Building a composition-microstructure-performance model for C–V–Cr–Mo wear-resistant steel via the thermodynamic calculations and machine learning synergy

To better understand the complex relationships between composition, microstructure, and performance in C–V–Cr–Mo wear-resistant steel, and to facilitate the rapid and efficient development of high-alloy steels for extreme working conditions, this study proposes a collaborative approach that integrates thermodynamic calculations and machine learning. Based on systematic composition design and friction-wear experimental data, microstructure data for the corresponding components were first derived through thermodynamic calculations. Subsequently, the strong coupling between microstructure composition and performance was explored using machine learning techniques. The results indicated that the relative content of vanadium (V), chromium (Cr), and carbon (C) directly influenced the content and distribution of carbides within the steel. By using phase content and experimental parameters as input features, the Gradient Boosted Tree model and Support Vector Regression model demonstrated strong applicability in predicting frictional performance and wear, respectively. Furthermore, feature importance analysis, conducted using the Random Forest algorithm, revealed significant differences in the factors affecting sliding friction and abrasive wear. Dry sliding friction was primarily governed by surface modification behaviors, while abrasive wear resistance was more dependent on the material's ability to resist plastic deformation. These findings contribute to the development of a composition-microstructure-performance model for C–V–Cr–Mo wear-resistant steel, supporting the rapid design and performance optimization of multi-component alloy steels, with significant potential for industrial application.