A study of second- and third-order spectral procedures and maximum likelihood in the identification of a bilinear system

Author

Brillinger, David R.

Author_Institution

Dept. of Stat., California Univ., Berkeley, CA, USA

Volume

38

Issue

7

fYear

1990

fDate

7/1/1990 12:00:00 AM

Firstpage

1238

Lastpage

1245

Abstract

An investigation is presented of two spectral moment-based identification procedures which are compared to each other and to the method of maximum likelihood, for the case of a bilinear system having observed input and corresponding output. Inputs to the system that are considered are pulse, pulse pair, and stochastic. A principle concern of the analysis is the examination of coupled frequencies. Relative advantages and disadvantages of the three identification procedures are mentioned. Advantages of the maximum-likelihood approach include the availability of expressions for standard errors and efficiency in an asymptotic sense

Keywords

parameter estimation; spectral analysis; bilinear system; coupled frequencies; efficiency; maximum likelihood; parameter estimation; pulse input; pulse pair input; second order procedure; spectral moment-based identification procedures; standard errors; stochastic input; third order procedure; Fourier transforms; Frequency; Information analysis; Maximum likelihood estimation; Moment methods; Nonlinear systems; Nuclear magnetic resonance; Robustness; Spectroscopy; Stochastic systems;

fLanguage

English

Journal_Title

Acoustics, Speech and Signal Processing, IEEE Transactions on

Publisher

ieee

ISSN

0096-3518

Type

jour

DOI

10.1109/29.57552

Filename

57552