针对复杂的控制系统对控制器完备性要求高的问题,提出了代理控制结构。将完备的集中式控制算法,分解为多个面向具体任务简单分布式控制代理;采用基于数据驱动的控制思想设计控制代理,充分利用系统输入输出计算最优控制律,从而避免了对复杂的动态系统建模困难的问题。仿真结果表明,数据驱动的网络化系统代理控制体系是有效的,且相对于未使用代理的网络体系结构具有更高的优越性,降低了网络控制器设计的复杂性。
As the complex system has a high demanding for the completeness of the controller, an agent control structure is proposed in this paper. It decomposes the complete and centralized control algorithm into more agent controllers for the specific and simple tasks. The agent controllers are designed based on the data-driven control theory which calculate the optimal control output by the input and output data. So the modeling problem is avoided when the plant is hard to model. Simulation results demonstrate the effectiveness of the data-driven agent control theory, Moreover the new approach has much better control performance than the network control system without the agent controllers, for it decreases the complexity of controller design.