Metalurgija, Vol. 41 No. 1, 2002.
Izvorni znanstveni članak
Modeling and the Reliability of Calculated Flow Curves
R. Turk
I. Peruš
M. Knap
Sažetak
Flow curves are very important input data for numerical modelling of industrial processes and for direct industrial applications. Precise thermal and mechanical testing of low carbon silicon steel showed obvious differences in yield stresses according to permissible oscillations of chemical composition. Since conventional Hajduk, Elfmark and Spittel equations for flow curve calculation are very rigid and cannot describe the local changes of yield stresses caused by phase transformations, a new neural network aproach for modelling the physical phenomena in materials science has been developed. The obtained results showed that neural-network method is a powerful tool, and it can be applied directly in solving problems of materials science (e.g. materials testing support, mathematical simulation of materials forming process).
Ključne riječi
Hrčak ID:
128602
URI
Datum izdavanja:
1.1.2002.
Posjeta: 942 *