总第 238 期 2009 年第 8 期 计算机与数字工程 Computer & Digital Engineering Vol. 37 No. 8 173 基于 MA TLAB 的广义最小二乘参数辨识与仿真3史贤俊 廖 剑 马长李 张 戎 (海军航空工程学院控制工程系 烟台 264001) 摘 要 针对参数辨识中最小二乘法(LS)存在的缺点 ,讨论了一种用迭代的松弛算法对最小二乘辨识的改进方法 - 广义最小二乘( GLS)辨识 ,并介绍了其基于 Matlab 的仿真和分析方法。首先简述参数辨识的概念、最小二乘法辨识存在的主要缺点和广义最小二乘法的基本原理 ,之后简要介绍了 Matlab 系统辨识工具箱及其中参数辨识的实现方法 ,最后结合实例给出相应的仿真程序及其结果分析 ,仿真结果表明 :该方法辨识精度高 ,明显优于最小二乘辨识。 关键词 参数辨识 广义最小二乘 仿真 中图分类号 TP301. 6 GLS Parameter Identification and Simulation Based on Matlab Shi Xianjun Liao Jian Ma Changli Zhang Rong (Department of Control Engineering , NAAU , Yantai 264001) Abstract In view of the shortcomings of LS in parameter identification , GLS identification is discussed , which is a new method improved on LS identification by using a kind of iterative flab algorithm. The method of simulation and analysis of parameter identification using GLS based on Matlab is also presented. Firstly , the paper introduces the concept of parameter identification , the primary shortcomings of LS identification and the basic principle of GLS. Secondly , it introduces the Matlab system identification toolbox and the realization method of parameter identification in it. Finally , according to a given example , the emulator and results are analyzed. The result indicates that the method is accurate and su2perior to LS identification apparently. Key w ords parameter identification , GLS , simulation Class Number TP301. 6 1 引言 在系统辨识领域中 ,最小二乘法是一种得到最广泛应用的估计方法 ,可用于动态、静态、线性、非线性系统。这种方法