H_âˆžControl For Networked Control Systems With Random Delays, Packet Dropouts And Quantizations  Posted on:20121213  Degree:Doctor  Type:Dissertation  Country:China  Candidate:D L Wen  Full Text:PDF  GTID:1228330467982691  Subject:Navigation, guidance and control  Abstract/Summary:  PDF Full Text Request  With the development of communication technologies, networked control has become one of the main streams of control systems design. Networkinduced time delays, packet dropouts and quantization due to the use of digital communication channels with limited bandwidth have become the research focus.The objective of the networked control is twofold:(i) to guarantee the closedloop system to be stable and,(ii) to ensure that the closedloop system satisfies some prescribed performance levels. In networked control systems (NCSs), time delays, packet dropouts will result in the degradation of the system performance or system stability, and the quantization will lead to such performance degradation. Therefore, NCSs performance analysis and controller design guaranteeing guaranteeing the performances are particularly important.There have been numerous investigations on stability analysis, controller design, and lots of results have been obtained. Quantization of NCSs has received increasing attention in recent years. The current emphasis mainly focuses on such problems as stability analysis and controller design for NCSs. When time delay, packet dropouts and quantization are considered simultaneously, the problems of Hâˆžcontrol for NCSs have not been taken into full consideration.This thesis, based on prior research, considers both the effects of time delays or packet dropouts and the need for signal quantization in network communication channels. This thesis presents Hâˆžcontrol methods and quantized control strategies via state feedback, dynamic output feedback and observerbased feedback, and the proposed methods can guarantee the exponential meansquare stability and Hâˆžperformance level of the closedloop systems. The plants studied in this thesis are linear timeinvariant discretetime systems. Networkinduced time delays and packet dropouts, resulting from the sensor to the controller and the controller to the actuator communication channels, are described by random variables satisfying the Bernoulli distributed sequence, and the random consecutive packet dropouts are described by the product of these random variables. Here, the quantizers are dynamic quantizers, and are conjuncted with static quantizers via dynamic scalings.First, based on LMI technique, the Hâˆžstate feedback controller design is given for NCSs with random delays such that the closedloop system is exponentially meansquare stable and with a prescribed Hâˆžperformance level. The Hâˆžstate feedback control problem for NCSs with random delays and simultaneous quantization is also considered. The proposed Hâˆžcontroller design and the quantized Hâˆžstrategy guarantee exponential meansquare stability of the closedloop system and the prescribed Hâˆžperformance level. The obtained conditions for quantized controller design are different from the ones that consider only the networkinduced random delays, which inflect the influence of quantization on design conditions.Secondly, the problems of Hâˆždynamic output feedback control with random delays and quantization, and Hâˆždynamic output feedback control with random packet dropouts and quantization. The methods of controller design are given for every problem, the nonconvex problem is converted to an LMI optimization problem, and the controller solution algorithm is presented by using the SLPMM. The proposed quantized Hâˆžstrategy can guarantee the exponential meansquare stability and Hâˆžperformance level of the closedloop system.Finally, the dynamic observerbased Hâˆžcontrol for systems with random delays and quantization, and dynamic observerbased Hâˆžcontrol for systems with random packet dropouts and quantization, are studied. The methods of the controller design are given for each problem, and the encountered nonconvex problem is converted into an LMI with matrix equation constraint. The proposed quantized Hâˆžcontrol strategies can guarantee exponential meansquare stability and Hâˆžperformance of the closedloop system.The details of this thesis are as follows:Chapter1summarizes the development and main research methods in time delays, packet dropouts, and quantization for NCSs.Chapter2provides preliminaries about the considered problems.Based on LMI technique, Chapter3studies the Hâˆžstate feedback control problem for NCSS with random delays and quantization. First, the Hâˆžstate feedback control problem is studied for systems with networkinduced random delays, which satisfies the Bernoulli distributed white sequence. The Hâˆžcontroller design is proposed such that the closedloop system is exponentially meansquare stable and with a prescribed H^performance level. Then, the state feedback Hâˆžcontrol problem for systems with random delays and quantization is studied. The quantizers are dynamic ones, which are composed of static quantizers and dynamic regulation parameters. The Hâˆžcontroller design and the quantized Hâˆžstrategy, which guarantee exponential meansquare stability and the Hâˆžperformance level, are presented. The obtained conditions for quantized controller design are different from the ones that consider only the networkinduced random delays, which inflect the influence of quantization on design conditions.Chapter4investigates the Hâˆždynamic output feedback control for discretetime NCSs with random delays and quantization, and Hâˆždynamic output feedback control for discretetime NCSs with random packet dropouts and quantization. The random delays and random packet dropouts, which occur in the sensortocontroller channel and controllertoactuator channel, are considered. The random delays or packet dropouts are described by the random variables taking0or1, which satisfies the Bernoulli distributed white sequence. The random consecutive packet dropouts are described by the product of random variables. The quantizers considered here are also dynamic ones. The Hâˆždynamic output feedback controller design for NCSs with random delays and quantization, and the Hâˆždynamic output feedback controller design for NCSs with random packet dropouts and quantization are proposed. The nonconvex problem is converted to an LMI optimization problem, and the controller solution algorithm is presented by using the SLPMM. The Hâˆžcontroller design and the quantized Hâˆžstrategy are proposed, which guarantee that the closedloop system is exponentially meansquare stable with the prescribed Hâˆžperformance level. The numerical examples have shown the effectiveness of the proposed approaches.Chapter5studies the dynamic observerbased Hâˆžcontrol for discretetime NCSs with random delays and quantization, and the dynamic observerbased Hâˆžcontrol for discretetime NCSs with random packet dropouts and quantization. The descriptions of random delays and packet dropouts during the digital transmission are based on Chapter4, and the dynamic quantizers are also used in this chapter. The nonconvex problem presented for each problem is converted into an LMI with matrix equation constraint. The Hâˆžcontrol design and the quantized Hâˆžstrategy, which guarantee exponential meansquare stability and the Hâˆžperformance level of the closedloop system, are presented. The numerical examples have shown the effectiveness of the proposed approaches.In Chapters6, the results of the dissertation are summarized and further research topics are suggested.  Keywords/Search Tags:  Linear timeinvariant discretetime systems, networkinduced random delays, networkinduced random packet dropouts, dynamic quantizer, static quantizaer, quantizer range, quantization error, H_âˆžcontrol, state feedback, dynamic output feedback  PDF Full Text Request  Related items 
 
