A Study Of Net-worked Systems Using Quantized Measurements

Networked control systems (NCSs), which are also called network-based control systems,are the closed-loop control systems where the plant, sensor, controller and actuator are con-nected through shared network. Compared with the traditional point-to-point control systems,the NCSs provide more advantages such as low cost, reduced weight and power requirements,simple installation and maintenance, etc. In realizing remote control at the same time,Networkcontrol system also has the information processing, management and decision-making func-tions.such that Network is Applied to control system deeperly. In addition,since the limitionof Network's communication bandwidth, carrying capacity, and service capacity , the Perfor-mance of control system is suppressed. Therefore, quantier become a indispensable part of theNCSs due to the bandwidth constraints. Thus, it is of great importance to study the quanti-zation effects. As a result, considerable attention has been pain to the problem of quantizedmeasurement control and estimation.In this thesis, we investigate the problems of network data packet dropout and quantizationfor the linear NCSs. Two well-known NCSs models, which are Bernoulli model , are discussed.The logarithmic quantizer is introduced. Our purposes are to develop conditions such thatthe closed-loop systems are stable and satisfy corresponding performance indices, and for thesingular systems are also regular, impulse free. The main work of this dissertation are outlinedas follows:1. The problem of H_∞filtering for singular systems is investigated. The information limi-tations, which include data packet dropout and measurement quantization, are analyzed.Based on the linear matrix inequality (LMI) approach, a reduction-order filter is designedto cope with the aforementioned information limitations such that the filtering error sin-gular system is regular, impulse-free and exponentially stable, and has a prescribed H_∞performance as well. Sufficient conditions that guarantee the existence of the desired fil-ter are also obtained. simulation examples are given to illustrate the effectiveness of theproposed approach.2. estimation problem for linear systems using quantized measurements is considered. Thecommunication channel we consider encounters inunreliable case. we introduce thestochastic variable satisfying Bernoulli random binary distribution to model the lossymeasurements. We design a estimator to cope with the losses and mitigate quantizationeffects simultaneously such that the estimation error system is stochastically stable, Fi-nally, a simulation example is given to illustrate that the proposed approach is effective and feasible.3. We address the problem of a reset state observer for linear systems using quantized mea-surements. According to the characteristic of the logarithmic quantizer, a reset state ob-server is presented based on the standard one to suppress sensor quantization effects. Byusing the Lyapunov approach, the closed-loop system is still stable when the reset tech-nique is introduced. The observer gains of the closed-loop systems are obtained via solv-ing linear matrix inequalities (LMIs). At last, a simulation example is given to illustratethe effectiveness of the proposed results.