有限时间迭代学习控制

针对任意初态情形, 借助于初始修正吸引子的概念,讨论不确定时变系统能够达到实际完全跟踪性能的迭代学习控制方法.闭环系统中含有限时间控制作用, 在预先指定的区间上实现零误差跟踪,且起始段的系统输出轨迹也可预先规划.分别讨论部分限幅学习与完全限幅学习, 证明闭环系统中各变量的一致有界性以及误差序列的一致收敛性. 变量有界性证明得益于提出的限幅学习算法,特别是完全限幅学习算法可确保参数估值的变化范围.

FINITE-TIME ITERATIVE LEARNING CONTROL

This paper deals with an iterative learning control approach for uncertain time-varying systems in the presence of initial state errors. A finite time controller is included in the developed learning system, where an initial rectifying action isintroduced for the controller design. The learning system is guaranteed to achieve practical complete tracking with which the tracking error converges to zero over a pre-specified interval, and the piece of trajectory from the starting position can be planned. Both partially- and fully-saturated learning algorithms are examined, respectively. The stability and convergence of the closed-loop system are established by using Lyapunov-like method. The way to carry out the boundedness proof is based on the saturated learning algorithms.