基于奇异值分解的非均匀采样系统最小二乘辨识
2014-04-11李楠张为
李 楠 张 为
(1.内蒙古科技大学包头师范学院物理科学与技术学院,内蒙古 包头 014030;2.内蒙古科技大学包头师范学院信息学院,内蒙古 包头 014030)
基于奇异值分解的非均匀采样系统最小二乘辨识
李楠1张为2
(1.内蒙古科技大学包头师范学院物理科学与技术学院,内蒙古包头014030;2.内蒙古科技大学包头师范学院信息学院,内蒙古包头014030)
摘要:针对非均匀周期多采样率系统,在状态估计为已知的情况下,提出了基于奇异值分解的模型参数的最小二乘辨识方法.首先,根据系统的连续时间状态空间模型,在满足因果关系基础上,推导了含有提升变量的离散状态空间模型.然后,为了克服辨识误差积累和传递,采用基于奇异值分解的递推最小二乘方法确定模型参数.最后,仿真结果表明提出方法的有效性.
关键词:状态空间模型;奇异值分;多采样率系统;非均匀采样
参考文献:
[1]倪博溢,萧德云.多采样率系统的辨识问题综述[J].控制理论与应用,2009,27(1).
[2]L.XIE,Y.J.LIU,H.Z.YANG,F.DING.Modelling and identification for non-uniformly periodically sampled-data systems[J].IET Control Theory and Applications.2010,4(5).
[3]吴瑶,罗雄麟.化工多采样率数字控制技术研究进展[J].化工进展,2008,27(9).
[4]FENG DING,LI QIU,TONG-WEN CHEN.Reconstruction of continuous-time systems from their non-uniformly sampled discrete-time systems[J].Automatica.2009,45(2).
[5]丁锋,陈通文,萧德云.非均匀周期采样多率系统的一种辨识方法[J].电子学报,2004,32(9).
[6]刘艳君,谢莉,丁锋.非均匀采样数据系统的AM-RLS辨识方法及仿真研究[J].系统仿真学报.2009,21(19).
[7]蒋红霞,丁锋.一类非均匀采样数据系统的状态估计[J].科学技术与工程.2008,8(2).
[8]Sheng J,Chen T,Shah S L.Generalized predictive control for non-uniformly sampled systems[J].Journal of Process Control.2002,12(8).
[9]蒋红霞,王金海.非均匀多率系统滤波的研究[A].2007年系统仿真技术及其应用学术会议论文集[C].中国科学技术大学出版社,2007.
[10]Wang W,Ding F,Dai J Y.Maximum likelihood least squares identification for systems with autoregressive moving average noise[J].Applied Mathematical Modelling,2012,36(5).
[11]Sheng J,Chen T,Shah S L.Generalized predictive control for non-uniformly sampled systems[J].Journal of Process Control.2002,12(8).
中图分类号:TP15
文献标识码:A
文章编号:2095-3771(2014)01-0093-07
收稿日期:2014-01-21
作者简介:李楠(1976—),女,汉族,河北省定县人,内蒙古科技大学包头师范学院物理科学与技术学院讲师,硕士。
基金项目:国家自然科学基金“用显卡通用计算方法设计超导/铁磁异质结构的磁通量子器件”(项目编号:11064008)。
The Identification of Least Squares in Non-Uniform Sampling System via Singular Values Decomposition
LI Nan1ZHANG Wei2
(1.Physics Science and Technology Department of Baotou Teachers’College,Inner Mongolia University of Science and Technology,Baotou 014030 Inner Mongolia;(2.Information Department of Baotou Teachers’College,Inner Mongolia University of Science and Technology,Baotou 014030,Inner Mongolia)
Abstract:The least-squares method is proposed via the model parameter of singular value decomposition(SVD)specific to the non-uniformly sampling system under the assumption that the state estimates are known.Firstly,the discrete state-space model with the lifting variables is derived from the continuous state-space model on the basis of realizing the causality constraints.Secondly,to overcome the accumulation and the transmission of the identification errors,the recursive least-squares method based on singular values decomposition is developed to determine the parameter of the identified model.Finally,the simulation results show the effectiveness of the proposed method.
Key words:state-space model;singular value decomposition;multi-rate sampled systems;nonuniform sampling