Due to the rapid development of the information science, the computer network has become the core of the modern communication technology. It is being applied into almost all aspects of everyday life. Networked control systems introduce the concept of network into control systems. The components in the networked control systems are connected by the network. Since networked control systems have many advantages, they have received considerable attention and are gradually replacing the role that traditional point to point control systems play in the control area. However, the introduce of the network also brings challenges, such as, networked-induce delay, packet dropouts and quantization errors. In this thesis, much efforts are done based on the three most common phenomena.Randomly multi-step networked-induce delay, multiple packet dropouts and quantization errors of the networked control systems are considered in the thesis. It also proposes new models to describe the three characteristics. In order to solve the problems of networked control systems with communication constraints and incomplete information, the unbiased minimum covariance filtering, the robust filtering, the set-membership filtering and the H∞control, the quantized H∞control methods are proposed. The main contents of the thesis are as follows:(1) In the practical networked control systems, randomly multi-step sensor delays arise much more frequently than one-step sensor delay or deterministic delays. According to the phenomenon, a new model is designed to describe the randomly multi-step sensor delays. Based on the proposed new model, an unbiased minimum covariance filtering method is presented for the networked control system with randomly multi-step sensor delays. Also, the conditions for the filter to be unbiased and the filtering algrithm are proposed. The effectiveness of the proposed method is proved by MATLAB simulations.(2) Since randomly multiple packet dropouts often happen in the practical networked control systems, a new model which is able to describe the randomly consecutive multiple packet dropouts with known maximum number of the dropouts is proposed. According to the new mathematical model, a modified H∞filtering is designed and the corresponding filtering algorithm is proposed. Simulation results show the effectiveness of the proposed H∞filtering method for networked control systems with randomly multiple packet dropouts. (3) In the real-world applications, the noise signal is normally unknown-but-bounded. According to this, the advantages of the set-membership filtering for systems with unknown-but-bounded noises are analyzed. The logarithmic quantizer is used to describe the quantization error. For the class of networked control systems in the presence of both unknown-but-bounded noises and the quantization errors, a set-membership filter is designed and a set-membership filtering algorithm is proposed. A simulation example is conducted to show the effectiveness of the proposed method.(4) Based on the proposed model for the randomly multi-step sensor delays, a mathematic model is designed for the randomly multi-step transmission delays. A logarithmic quantizer is added into the control loop before the measurement enters into the transmission network. A quantized H∞control problem is proposed for networked control systems with randomly multi-step transmission delays. A quantized H∞, control strategy is designed such that the closed-loop system is asymptotically mean square stable and the control output meets the H∞performance. According to the design of the controller, a quantized H∞, control algorithm is proposed and a simulation example is conducted to verify its effectiveness.(5) The linear parameter varying method has advantages in dealing with nonlinear systems by applying the linear methodologies to the nonlinear systems. A new control problem for networked control systems with randomly multi-step sensor delays under the linear parameter varying framework is proposed. A modified H∞, controller is designed and the controller parameters are calculated by the extended CCLM algorithm such that, the closed-loop system is asymptotically mean square stable and the controlled output meets the improved H∞performance. Simulation results show the effectiveness of the proposed method. |