【作者】 樊卫华；

【导师】 胡维礼；

【作者基本信息】 南京理工大学 ， 控制理论与控制工程， 2004， 博士

【摘要】 本文研究了网络控制系统的建模与控制，主要内容如下： (1) 研究了短时延网络控制系统的模型描述，证明了传感器节点为时间驱动，控制器节点和执行器节点为事件驱动的不确定时延网络控制系统可等效为一类具有参数不确定性的线性离散系统；利用Lyapunov函数和线性矩阵不等式(LMIS)方法，给出了此类网络控制系统渐近稳定的充分条件以及控制器的设计方法，研究了此类网络控制系统的H_∞控制问题。 (2) 研究了多包传输网络控制系统(MNCS)的建模与控制。针对控制器节点和执行器节点为事件驱动，传感器节点为时间驱动且采用多包传输的网络控制系统，将传感器节点采用静态调度策略的MNCS建模为周期系统和不确定周期系统，将传感器节点采用动态调度策略的MNCS建模为切换系统；利用周期Lyapunov函数和LMIs方法，给出了周期网络控制系统和不确定周期网络控制系统的渐近稳定的充分条件以及控制器的设计方法，研究了周期和不确定周期网络控制系统的H_∞控制问题；利用共同Lyapunov函数和LMIs方法，给出了切换网络控制系统渐近稳定且满足给定H_∞性能指标的充分条件以及控制器的设计方法。 (3) 研究了短时延MIMO网络控制系统的控制问题。首先给出了传感器节点和控制器节点均为时间驱动的短时延MIMO网络控制系统的模型描述，将一个周期内未能成功传输数据的传感器看成暂时失效，针对网络诱导时延的影响，提出了利用状态观测器的预测功能削弱时延的影响，利用容错控制以及切换系统的理论，给出了MIMO网络控制系统渐近稳定的充分条件以及控制器和观测器的协同设计方法。 (4) 研究了长时延网络控制系统的建模问题。针对在网络未介入时，渐近稳定的闭环系统，利用Lyapunov函数和LMIs方法，给出了网络介入后，闭环系统渐近稳定的充分条件以及使得系统稳定的最大时延的求取方法；针对只具有控制时延的网络控制系统，假设传感器和执行器节点的时钟完全同步，利用接收缓存技术将时变控制时延转化为固定时延，给出了闭环系统渐近稳定的充分条件以及最优控制器的设计方法；在此基础上，针对传感器和执行器节点之间存在的时间差给系统引入的分数时延，利用第2章的有关结论，将此类网络控制系统建模为一类具有参数不确定性的系统，给出了闭环系统渐近稳定的充分条件以及最优控制器的设计方法；针对只具有时变输出时延的网络控制系统，本文引入具有时延补偿功能的观测器，利用切换系统的稳定性理论和LMIs方法，给出了观测器的设计方法；最后针对同时具有控制时延和输出时延的网络控制系统，利用缓存技术将时变的控制时延转化为固定时延，观测器补偿时变输出时延，基于分离原理，给出了观测器和控制器的设计方法。 (5)将同时存在时延和数据包丢失的网络控制系统建模为异步动态系统，针对无数据包丢失时稳定的网络控制系统，给出了数据包丢失率已知时，网络控制系统指数稳定的充分条件;研究了此类网络控制系统的控制器的设计方法;并针对数据包丢失率不确知的情形，给出了网络控制系统指数稳定的充分条件。关键词:网络控制系统，网络诱导时延，多包传输，数据包丢失， Lyapunov函数，线性矩阵不等式，切换系统更多还原

【Abstract】 The Modeling and Control of the Networked Control Systems (NCSs) are studied in this dissertation. The main parts are concluded as follow:(1) The models of the NCSs with delay shorter than one sample period are given. The NCS with time-driven sensor node, event-driven controller node and actuator node and uncertain delay is proved to be modeled as a class of uncertain discrete linear system. Based on the Lyapunov function and the Linear Matrix inequalities (LMIs), the sufficient condition for the NCS which is subject to the asymptotic stability and H∞ performance are derived. The design approach of the controller is given.(2) Supposing the controller node and the actuator node is event driven, the sensor node is time driven and uses multi-packets transfer; the model of the Multi-packets -transfer NCS (MNCS) is given. On condition that the sensor node using static scheduling strategy; the MNCS is modeled as a periodic system or an uncertain periodic system. The MNCS is modeled as a switched system when the sensor node using dynamical scheduling strategy. Based on the periodic Lyapunov function and LMIs, the sufficient condition for the periodic and uncertain periodic NCS which is subject to the asymptotic stability and H ∞ performance are derived, and the design approach of the controller is given. Based on the common Lyapunov function and LMIs, the sufficient condition for the switched NCS which is subject to the asymptotic stability and H∞ performance are derived, and the design approach of the controller is given.(3) The control problem of the MIMO NCSs with short delay is studied. At first, the model of the short-delay MIMO NCSs with time-driven sensor nodes and controller node is given. These sensors failed transferring the data packet in the sampling period are treated as invalidated temporarily, so the NCS is a system with sensors failure. In order to eliminate the effect of the network-induced-delay, the state observer is introduced. Based on the theory of the fault-tolerant control and the switched system, the sufficient condition for the MIMO NCSs which is subject to the asymptotic stability and H∞ performance are derived, and the cooperative design approach of the controller is given.(4) Firstly, the modeling of the long-delay NCSs is studied. Supposing the close loop system without network is asymptotic stable, the sufficient condition for the stability of the close loop system including network is derived based on the Lyapunov function andLMIs. And the approach to calculate the maximum time delay which guarantees the stability of the NCSs is given. For the NCSs with only control delay, supposing the sensor and the actuator node are synchronous, the storage is setup to fix the time-vary delay. The sufficient condition for the NCSs which is subject to the asymptotic stability is derived and the optimal controller is designed. Then supposing there are fixed time difference between the sensor and the actuator node, which takes the fractional time delay to the NCSs, the NCSs is modeled as a class of uncertain system with the conclusion in the chapter 2, and the sufficient condition for this NCSs which is subject to the asymptotic stability is derived and the optimal controller is designed. For the NCSs with only time varying output delay, an observer with delay compensation is introduced, and the design approach is studied using the theory of the switched system and LMIs. Finally, for the NCSs with both control delay and output delay, the storage is used to make the control delay fixed, and the observer is used to compensate the output delay, the design approach of the controller and the observer is derived based on the separate principle.(5) The NCSs with both data packet dropout and the delay are modeled as an Asynchronous Dynamical Systems (ADSs). Supposing the NCS without data packet dropout is stable, the sufficient condition for the exponential stability of the NCS with fixed data packet rate is derived. The design approach of the controller for the NCS with fixed data packet rate is studied. And the更多还原