Feile K, Bartz M, Wartzack S (2024)
Publication Language: English
Publication Type: Conference contribution, Conference Contribution
Publication year: 2024
Event location: Erlangen
URI: https://junge-tribologen.de/2024-3/
Friction losses are, among other factors, significantly caused by manufacturing-related shape deviations of surface topographies, e.g. due to tool vibrations. Until now, the consideration of these shape deviations within the calculation of EHL contacts in the design process of low-friction machine elements with lubricated contacts has only been possible by means of complex and time-consuming numerical simulations.
Consequently, the objective of this contribution is to utilize machine learning (ML) approaches in order to significantly accelerate the prediction of lubricant film parameters in 2D line contacts, while accounting for shape deviations in the form of waviness. The numerical simulations used to generate the training and test data sets are discussed first. Subsequently, Gaussian process regression and neural network models are compared, whereby the number of input parameters is reduced due to the use of dimensionless parameters. The results demonstrate the potential of ML models to predict lubricating film parameters quickly and precisely, even when taking shape deviations into account.
APA:
Feile, K., Bartz, M., & Wartzack, S. (2024). Prediction of Lubricant Film Parameters of Deviated EHL Line Contacts using Machine Learning Approaches. In Proceedings of the 7th YOUNG TRIBOLOGICAL RESEARCHER SYMPOSIUM. Erlangen.
MLA:
Feile, Klara, Marcel Bartz, and Sandro Wartzack. "Prediction of Lubricant Film Parameters of Deviated EHL Line Contacts using Machine Learning Approaches." Proceedings of the 7th YOUNG TRIBOLOGICAL RESEARCHER SYMPOSIUM, Erlangen 2024.
BibTeX: Download