Predicting multi-component oil viscosity using machine learning methods

Vladimir I. Deshchenya

doi:10.1016/j.ces.2026.124616

Predicting multi-component oil viscosity using machine learning methods

2026-07-11

0

PRE

AI

V

Vladimir I. Deshchenya *

DOI：10.1016/j.ces.2026.124616

Original

Original request for help

Save

Abstract

En 中文

• Robust viscosity models are trainable on small, anonymized proprietary datasets. • Gradient boosting achieves 16.2% median error, outperforming physics-based models. • Walther equation screens experimental outliers in complex viscosity datasets.

Keywords:

Kinematic viscosity

Lubricating oils

Machine learning

Materials science

Predicting multi-component oil viscosity using machine learning methods

Abstract

Journal

Organization