Modeling of the Stress in 13Cr Supermartensitic Stainless Steel Welds by Artificial Neural Network

Abstract:

In this paper, artificial neural networks (ANN) has been proposed to determine the stresses of 13Cr supermartensitic stainless steel (SMSS) welds based on various deformation temperatures and strains using experimental data from tensile tests. The experiments provided the required data for training and testing. A three layer feed-forward network, deformation temperature and strain as input parameters while stress as the output, was trained with automated regularization (AR) algorithm for preventing overfitting. The results showed that the best fitting training dataset was obtained with ten units in the hidden layer, which made it possible to predict stress accurately. The correlation coefficients (R-value) between experiments and prediction for the training and testing dataset were 0.9980 and 0.9943, respectively, the biggest absolute relative error (ARE) was 6.060 %. As seen that the ANN model was an efficient quantitative tool to evaluate and predict the deformation behavior of type 13Cr SMSS welds during tensile test under different temperatures and strains.

• Periodical: Materials Science Forum (Volume 658)
• Pages: 141-144
• DOI: https://doi.org/10.4028/www.scientific.net/MSF.658.141
• Online since: July 2010
• Authors: Jun Hui Yu, De Ning Zou, Ying Han, Zhi Yu Chen
• Keywords: Artificial Neural Network (ANN), Stress, Supermartensitic Stainless Steel, Weld

References

[1] A. Turnbull, B. Nimmo: Corros. Eng. Sci. Techn. Vol. 40 (2005), p.103.

[2] E. Ladanoba, J.K. Solberg and T. Rogne: Corros. Engi. Sci. Techn. Vol. 41 (2) (2006), p.143.

[3] P.D. Bilmes, M. Solari, C.L. Llorente: Mater. Charact. Vol. 46 (2001), p.285.

[4] A. Griffiths, W. Nimmo, B. Roebuck, et al: Mater. Sci. Eng. A. Vol. 383 (2004), p.83.

[5] A. Turnbull, B. Nimmo: Corros. Eng. Sci. Techn. Vol. 40 (2) (2005), p.103.

[6] K.V.S. Ramana, T. Anita, S. Mandal: Mater. Des. Vol. 30 (2009), p.3770.

[7] H. Mirzadeh, A. Najafizadeh: Mater. Charact. Vol. 59 ( 2008), p.1650.

[8] M. Vasudevan, B.P.C. Rao, B. Venkatraman, et al: J. Mater. Process. Tech. Vol. 169 (2005), p.396.

[9] K.L. Zhou, Y.H. Kang: Neural Network model and design of MATLAB simulate. Tsing-Hua University Press, Beijing, (2005) (in Chinese).

[10] M. Yazdanmehr, S.H. Mousavi Anijdan, A. Bahrami: Comput. Mater. Sci. Vol. 44 (2009), p.1218.

[11] A. Bahrami, S.H. Mousavi Anijdan, H. R Madaah Hosseini, et al: Comput. Mater. Sci. Vol. 34 (2005), p.335.

[12] M. Zakeri, A. Bahrami, S.H. Mousavi Anijdan: Mater. Design. Vol. 28 (2007), p. (2034).

[13] P. Sathiya, S. Aravindan, A. Noorul Haq, et al: J. Mater. Process. Tech. Vol. 209 (2009), p.2576.