【作者】 黄鹤；

【导师】 王执铨；

【作者基本信息】 南京理工大学 ， 控制科学与工程， 2009， 博士

【摘要】 通过通信网络实现地域上分布的系统部件之间的信息交互,形成闭环反馈的控制系统被称为网络控制系统(Networked Control System, NCS)。网络控制系统具有连线少、占用空间少、实现成本低、安装维护方便等优点,但网络本身的复杂性,给系统的分析和设计带来了困难,也使得网络控制系统在工程应用中对安全性、可靠性的要求更高。在含有时延和数据包丢失的网络环境下,本文以几种不同的网络控制系统为研究对象,研究了它们的故障检测与容错控制问题,以期提高NCS的安全性和可靠性,为其能更好的应用于工程实际做好理论铺垫。本文的主要研究成果如下：(1)针对包含时延且丢包率已知的网络控制系统,提出了一种基于观测器的故障检测方法。利用异步动态系统理论,给出了故障观测器指数稳定的充分条件及其设计方法。针对丢包率未知的NCS,将系统建模为含有四个模态的马尔可夫跳变系统,提出了一种鲁棒故障检测方法。利用马尔可夫跳变系统理论设计了故障检测滤波器,使得残差动态系统均方稳定且具有一定的扰动抑制水平。考虑参数不确定性和执行器连续增益故障的情况,给出了闭环故障系统均方稳定的充分条件和容错控制器的设计方法。(2)针对包含马尔可夫随机时延的NCS,提出了一种鲁棒故障检测方法和一种容错控制器设计方法。通过构造系统的残差发生器,将相应的故障检测问题转化为H∞滤波问题,给出了基于线性矩阵不等式形式的故障检测滤波器存在的充分条件及其设计方法。采用锥补线性化技术,给出了基于动态输出反馈的容错控制器的设计方法。(3)考虑丢包过程满足有限状态马尔可夫链性质,针对多执行器失效故障,提出了一种NCS主动容错控制策略。设计了系统的保性能被动容错控制器：在此基础上,为提高控制性能,分别针对不同的执行器失效模式,给出了丢包网络环境下系统的主动容错控制策略和设计方法。(4)针对包含时延和丢包的一种连续时间NCS,提出了一种保性能容错控制器和一种鲁棒H∞容错控制器的设计方法。通过引入积分不等式和松弛矩阵变量,给出了此种系统基于矩阵不等式表示的保性能容错控制器存在的充分条件及其设计方法；考虑系统具有参数不确定性,通过引入新的Lyapunov-Kraso vskii泛函,得到了保守性较小的故障系统鲁棒稳定性条件,并给出了使故障系统满足给定的扰动抑制性能指标的鲁棒H∞容错控制器的设计方法。(5)针对一种由T-S模糊模型描述的非线性NCS,研究了包含时延和丢包情况下系统的故障检测与容错控制问题。给出了系统故障观测器增益的设计方法,保证了残差信号对故障信号敏感且对干扰具有一定的鲁棒性；针对包含参数不确定性的非线性NCS,提出了一种鲁棒H∞模糊容错控制器设计方法。采用时滞依赖的分析方法,获得了系统鲁棒稳定条件。并引入锥补线性化方法,给出了求解控制器的迭代算法。最后,在总结全文工作的基础上,给出了本文后续需进一步探讨的一些问题。更多还原

【Abstract】 The control systems that control loops are closed through communication network are called networked control systems (NCS). It aims to ensure data transmission and coordinating manipulation among spatially distributed components of the system. NCS have enormous advantages, such as less wire and volume, lower cost, easy installation and maintenance, etc. However, the complexity introduced by communication networks makes the analysis and design of the system complicated, and this kind of system has higher level of demands for safety and reliability. This dissertation focuses on the fault detection and fault-tolerant control problems for several kinds of NCS with network delay and data packet dropout. The model-based fault detection method and the fault-tolerant controller design method for the system are proposed so as to increase safety and reliability of the NCS and to provide a theoretical backbone for its engineering applications. The main contributions can be enumerated as follows:First, the observer-based fault detection method for NCS with networked introduced delay and fixed rate of data packet dropout is proposed. Based on the asynchronous dynamical system theory, the sufficient condition of exponential stable for the fault observer system is presented, and the fault observer design method is also provided. In addition, for the NCS with unknown rate of data packet dropout, the robust fault detection approach is provided. The system is modeled as Markovian jump system with four operation modes. The design method for fault detection filter is proposed to insure the mean-square stability for the residual generation system and attenuate the disturbance to a prescribed level. Furthermore, by considering parametric uncertainties and actuator continuous gain faults, the necessary and sufficient condition of mean-square stable for the closed loop faulty system is given and the fault-tolerant controller design method is proposed.Second, the robust fault detection method and fault-tolerant controller design approach is provided for NCS with Markovian delays. The residual generator is constructed so that the fault detection problem can be formulated as an H∞filtering problem. A sufficient condition to solve this problem is established in terms of the feasibility of certain linear matrix inequalities (LMI), and the fault detection filter gain design procedure is proposed. Based on the same NCS model, the dynamic output feedback fault-tolerant controller is designed by using cone complementary technology.Third, for multi-actuator failures, the fault-tolerant control strategy for NCS with Markovian packet dropout is given with stability guaranteed. The guaranteed cost passive fault-tolerant controller for the system is designed. Then, with consideration of control performance and different actuator failure model, the design procedure of the active fault-tolerant controller for NCS with packet dropout is proposed.Fourth, the design approach of guarantee cost fault-tolerant controller and robust H∞fault-tolerant controller is proposed for continuous-time NCS with time-delay and packet dropout. By introducing integral-inequality and slack variables methods, the sufficient condition for the existence of the guaranteed cost controller in terms of matrix inequalities is obtained and the guaranteed cost robust fault-tolerant controller is designed. Furthermore, by introducing new Lyapunov functional, the less conservative robust stability conditions for system with parametric uncertainties are obtained, and the robust H∞fault-tolerant controller that achieves the prescribed disturbance attenuation level is also designed.Fifth, based on a nonlinear NCS approximated by T-S fuzzy models, the fault detection and fault-tolerant control problems for a class of nonlinear NCS with time-delay and packet dropout are studied. The fault detection observer is designed such that the residual signal is sensitive to the fault but robust to exogenous disturbance for a certain level. Then, the fuzzy robust H∞fault-tolerant controller is designed for the nonlinear NCS with parametric uncertainties. The robust stability conditions are obtained based on delay-dependent approach. An algorithm is given to get a feasible solution to the fuzzy fault-tolerant controller gain by using a cone complementary technology.Finally, some open issues and the future work in fault detection and fault-tolerant control of NCS are discussed.更多还原