Parametric and nonparametric curve fitting

被引:30
作者
Hsu, Kenneth [1 ]
Novara, Carlo
Vincent, Tyrone
Milanese, Mario
Poolla, Kameshwar
机构
[1] Univ Calif Berkeley, Dept Mech Engn, Berkeley, CA 94720 USA
[2] Politecn Torino, Dipartimento Automat & Informat, I-10129 Turin, Italy
[3] Colorado Sch Mines, Div Engn, Golden, CO 80401 USA
基金
美国国家科学基金会;
关键词
curve fitting; static nonlinearity; persistence of excitation; identifiability; convergence;
D O I
10.1016/j.automatica.2006.05.024
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
We are concerned with convergence issues in the identification of a static nonlinear function. Our investigation focuses on properties of the input signal that ensure convergence of the estimate. Both parametric and nonparametric approaches are considered. In the parametric case, we offer sufficient conditions under which the estimated parameters converge to their true values almost surely. For the nonparametric case, we offer necessary and sufficient conditions under which the estimated function converges almost surely to the true nonlinearity. (c) 2006 Elsevier Ltd. All rights reserved.
引用
收藏
页码:1869 / 1873
页数:5
相关论文
共 20 条