Input Parameters Determination for Predicting Ram Speed and Billet Temperature for the First Billet


Article Preview

The aim of this paper is to present the results of the first step of a defined methodology for the neural network tool development. That first step is to studying the variables that have influence on extrusion process, especially in those that affect billet temperature and extrusion speed. In order to determine those parameters, a preliminary analysis was conducted with experimental data from real industry. Then, a multiple regression analysis was carried out to define which parameters will be the inputs of the neural network prediction tool.



Edited by:

Luca Tomesani and Lorenzo Donati




M. Sabater et al., "Input Parameters Determination for Predicting Ram Speed and Billet Temperature for the First Billet", Key Engineering Materials, Vol. 367, pp. 161-168, 2008

Online since:

February 2008




[1] M. Reddy, H. Bertolini and H. Biel. HyperXtrude/Process. Extrusion Process Optimization Software, Proceedings of the Eighth International Aluminum Extrusion Technology Seminar Vol. 1(2004) p.23.

[2] Anonymous Fichas técnicas A-GS (Aluminium Pechiney, 1987).

[3] P. Saha and ASM International. Aluminum extrusion technology (ASM International, 2000).

[4] M.L. Garcia-Romeu and J. Ciurana. Springback and geometry prediction - neural network applied to air bending process. Lecture Notes in Computer Science, Vol. 4113(2006), pp.470-475.


[5] Z. Lozina, I. Duplancic and B. Lela. Optimization of aluminium extrusion and die design using neural networks and genentic algorithms, Aluminium Two Thousand 5th world congress, (2003).

[6] S. Kalpakjian. Manufacturing processes for engineering materials (Prentice Hall, 2003).

[7] Committee under ASM direction. Metals handbook-ASM Handbook: Forming and Forging. (American Society for Metals, 1988).

[8] K. Laue. Moglichkeiten werkstoffgerechter. (Gestaltung von Leichtmetmetallprofilen, Z. f. Metallkunde 54., 1963).

[9] J. Hair, R. Anderson, R. Tatham, et al. Multivariate data analysis (Prentice Hall, 1998) (a) (b).