【作者】 常玲芳；

【导师】 李惠光；

【作者基本信息】 燕山大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 随着计算机技术与网络通信技术的高速发展,网络化控制系统的研究得到了广泛的关注。它具有信息资源共享,接线减少和易于扩展和维护等优点。同时网络的引入给控制系统带来了网络诱导时延和数据包丢失两大主要问题,因而原有的控制理论不能直接应用于网络化控制系统中。本文对非线性时延网络化控制系统、基于模型的时延网络化控制系统及基于变采样理论的时延网络化控制系统进行了建模、控制器设计及稳定性等方面的研究。本文将模糊控制理论与线性网络化随机控制理论相结合,解决了非线性网络化控制系统的建模、控制器设计及稳定性分析问题。运用T-S模糊理论,将一个整体非线性系统精确地表示为多个局部线性模型的模糊逼近。在设计线性子系统的控制器时,首先在已知网络时延分布概率特性的条件下,求取状态转移矩阵;然后采用动态规划和线性随机最优理论设计出线性子系统的控制器并进行了稳定性分析。仿真结果表明用本文提出的控制方案来设计非线性网络化控制系统控制器是可行而且有效的。结合Delta域离散理论,本文对仿射非线性网络化控制系统进行了研究。首先运用泰勒级数展开法,将仿射非线性被控对象表示为多个局部线性模型的模糊逼近;然后将Delta算子引入到网络化控制系统中,运用动态规划理论和随机最优理论设计出Delta域随机最优网络控制器和Delta域随机最优跟踪器,并进行了稳定性分析。仿真结果表明所设计出的Delta域无限时间随机最优控制器是有效的。为了减少网络对控制系统的影响,本文采用了仅在传感器到控制器之间存在网络的网络化控制系统结构,研究了基于模型的网络化控制系统的确定性控制器的设计和稳定性分析问题。对于基于模型的网络化控制系统的结构,本文进行了两处改进:第一,将观测器放在控制器节点侧,用数字观测器来有代替传统的观测器,大大地节约了成本;第二,在传感器信号到达控制器前增加了信号处理单元,增大了系统所能容忍的最大采样周期值。基于模型的网络化控制系统的设计思想是:先根据被控对象的连续模型及所期望的极点配置值,在不考虑网络影响的情况下,求出连续控制器的增益矩阵;然后用此控制器来控制系统,在考虑网络影响的情况下,求出保证系统稳定所能容忍的采样周期值。本文在基于模型的无时延和时延网络化控制系统新结构模型的基础上,分别提出了保证系统稳定的充要条件和保证Lyapunov稳定的充分条件,给出了保证系统稳定的采样周期的变化区间的求取方法。采用事件-时间驱动方式,本文研究了基于变采样理论的网络化控制系统的控制器设计和稳定性分析问题。首先,研究了连续系统所容许的最大采样周期的选取原则;然后在变采样离散化模型的基础上,提出了保证系统稳定的充要条件,进而给出了求取控制增益的步骤;接着研究了带观测器的变采样离散化模型,提出了系统控制器和观测器的设计方法;最后基于变采样理论对时延网络化控制系统进行了控制器的设计,并在基于Lon技术的网络化实验平台上进行了实验,实验结果表明所设计的控制器能保证网络化控制系统是稳定的。更多还原

【Abstract】 The networked control system is paid more attention to by scholars with the rapid development of computer technology and network communication technology. It has many advantages such as sharing information resources, reducing wire and extending and maintaining easily. On the other hand, the two main problems, the network induced time delay and data packet dropout, are brought into the control system. Therefore, it is impossible that traditional control theories are applied to networked control systems. The problems of modeling and controlling for the nonlinear networked control system, the model-based networked control system and the variable-sampling networked control system are studied in the paper.In the paper, the problems of modeling, controller designing and stability analyzing for the nonlinear networked control system are settled by combining the fuzzy control theory with the linear networked stochastic control theory. A nonlinear system is exactly described as several local linear fuzzy models by using T-S fuzzy theory. When designing controller for the linear subsystem, firstly, the state transition matrix is derived on condition that distributing probability of the network-induced delay is known; Afterwards, the stochastic optimal controller for the linear subsystem is designed and stability is analyzed adopting dynamic programming and linear stochastic optimal theory. The simulation results show that the control scheme for the nonlinear networked control system presented in the paper is feasible and effective.The affine nonlinear network control system is further studied using the Delta domain discrete theory. Firstly, the nonlinear plant is described as several local linear fuzzy models by using the method of Taylor series expansion. Afterwards, Delta operator is introduced into the networked control system, the stochastic optimal controller and the tractor for the linear subsystem are designed in Delta domain, and stability is analyzed by using dynamic programming and linear stochastic optimal theory. The simulation results show that infinite-time stochastic optimal controller in Delta domain designed in the paper is effective.In order to lessen the influence of network on control systems, the structure for the NCSs is network connecting only between the sensor and the controller. The certain controller for the model-based networked control systems is designed and stability is analyzed. Two innovations for the structure of model-based networked control systems are done in the paper. Firstly, the observer is laid on the side of controller node, and digital observer instead of traditional observer can save the cost effectively. Secondly, the signal processing unit is added before the signal from the sensor reaches the controller, thus the tolerance value of the maximal sample period of the system is greatly increased. The model-based networked control system design is designed as follows. Firstly, the gain matrix of continuous controller is obtained according to continuous model of the plant and the expected pole values without considering the influence of the network. Then, by using the controller to control the system, considering the influence of network, the tolerated maximal sampling period is obtained. The necessary and sufficient conditions for ensuring the system’s stability and sufficient conditions for ensuring Lyapunov stability are both put forward for the model-based networked control system based on new structure with time delay and without time delay, and the change zone of the sampling period for ensuring the system’s stability is given.By adopting event-time drive method, the controller design and stability analysis for the networked control system based on variable-sampling theory is explored in the paper. Above of all, the selecting principle for the continuous system’s tolerance maximum sampling period is studyed. According to variable-sampling discrete model, the necessary and sufficient condition for ensuring the system’s stability is putted forward, and the steps for control gain is given. Then, the variable-sampling discrete model with observer is studied, and design method for the controller and the observer are given. Finally, the controller for the time delay networked control system is designed based on variable-sampling theory, and the experiment is conducted on the networked platform based on the Lon technology. The results of the experiment prove that the designed controller can ensure the networked control system’s stability.更多还原