Modeling of the effect of the hot-deformation parameters on the strength of A1-base metal-matrix composites by the use of a radial-base function (RBF) network

Citation
Is. Jalham et M. Barghash, Modeling of the effect of the hot-deformation parameters on the strength of A1-base metal-matrix composites by the use of a radial-base function (RBF) network, COMP SCI T, 61(9), 2001, pp. 1225-1231
Citations number
19
Categorie Soggetti
Material Science & Engineering
Journal title
COMPOSITES SCIENCE AND TECHNOLOGY
ISSN journal
02663538 → ACNP
Volume
61
Issue
9
Year of publication
2001
Pages
1225 - 1231
Database
ISI
SICI code
0266-3538(200107)61:9<1225:MOTEOT>2.0.ZU;2-K
Abstract
Empirical models such as the power Arrhenius relationships or the relations hips that depend on the Zener-Hollomon parameter have failed or been found to be followed only roughly when applied to composite materials. The reason may be the unclear mechanism of deformation in composites which comes from the input physical quantities that are thought to be the governing variabl es of the process. The aim of this investigation is to determine the physic al quantities that affect the deformation process and to use them to predic t the behavior of aluminium-based metal-matrix composites (MMCs) of the sam e composition under hot deformation. The two-layered RBF network structure with one hidden layer was used to evaluate the deformation process, to expl ore the input physical quantities that may govern the process, and to predi ct the behavior. In this approach, three schemes of different inputs were u sed. The first scheme contains the variables that concern the existing mode ls without the activation energy, since this was questionable, the second c ontains the variables of the Zener-Hollomon parameter, while the third is t he proposed scheme. The third scheme showed that the activation energy did not show a significant effect on the determination of the stress. Strain, s train rate, and temperature are found to be the only physical quantites tha t affect the process. The best agreement of the results of predictions with the experimental ones was found using the third scheme. (C) 2001 Elsevier Science Ltd. All rights reserved.