基于二维混合模型的鲁棒迭代学习控制设计  

Design Method for Iterative learning Control Based on 2D Hybrid Model

作　　者：兰永红[1] 吴敏[1] 余锦华[2] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]东京工科大学仿生学部,东京192-0982

出　　处：《系统仿真学报》2008年第9期2364-2367,共4页Journal of System Simulation

基　　金：国家自然科学基金项目(60674016);国家杰出青年科学基金项目(60425310)

摘　　要：针对一类正则线性不确定系统,提出一种基于连续/离散二维混合模型的迭代学习控制设计方法。首先,通过独立地考虑迭代学习控制系统连续的控制行为与离散的学习行为,建立迭代学习控制系统的连续/离散二维混合模型,将迭代学习控制器设计问题转化为一类连续/离散二维系统的状态反馈控制问题。然后应用二维连续/离散系统方法,获得迭代学习控制系统稳定性条件。最后根据稳定性条件,利用线性矩阵不等式方法,求得迭代学习控制器参数,它通过Matlab工具箱可以方便的获得。与现有方法相比,所提出的迭代学习控制器利用了输出误差以及状态变化的信息来修正当前的控制,因而设计方法更加符合其本质特征,具有简单实用、直观明了的特点。数值仿真实例验证了所提方法的有效性。A method for an iterative learning control system containing a class of regular linear uncertain plants was proposed based on a hybrid continuous discrete two dimensional （2D） model. First, by analyzing the control and dynamic learning processes independently, a hybrid continuous-discrete 2D model for the iterative learning control system was established and the iterative learning controller design problem was converted to a state feedback design problem for the continuous-discrete 2D systems. Next, under the 2D systems theory, a stability condition for the iterative learning control systems was obtained by using 2D system theory. Finally, based on the stability condition, the parameters of the iterative learning controller were resolved by Matlab Toolbox via linear matrix inequalities （LMI）. Unlike previous methods, the new method is very simple, intuitive and practical. The information of tracking errors and the state feedback were used in the control input. The validity of the method was verified by a numerical example.

关 键 词：正则线性系统 迭代学习控制 二维系统 线性矩阵不等式 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]