基于NetCon的倒立摆系统的网络控制

【作者】 裴灵犀；

【导师】 彭晨；

【作者基本信息】 南京师范大学 ， 控制理论与控制工程， 2011， 硕士

【摘要】 随着控制科学和计算机网络的日益发展,各种基于网络的控制研究越来越受到人们的关注。与传统的点对点式的控制系统比较,NCS (Networked Control System)具有连线少、易于系统扩展和维护、灵活性高等优点；总而言之,NCS使某个区域内不同地点的用户实现资源共享和协调操作。结合网络控制技术的发展,倒立摆是理想的基于网络的控制教学实验设备。然而由于网络通信带宽的存在,数据传输不可避免的存在延时、丢包和多包传输等问题,导致控制性能下降甚至使系统不稳定,所以针对基于网络的控制系统,必须采取有效的控制策略。本文主要针对网络控制系统中的时延补偿控制器的设计进行了研究,并进行了实时仿真和在线实验。首先回顾了从最初的计算机控制系统的引进到目前网络控制系统的发展过程,同时对网络控制系统中基本概念以及国内外基于网络控制的倒立摆系统的研究现状进行了综述。然后分析了产生网络时延的原因和类型；接着利用Matlab/TrueTime工具箱以一个典型的网络控制系统为例进行对比研究,分析了在不同的控制网络模型下的传输特性以及在同一种控制网络模型下的不同时延的输出响应特性；仿真结果表明了时延会大大降低控制系统的性能,甚至使之不稳定。接着搭建了基于NetCon系统的倒立摆网络控制平台。通过系统辨识方法得到倒立摆模型,通过NetCon控制器实现本地离线仿真控制；进而建立基于倒立摆实物的本地实时控制系统；并以此为基础,利用两台NetCon网络控制器建立了一个实际网络控制系统,由于NetCon系统对采样时间的精确度不理想,只能控制慢变过程,所以仅对倒立摆小车的位移进行控制；通过离线仿真和在线实验,表明必须要采取有效措施才能保证系统的稳定性。最后利用基于模型参数的广义预测控制(GPC)方法,提出一种网络预测控制策略,实现了在网络环境下对倒立摆小车位移的控制；通过实时仿真和在线试验,验证了控制策略的有效性。相对于没有采用预测的控制方法,GPC减弱了网络时延对倒立摆小车位移的影响,优化了控制效果。更多还原

【Abstract】 With increasingly development of the control science and computer network system, a lot of researches for control system which is based on network have been more and more concerned about by people. Compared with the traditional point-to-point control system, NCS has some enormous advantages such as less line connecting, easy to system extension and maintenance, higher reliability and flexibility and so on. In a word people at different locations can share the information resources and operate with others coordinately. Since the development of the network control technology, the pendulum is a proper experiment installation based on the network. It is used for many experiments referred by lots of references on network research. Since the constraints of the communication bandwidth, some new issues such as random delays, packet dropout, multiple-packet transmission and so on, can degrade the performance of control system and even make it unstable. So some strategies should be used for the control system based on network.Considering the network delays, this paper discusses and designs a method concerning network predictive control and does the simulation and experiment. First, the process from introduction of computer-control system to the development of networked control system nowadays is reviewed, and some basic contents and recent research results of NCS based on the pendulum at home and aboard are summarized. Then the causes and the kinds of network delays are analyzed. Then a typical example of Matlab/TrueTime is used to analysis the output characteristic of some different kinds of network models, and the output characteristic of different delays based on the same network model. The simulation result makes clear that the network delays degrade the performance of control system and even make it unstable.Subsequently, the network control platform based on NetCon control system which is designed by Chinese Academy of Science is established. The pendulum model computed by identification method is used to complete the local off-line simulation system. Then the real-time local control system of pendulum is established. Basing on above foundation the real-time remote control system of pendulum is completed by two NetCon controllers. Since the sample time of NetCon system is not satisfied, it only can be used to deal with the process of slow alteration. It is changed to control the location of the car of the pendulum. The off-line simulation and on-line experiment are conducted. The results indicate that some valid control strategy must be adopted to compensate the network delays, in order to ensure the system performance and stability.Finally a method of network delay predictive control strategy is proposed, by the theory of Controller Prediction Generator (GPC) based on model parameters, to control the location of car in the network. The result of real-time simulation and remote real-time experiment prove the effectiveness of the control strategy. Comparing with the control which doesn’t use prediction method, it reduces the bad effect of the delay to the location of the car, so that the control process is optimized.更多还原