Iterative Learning Control With Input Saturation

The presence of actuator saturation has a significant effect on system performance and may destroy the stability of the system undertaken. Especially, in the practiced implementation of iterative learning control (ILC), input saturation is of particular importance, due to its integral nature along the iterative axis and the existence of non-repeatable components in the learning process.This thesis presents ILC controller designs for systems with saturating actuator by lyapunov approach. On the basis of adaptive control and robust control theories, in the presence of parametric uncertainties and norm-bounded uncertainties, adaptive and robust ILC controller designs are presented for classes of uncertainties systems with input saturation. The contribution of this thesis lie in the following aspects: 1.For two classes of nonlinear systems, two iterative learning control strategies with input saturation are presented by using distinct design methods. Perfect tracking over the specified time interval is achieved while boundedness of all signals of the closed-loop systems is ensured. 2. Two design strategies of adaptive iterative learning controlwith input saturation are proposed for two classes of parametric nonlinear systems, the unknown parameters are estimated. Perfect tracking over the specified time interval is achieved while boundedness of all signals of the closed-loop systems is guaranteed. 3. For a class of unknown nonlinear systems, a robust iterative learning control design strategy is proposed. By using robust control into iterative learning control, the periodic uncertainties unknown bounds of system uncertainties are learned, and the overall closed-loop stability is ensured. For a class of n joint manipulator systems, an adaptive iterative learning control design with input saturation and a robust iterative learning control design with input saturation are proposed. Both of the controllers guarantee global stability of the closed-loop systems. The simulation researches are done to every method, which illustrate the effectiveness and feasibility of the proposed algorithms.