【作者】 王艳；

【导师】 胡维礼； 陈庆伟；

【作者基本信息】 南京理工大学 ， 控制理论与控制工程， 2006， 博士

【摘要】 本文研究了网络控制系统(networked control systems，NCS)的控制与调度问题，主要内容如下：(1)研究了线性网络控制系统的补偿与控制问题。首先针对存在数据包丢失和不确定短时延的网络控制系统，构造丢包补偿器补偿时延的影响，将NCS建模为一类具有参数不确定性的离散切换系统，利用切换系统控制理论和线性矩阵不等式(linear matrix inequality，LMI)方法，给出了丢包补偿器和控制器设计的参数化方法。然后针对存在数据包丢失和时变长时延的网络控制系统，设计基于开环观测和闭环观测相结合的状态观测器，补偿数据包丢失和长时延对系统性能的影响，将观测误差系统建模为一类离散切换系统，分别给出了状态观测器和区域极点配置控制器的设计方法。(2)研究了网络控制系统中的动态调度策略设计问题。提出了一种基于网络运行状态的动态调度器，通过实时调节控制系统的采样周期，调度网络中的控制数据，以适应网络中信息流的变化。动态调度器由网络状态监测器、网络状态预测算法、采样周期调节算法组成。监测器等间隔采集当前的网络状态，预测算法利用获取的网络状态，估计下一监测周期内的网络利用率和数据包执行时间。针对设定网络利用率期望值的NCS，基于反馈控制机理，预测下一监测周期内的网络利用率；对于未设定网络利用率期望值的NCS，建立网络利用率AR模型，采用最小方差预报估计下一监测周期内的网络利用率。同样建立数据包执行时间的AR模型，求得下一监测周期内数据包执行时间的最小方差预报值。采样周期调节算法基于网络利用率的预估值和数据包执行时间的预估值及当前的网络诱导误差与控制误差，计算控制系统新的采样周期。利用Truetime仿真软件对提出的动态调度器进行仿真研究，研究结果表明，所设计的动态调度器能够获得较高的网络资源可调度性和利用率。(3)研究了网络控制系统中控制与调度协同设计问题，提出了一种变采样周期系统的建模与控制方法。采用动态调度器的NCS，其采样周期是时变不确定的，此时NCS为一变采样周期系统。首先针对理想调度，忽略数据包丢失和网络诱导时延对系统的影响，利用数学变换和矩阵理论，将变采样周期系统等效为一类具有参数不确定性的离散时间系统，实现了变采样周期系统的建模。基于Lyapunov方法和LMI方法，给出了变采样周期NCS状态反馈控制器设计的参数化方法。进而，针对非理想调度，考虑不确定短时延和数据包丢失对系统的影响，将变采样周期系统等效为一类具有参数不确定性的离散切换系统，基于切换系统理论和LMI方法给出了非理想调度下变采样周期NCS的控制器设计方法，从而实现网络控制系统中的控制与调度的协同设计。(4)研究了一类非线性网络控制系统的建模与控制问题。针对被控对象可由Takagi-Sugeno(T-S)模糊模型描述的非线性网络控制系统，当网络中存在不确定短时延，系统可建模为一类具有参数不确定性的T-S模糊系统，利用模糊控制理论与方法，给出了系统渐近稳定的充分条件和模糊控制器设计的参数化方法，在此基础上，进一步研究了系统的H_∞控制问题，给出了使系统渐近稳定且满足扰动抑制性能的控制器设计方法。更多还原

【Abstract】 The methods of control and scheduling for networked control systems (NCSs) arestudied in this dissertation. The main research works are concluded as follows:(1) The compensation and control methods for linear networked control systems arestudied. Firstly, the NCS existing uncertain short delay and data packets dropout is concerned.A dropout compensator is designed to predict state when data packet is dropout. The NCS canbe equivalent to a discrete switched system with parametrical uncertainties. Then based on themodel, the design method of state feedback controller and compensator is given by using theapproaches of Lyapunov and Linear Matrix Inequality (LMI). Secondly, considering the NCSwith time-vary long delay and data packet dropout, an observer composed by open-loopprediction and close-loop prediction is proposed to compensate the effect of delay and datapacket dropout. Based on separation principle, the methods of controller design and obserberdesign are given respectively.(2) A dynamic scheduler for NCS is studied. By on-line adjusting sample periods ofthe control systems sharing network resource, the network bandwidth are allocate to eachcontrol system dynamically so as to adapt to the variation of network load. The dynamicscheduler is composed of network state monitor, prediction algorithm and sample periodadjusting rule. The monitor samples the network state with equivalent interval, then thepredict algorithm predicts the network utilization and data packets execution times by usingthe current network state acquired by the monitor. For different demands of system utilization,the prediction algorithms of network utilization are different. For those NCSs given utilizationsetting value, the prediction algorithm is designed based on feedback control mechanism. Forthe NCSs without utilization setting value, a AR model is established to predict the networkutilization. Both NCSs are using AR model to predict data packets execution times. Based onthe predictive values of network utilization and data packets execution times, sample periodsadjusting rule allocates network resource and computes the new sample periods according tothe network-induced error and control error. The effectiveness and priorities of dynamicscheduler we proposed are validated by a group of simulation.(3) The methods of control and scheduling co-design for NCSs are discussed. TheNCS scheduled by the above dynamic scheduler is a vary-period system. Firstly, consideringthe ideal scheduling situation under which delay and data packet dropout can be ignored, theNCS is equivalent to a class of discrete time system with parametrical uncertainties bymathematical transformation and matrix theory. Then a design method of state feedback controller is studied by using the approaches of Lyapunov and LMI. Furthermore, the NCSunder non-ideal scheduling is considered. It is assumed that uncertain short delay and datapacket dropout occurred in the network, then the NCS is modeled as a discrete time switchedsystem with parametrical uncertainties. With the theory of switched systems control and LMIapproach, a state feedback controller is designed to stabilize the NCS. So, the control andscheduling can be co-designed based on the methods of modeling and control for vary-periodsystems we proposed.(4) The modeling and control approaches for nonlinear NCSs are studied. Thecontrol methods for the NCSs which plant can be described by T-S fuzzy models are studied.Considering the effect of uncertain short delay, the NCS is modeled as a discrete time T-Sfuzzy model. With the theory of fuzzy control and LMI, a method of fuzzy state feedbackcontroller is given. Then the H_∞controller method is also studied. A fuzzy controller isdesigned to guarantee the system stability as well as satisfying the H_∞performance index.更多还原