H_∞Filtering For Network-based Systems With Time Delay

With the widely applications of computer networks and the development ofautomation and control technology, Network Control Systems (NCSs) are becoming atrend of future control systems. When comparing with the traditional control system,which the controller signals receive sensor signals by establishing point-to-pointconnection, NCSs have many advantages, such as costs lower, more reliability andmore flexibility. H∞filtering has good robustness, even if there exist system modeluncertainties and external noise signals or noise signals from the equipment itself. Inthe thesis, H∞filtering is given to signal estimate for the network control systems,which benefits the realization of the control strategy.This thesis is modelled the NCS with time delay and disorder. The design of H∞filtering in the NCS with time varying delay and network-based Markov jump systemare researched respectively. The major contents of the thesis are concluded as follows:First, the state equation of NCSs with time-varying delay is established. The H∞filtering state equation is formulated based on the characteristic of H∞filtering. Thefiltering error equation is listed by combining two groups of state equation above.Choosing a new Lyapunov-Krasovskii functional analyzes the stability of system. Theintegral terms with time-delay in the Lyapunov-Krasovskii functional are transformedby Newton-Leibniz formula and inequality approaches. Finally, the quadratic form isturned into linear matrix inequality by Schur complement lemma. By solving the LMI,we can achieve the stability conditions, the performance indicators and the parameterof filtering.Second, the design of H∞filtering for Markov jump system based network isconcerned. Since the model of Markov jump system can formulate the randomvariation, external noises and information time delay in the real systems effectively,the model of Markov jump system is established, in which the system parameter isdetermined by the network information parameters. Using zero order hold chooses thenewest data information, which can avoid out of order phenomenon. The sufficient condition of stochastic stability is derived by a bounded real lemma and somepractical lemmas. In terms of the simulation results by LMI toolbox, the design offilter system is asymptotic stability, and can estimate the signal state accurately.