Title :
A family of norms for system identification problems
Author :
Massoumnia, Mohammad-Ali ; Kosut, Robert L.
Author_Institution :
Integrated Syst. Inc., Santa Clara, CA, USA
fDate :
5/1/1994 12:00:00 AM
Abstract :
Introduces a family of norms that may prove useful in system identification problems. The important property of the new norm is that for a given sequence, its value in the limit will converge to the supremum over all frequencies of the spectrum of the sequence. Using this property, a procedure is outlined to approximately minimize the weighted L∞ norm of the frequency response estimation error. In addition, a test for checking the whiteness of a given sequence is proposed
Keywords :
frequency response; frequency-domain analysis; identification; optimisation; L∞ norm; frequency response estimation error; spectrum; system identification; whiteness test; Autocorrelation; Eigenvalues and eigenfunctions; Estimation error; Frequency domain analysis; Frequency response; Predictive models; Symmetric matrices; System identification; Testing; Virtual manufacturing;
Journal_Title :
Automatic Control, IEEE Transactions on