Abstract: A novel control scheme combining disturbance observer technique and generalized minimum variance method is investigated for a class of linear stochastic systems. The unknown external disturbance is supposed to be generated by a exogenous system with perturbation, and the reduced-order and full-order disturbance observers based on minimum variance benchmark, which can be designed separately from the controller design, are constructed, respectively, for the cases that the state can be measured and the state cannot be measured. By integrating the disturbance observers with generalized minimum variance control laws, the disturbances can be rejected and the desired dynamic performances can be guaranteed. Finally, a numerical example is proved to show the effectiveness of the approach.

Key words: Stochastic systems, disturbance observer, minimum variance, composite control

摘要： 为提高控制精度，针对一类带有附加干扰的线性随机系统，提出了一种基于干扰观测器的复合分层控制方法。假定位于输入通道的干扰可由一个带有摄动的外部动态方程来描述。基于最小方差准则，就系统状态可量测和系统状态不可量测两种情形来分别设计干扰观测器，并且可以依据分离原理，将观测器设计与控制器设计分解成两个独立的步骤先后进行。通过将基于干扰观测器的控制方法(DOBC)与广义最小方差控制(GMVC)方法相结合，不仅能够抵消可建模干扰，而且能够提高系统的动态性能。最后，仿真算例验证了该方法的有效性。

关键词: 随机系统, 干扰观测器, 最小方差, 复合控制