Non-fragile Iterative Learning Control

Learning is a main method of getting knowledge, a remarkable symbol of intelligence and a basic method for human being. So, learning is also an important attribute of intelligent control. If information of a process (or inherent information of environmental location) are learned, getting experience is estimated, classified and decided(or controlled) further and performance of the systems is improved, the systems are called learning systems. Control systems of learning function are called learning control systems. Iterative learning control is also an advanced control method and was applied to systems executing repetitive task, such as industry robot. Iterative learning control is to perfect tracking goals over a finite interval. During learning, learning controllers have parametric variation of controllers because of soft and hardware. This uncertainty leads to performance reduction of closed-loop systems and can't make systems tracking goals well. But very few people study this uncertainty in iterative learning controllers. Non-fragile control is an effective method for uncertainties of controllers. Its intention is to design a non-fragile controller that stabilize closed-loop systems and satisfy certain performance .This is main work in this paper.The problem of non-fragile guaranteed cost iterative learning control and its optimization is studied for discrete linear time-varying systems. In terms of quadratic performance function in iteration domain and linear matrix inequalities, design of non-fragile guaranteed cost iterative learning controllers and its optimization are presented for additive and multiplicative uncertainties.The problem of robust non-fragile guaranteed cost iterative learning control and its optimization is studied for discrete linear time-invariant systems. Consider parametric uncertainties of discrete linear time-invariant system and parametric uncertainties of learning controller, design of robust non-fragile guaranteed cost iterative learning controllers and its optimization are presented for additive and multiplicative uncertainties in terms of quadratic performance function in iteration domain and linear matrix inequalities.