【作者】 朱其新；

【导师】 胡寿松；

【作者基本信息】 南京航空航天大学 ， 控制理论与控制工程， 2003， 博士

【摘要】 本文基于线性时不变的被控对象，研究了网络控制系统的建模、分析与控制问题。 首先，研究了网络控制系统(networked control system，简记为NCS)的建模问题。建立了短时延网络控制系统在单包传输、多包传输以及有数据包丢失时的数学模型；建立了长时延网络控制系统在四种驱动方式下的开环数学模型，提出了长时延网络控制系统的统一建模方法；建立了多输入、多输出NCS的数学模型。 其次，研究了网络控制系统的性质。提出了网络控制系统的均值能控、均方能控、均值能观、均方能观、均值可镇定、均方可镇定、均值可检测和均方可检测等新概念；得到了网络控制系统具有相应性质的充分或必要条件。 再其次，研究了网络控制系统的确定性控制问题。提出了将网络控制系统中的随机时延转化为确定性时延的实现条件和实现方法；设计了确定性网络控制系统的分步变换状态反馈控制器和基于状态预测的状态反馈控制器；设计了确定性网络控制系统的输出反馈控制器和具有时延补偿功能的多步预测控制器；仿真结果表明了各种确定性控制算法是正确的和有效的。 然后，研究了网络控制系统的随机控制问题。设计了短时延网络控制系统的随机状态反馈控制器和随机输出反馈控制器，若不考虑系统噪声，则所设计的状态反馈控制律可以使短时延网络控制系统均方指数稳定；设计了长时延网络控制系统的随机状态反馈控制器和随机输出反馈控制器，证明了分离原理在长时延网络控制系统中依然成立，若不考虑系统噪声，则所设计的状态反馈控制律可以使长时延网络控制系统均方指数稳定；设计了短时延网络控制系统的随机分步变换控制器，该控制律可以使网络控制系统均方指数稳定；研究了网络控制系统的等价离散性能指标问题；仿真结果表明了各种随机控制律是正确的和有效的。 随后，研究了长时延Markov网络控制系统的随机控制问题。设计了长时延Markov网络控制系统的随机状态反馈控制器和输出反馈控制器，证明了分离原理在长时延Markov网络控制系统中依然成立。 最后，研究了网络控制系统的稳定性问题。分析了长时延网络控制系统的二阶矩稳定性和随机稳定性；针对网络传输中的数据包的时序错乱问题，提出了第二缓冲器的方法；分析了网络诱导时延的Markov特性，并给出了时延Markov链的状态转移矩阵中元素的求取方法；建立了存在数据包时序错乱时长时延NCS的数学模型，并给出了对应的长时延NCS随机稳定的充分必要条件。更多还原

【Abstract】 The modeling, analysis and control of networked control systems are studied in this dissertation when the controlled process is linear time-invariant system.Firstly, the modeling of networked control systems is presented. The mathematic models of networked control systems with single packet transmission, multiple packet transmission or data packet dropout are developed when the network-induced delay is short time delay. The open-loop mathematic models of networked control systems with long time delay are developed when the system is driven by anyone of the four different modes. The uniformed modeling method for networked control systems with long time delay is proposed. The mathematic model of networked control systems with multiple-input and multiple-output is developed.Secondly, some new concepts including mean value controllability, mean square controllability, mean value observability, mean square observability, mean value stabilization, mean square stabilizability, mean value detectability and mean square detectability of networked control systems are defined in this dissertation. Some sufficient or necessary conditions for networked control systems with corresponding properties are presented.Thirdly, the deterministic control of networked control systems is studied. The realizing methods and conditions of transforming the random time delay to a deterministic time delay are proposed. The step-by-step transform state feedback controller and the controller basing on state prediction are designed. The output feedback controller and multiple-step predicting controller for compensating time delay are designed. The simulation results show the proposed deterministic controllers are right and effective.Fourthly, the stochastic control of networked control systems is studied. The stochastic state feedback controller and the stochastic output feedback controller for networked control systems are designed. Without considering the noise of the system, this stochastic state feedback controller can render networked control systems Exponentially Mean Square Stable. The stochastic state feedback controller and the stochastic output feedback controller for networked control systems with long time delay are designed. The separation principle is proved to still hold in networked control systems with long time delay. Without considering the noise of the system, this stochastic state feedback controller can render corresponding networked control systems Exponentially Mean Square Stable. The step-by-step transform controller for networked control systems with short time delay is designed. This controller can render corresponding networked control systemsExponentially Mean Square Stable. Equivalent discrete optimal performance index problem for networked control systems is studied. The simulation results show the proposed stochastic controllers are right and effective.Fifthly, the stochastic control of Markov networked control systems with long time delay is studied. The state feedback controller and the output feedback controller of Markov networked control systems with long time delay are designed. The separation principle is proved to still hold in such networked control systems.Finally, the stability of networked control systems is analyzed. The second moment stability and stochastically stability of networked control systems are analyzed. The second buffer method is put forward to deal with the wrong order of data packets. The Markov characteristic of network-induced delay in networked control systems is analyzed. The evaluation method of element in state transition matrix is given when the wrong order of data packet is considered. Considering the wrong order of data packets, the mathematic model of networked control systems with long time delay is developed. The sufficient and necessary conditions for stochastic stability of such networked control systems with long time delay are given.更多还原