分布式系统预测控制的协调策略与系统综合

With the development of industrial technologies and the field-bus technologies, the centralized control has been gradually substituted by the distributed control in many industrial fields. The informatiaon structure and the control mode changed, each subsystem can be well controlled by corresponding subsystem-based controller, but how to realize the optimal control of entire system is still an important problem to be solved. In distributed control framwork, each controller calculates its control law according to non-globle information, to design a effective coordination strategy which could improve the performance of whole system is an important way to deal with the problem mentioned obove. In this project, the control structure, coordination strategy and the performance of distributed Model Predictive Control will be studied, and it includes: to analyze the relationship between the control structure and the closed-loop system's performance under the control of distributed controllers; to design a strategy for coordinating distributed Model Predictive Control, comparing to the existing methods, which can be able to achieve better global performance of closed-loop system with the same control structure and available information; to analyze the stability and optimality of closed-loop system under the control of proposed cooperative distributed Model Predictive Control. Then, an systematic methodology is developed for designing the distributed Model Predictive Control with non-global information, which will promote the development of the theory of distributed Model Predictive Control. Due to directly face the industrial demands, the proposed approaches demonstrate the strong innovation character and have a high potential in industrial applications.

随着工业技术的发展和现场总线技术的成熟，工业系统控制方式已由集中式向分布式转变。系统信息结构和控制方式都发生了改变，各局部控制器可以实现子系统的控制，但如何实现全局系统的最优控制仍是主要问题。在分布式结构下，对信息的利用由全局变为非全局，因而，通过有效的协调策略提高系统的全局性能是解决上述问题的主要途径。本课题针对这个问题深入分析分布式系统控制结构与闭环系统性能之间的内在联系和本质规律，并提出最优控制结构的设计方法；设计非全局信息模式下分布式预测控制器的协调策略，使得在同等控制结构下，与现有方法相比系统全局性能得到改善；分析本课题提出的分布式预测控制的稳定性和全局最优性，并给出性能指标及保稳定性控制器的设计方法。形成设计非全局信息模式下协调预测控制的系统理论和方法，将分布式预测控制理论向更深的层面推进。同时，本课题的研究内容面向工业过程控制的实际需要，具有很强的应用价值。

本项目在分布式预测控制框架下，针对如何在不降低系统容错性、不提高网络通信复杂度的情况下，改善系统全局性能这一分布式控制的关键问题： 提出基于N步临界矩阵的系统结构划分方法，使得根据该方法设计得到的分布式预测控制能够在不需要全局信息的情况下，得到 “帕累托”最优；提出非全局信息模式下基于敏感度函数预估和作用域优化的协调分布式预测控制器的协调策略，使得在相同信息条件下与现有方法相比，提高系统的协调度，进而改善系统的全局优化性能；给出了适用于采用协调度来提高系统性能的这一大类分布式预测控制的保稳定性的控制器综合方法；并以上海武宁科技园区微能源网为背景进行算法验证。形成了系统的非全局信息模式下分布式标称系统的分布式预测控制理论。