鲁棒迭代学习控制律的优化设计方法  

Optimal design method of robust iterative learning control

作　　者：姜晓明[1] 王岩[1] 陈兴林[1] 

机构地区：[1]哈尔滨工业大学航天学院,黑龙江哈尔滨150001

出　　处：《电机与控制学报》2013年第3期110-116,共7页Electric Machines and Control

基　　金：国家自然科学基金(61174037);国家科技重大专项资助项目(2009ZX02207)

摘　　要：针对非参数不确定性系统难以实施非因果学习控制的问题,提出一种鲁棒迭代学习控制律的优化设计方法。采用基于名义性能指标的优化设计方法,预留调整加权,并将其演变成频域描述下具有系统伪逆形式的迭代学习控制律。进一步建立μ分析下的系统框架,分析加权系数对系统收敛速度和鲁棒性的影响。根据上述分析来确定初始的加权系数,并采用零相位滤波器和μ分析来修正加权,使系统满足鲁棒收敛性条件,从而得到优化的鲁棒迭代学习控制律。文中附有算例,将时域设计应用到频域分析框架,可以实现非因果稳定系统的鲁棒收敛性分析,仿真结果验证了方法的有效性。An optimal design method of robust iterative learning control is presented in order to solve the problem of which it is difficult to design non-causal learning control for systems with non-parameter uncertainties.Optimal design based on nominal performance index was employed,and it owned the variable weight.The method was turned into iterative learning control with the style of frequency-domain pseudoinverse of the system.Then the system framework based on μ-analysis was built up,and an effect on convergent rate and robustness that the weight has was analyzed.According to the analysis,the initial weight was given,and then modified to make the system satisfy the robust convergent condition by using zero phase filters and μ-analysis.Finally an optimal robust iterative learning control was obtained.An example was given,and time-domain design was applied to the frequency-domain analysis framework.Thus,robust convergence analysis of the non-causal stable system was made.Simulation results verify the effectiveness of the proposed method.

关 键 词：鲁棒迭代学习控制 不确定性系统 鲁棒收敛性 Μ分析 优化设计 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]