多变量时滞系统的闭环渐进辨识研究与应用  

Closed-loop Identification Research and Application for Multivariable Processes with Time Delays

作　　者：靳其兵[1] 蒋文春[1] 龙萍[2] 

机构地区：[1]北京化工大学信息科学与技术学院,北京100029 [2]中国石油西南油气田分公司川中油气矿,四川遂宁629000

出　　处：《化工自动化及仪表》2009年第5期74-78,共5页Control and Instruments in Chemical Industry

基　　金：国家"863"计划项目(2008AA042131);国家"973"计划项目(2007CB714300)

摘　　要：多变量时滞系统由于滞后以及相互之间的耦合而难以得到理想的辨识效果,因此基于渐进理论,提出一种多变量的闭环辨识方法。采用阶跃作为测试信号,利用两阶段闭环辨识算法进行辨识。首先采用递推ARX算法实现高阶ARX模型对真实模型的逼近,消除耦合,然后利用二阶泰勒近似实现对滞后的辨识,最后采用渐进准则(ASYC)对模型进行降阶,利用输出误差(OE)准则对降阶后的模型进行参数估计。该方法在实际项目的应用中得到了较好的效果,应用前景广阔。It is difficult to achieve a good identification result from multivariable processes with time delays because of the time delay and the coupling between each other.An identification method for multivariable processes in closed-loop situation from step response was proposed.This method used two-stage closed-loop method to identify multivariable system based on the step-response data.Firstly,an iterative method was taken to obtain an high-order ARX model which fits the real model most,through this step the coupling could be eliminated.Then through the method of second-order Taylor approximation for time delays,an estimation of delay was gained.Finally,each sub model was reduced by the output error(OE) structure and asymptotic criterion(ASYC).This method displays an approving effect in practical project and has an extensive foreground.

关 键 词：多变量时滞系统 渐进准则 两阶段法 闭环辨识 阶跃响应 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]