加速抑制随机初态误差影响的迭代学习控制  

Iterative Learning Control for Accelerated Inhibition Effect of Initial State Random Error

作　　者：吕庆[1] 方勇纯[1] 任逍 

机构地区：[1]南开大学机器人与信息自动化研究所,天津300071

出　　处：《自动化学报》2014年第7期1295-1302,共8页Acta Automatica Sinica

基　　金：国家自然科学基金(61127006;61325017)资助~~

摘　　要：针对一类具有不确定性的多输入多输出非线性系统,提出一种迭代学习控制算法.该算法具有的特点是:针对任意初态情形,结合开环D型迭代学习控制器的优点,在时间轴上设计了一个随迭代次数增加而缩短的时间段.在该时间段上,控制算法对状态偏差进行修正,以使系统输出在此段时间后跟踪期望输出,且系统跟踪误差收敛到一个界内.这个界仅由系统自身不确定性和不确定的外界干扰决定,与初态误差无关.当外界扰动为0,以及迭代次数趋于无穷时,经过上述时问段后,系统输出精确跟踪期望输出.理论证明和仿真结果都说明了该算法的有效性.A novel iterative learning control technique is presented for a class of multi-input and multi-output nonlinear systems with uncertainty. It is well known that the system output cannot track the desired output accurately and immediately due to the initial state error. In this paper, a special time interval, with the characteristics of more iterations and less time interval, is designed. By combining an open-loop D-type iterative learning controller, the random state errors are rectified in the special time interval. Subsequently, system outputs track the reference trajectory. Uniform bounds for the final tracking errors are obtained, which are only dependent on the system uncertainties and disturbances but independent of the initial errors. After the special time interval, the system outputs will accurately track desired outputs with infinite iteration and without disturbances. Both theoretical proof and simulation results show the effectiveness of the proposed algorithm.

关 键 词：迭代学习控制 初始状态误差 快速学习 收敛性 鲁棒性 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]