Analysis And Design Of Networked Control System Based On Free-weighting Matrices

With the rapid developments of computer, network and communication technologies, networked control systems (NCSs) have been widely applied in many control fields. In NCSs, sensors, actuators, and controllers interconnect by communication networks, which always cause time delays. Network-induced delays degrade the control performance, even make NCSs unstable. Moreover, since network-induced delays are usually time-varying, it makes the analysis and design of NCSs more complex. The purpose of this dissertation is to find the maximum allowable delay bound (MADB) and design valid controllers to guarantee the stability and control performance of NCSs within the MADB. By using Lyapunov stability theory and free-weighting matrix approach, the issues, including stability analysis, stabilization control, guaranteed cost control, robust H?control and robust H?filter design of NCSs, are investigated. The main achievements in the dissertation are outlined as follows:(1) The stability conditions of both continuous-time and discrete-time NCSs, and the design methods of state feedback stabilizing controllers are proposed.For the continuous-time NCSs with time-varying delay, using Lyapunov-Krasovskii functional and the improved free-weighting matrix approach, the stability conditions of NCSs are presented with all terms in the derivative of Lyapunov-Krasovskii functional reserved. Moreover, the conditions are extended to the NCSs with time-varying structured uncertainties. Based on the stability criteria derived, the design method of state feedback stabilizing controller is proposed. The improved cone complementarity linearization (ICCL) algorithm based on linear matrix inequalities (LMIs) is employed to deal with nonlinear matrix inequalities (NLMIs), so that the MADB guaranteeing the stability of NCSs and the parameter of stabilizing controller are obtained. Similarly, the stability conditions for discrete-time NCSs and the design method of discrete stabilizing controller are proposed. The numerical examples demonstrate the effectiveness and less conservativeness over the existing methods.(2) The design methods of guaranteed cost controllers of both continuous-time and discrete-time NCSs are presented.According to the given control performance conditions of continuous-time NCSs, the design method of guaranteed cost controller of NCSs is investigated by using Lyapunov-Krasovskii functional and the improved free-weighting matrix approach. The parameter of controller to guarantee given performance is obtained by employing the ICCL algorithm. Meanwhile, the design method of guaranteed cost controller for discrete-time plants is proposed. The numerical examples show the less guaranteed cost performance comparing with the existing methods.(3) The design methods of robust H?controllers of both continuous-time and discrete-time NCSs are proposed.For the continuous-time NCSs with time-varying delay, the criteria for H?performance of NCSs are presented using Lyapunov-Krasovskii functional and the improved free-weighting matrix approach. Based on the criteria derived, the design method of H?state feedback controller is obtained, and it is extended to robust H?controller. The discrete H?controller can be designed by the similar method. The results of the numerical examples demonstrate the better H?performance over the existing methods.(4) The design methods of robust H?filters of both continuous-time and discrete-time NCSs are proposed.Lyapunov-Krasovskii functional and the improved free-weighting matrix approach are employed to design robust H?filter of continuous-time NCSs. The criteria for robust H?performance analysis of the filtering-error systems are proposed by reserving all terms in the derivative of Lyapunov-Krasovskii functional and dealing with the different parts of uncertainties separately. Based on the derived criteria, the design method of robust H?filter is obtained in term of LMIs. Similarly, the design method of robust H?filter for discrete-time NCSs is proposed. Numerical examples are given to demonstrate the validity and the less conservativeness over the existing methods.