确定性网络控制系统的多步预测控制器设计  

The Multi-step Predicting Controllers for Deterministic Networked Control Systems

作　　者：朱其新[1] 刘红俐[2] 胡寿松[3] 

机构地区：[1]华东交通大学电气与电子工程学院,江西南昌330013 [2]华东交通大学基础科学学院,江西南昌330013 [3]南京航空航天大学自动化学院,江苏南京210016

出　　处：《兵工学报》2009年第8期1124-1128,共5页Acta Armamentarii

基　　金：国家"863"计划项目(2008AA04Z129);国家自然科学基金资助项目(60474076;60864004;60804066);江西省自然科学基金自筹项目(2007GZS0899)

摘　　要：由于网络诱地时延的随机性,所以网络控制系统是一随机系统。对这样的随机系统,在网络诱导时延服从某一确定分布且其统计特性已知的情况下可以设计其随机控制器,否则只能设计其确定性控制器。在一定的假设条件下,通过在网络控制系统的源节点和目标节点分别设置一定长度的缓冲器的方法将网络控制系统中的随机时延转化为确定性时延,从而将原随机的网络控制系统转化为确定性的网络控制系统。基于线性时不变的被控对象,设计了确定性网络控制系统的多步预测控制器,所得到的闭环系统是可镇定的,并且其控制器和观测器满足分离原理。以不稳定的倒立摆为对象进行了仿真研究,仿真结果表明本文方法的有效性。As network-induced delay is a stochastic variable, networked control systems are stocnasuc systems. For this kind of stochastic system, one can design its stochastic controllers when network-induced delay has a known probability distribution function, otherwise one can only design its deterministic controllers. On some assumptions, the method of placing a specific amount of buffer at source node and object node of the systems was used to transform stochastic time delay to deterministic time delay. The systems with stochastic delay were transformed to a deterministic one. Based on linear time-invariation controlled object, multi-step predicting controllers of the systems were designed, which makes the closed loop system stabilizable and satisfies the separation principle with the observer. The simulation based on the unstable inverted pendulum was performed. The simulated results show that the proposed method is effective

关 键 词：控制理论 网络控制系统 时延 预测器 控制器 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]