文献类型：期刊文献

中文题名：基于数据滤波的子空间辨识算法

英文题名：Subspace Identification Algorithm Based on Data Filtering

作者：闫小东[1,2];祁荣宾[1,2];梅华[1,2]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]华东理工大学化工过程先进控制与优化技术教育部重点实验室,上海200237

年份：2017

卷号：24

期号：1

起止页码：72

中文期刊名：控制工程

外文期刊名：Control Engineering of China

收录：CSTPCD;;北大核心:【北大核心2014】；CSCD:【CSCD_E2017_2018】；

基金：国家自然科学基金项目(61333010;61174118);上海市自然科学基金项目(15ZR1408900);中央高校基本科研业务费;上海市重点学科建设项目(B504)

语种：中文

中文关键词：子空间辨识;均值滤波;中值滤波;有色噪声

外文关键词：Subspace identification; average filtering; median filtering; colored noise

摘要：近年来基于子空间方法的多变量系统辨识得到越来越多的关注。子空间辨识算法当前主要的研究是基于白噪声环境,而实际现场数据多是受有色噪声干扰的。针对有色噪声问题,提出了基于数据滤波的子空间辨识算法。利用经典的均值滤波与中值滤波方法对输入输出数据处理,从而大大减弱有色噪声对数据的影响,提高系统的辨识精度。从方差的角度在理论上证明了两种滤波方法的有效性,而且此方法得到的模型是无偏的,其精度优于常见的子空间辨识算法。最后又通过实例仿真表明了该方法的可行性和有效性。
Multivariable system identification based on the subspace method is getting more and moreattention recently. Currently, the main subspace identification algorithm is based on the white noiseenvironment, however, most data in the actual field is affected by the colored noise. For the colored noiseproblem, this paper puts forward a kind of subspace identification algorithm based on the data filtering.Processing the input and output data with the classic average filtering and median filtering greatly weakensthe influence of the colored noise on data and improves the identification precision of the system. Based onthe variance, this paper theoretically proves that the effectiveness of the two methods of filtering, and we canget unbiased model of the object with this filter method, and its precision is better than the subspaceidentification algorithm usually adopted. Finally, the simulation example shows the feasibility and validity ofthis method.