中文摘要：

网络时延的时变特性导致只有某些时刻的对象输出被控制器端接收,回路时延的形式较复杂.采用开关式的状态观测器,基于接收到的输出估计系统状态,建立状态估计误差系统.为讨论切换系统均方稳定的条件,推导了多随机变量序列的Markov特性,进而在Markov跳跃系统的基础上得到了闭环系统均方稳定的充分条件.该条件由状态估计误差系统与时延状态反馈控制系统的均方稳定条件组成,所以控制器和观测器满足分离特性.此外,通过增广状态向量利用系统的Markov切换特性得到了系统均方稳定的充分必要条件.最后用数值实例验证了预估补偿控制策略的有效性.

英文摘要：

The networked predictive control is studied. Due to time-varying delays, not all of the plant outputs are available for the controller and the form of round-trip delay is complicated. By applying a switching observer,the system states are estimated based on the available outputs and thus the state estimation error system is obtained. To discuss the mean square stability of the switching system, the Markov switching property of multiple random variables is derived. Based on the Markov jump linear system, a sufficient condition of mean square stability is proposed, which is composed of the mean square stability conditions of both state estimation error system and delayed state feedback control system. Thus the observer and controller have separation properties. Moreover, by augmenting state vectors, the sufficient and necessary condition of mean square stability is obtained. The validity of the predictor-based compensator is verified via a numerical simulation.