and eigenvectors of the powers of A when A is an interval matrix. These bounds are then used for calculation of the interval uncertainty of the Markov matrix. The bounds on the Markov matrix are then used to design an iterative learning controller that ensures monotonic convergence for all systems in the interval plant.