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。 Combining learning theory with robust control, a stochastic algorithm is proposed to solve the problem of robust controller design for uncertain real parameters. The research shows that the randomization algorithm can significantly reduce the computational complexity without considering the worst case. In addition, when the uncertain interval parameters appear in the characteristic polynomial coefficients in a multi-linear or nonlinear way, the randomization algorithm Has obvious advantages and is very effective. The calculation example is given in this paper.