ejmse.2025.10.01.011

PREDICTION OF HARDNESS CHARACTERISTICIS OF TIG WELDED Co-Mo STEEL FLAT BARS USING NEUTRAL NETWORKS

European Journal of Materials Science and Engineering, Volume 10, Issue 1, 2025
PDF Full Article,  DOI: 10.36868/ejmse.2025.10.01.011,   pp. 11-22
Published: March 20, 2025

Reuben Adebare ADEWUYI1,*
1 Department of Mechanical Engineering, The Federal Polytechnic, Ado-Ekiti, Nigeria

* Corresponding author: adewuyi_ra@fedpolyado.eudu.ng

Abstract

This study investigates the development and application of a neural network (NN) model to predict hardness characteristics in Cr-Mo steel welded joints, utilizing the SPSS software suite. The model incorporates four key welding parameters; material thickness, welding current, the number of weld passes, and electrode diameter as inputs, with hardness values in the weld zone (WZ) and heat-affected zone (HAZ) as outputs. The neural network features a single hidden layer with eight neurons, using the Softmax activation function for non-linear regression tasks. A comprehensive dataset comprising 18 combinations of the input parameters was employed to train, validate, and test the model, ensuring it could generalize across diverse welding conditions. Results demonstrate the model’s high predictive accuracy, particularly in the HAZ, where an R² value of 0.997 and a low Mean Squared Error (MSE) of 0.94 indicate minimal prediction error. The analysis also reveals that material thickness is the most influential parameter, significantly affecting hardness outcomes, while welding current, number of weld passes, and electrode diameter play secondary roles. However, the model’s performance varies between zones, with greater dispersion observed in the HAZ, suggesting complexities in predicting hardness due to microstructural changes in this region. Overall, the study confirms that the SPSS-developed neural network is a robust tool for predicting hardness in welded joints, offering valuable insights for optimizing welding parameters to achieve desired mechanical properties. This approach can reduce the need for extensive physical experimentation, streamlining the welding process in industrial applications.

Keywords: Cr-Mo steel bar, welding parameters, weld joint, hardness prediction, neural network

References:

  1. Zhang, J., et al. Prediction of mechanical properties in welded joints using artificial neural networks and regression analysi Journal of Manufacturing Processes, 63, 2021, pp. 678-688.
  2. Lee, S., & Kim, D. Using deep neural networks to predict weld quality in high-strength steel welding. Materials Science and Engineering A, 830, 2022, p. 142221.
  3. Nguyen, T. T., & Tran, Q. L. (2020). Evaluation of hardness properties of welded Cr-Mo steel under varying heat input conditions. Journal of Materials Processing Technology, 285, 2020, p. 116802.
  4. Santos, F. M., et al. Effect of welding speed and current on the mechanical properties of low-alloy steels. International Journal of Advanced Manufacturing Technology, 125, 2023, pp. 317-330.
  5. Kumar, S., & Gupta, R., Optimization of TIG welding parameters using the Taguchi method for Cr-Mo steel. Materials Today: Proceedings, 45, 2021, pp. 568-575.
  6. Wang, Y., et al. Application of machine learning for microstructural prediction in heat-affected zones of welded steels. Computational Materials Science, 206, 2022, pp. 111-118.
  7. Patel, D., & Singh, P. Comparison of response surface methodology and ANN in the prediction of weld hardness. Welding Journal, 99(12), 2020, pp. 525-533.
  8. Hosseini, A., et al. Microstructural evolution and hardness prediction in TIG-welded Cr-Mo steels. Metallurgical and Materials Transactions A, 54, 2023, pp. 1351-1360.
  9. Gupta, A., & Chawla, V. Hardness distribution in multi-zone welds: A case study on high-temperature steels. Journal of Materials Research and Technology, 12, 2021, pp. 1556-1567.
  10. Chen, H., et al. The influence of electrode diameter on the hardness of Cr-Mo steel welds. Materials & Design, 238, 2021, p. 111634.
  11. Adewuyi, R. A., & Aweda, J. O. Predicting thermal behaviour of welded Cr-Mo steel flat bar: A sensitivity analysis. The Engineer Story, Faculty of Mechanical Engineering, University Technology Malaysia, 8(9-14), 2024, e-ISSN: 3009 – 0792.
  12. Adewuyi, R.A., Aweda, J.O., Ogunwoye, F.O., Omoniyi, P.O., & Jen, T.C. Prediction of Impact Strength of TIG Welded Cr-Mo Steel Using Artificial Neural Networks. Metallurgical and Materials Engineering Research Paper. 2023, DOI: 10.56801/MME1031.
  13. Chauhan, R., Jha, S. K., & Reddy, G. M. Prediction of tensile strength of friction stir-welded joints of Al-Mg alloys using artificial neural network. International Journal of Advanced Manufacturing Technology, 83(1-4), 2016,215-225.
  14. Sharma, V., & Mahesh, V. Optimization of process parameters in TIG welding of AISI 304 stainless steel using response surface methodology. Materials Today: Proceedings, 5(2), 2018, pp. 5686-5695.
  15. Raj, V., Kumar, V., & Singh, G. A study on welding defects in TIG welding and its remedies. Materials Today: Proceedings, 26(2), 2020, pp. 731-736.
  16. Kumar, A., & Singh, R. Microstructure and mechanical properties of Cr-Mo steel by heat treatment and welding. Journal of Materials Engineering and Performance, 26(8), 2017, pp. 3785-3794.
  17. Rathod, D. G., & Saini, M. L. Prediction of Welding Process Parameters using Artificial Neural Networks: A Review. Journal of Materials Science and Surface Engineering, 7(2), 2021, pp. 839-846.
  18. Sharma, R., Singh, D., & Kumar, M. Influence of welding parameters on the hardness and tensile strength of dissimilar metal joints. Journal of Manufacturing Processes, 45, 2019, pp. 420-429.
  19. Singh, A., Bhattacharya, S., & Pandey, C. Microstructure and mechanical property analysis of welded joints with varying welding parameters. Materials Today: Proceedings, 28, 2020, pp. 2161-2168.
  20. Kalita, K., & Dhiman, M. Application of Artificial Neural Network (ANN) for prediction of mechanical properties of materials. Journal of Materials Research and Technology, 7(4), 2018, pp. 573-581.
  21. Haykin, S. (1998). Neural Networks: A Comprehensive Foundation (2nd ed.). Prentice Hall.
  22. Gunaraj, V., & Murugan, N. (1999). Application of response surface methodology for predicting weld bead quality in submerged arc welding of pipes. Journal of Materials Processing Technology, 88(1-3), 266-275.
  23. Adewuyi R. A., and Aweda J.O., (2021). Characterization of weld heat input effects on Mechanical Properties of Cr-Mo Steel bar using TIG welding process, European Journal of Materials Science and Engineering (EJMSE), 6(4), pp 192-204. DOI: 10.36868/ejmse.2021.06.04.192 www.ejmse.ro.
  24. Adewuyi R. A., and Aweda J.O., (2021), Modeling and simulation of welding temperature fields in Cr-Mo steel bar, European Journal of Materials Science and Engineering (EJMSE), 6(1), pp 003-018, DOI:10.36868/ejmse.2021.06.01.003.