Researches On H_∞ Filter Designed Method For Network Systems Based On Small Gain Theory

Networked control systems(NCSs) are the closed loop systems which are formed by the network connection controller, sensor and actuator. They have good fault-tolerant and fault detection ability, and easy maintenance, high reliability, etc. But, the time-delay phenomenon occurs when the signal are transmitted in the network communication channel. Due to the network bandwidth, the signal transmission need to be quantified, and it may cause packet loss. The nonlinear factors ane caused by the device itself or the field environment, and the state estimation problem is caused by the state signal of the system can not be directly measured, etc. The analysis and Research on the influence of these factors on the performance of the system have become a hot research topic in the control field. But, the H_∞filtering problem of network control systems has not been fully considered based on small gain theory.The small gain theory is the theoretical basis of the input and output stability method. The conclusion based on this method has less conservative, and it is more effective to deal with time-varying delay. This thesis studies the filter problem for networked control systems with time-delay, quantization and nonlinear according to the small gain theory.(1) The quantized H_∞filtering problem for discrete-time systems are analysed.The quantized H_∞filtering problem of systems is transformed into the scaled small gain problem through a process two-term approximation. By using the Scaled Small Gain(SSG) Theorem and LMI, sufficient conditions of the stability for the quantized filtering error system with a prescribed H_∞performance level can be obtained.(2) The H_∞filtering problem for discrete time-delay systems with quantization and stochastic sensor nonlinearity are considered. The nonlinearity is supposed to occur randomly on the basis of a stochastic variable obeying the Bernoulli distribution and the nonlinear decomposed technique is introduced to deal with the nonlinear term. The LMI-based sufficient conditions are presented to guarantee the error system mean square stable.(3) The quantized H_∞filtering problem for discrete time-delay systems with sector-bounded nonlinearity are discussed. The small gain method is also used to deal with the uncertainties obtained by the quantization error. In the case of the sector-bounded nonlinearity, the design method of the H_∞filter is given.