Guaranteed Cost Controller Design For Two Classes Of Networked Control Systems

With the development of science and technology, computer technology, network technology and intelligent sensor technology etc. have been integrated into the industrial control systems gradually. Control systems combine with network communications have become a new direction of control theory, so Networked Control Systems (NCS) is proposed. The system which based on network environment can be remote controlled, easily extended, easily maintained, flexibility and low-cost. Those advantages cause the research of NCS to hold a very important practical background. However, the communication network in control systems must make them have some unexpected problems, such as, network-induced delay, data missing, packet disorder etc. These negative factors affect the performance seriously and even cause system instability. As a result, the stability research of NCS has very important practical significance.Basing on Lyapunov stability theory, this thesis investigates guaranteed cost controller design problem for two classes of NCSs, which uses linear matrix inequality (LMI) to design guaranteed cost controller. This thesis is organized as follows.The first chapter brings in background, significance, problems considered, and recent research achievements of NCS.The second chapter introduces mathematical basis of control theory and preliminary knowledge for this paper, such as, definition of exponentially mean-square stable, the discrete system model of short delay NCS, summary on non-fragile control, the basis of linear matrix inequalities and some lemmas in this paper, etc.The third chapter studies a class of short delay NCS. By viewing the time-varying delays as time-varying parameter uncertainties, and considering control gain perturbations, non-fragile state feedback controller is designed so that the system is asymptotic stability and satisfies the quadratic performance index.The forth chapter analyses a class of uncertain NCS which both have measurement data missing and controi data missing and does not consider network-induced delay. The random data missing is modeled as a stochastic variable satisfying Bernoulli distribution, therefore, dynamic output feedback controller is designed for general system and time-delay system respectively so that the system is exponentially mean-square stable and satisfy the quadratic performance index.The fifth chapter concludes this paper and looks forward to the research of NCS.