【作者】 尹姝；

【导师】 俞立；

【作者基本信息】 浙江工业大学 ， 系统工程， 2009， 硕士

【摘要】 网络化控制系统(Networked Control Systems,NCSs)是将控制系统的传感器、执行器和控制器等单元通过网络连接起来的分布式控制系统。NCSs具有连线少、灵活性强、便于故障诊断和信息资源共享等众多优点。当然,在网络化控制系统带来诸多优点的同时,也给控制学科带来了新的挑战。由于网络的介入,控制回路中不可避免地出现网络诱导时延、数据包丢失等问题,并且在很多情况下其影响不可忽略。具有不确定丢包的网络化控制系统可描述为具有有限个子系统的切换系统,其各个子系统可以描述不同丢包情况影响下网络化控制系统的动态特性。因此,本文利用切换系统理论和线性矩阵不等式(Linear Matrix Inequalities,LMI)方法,研究了同时带有时延和丢包的NCSs的建模、稳定性、控制与风滤波等问题,主要工作如下:1)对网络化控制系统的一些基本概念、问题及目前的研究现状做了一个比较详细的综述,并概括介绍了本论文的主要研究工作。2)针对一个同时带有时延和不确定丢包的网络化控制系统(时延仅存在于传感器和控制器之间),将其建模为一类具有定常时延和数据包丢失的离散时间切换系统模型,基于切换系统的平均驻留时间方法给出保证闭环网络控制系统指数稳定的充分条件,并给出丢包补偿器和模式依赖控制器的设计方法,同时建立丢包率与网络化控制系统指数稳定性之间的定量解析关系。3)研究了一类具有丢包的网络化系统的H_∞滤波问题。在考虑数据传输发生丢包的情况下,将网络化滤波系统描述为一个离散时间切换系统。利用非连续分段Lyapunov函数方法给出了一个保证该系统指数稳定和具有给定H_∞性能指标的充分条件,并建立起丢包率和系统衰减率之间的定量解析关系,进而利用切换系统的平均驻留时间方法给出了丢包依赖全维滤波器的设计方法。4)研究了一类同时存在时延和丢包的多输入多输出(MIMO)NCSs的建模和稳定性分析问题。用两个马尔可夫链来描述传感器-控制器通道和控制器-执行器通道的丢包过程,从而将整个闭环NCSs描述为一个具有两个模式的马尔可夫切换系统。基于Lyapunov稳定性理论和变量替换法,给出了保证闭环NCSs随机稳定的充分条件。5)最后,对全文进行总结,并对进一步的研究提出一些展望。更多还原

【Abstract】 Networked Control Systems (NCSs) are distributed control systems in which sensors, actuators, controllers, and other units are connected together. The primary advantages of NCSs are reduced system wiring, ease of system diagnosis and maintenance, increased system agility and sharing information resources. While the networked control system brings many merits, it also gives the control discipline a new challenge. That is to say, as a result of network involvement, network-induced delay, data packet dropout and many other problems appear in the control loops inevitably, and their impact on the system performance can not be ignored in many cases. Networked control system with uncertain packet dropouts can be naturally described as a switched system with limited subsystems, and the subsystems can describe the various dynamics of the networked control system under the influence of different packet dropout status.This thesis is concerned with the modeling, stability analysis, control and H_∞filtering for NCSs with network-induced delay and packet dropout phenomenon by using the switched system theory and the Linear Matrix Inequality (LMI) technique. The main achievements are as follows:1) The detailed summary of some basic contents、issues and recent works on networked control systems is made, then the main results of the thesis are described.2) Networked control system (NCS) with time-delay and uncertain packet-dropout is studied (time-delay only exists between the sensor and controller), and the NCS is modelled as a discrete-time switched system with one step delay and data packet-dropout. Then, sufficient conditions for the exponential stability of the closed-loop NCS are derived by using the average dwell-time method, and the design method is also presented for the packet-dropout estimator and mode-dependent controller. The obtained stability conditions establish the quantitative relation between the packet dropout rate and the stability of the closed-loop NCS. 3) The problem of H_∞filtering for a class of networked systems with packet-dropout is investigated. The concerned networked system with packet-dropout is described as a discrete-time linear switched system. The exponential stability result and an exponential H_∞performance is first explored using a discontinuous piecewise Lyapunov function approach, and the quantitative relation is established between the packet-dropout rate and the stability of the networked systems with a prescribed H_∞filtering performance level. Then a packet-dropout-dependent full-order filter is designed by using the average dewell time method.4) The modeling and stability analysis problem for a class of multiple input and multiple output (MIMO) networked control systems (NCSs) with packet-dropouts and delays are presented. The packet-dropout processes in the channel of sensor-controller and controller-actuator are described as two Markov chains, and then the closed-loop NCS is finally modeled as a Markovian switching system with two switching modes. Based on Lyapunov stability theory and variable substitution method, a sufficient condition is derived for the closed-loop NCS to be stochastically stable.5) Finally, put up the summary to the full text and propose some forecasts to the future research.更多还原