DocumentCode
2941027
Title
System identification using frequency domain methodology
Author
Buss, Arnold
Author_Institution
John M. Olin Sch. of Bus., Washington Univ., St. Louis, MO, USA
fYear
1990
fDate
9-12 Dec 1990
Firstpage
360
Lastpage
363
Abstract
A method for using the frequency domain methodology to identify a linear meta-model is proposed. This involves multiple frequency domain runs to eliminate the effect of gain. The method uses the Fourier transform of the output process, rather than the periodogram or smoothed periodgram estimators. The identification focuses on estimating the coefficients of the model, and is therefore similar in spirit to response surface methodology. The method is tested on two models, one with known structure and the other with unknown structure. The procedure does well as fitting the coefficients, including obtaining the correct sign
Keywords
Fourier transforms; frequency-domain analysis; identification; Fourier transform; frequency domain methodology; gain; linear meta-model; model coefficients estimation; response surface methodology; sign; system identification; Autocorrelation; Fourier transforms; Frequency domain analysis; Frequency estimation; Jacobian matrices; Response surface methodology; Stochastic processes; Surface fitting; System identification; Testing;
fLanguage
English
Publisher
ieee
Conference_Titel
Simulation Conference, 1990. Proceedings., Winter
Conference_Location
New Orleans, LA
Print_ISBN
0-911801-72-3
Type
conf
DOI
10.1109/WSC.1990.129541
Filename
129541
Link To Document