Study On Modeling And Control Methods Of Networked Control Systems

With the development of computers and network technology, there is an increasingtendency to apply communication network to connect all parts of the control systems (sensors,controllers and actuators, etc). In this dissertation, we discuss the modeling, robustperformance, stability and controller design for NCSs in view of the network issues such astime-varying sampling periods, network-induced time delays and data packet dropouts. Theanalysis mainly base on Lyapunov stability theory, using LMI method and combined with avariety of mathematical tools like Schur complement lemma, projection lemma, asynchronousdynamic system theory, Markov chain system theory.etc. The main contents are as follows:Firstly, in view of the NCS with time-varying sampling period and random packet loss,discrete interval dynamic model of NCS is built, in which the randomness of the data packetloss is described via a stochastic Bernoulli distribution sequence, and the time-varyingcharacteristics of sampling period and transmission time-delay are represented by uncertaininterval matrices. Moreover, sufficient conditions for asymptotically stable of closed-loopsystem and the existence of robust H_∞controller for NCS with time-varying sampling periodand random packet dropouts are established in terms of linear matrix inequalities. Theadmissible controller guarantees robust asymptotic stability and a desired H_∞noise attenuationlevel.Secondly, for both cases of short-delay and long-delay, respectively, we consider thestabilization of NCS with random network-induced delay and asynchronous packet dropout.The uncertainty of network-induced delay is transformed into the uncertainty of state spaceequation coefficients, and we describe the data packet dropouts as a asynchronous dynamicalsystem with two events constraints. According to the theory of asynchronous dynamicalsystems, the sufficient conditions and the design methods of NCS state feedback controllerand dynamic output feedback controller are obtained. And then expressed by a form of thelinear matrix inequality (LMI).In order to explore further characteristics of the data packet dropouts, the state feedbackrobust H_∞control of NCS with short random time-varying delay and packet dropouts whichgoverned by Markov chain characteristic is discussed. The uncertainty of network-induceddelay is transformed into the uncertainty of state space equation coefficients, and the NCS ismodeled as Markovian jump linear system with two modes. According to the theory ofMarkovian jump linear system, the sufficient conditions of robust Hcontroller are presented based on linear matrix inequalities.Finally, considering the characteristics of actual distillation column control in petroleumand chemical, the network control is combined with the distillation tower temperature controlsystem. The design method in this dissertation is applied, and the simulation resultsdemonstrate the validity of this method. This part of content is seen as a replication tentativeapplication for practical problems.Furthermore, simulations are made for main designed schemes and demonstrate theeffectiveness of the proposed approaches.