Machine learning application for wear rate prediction of WC/Co-based cermet with different content of Ni, Cr, TiC, TaC, and NbC

• Published in: International journal of advanced manufacturing technology Vol. 135; no. 11; pp. 5945 - 5959
• Main Authors: Harouz, Riad, Zelmatı, Djamel, Khelil, Khaled
• Format: Journal Article
• Language: English
• Published: London Springer London 01.12.2024; Springer Nature B.V
• Subjects: Algorithms; CAE) and Design; Cermets; Coefficient of friction; Computer-Aided Engineering (CAD; Data points; Dry friction; Engineering; High temperature; Impact analysis; Industrial and Production Engineering; Machine learning; Mechanical Engineering; Mechanical properties; Media Management; Niobium carbide; Original Article; Performance assessment; Physical properties; Powder metallurgy; Thermomechanical properties; Tribology; Tungsten carbide; Wear rate; Cermets; Wear rate; Sliding speed; XGBoost; Machine learning (ML); Tribological behavior; Random Forest
• ISSN: 0268-3768; 1433-3015
• DOI: 10.1007/s00170-024-14862-4

Wear rate of WC/Co-based cermet materials under severe tribological conditions is a critical thermomechanical property that can limit the practical application of various tools and industrial machinery. In this paper, three machine learning (ML) algorithms including Random Forest, Gradient Boosting Regressor, and XGBoost are employed to predict the wear rate of WC/Co-based cermets elaborated through powder metallurgy, utilizing dry friction under severe pin-on-disk conditions at elevated temperatures. The study analyzes 116 experimental tribological data points to assess the impact of sliding speed, additive content, hardness, friction coefficient, density, and temperature on estimating the wear of various cermet samples, all tested under a constant normal load of 20 N. The performance assessment shows that ML-based models could effectively predict the wear rate, with the Random Forest algorithm outperforming the others, achieving a coefficient of determination ( R 2 ) of 0.8438. Additionally, a comparative analysis is conducted to assess the performance of the ML-based models relative to one another. The models successfully predicted the wear rate of WC/Co-based cermets across various grades, tribological parameters, and physical and mechanical properties, achieving satisfactory accuracy.