The Research on Filtering with Random Packet Loss and Delay in Networked Control System The Research on Filtering with Random Packet Loss and Delay in Networked Control System
|School||Zhejiang University of Technology|
|Course||Control Theory and Control Engineering|
|Keywords||Network control system packet loss time delay linear quadratic Gaussian control(LQG) problem Kalman filtering|
The real control systems currently have a deep connection with computer technology andnetwork communication technology development to form network control system(NCS).However, the instability characteristics of communication networks always lead to some negativephenomenon, which include the induced delay, data packet loss caused by multi-packet datatransmission, data congestion and the sequential disorder. These phenomenons affect inevitablythe performance of the control system, and even lead these systems to be unstable. Therefore,many scholars have paid their attentions to study the networked control systems with randompacket loss and delay over the years.For network control system with time-delay or packet loss, the problem of robust stabilityand controlling performance is of great importance. The robust stability of a class ofvarying-delay uncertain network control system using delay segmentation method is studied andthe delay-dependent stability criteria are obtained. And this dissertation also considers theproblem of the guarantee cost control (GCC) with the packet loss compensate for networkedcontrol systems under unstable communication links with packet dropouts and gets the stabilityregion of packet loss compensation. The filtering of network control system with both packet lossand time delay is also investigated to get a Kalman filter finally.The main parts of this paper are outlined as follows.(1) The problem of robust stability for a class of network control system with uncertainvarying-delay is investigated, the uncertainties include nonlinear perturbation and norm-boundeduncertainty. By constructing suitable Lyapunov function and using Jensen inequality conversiontechnology, some sufficient conditions for stability are given in terms of LMI.(2) The problem of filtering of linear quadratic Gaussian control for network control systemwith random packet loss is studied. Using the filtering estimates to design the optimal controllerand by the optimal cost function to get the optimal controller gain, the optimal controller law andcritical packet loss probability between sensors and controllers are presented. (3) The problem of filtering of network control system with both packet loss and time delay isalso investigated, using Kalman filtering theory and recursive projective theorem to design afilter.Finally, the applications of delay segment Lyapunov function for the delay-dependent robuststability of network control systems, two packet loss compensation methods on linear quadraticGaussian control problem and the filter designed for network control system with packet loss andtime delay are summarized. And the perspective of future studies is discussed at the end of thedissertation.