Moving total least squares for parameter identification in mathematical model (CROSBI ID 464761)
Prilog sa skupa u zborniku | izvorni znanstveni rad | međunarodna recenzija
Podaci o odgovornosti
Scitovski, Rudolf ; Ungar, Šime ; Jukić, Dragan ; Crnjac, Miljenko
engleski
Moving total least squares for parameter identification in mathematical model
We consider a parameter identification problem in a mathematical model which is described by a system of differential equations, Usually it is not possible to expres s the solution of such a system by means of elementary functions, but nevertheless, one has to estimate the optimal parameter values using the experimental data. In our paper we apply the smooth-the-data method. In order to obtain smooth function we are using the moving total least squares method (TLS). The local approximants are linear functions obtained by applying the TLS method usin appropriate weight functions. This means that for each local approximant one has to find the smallest singular value for a second order matrix. The reason for using the TLS is the fact that the reeors may be expected in experimental measurements of both the independent and the dependent variables in the model. Once the optimal parameters are estimated, the ooptimal initial conditions are going to be estimated using the quasilinearisation method. The method is going to be tested on the mathematical model of insulin sensitivity.
surface generating; moving least squares; smoothing
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
nije evidentirano
Podaci o prilogu
196-201-x.
1996.
objavljeno
Podaci o matičnoj publikaciji
Operations Research Proceedings
Kleinschmidt, P. ; Bachem, A. ; Leopold-Wildburger, U. et al.
Berlin: Springer
Podaci o skupu
Nepoznat skup
predavanje
29.02.1904-29.02.2096