Abstract: This paper combines learning theory with robust control and discusses robust control design problems involving real parameter uncertainty in control systems based on randomized algorithms. It is shown that randomized algorithms can decrease the computational complexity dramatically instead of seeking worst case guarantees. In addition, examples in this paper show that employing randomized algorithms is very efficient and has obvious advantages especially when uncertain interval parameters appear multilinearly or nonlinearly in the characteristic polynomial coefficients.

Key words: randomized algorithms, learning theory, robust control

摘要： 将学习理论与鲁棒控制相结合，采用随机化算法针对实参数不确定系统讨论了鲁棒控制器的设计问题。研究表明，在不考虑最坏情况的意义下，随机化算法可以显著降低计算复杂性，另外，当不确定区间参数以多线性或非线性的方式出现在特征多项式系数中时，采用随机化算法具有明显的优点并且是非常有效的，文中给出了计算实例。

关键词: 随机化算法, 学习理论, 鲁棒控制