Power Forecasting in Continuous Extrusion of Pure Titanium Using Naïve Bayes Algorithm

Authors

Amar S. Abdul-Zahra, Mechanical Engineering Department, Mustansiriyah University, Baghdad, IraqFollow
Ehsan Ghane, Department of Physics, University of Gothenburg, Gothenburg, SwedenFollow
Aslan Kamali, Gropyus Technologies GmbH, Haupstr. 27, 10827 Berlin, GermanyFollow
Ahmed Ali Farhan Ogaili, Mechanical Engineering Department, College of Engineering, Mustansiriyah University, Baghdad 10052, IraqFollow

Abstract

Effective power estimation is key in streamlining continuous extrusion processes by reducing computational intensity and maximizing operational efficiency. In this paper, Naïve Bayes algorithm application in predicting commercially pure (CP) Titanium Grade 2 continuous extrusion power consumption is investigated. The model is trained using a theoretical, numerical, and experimental calculation dataset, enabling a probabilistic forecasting model. The model takes into account influential processing parameters such as feedstock temperature and extrusion wheel velocity in assessing their contribution towards powering requirements. Naïve Bayes model returns an RMSE of 0.46, demonstrating competitive performance in power consumption estimation. By leveraging probabilistic classification, this study recognizes Naïve Bayes’ potential in providing a lightweight and interpretable predictive model for metal forming processes. Contributions towards the development of data-driven methodologies in manufacturing by a real-life alternative to computationally intensive approaches render its application a viable and valuable tool in metal forming processes.

Recommended Citation

Abdul-Zahra, Amar S.; Ghane, Ehsan; Kamali, Aslan; and Farhan Ogaili, Ahmed Ali (2024) "Power Forecasting in Continuous Extrusion of Pure Titanium Using Naïve Bayes Algorithm," Terra Joule Journal: Vol. 1: Iss. 1, Article 2.
Available at: https://tjj.researchcommons.org/journal/vol1/iss1/2