Contrasting Models to Determine the Approximate Extrusion Process Conditions for the First Billet

Abstract:

This paper is focused on determining extrusion process conditions for the first billet and, particularly, on selecting an adequate billet temperature and extrusion speed. It is important to predict these values before adjusting the die to avoid lengthy and unproductive preparation times, and to make die adjustment easier during the tests. With that in mind, this study employs two models to determine the first billet temperature (TFB) and the extrusion speed (SE) parameters: one that applies regression techniques and another that uses neural networks. The aim of this paper is to present the results obtained by comparing the performance of the two models to determine which one offers the best results. We also analyze whether changing the number of patterns for construction of the models will improve results. The models are based on extrusion process parameters taken from real industry contexts. The results indicate that the variables chosen as predictor variables for the extrusion speed output must be refined.

• Periodical: Key Engineering Materials (Volume 424)
• Pages: 249-256
• DOI: https://doi.org/10.4028/www.scientific.net/KEM.424.249
• Online since: December 2009
• Authors: Marc Sabater, Maria Luisa Garcia-Romeu
• Keywords: Billet Temperature, First Billet, Neural Network Tool, Process Parameter, Ram Speed

References

[1] K. Muller, A. Ames, O. Diegritz, et al.: Fundamentals of extrusion technology, edited by Giesel Verlag. Isernhagen. (2004).

[2] M.P. Reddy, H.E. Bertolini, and B.W. Biel,: HyperXtrude/Process - extrusion process optimization software, edited by Proceedings of the Eighth International Aluminum Extrusion Technology Seminar. Vol. I. 231-235 (2004).

[3] Fichas técnicas A-GS, edited by Aluminium Pechiney (1987).

[4] P.K. Saha: Aluminum extrusion technology, edited by ASM International. United States of America (2000).

[5] M.L. Garcia-Romeu and J. Ciurana: Springback and geometry prediction - neural networks applied to the air bending process, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4113 LNCS - I 470-475. (2006).

DOI: 10.1007/11816157_58

[6] Z. Lozina, I. Duplancic and J. Prgin: Evolution- and biology-based approach to the optimization of the extrusion process, edited by Proceedings of the Seventh International Aluminum Extrusion Technology Seminar. (2000).

[7] Z. Lozina, I. Duplancic, B. Lela, et al.: Optimisation of aluminum extrusion and die design using neural networks and genetic algorithms, edited by Aluminium Two Thousand 5th World Congress. Rome. (2003).

[8] D. Karayel and S.S. Ozkan: Simulation of direct extrusion process and optimal design of technological parameters using FEM and artificial neural network, edited by Key Engineering Materials. Vol. 367. (2008).

DOI: 10.4028/www.scientific.net/kem.367.185

[9] R. Camargo and R. Lesley: Use of adaptive neural networks (A*) in aluminum extrusion process control. Proceedings of the Ninth International Aluminum Extrusion Technology Seminar. Orlando. (2008).

[10] M. Sabater, M.L. Garcia-Romeu and J. Ciurana: Input parameters determination for predicting ram speed and billet temperature for the first billet, edited by Key Engineering Materials. Vol. 367. 161-168. (2008).

DOI: 10.4028/www.scientific.net/kem.367.161