Research On Networked Control Systems With Uncertain Data Losses And Time Delays

Feedback control systems in which the control loops are closed through data and command communicating networks are called networked control systems(NCS). Compared to traditionally point-to-point control systems, the primary advantages of an NCS are reducing system wiring, easiness of system diagnosis and maintenance, and increasing system agility. However, with the insertion of communicating networks in the feedback control loop, it makes the analysis and design of an NCS complex. As we know, due to the limit of the capacities and bandwidth of the physical network communication, packet loss and time delays are unavoidable and mostly random in a network communication channel. Consequently, the NCS designed under idealized network conditions would be disturbed in practical network with time delay or(and) packet loss, and its performance could be most likely decreased. Therefore, the analysis and stabilization issues of networked control systems under non-ideal conditions have been illustrated extensively by interested researchers in world wide. In this thesis, the issues such as modeling, stability analysis, and stabilization of NCS under non-ideal communication conditions are investigated.Using a gain scheduler algorithm to modulate the output of an existing controller externally by middleware, a novel method is proposed to compensate for stochastic network delay, without changing the existing controller. Under discrete linear quadratic regulation algorithm, the discrete linear quadratic optimal gain scheduling sequences and the lookup table of optimal scheduling sequences corresponding to stochastic time delays are both designed offline. At the same time, the steps using the lookup table method to adjust the gains of the existing controller online externally are introduced. Finally, a simulation example shows the validity of this method.Furthermore, the modeling and stochastically controller design problems of the NCS with Markovian packet-loss are investigated in this thesis. In particular, we consider a more general case where there exist some unknown or unavailable elements in the transition probabilities matrix of the Markovian packet-loss process. Firstly,the NCS with packet-loss is modeled as a Markovian jump linear system(MJLS) by discrete system method. Then, sufficient conditions for stochastically stable of closed-loop NCS are obtained via MJLS theories and packet-loss Lyapunov function approach. Moreover, packet-loss dependent stabilizing controller is designed with linear matrix inequalities(LMIs) formulation. Finally, a numerical example for comparison between two methods is given to illustrate the positive effectiveness of the proposed theory.Finally, the modeling and stabilizing control of networked control systems with stochastic packet loss and time-varying delays are addressed. In specific, a novel Markovian jump linear system with time-varying delays is used to model the networked control system (NCS). Without using the augmented state method, sufficient conditions for the stochastic stabilization of the NCS with packet loss and time-varying delays are obtained via the mode-dependent Lyapunov function method. The mode-dependent controller for the closed-loop NCS is presented in the linear matrix inequalities(LMIs) formulation via the Shur complement theory. A numerical example is given to illustrate the effectiveness of the proposed method.