Investigation Of Iterative Learning Control Algorithm For A Class Of Nonlinear Time-delay Systems

Iterative learning control (ILC) is a new intelligent control method in these two decades. In industrial process control, most complex industrial process has some repeatable properties. Based on these properties, ILC uses previous control experience and output error to modify the current control information so as to make system output converge expecting output as soon as possible. ILC is an off-line learning process which doesn't depend on the accurate mathematical model of dynamic systems (ILC is a self-study method based on the quality which needn't to identify system parameters.). ILC can deal with dynamic systems with high uncertainty in simple mode, and need less previous knowledge and less calculation, besides, ILC have high adaptability and realize easily. So it is applied widely in control process.All practical control systems are nonlinear, so the absolute linear systems don't exist. Because static characteristics of control system components have nonlinear with different degree, the ordinary systems are all nonlinear systems. Linear system is extreme situation, so the research of nonlinear systems can discover essence of nature as well as possible.In all practical processes, phenomena of time-delay exist widely. Because of factors of measure methods of system variables, physical characters of equipment and transport of substance and signal, therefore, the performance parameters decline and systems are unstable as well, so the research of time-delay system has academic and practical value, but there are many difficulties for this research on mathematical theory and practical applying, at present so many questions need be solved in this field.Aiming at a class of nonlinear systems with time-delay, this paper presents a robust PD-type iterative learning control algorithm considering uncertainty of the initial conditions. This algorithm not only considers effect of system feed-forward and real-time feedback but also can modify initial error. This paper discusses the output limited track of this kind of system and milder sufficient condition of algorithm output approaching this limit. Furthermore, this paper broadens the initial condition to the available repeated initial state function at will. Finally this paper proves robust convergence of this algorithm. Simulation results farther demonstrate the effectiveness of proposed algorithm.