Decentralized Control for Networked Control Systems With Asymmetric Information
Decentralized Control for Networked Control Systems With Asymmetric Information
复制标题
DOI：
10.1109/tac.2021.3073069
复制
发表时间：
2022-04
影响因子：
6.8
通讯作者：
Xiao Liang;Qingyuan Qi;Huanshui Zhang;Lihua Xie
中科院分区：
文献类型：
--
作者：
Xiao Liang;Qingyuan Qi;Huanshui Zhang;Lihua Xie
研究了信息不对称网络控制系统的分散控制问题。在该模型中，控制器2（C2）与控制器1（C1）共享其观测值和部分历史控制输入，而C2由于网络约束而无法获得C1的信息。在线性控制策略假设下，基于非对称观测，分别给出了C1和C2的最优估计。由于C1和C2的信息是不对称的，估计误差协方差（EEC）与控制器耦合，这意味着经典的分离原理失效。应用庞特里亚金极大值原理，得到了正倒向随机差分方程的一个解。基于此解，我们导出了最优控制器，以最小化一个二次成本函数。将线性最优控制器与EEC相结合，控制器C1与EEC解耦。应当强调的是，控制增益取决于估计增益。同时，估计增益满足前向Riccati方程，控制增益满足后向Riccati方程，使得问题更具挑战性。我们提出了迭代解的Riccati方程，并给出了一个次优解的最优分散控制问题。
This article considers the decentralized control for networked control systems (NCSs) with asymmetric information. In this NCSs model, the controller 2 (C2) shares its observations and part of its historical control inputs with the controller 1 (C1), whereas C2 cannot obtain the information of C1 due to network constraints. Under the linear control strategies assumption, we present the optimal estimators for C1 and C2 respectively based on asymmetric observations. Since the information for C1 and C2 are asymmetric, the estimation error covariance (EEC) is coupled with the controller which means that the classical separation principle fails. By applying the Pontryagin’s maximum principle, we obtain a solution to the forward and backward stochastic difference equations. Based on this solution, we derive the optimal controllers to minimize a quadratic cost function. Combining the linear optimal controllers with the EEC, the controller C1 is decoupled from the EEC. It should be emphasized that the control gain is dependent on the estimation gain. What is more, the estimation gain satisfies the forward Riccati equation and the control gain satisfies the backward Riccati equation which makes the problem more challenging. We propose iterative solutions to the Riccati equations and give a suboptimal solution to the optimal decentralized control problem.