中文摘要：

针对分布式控制系统的特点，提出了一种新型的基于数据的分布式预测控制优化算法。由输入、输出数据直接设计分布式控制器，控制器在各个子系统通信的条件下采用基于纳什最优的分布式控制优化算法，以较低的成本达到整个大系统的性能优化。这种基于数据的方法使传统的预测控制器设计过程中的系统辨识和基于状态空间模型的预测控制简化为直接的一步，即直接利用数据设计分布式控制器。本文给出了这种新型算法的收敛条件，仿真结果证明了该方法的有效性。

英文摘要：

Aiming at the characteristics of distributed control systems, this paper proposed a new data- driven distributed predictive control strategy. The input-output data are directly utilized to design the distributed controller of each subsystem. The input-output information is exchanged among subsystems via network, which can make each controller work in coordination with the others. By this way, the control performance of each subsystem is improved by considering the interactions among subsystems. An iteration algorithm is utilized to achieve the Nash optimality and the convergence condition for iteration is also given. The simulation results demonstrate the effectiveness of the proposed approach.