基于广义最小二乘法的系统模型辨识及应用

Systematic Model Identification and Application Based on Generalized Least Square Method

为了给控制系统设计提供一个有力的依据,提出了一种基于广义最小二乘法的模型参数辨识方法。首先,由CV I结合PCL-1800高速数据采集卡对被辨识对象输出进行数据采集,获得大量实用的观测数据,在完成数据截取、滤波、归一化等预处理后,对预处理后的数据按照广义最小二乘法进行模型辨识——先引入一个白化滤波器,把相关噪声转换为白噪声,基于对观测数据先进行一次滤波处理,然后利用普通最小二乘法对滤波后的数据进行辨识。为了证明该方法的有效性,采用该文方法对某系统的直流力矩电机进行建模研究,实际结果表明该方法辨识精度高,应用性强。

In order to provide the design of the controlling system with a strong basis,this article proposes a model parameter identification method on the basis of generalized least square method.Firstly,combining CVI with PCL-1800,the high-speed DAS card,and using them to adopt data from the output signals of the identified subject and obtain a great number of practical observation data.After the data preprocessing,such as： data interception,filtering and normalization,according to the theory of generalized least square method to carry out the model identification of the preprocessed data--first,introduce a whitening filter to transform the correlated noise into the white noise.Then on the basis of the first filter disposal of the observational data,the least square method can be used to indentify the disposed data.In order to test the effectiveness of the method introduced in this article,it is adopted to conduct modeling research of the direct current torque motor of some system.The actual result indicates that the method is accurate and applicable.