Stability Analysis of Continuous-time Iterative Learning Control Systems with Multiple State Delays
-
摘要: 探讨了含多状态时滞连续时间迭代学习控制系统的稳定性分析问题, 尤其是当系统参数带有多面体不确定性时的鲁棒稳定性分析问题. 通过引入一个扩展算子, 利用迭代学习控制中的二维分析方法给出了时滞系统整个学习动态过程的连续离散Roesser系统描述. 基于所得的Roesser系统, 首先利用二维系统理论给出了保证迭代学习控制系统渐近稳定的充要条件, 然后结合鲁棒H∞控制理论提出了以线性矩阵不等式形式描述的充分条件来保证迭代学习控制系统的单调收敛性. 结果表明, 通过求解线性矩阵不等式确定的学习增益可以使控制输入误差随着迭代次数的增加单调收敛于零. 仿真结果表明, 通过增加满足一组线性矩阵不等式条件的P型学习增益能够使得一个鲁棒渐近稳定的迭代学习控制方案变为鲁棒单调收敛的, 同时还可以大大提高收敛速率.Abstract: This paper presents a stability analysis of the iterative learning control (ILC) problem for continuous-time systems with multiple state delays, especially when system parameters are subject to polytopic-type uncertainties. Using the two-dimensional (2-D) analysis approach to ILC, the continuous-discrete Roesser's type linear systems are employed to describe the entire learning dynamics of time-delay systems (TDS) with the development of an expanding operator. Based on such Roesser systems, the 2-D system theory is first used to develop a necessary and sufficient condition for the asymptotic stability of ILC, and then the robust H∞ control theory is combined to provide a sufficient condition in terms of linear matrix inequalities (LMIs) for the monotonic convergence of ILC. It is shown that learning gains can be determined by solving LMIs, which ensure the control input error converges monotonically to zero as a function of iteration. Simulation results show that a robust asymptotically stable ILC scheme can become robustly monotonically convergent by adding the P-type learning gains that satisfy a set of LMIs, which can also improve the convergence rate greatly.
点击查看大图
计量
- 文章访问数: 2464
- HTML全文浏览量: 58
- PDF下载量: 1244
- 被引次数: 0