【作者】 王天堃；

【导师】 徐大平； 周黎辉；

【作者基本信息】 华北电力大学（北京） ， 热能工程， 2009， 博士

【摘要】 网络控制系统(Networked Control Systems,简称NCS)是近十几年随着控制技术、网络技术和计算机应用技术的飞速发展而出现的一种新的控制系统理论,现已成为自动化领域的一个前沿课题。本文在近年国内、外关于网络控制系统理论研究成果的基础上,研究了网络控制系统的控制算法、调度策略、仿真工具与工程应用方面的部分问题,主要工作如下:1.详细分析了影响网络控制系统建模的主要因素,如网络传输、节点驱动和时钟同步等。根据网络节点的不同驱动组合方式给出了网络控制系统中被控对象的离散模型;提出了一种优化建模方法,考虑了系统输出的变化趋势,提高了模型精度。2.研究了广义预测控制在网络控制系统中的应用,建立了网络控制系统分布式模型,采用BP神经网络误差预测值补偿控制器预测输出与被控对象实际输出之间的误差,修正了控制器输出的控制增量值,提高了控制精度;利用隐式广义预测控制算法直接辨识最优控制器参数,减少了在线计算工作量,提高了效率。3.在特定条件下,建立了具有迟延特性的状态空间预测模型,研究了预测函数控制算法在网络控制系统中的应用,算法对克服随机长时间迟延的影响效果更好。将具有不确定性传输迟延和数据丢包的网络控制系统视为灰色系统,采用GM(1,2)灰色预测模型,改进了控制算法,使计算参数减少到两个,简化了计算步骤。4.设计了一种基于综合调度指标的混合调度方法,给出了基于RM静态调度和MEF-TOD动态调度的混合调度策略,同时发挥了两种调度算法的优越性;设计了一种基于BP神经网络和模糊反馈的两级调度策略,一级调度通过系统的结构设计减小了网络负荷;二级调度设计了模糊反馈信息调度器,根据不同的参数分配各个节点的发送信息优先级。两级调度策略在减小网络流量的基础上优化了信息优先级配置,提高了系统的控制品质。5.研究了TrueTime工具箱的特性及其使用方法;设计了基于以太网的网络控制系统仿真平台,给出了仿真平台的设计结构、工作原理、程序流程和时钟同步、迟延测试等关键技术的实现方法,对基于以太网的网络控制系统理论研究与分析有一定帮助。6.介绍了笔者参与研发的一套用于大型火力发电厂生产过程控制的现场总线控制系统的基本功能和整体结构,着重描述系统研发中的两项关键技术。首先介绍了事件-时间驱动方式工程实现的技术细节;然后介绍了采用混合任务集动态调度算法设计的高效率OPC服务器软件。该项技术实现了实时数据的双向交换,使不同网络控制系统之间的连接更加方便快捷,且开放性强,具有广泛的实际工程意义。更多还原

【Abstract】 Networked Control System is the novel frame of control system theories born with control,network and computer application technologies.The paper studied on control algorithms,scheduling strategies,simulation tools and engineering applications based on the existing theoretical achievements.The main contents are:1.Discussed in detail the main factors influencing networked control,such as transmission efficiency,network node drives,clock synchronization,and sampling period.Besides,according to the different combinations of controller,sensor and actuator drive methods,the discrete model of networked control system was deduced.2.Studied the application of GPC in NCS and built up the NCS distributional model.BP neural network was adopted to predict the errors between the controller and object outputs for their compensation and thus achieving the enhanced precision.Then the Implicit Generalized Predictive Control was adopted to identify the optimal controller parameters,therefore decreasing the online calculation and improving the efficiency.3.The State Space model was built up under certain specifications for the study of Predictive Functional Control applied in NCS.The network with uncertain transmission delay and data drop-off can be regarded as grey system.The paper adopted the GM(1,2) grey predictive control model to do research on the application of grey control theory in NCS.4.A mixed scheduling method was designed based on SSI in which the strategy combining RM static scheduling and MEF-TOD dynamic scheduling algorithm was introduced for QoS scheduling.Besides,a two-layer scheduling strategy was proposed based on BP neural network and fuzzy feedback,one layer being intended for decreasing network load through optimal system structure.Fuzzy feedback information scheduling was devised on the other layer in which the priority of information inputted into controllers was decided according to response error and the error differences while the priority of information entering sensors was made by testing sampling time and sampled value changes.5.Studied the characteristics and usage of TrueTime toolbox and designed the simulation platform of NCS based on Ethernet.The key techniques of structure building, working principles,program flow,clock synchronization,delay test and so forth were introduced and realized so that it would be helpful for the theoretical study of Ethernet-based NCS.6.A set of FCS which the author has developed as a participant was introduced by describing its basic functions and entire structure while emphasizing two key techniques. The event-clock driven way was explained in detail.The OPC server was designed using dynamic scheduling algorithm of complex task collection,attaining real-time data exchanges.This technique made the connection between different network control systems more convenient;besides,its openness would ensure the extensive the uses in practice.更多还原