Analysis And Control Of Networked Control Systems With Markovian Characteristics

With the rapid development of computer and network technology, the control system based on the high speed communication network, which is so called networked control systems (NCSs), has become one of the focus issues in current international control field. The research on NCSs has important theoretical meanings and extensive application prospects. NCSs are feedback control system wherein the sensors, actuators and controllers as the network nodes connected through the computer network or fieldbus to accomplish the control tasks. The enormous advantages of NCSs, including lower cost, reduced weight and power, simpler installation and maintenance, ease of system diagnosis, and increased system flexibility, etc., have made NCSs increasingly popular in many real-time distributed control applications such as industrial automation, intelligent vehicle systems, mobile robotics, advance aircraft and spacecraft and national defense, etc.. However, the insertion of the network in the closed-loop systems brings many new issues inevitably, such as network-induced delays, data packet dropouts, multiple-packet transmissions, clock synchronization, etc.. These issues can degrade the performance of control systems and can even destabilize the system, which make the analysis and synthesis of NCSs complex. Therefore, modeling and control of NCSs have become one of the most important issues in control theories study, and also have received considerable attention and interest by researchers in recent years.This thesis has systematically analyzed the characteristics of networked control system, network-induced time delays and data packet dropouts etc., and the effects upon the control performances, then propose newly designed methods which solve the stability and stabilization problems of the control system brought by the network-induced time delays and the data packet dropouts. Firstly, in terms of the networked-induced time delays and data packet dropouts, the modeling and control problems of NCSs with continuous-time plant are investigated. Based on Markovian characteristics of the networked-induced delays, we propose the Markov jump model of NCSs, which has successfully transformed the uncertainty of the delays into the uncertainty of the systems'input matrices during the process of discretization, then present the sufficient conditions for the stability of the NCS, and give the state feedback controller design methods based on LMI technique. Furthermore, we consider both networked-induced delays and packet dropouts events happened in the network, describe them as two independent Markov chains, and set up corresponding model, utilize the linear jump theories to analyze the stability and the stabilization of the control systems. Finally, numerical examples are given to illustrate the validity of the results.Secondly, in terms of the networked-induced time delays, the modeling and control problems of NCSs with discrete-time plant are discussed. We set up a closed-loop NCS model based on the Markovian characteristics of networked-induced delays, then analyze the stability of the system, and propose an LMI method for the controller design. To be more general, owning to the network's randomness and unpredictability, it can be argued that in the actual network, all the elements in the desired transition probabilities matrix are hard to obtain, in order to solve the problem, we design a more general model with partly unknown transition probabilities, give the sufficient conditions for the stability of the closed-loop system, and propose the method for the controller design. Eventually, numerical examples are given to demonstrate the effectiveness of the proposed methods.