Back close

Roughness prediction in end milling of Ti1023 alloy using machine learning-based Gaussian Process Regression

Publication Type : Conference Proceedings

Publisher : Elsevier BV

Source : Procedia Computer Science

Url : https://doi.org/10.1016/j.procs.2025.01.107

Keywords : Ti1023, Gaussian Process Regression, surface roughness, non-linear feature, machine learning

Campus : Coimbatore

School : School of Engineering

Department : Mechanical Engineering

Year : 2025

Abstract : The present investigation explores the use of machine learning-based Gaussian Process Regression (GPR) to predict the surface roughness of Ti1023 alloy components during the end milling process. Cutting force and vibration sensor signatures were acquired under different cutting conditions to establish the GPR model. The GPR models were trained using non-linear features extracted from the sensor signatures acquired during the machining of the titanium alloy. The prediction ability of GPR models using non-linear features, namely Approximate Entropy, Holder Exponent, and Lyapunov Exponent, was investigated using different kernel functions. The GPR model trained with non-linear features of force signatures was observed to provide the best performance with an R2 of 0.9175. The findings suggested that incorporating non-linear features improved the capability of the model to dynamically predict roughness under optimal and sub-optimal machining conditions. The probabilistic GPR model is inferred to have good reliability and generalization capability and can serve as a support system for digital twin-enabled machining.

Cite this Research Publication : Viswajith S. Nair, Rameshkumar K., Saravanamurugan S., Roughness prediction in end milling of Ti1023 alloy using machine learning-based Gaussian Process Regression, Procedia Computer Science, Elsevier BV, 2025, https://doi.org/10.1016/j.procs.2025.01.107

Admissions Apply Now