Study On Networked Control Systems And Multi-agent Systems

Applying the network into control systems has already become an active re-search field, which has extensive applications in many engineering fields, such asaerospace, equipment manufacturing, homemade highway system, etc. In thisdissertation, we mainly study stabilization of networked control systems and con-sensus of multi-agent systems with reference signal. The main results and contri-butions are as follows:1. By introducing the concept of efective sampling data-packet and assumingthe controllers to be time-division-driven, this paper first models the networkedcontrol systems with both time-delay and dropouts as a new switched system withtwo switching signals, indexed by both γ(im) and η(im), where γ(im) is generatedby time-delay, and η(im) is generated by the packet dropout process. Moreover,our modelling approach is more extensive in the application range since time-delayinvestigated in this paper can be constant, time-varying, random, smaller/largerthan one sampling period and the packet dropout process can be arbitrary orrandom. Then, we establish the corresponding stability and stabilization theoryfor NCSs with arbitrary and stochastic dropouts, respectively. Finally, numericalexample is given to illustrate the efectiveness of our results.2. For discrete-time multi-agent systems with time-varying reference signalunder directed fixed topology, this paper investigates its consensus problem in twoways: matrix method and stochastic matrix method. The matrix method focuseson designing the parameters to guarantee that the system matrix Φ has only one1eigenvalue and all the other eigenvalues lie in the unit circle. The stochasticmatrix method mainly designs the parameters to guarantee that the system matrixΦ is a stochastic matrix and then use the ergodic theory of stochastic matrix toestablish the consensus criteria. Finally, numerical examples are given to showthe efectiveness of our results.