Time series analysis in the frequency domain

Author

Pintelon, R. ; Rolain, Y. ; Schoukens, J.

Author_Institution

Dept. ELEC, Vrije Univ., Brussels, Belgium

Volume

1

fYear

1997

Firstpage

554

Abstract

This paper presents a parametric frequency domain identification algorithm for autoregressive moving average (ARMA) processes that does not suffer from spectral leakage errors. It is based on an extended transfer function model that takes into account the begin and end effect of the finite data record. The relationship with the one step ahead prediction error method is established. The advantages of the proposed method are easy prefiltering and leakage free spectral representation of the raw data

Keywords

autoregressive moving average processes; frequency-domain analysis; parameter estimation; prediction theory; time series; transfer functions; autoregressive moving average processes; extended transfer function model; finite data record; frequency domain analysis; leakage free spectral representation; one step ahead prediction error method; parametric frequency domain identification; prefiltering; time series analysis; Autoregressive processes; Discrete Fourier transforms; Frequency domain analysis; Frequency estimation; Least squares approximation; Maximum likelihood estimation; Parameter estimation; Predictive models; Signal processing algorithms; Time series analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Instrumentation and Measurement Technology Conference, 1997. IMTC/97. Proceedings. Sensing, Processing, Networking., IEEE

Conference_Location

Ottawa, Ont.

ISSN

1091-5281

Print_ISBN

0-7803-3747-6

Type

conf

DOI

10.1109/IMTC.1997.604010

Filename

604010