Molecular dynamics simulation and machine learning models for predicting welding and tensile properties of diffusion-welded Aluminum-Nickel

Articles in Press

Document Type : Article

Authors

1 Department of Mechanical Engineering, Faculty of Engineering and Computer Science, Jakarta Global University, Boulevard Raya St. No 2, Tirtajaya, Sukmajaya, Depok 16412, Indonesia

2 Department of Informatics Engineering, Faculty of Engineering and Computer Science, Jakarta Global University, Boulevard Raya St. No 2, Tirtajaya, Sukmajaya, Depok 16412, Indonesia

3 Department Mechanical Engineering, College of Engineering, Gulf University Sanad 26489, Bahrain

10.24200/sci.2025.64612.9034

Abstract

In this study, machine learning (ML) models are developed to predict the value of interfacial region thickness (IRT) and ultimate tensile strength (UTS) of diffusion-bonded Al-Ni based on molecular dynamics simulation data. Molecular dynamics simulations are performed to simulate the diffusion bonding of Al-Ni with three parameters with three to four level for each parameter. The results of the simulations that are used to generate the ML models are the value of IRT and UTS. The temperature have influenced the performance of ML models with significant impact, indicated by the value of MSE and R2 that used temperature as the input parameters with an excellent performance. However, the combination of the three parameters as the input shows the best performance, indicated by the MSE value of 0 and the R2 value of 1, showing that the ML models performance will increase with the increase of the input data. Furthermore, the models with the highest performance throughout the test are the NN models, followed by the kNN, whereas the other three models showing average performance. This study has successfully developed ML models to predict the IRT and UTS from the molecular dynamics simulation data of diffusion bonding of Al-Ni.

Keywords

Main Subjects

Scientia Iranica

Articles in Press, Accepted Manuscript
Available Online from 27 January 2025

Files

History

  • Receive Date: 20 May 2024
  • Revise Date: 11 December 2024
  • Accept Date: 27 January 2025

Share

How to cite

Statistics