Department of Automation, Zhejiang University of Technology, Hangzhou, 310023, People's Republic of China.
ISA Trans. 2011 Jan;50(1):44-52. doi: 10.1016/j.isatra.2010.10.001. Epub 2010 Oct 30.
The receding horizon H(∞) control (RHHC) problem is investigated in this paper for a class of networked control systems (NCSs) with random delay and packet disordering. A new model is proposed to describe the NCS with random delay which may be larger than one sampling period. The random delay is modeled as a Markov chain while the closed-loop system is described as a Markovian jump system. Sufficient conditions for the closed-loop NCS to be stochastically stable and the performance index to be upper bounded are derived by using the receding optimization principle. Furthermore, by solving a semi-definite programming (SDP) with linear matrix inequalities (LMIs) constraint, a piecewise-constant receding horizon H(∞) controller is obtained, and the designed piecewise-constant controller ensures that the closed-loop NCS achieves a prescribed H(∞) disturbance attenuation level. Finally, an illustrative example is given to demonstrate the effectiveness of the proposed method.
本文针对一类具有随机时滞和数据包丢失的网络控制系统（NCSs），研究了预测控制（RHHC）问题。提出了一种新的模型来描述具有随机时滞的 NCS，该随机时滞可能大于一个采样周期。随机时滞被建模为马尔可夫链，而闭环系统被描述为马尔可夫跳跃系统。利用预测优化原理，推导出了闭环 NCS 随机稳定和性能指标上界的充分条件。此外，通过求解具有线性矩阵不等式（LMI）约束的半定规划（SDP）问题，得到了分段定常预测控制律，并证明了所设计的分段定常控制器能使闭环 NCS 达到预定的 H（∞）干扰衰减水平。最后，通过一个实例说明了该方法的有效性。
