Page 65 - IJAMD-1-1

International Journal of AI
for Material and Design Integrating physics data for DL in DED

Programme and the National Research Foundation, Prime
Minister’s Office, Singapore, under its Medium-Sized
Centre funding scheme. In addition, this work receives
support from the Singapore Centre for 3D Printing at
Nanyang Technological University, Singapore, which
provides access to its AM facilities.

Conflict of interest
The authors declare that they have no known competing
financial interests or personal relationships that could have
appeared to influence the work reported in this paper.

Author contributions
Conceptualization: Stanley Jian Liang Wong, Chengxi
Chen, Hua Li
Figure 22. Total root mean square error values for deep learning Data generation: Stanley Jian Liang Wong
prediction models.
Formal analysis: Stanley Jian Liang Wong, Cheng Xi Chen
model, demonstrating its superior capability in predicting Investigation: Stanley Jian Liang Wong
melt pool geometry with variations in sulfur content. Methodology: Stanley Jian Liang Wong, Cheng Xi Chen
Results: Stanley Jian Liang Wong
5. Conclusion Writing – original draft: Stanley Jian Liang Wong

This study undertakes the integration of pre-process, Writing – review & editing: Hua Li, Eddie Zhi’En Tan
in-process, and post-process information through the Availability of data
application of ML. A high-quality in-process dataset was Data used in this work are available from the corresponding
generated using physics-based simulation to predict track author upon reasonable request.
geometry. The physics-based simulation provided input data
that cannot be accurately captured through experimental References
methods, such as sulfur content, thereby offering further
insights into the effect of these input parameters. 1. Tsotridis G, Rother H, Hondros ED. Marangoni flow and the
shapes of laser-melted pool. Sci Nat. 1989;76(5):216-218.
The main conclusions drawn from this study are as doi: 10.1007/BF00627691
follows:
• Among the various combinations of training 2. Zhang YM, Lim CWJ, Tang C, Li B. Numerical investigation
sequences in this study, the DL-AugExp-Sim-Exp on heat transfer of melt pool and clad generation in
model demonstrated the best-performing prediction directed energy deposition of stainless steel. Int J Therm Sci.
accuracy, as evidenced by R and RMSE values. 2021;165:106954.
2
• Through the integration of in-process information, doi: 10.1016/j.ijthermalsci.2021.106954
additional prediction capability of sulfur content and track 3. Guan X, Zhao YF. Modeling of the laser powder-
geometry was established with low tradeoffs in R value. based directed energy deposition process for additive
2
As the results presented in this work were exclusively manufacturing: A review. Int J Adv Manuf Technol.
tested on a general neural network, it lays the groundwork 2020;107(5-6):1959-1982.
for the integration of simulation and experimental data doi: 10.1007/s00170-020-05027-0
in ML training. Future research can use this study as a 4. Yao L, Xiao Z, Huang S, Ramamurty U. The formation
reference for further exploration in this evolving field. mechanism of metal-ceramic interlayer interface
during laser powder bed fusion. Virtual Phys
Acknowledgments Prototyp. 2023;18(1):e2235324.
None. doi: 10.1080/17452759.2023.2235324
Funding 5. Yao L, Huang S, Ramamurty U, Xiao Z. On the formation
of ‘Fish-scale’ morphology with curved grain interfacial
This research is supported by Makino Asia Pte Ltd through microstructures during selective laser melting of dissimilar
the Economic Development Board Industrial Postgraduate alloys. Acta Mater. 2021;220:117331.

Volume 1 Issue 1 (2024) 59 https://doi.org/10.36922/ijamd.2355