Robust Learning Control Systems Design And Application

For plants performing repetitive or periodic tasks, learning control technology can be applied to realize the perfect tracking performance over the finite time interval or whole period, through updating input repetitively, and the deterministic disturbances can be compensated completely. Learning algorithms deal with dynamic systems with high uncertainties and improve tracking performance through learning. However, learning control requires that the uncertainties are repetitive or periodic, while robust control can solve the problems caused by non-repetitive or non-periodic uncertainties effectively. Linear motor servo systems often operate repetitively over a finite time interval, and learning control could be an efficient method for trajectory tracking tasks of such systems. The robust learning control systems design deserve more studies. Take into account the issues above mentioned, this thesis focuses on the following aspects:(1) For uncertain nonlinear systems running repetitively over a finite interval, iterative learning control is not applicable due to non-repetitive uncertainties, while both variable structure control and robust control can deal with the uncertainties effectively. By incorporating these two control methods with iterative learning control, respectively, both simple form variable structure iterative learning controller and robust iterative learning controller are presented to avoid the limitation of single control mode. The parts of variable structure control and robust control replace the sign function with a continuous function which is effective to eliminate the chattering and guarantee all the system variables to be bounded. At the same time, the part of iterative learning control can improve the system tracking performance. At last, simulations are presented to demonstrate the effectiveness of the proposed algorithms.(2) For uncertain nonlinear systems running periodically in an infinite interval, repetitive control works well in the face of periodic uncertainties, but not for the non-periodic ones. Utilizing variable structure control and robust control to effectively deal with uncertainties, both variable structure repetitive controller and robust repetitive controller are designed . These two proposed algorithms can both compensate for the shortcoming as either the variable structure control or robust control or repetitive control is applied separately. For two proposed algorithms, replacing the signum function with a continuous function, the parts of variable structure control and robust control both are effective to eliminate chattering. Two proposed algorithms both can guarantee all the system variables to be bounded and achieve the perfect tracking. Booundedness of closed-loop system variables and learning convergence are established theoretically, and simulation results are presented to demonstrate effectiveness of the proposed algorithms.(3) Variable structure control, variable structure iterative learning control, variable structure repetitive control, robust control, robust iterative learning control and robust repetitive control are presented for controller designs of Linear motor servo systems. Computer simulation are conducted and numerical results are given to verify the effectiveness of the proposed algorithms.