广义输出误差模型的两阶段最小二乘递推辨识  

Two-stage least squares recursive identification for generalized output error models

作　　者：贾杰[1] 陈晨[1] 曹姣[1] 罗小娜[1] 丁锋[2] 

机构地区：[1]南昌航空大学信息工程学院,江西南昌330063 [2]江南大学物联网工程学院,江苏无锡214122

出　　处：《控制理论与应用》2014年第2期195-200,共6页Control Theory & Applications

基　　金：中国国家自然基金资助项目（61263012,61263040）;中国博士后基金资助项目（2012M510593）;航空科学基金资助项目（20120156001）.

摘　　要：在有色噪声干扰系统中有一类系统,它具有广义输出误差模型（OEARMA）,本文提出一类广义输出误差模型的两阶段递推最小二乘参数估计算法.该算法基本思想是结合辅助模型辨识思想和分解技术,将系统分解成两个子系统,每个子系统包含一个参数向量.借助基于辅助模型和递推最小二乘理论,用辅助模型的输出代替辨识模型信息向量中未知中间变量,用估计残差代替信息向量中不可测噪声项,从而可以运用递推辨识思想来估计系统所有参数.该算法具有较高的计算效率,仿真例子说明提出算法的有效性.A class of the colored noise interference systems is with the generalized output error model （OEARMA）.This paper presents a two-stage recursive least squares algorithm for their parameter identifications.The basic idea is to combine the auxiliary model identification idea and the decomposition technique to decompose a system into two subsystems,each of which contains one parameter vector.When applying the auxiliary model-based recursive extended least squares theory,we employ the auxiliary model output to replace the unknown intermediate variables in the identified model information vector,and use the estimated residuals to replace the immeasurable noise terms in the information vector.This makes it possible to apply the recursive identification idea to estimate all the parameters of the system with a high computational efficiency.The simulation examples validate the effectiveness of the proposed algorithm.

关 键 词：随机系统 最小二乘 两阶段递推 辅助模型 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]