journal article·2022-09-05Open access

A Comparative Study of the Accuracy of Machine Learning Models for Predicting Tempered Martensite Hardness According to Model Complexity

Junhyub Jeon, DongEung Kim, Jun-Ho Hong, Hwi-Jun Kim, Seok-Jae Lee

Abstract

We investigated various numerical methods including a physical-based empirical equation, linear regression, shallow neural network, and deep learning approaches, to compare their accuracy for predicting the hardness of tempered martensite in low alloy steels. The physical-based empirical equation, which had been previously proposed with experimental data, was labelled and used in the present study. While it had a smaller number of coefficients, the prediction accuracy of the physical-based empirical equation was almost similar to that of the regression model based on the response surface method. The prediction accuracy of the machine learning models clearly improved as the number of layers increased and became more complicated in structure before the model began to overfit. The key point we found was that a single layered neural network model with optimized hyperparameters resulted in similar or better hardness prediction performance compared to deep learning models with a more complex architecture. We also analyzed 18 research papers from the literature which used neural network models to predict the hardness of steels. Only two recent papers adopted a convolutional neural network, as a kind of deep learning model, in a new attempt to predict hardness. The other 16 papers from 1998 to 2021 commonly chose shallow neural network models because a more complicated model is less effective than a simple model for regression problems with well-labeled experimental data in materials science and engineering.
Source: crossref · Quality: baseline

Research context

OpenAlex topics: Microstructure and Mechanical Properties of Steels · Non-Destructive Testing Techniques · Hydrogen embrittlement and corrosion behaviors in metals

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

References (45)

Show source-provided references
1. 10.1016/j.heliyon.2018.e0093810.1016/j.heliyon.2018.e00938 ↗
2. Shen · 2019 · 201
3. Han · 2021 · 1833
4. 10.3390/met1105072910.3390/met11050729 ↗
5. 10.1007/s12540-020-00854-y10.1007/s12540-020-00854-y ↗
6. Jeon · 2022 · 129
7. Hansen · 2004 · 801
8. Liu · 2013 · 526
9. Lee · 2020 · 86
10. 10.1007/s12540-020-00713-w10.1007/s12540-020-00713-w ↗
11. 10.3365/kjmm.2021.59.6.43010.3365/kjmm.2021.59.6.430 ↗
12. 10.1007/s12540-021-00982-z10.1007/s12540-021-00982-z ↗
13. Lee · 2021 · 11012
14. 10.3365/kjmm.2021.59.11.83810.3365/kjmm.2021.59.11.838 ↗
15. 10.1007/s12540-020-00854-y10.1007/s12540-020-00854-y ↗
16. 10.2320/matertrans.m201400410.2320/matertrans.m2014004 ↗
17. Hollomon · 1945 · 223
18. Datta · 1999 · 986
19. Wang · 2001 · 707
20. Saeglitz · 1997 · 377
21. Delagnes · 2005 · 435
22. 10.3390/met1108115910.3390/met11081159 ↗
23. 10.1007/s10462-021-09975-110.1007/s10462-021-09975-1 ↗
24. Waring · 2020 · 101822
25. Zöller · 2021 · 409
26. Solutions · 2022
27. Green · 1991 · 499
28. Dobrzański · 1998 · 59
29. Dobrzański · 1999 · 8
30. Guo · 2004 · 12
31. Sitek · 2004 · 245
32. Sterjovski · 2005 · 536
33. Xu · 2007 · 1425
34. Ozerdem · 2008 · 437
35. Mirzadeh · 2009 · 119
36. Sidhu · 2012 · 99
37. Taghizadeh · 2013 · 530
38. Powar · 2015 · 89
39. Jiang · 2016 · 191
40. Razavi · 2016 · 147
41. 10.1016/s1006-706x(18)30025-610.1016/s1006-706x(18)30025-6 ↗
42. Qiao · 2020 · 139845
43. Khalaj · 2021 · 101806
44. Oh · 2019 · 583
45. Jia · 2021 · 110126