Several multi-innovation identification methods

被引:217
作者
Ding, Feng [1 ]
机构
[1] Jiangnan Univ, Sch Commun & Control Engn, Wuxi 214122, Peoples R China
基金
中国国家自然科学基金;
关键词
Recursive identification; Parameter estimation; Signal processing; Filtering; Multi-innovation identification; Stochastic gradient; Least squares; Missing data; GRADIENT PARAMETER-ESTIMATION; NONSTATIONARY ARMA PROCESSES; DUAL-RATE SYSTEMS; AUXILIARY MODEL; LEAST-SQUARES; PERFORMANCE ANALYSIS; ESTIMATION ALGORITHMS; OUTPUT ESTIMATION;
D O I
10.1016/j.dsp.2009.10.030
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
This paper considers connections between the cost functions of some parameter identification methods for system modelling, including the well known projection algorithm, stochastic gradient (SG) algorithm and recursive least squares (RLS) algorithm, and presents a modified SG algorithm by introducing the convergence index and a multi-innovation projection algorithm, a multi-innovation SG algorithm and a multi-innovation RLS algorithm by introducing the innovation length, aiming at improving the convergence rate of the SG and RLS algorithms. Furthermore, this paper derives an interval-varying multi-innovation SG and an interval-varying multi-innovation RLS algorithm in order to deal with missing data cases. (C) 2010 Elsevier Inc. All rights reserved.
引用
收藏
页码:1027 / 1039
页数:13
相关论文
共 30 条