迭代学习控制系统的鲁棒性分析

Robustness Analysis for Iterative Learning Control Systems

讨论了在初态偏离、状态输出扰动和非线性扰动同时存在的干扰环境中运行的迭代学习控制系统的鲁棒性问题．通过更精确的误差渐近界估计，结合迭代学习控制算法中的开环和闭环方案，给出了算法的鲁棒性条件，以及算法收敛性所要求的渐近干扰条件．

Combining the open-loop algorithm and closed-loop algorithm,the robustness of the improved iterative learning control schemes with respect to initial state bias,disturbances of state and output, and nonlinear fluctuation is studied extensively. Via more precise estimation for asymptotic bounds of iterative errors, the sufficient conditions for convergence are provided. It exhibits that motion trajectories converge to the desired one in the existence of asymptotic invariant disturbances under P-type learning law, and the convergence of D-type learning law can not be guaranteed in the presence of these kinds of disturbances. In addition, the uniform boundedness of motion trajectories is also proved.