Identyfikatory
Warianty tytułu
Języki publikacji
Abstrakty
The paper presents results of investigation on the method of determination of significant properties in the case of virtual computational objects like FEM models. Typical approach is very time-consuming and involves the sequence of meshing and high performance computing. The method proposed in the paper is time-saving by utilization of the experimental design methodology, in particular the screening design analysis. The analysis based on Plackett-Burman designs and fractional factorial designs is focused on the maximum cost reduction. It provides to the main effects analysis. In the mentioned case of the virtual computational object it allows determining significant properties of the model and to focus on the selected most important parameters affecting the key properties of the object. Time-saving is obtained by eliminating insignificant factors from the study area. The mathematical basis for this approach is well known from the experimental design area, and the novelty is the use of it to the particular deterministic computational object. Specific metrics are defined for approximation accuracy, computational cost and their relationship to show the benefits of the method. Identification was carried out in two ways. The first method is used when the two elements are available: the set of output feature values obtained experimentally and chosen error criterion for comparing the predicted and measured values of output characteristics. The second method requires a binding equation or equations, which must be satisfied.
Słowa kluczowe
Wydawca
Czasopismo
Rocznik
Tom
Strony
339--346
Opis fizyczny
Bibliogr. 5 poz., rys.
Twórcy
autor
autor
- Krakow University of Technology Department of Applied Informatics JanaPawla II Ave. 37, 31-864 Krakow, Poland tel.: +48 12 6283580, fax: +48 12 648 82 67, pmpietra@mech.pk.edu.pl
Bibliografia
- [1] Montgomery, D. C., Design and Analysis of Experiments, Wiley, New York 1997.
- [2] Moré, J. J., Garbow, B. S., Hillstrom, K. E., Algorithm 566: FORTRAN Subroutines for Testing Unconstrained Optimization Software, ACM Trans. Math. Software, 7(1), 136-140, 1981.
- [3] Mühlenbein, H., Schomisch, D., Born, J., The Parallel Genetic Algorithm as Function Optimizer, Parallel Computing, Vol. 17, No. 6,7, 619-632, 1991.
- [4] Rastrigin, L. A., Extremal Control Systems, Znanie-Sila, Moscow 1974.
- [5] Törn, A., Zilinskas, A., Global optimization, LNCS 350, 1-255, Springer-Verlag, New York 1989.
Typ dokumentu
Bibliografia
Identyfikator YADDA
bwmeta1.element.baztech-article-BUJ5-0040-0044