Modeling, Analysis And Decentralized Control Of Networked Large-scale Systems

Decentralized networked control systems are complex control systems with mul-tiple control stations while transmitting control signals through network. Not only could it keep the superiority that decentralized control systems hold at processing control systems that distribute in extensive regions, but also it can send over real-time control signals with the help of network, which greatly enhances the flexibility and reliability of the systems. As the product of systemic control theory and networked communication technique, there is no denying that it embraces enormously applied value. This thesis focuses on the problems of modeling, analysis and decentralized control of networked large-scale systems.Firstly, we introduce the study purposes and significations of decentralized net-worked control systems. Then some basic issues and recent progress of networked control systems and decentralized control systems are given respectively. Besides, the fundamental knowledge cited in this paper is also concisely offered.Secondly, the mathematical models of networked large-scale systems which con-sider the network-induced delay and transmitted data dropout are developed respec-tively. For one thing, according to the characteristics of delay, the input of each chan-nel of the controlled plant is deemed as segmental constant value in order to acquire an uncertain large-scale system model in the course of converting continuous system into discrete one. For another, we equate the data dropout with the network's discon-nection. As a result, by defining some Bernoulli stochastic variables which are used to describe each channel's data dropout during information delivering process and then applying the feedback controller to the open-loop system, the issue can be modeled as a closed-loop stochastic system.Thirdly, the problems of stability analysis and controller synthesis for networked large-scale systems are investigated primarily. For a class of networked large-scale system with induced delay, the dynamic decentralized robust controller is derived by Lyapunov stability theory, which make sure the closed-loop system is asymptotic sta-ble and have a specified disturbance attenuation level. While for a class of networked large-scale system with data dropout, we set the Lyapunov matrix as special block di-agonal structure. Accordingly, by using some matrix substitute techniques, the bilinear matrix inequality can be transformed to linear matrix inequality that can be easily solved by MATLAB/LMI. Then we obtain output feedback decentralized controllers which ensure the closed-loop system to be exponentially mean-square stable and also have a specified disturbance attenuation level, when considering the controlled plant is interconnected large-scale system and multi-channels one respectively. Subse-quently, several numerical examples which are relative to each situation above are presented. Furthermore, we acquire the controller coefficients by means of defining some complex matrixes based on linear matrix inequality toolbox. Then we design appropriate MATLAB/Simulink module to incarnate network's property of delay and data dropout; therefore, the feasibility and effectiveness of the proposed methods can be verified by taking simulation experiments.Lastly, a summary of the main research methods used in this paper is presented, coupled with some prospection to decentralized control of large-scale systems based on network environment. The drawbacks exist in this article and the possible im-provement directions are pointed out simultaneously.