Title: A comparative study of different machine learning approaches for predicting cutting force and surface roughness during ultrasonic-assisted milling

Authors: Yasmine El-Taybany; Ghada A. Elhendawy

Addresses: Production Engineering and Mechanical Design Department, Port Said University, Port Said, Egypt ' Mechanical Engineering Department, Damietta University, Damietta, Egypt

Abstract: This study utilises machine learning (ML) approaches to predict the performance of ultrasonic-assisted milling (UAM) of soda-lime glass based on the experimental results. The main goal is to model UAM performance and investigate the potential of ML in predicting the cutting force and surface roughness. Four different ML algorithms, mainly multilayer perceptron (MLP), self-organising maps (SOM), support vector machine (SVM), and principal component analysis (PCA), have been used and compared with the experimental data. To evaluate the accuracy of each ML technique, the mean square error (MSE), normalised root mean square error (NRMSE), and correlation coefficient (r) for each model have been obtained and compared. The findings prove that the proposed ML approaches can model UAM process, as the predicted results indicate good agreement with the experimental values. In particular, the results show that MLP and SOM are more reliable and accurate methods for predicting the cutting force and surface roughness. [Submitted 22 December 2023; Accepted 27 August 2024]

Keywords: cutting force; surface roughness; ultrasonic-assisted milling; machine learning; modelling; prediction.

DOI: 10.1504/IJMR.2024.143430

International Journal of Manufacturing Research, 2024 Vol.19 No.3, pp.338 - 357

Received: 22 Dec 2023
Accepted: 27 Aug 2024
Published online: 19 Dec 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article