Machine Learning-Driven Prediction of Wear Rate and Phase Formation in High Entropy Alloy Coatings for Enhanced Durability and Performance

Abstract

High Entropy Alloys (HEAs) are widely recognized for their excellent microstructure and properties, enhancing their effectiveness in surface modification through coatings techniques. These HEA coatings exhibit superior wear and corrosive resistance, making them suitable for various industries. However, accessing the wear behaviour and phase evolution of HEA coatings is complex and time-consuming due to their multiple element’s nature. To address this, Machine Learning (ML) techniques were integrated to predict the wear rate and phase formation in HEA coatings processed through thermal spray methods. Ten ML models such as AdaBoost, XGBoost, CatBoost, GBRT, DT, SVM-RBF, MLP, BNN, MLR and HR were utilized to predict wear rate, Feature engineering was conducted using Mutual Information (MI) and Pearson Correlation Coefficient (PCC) to access feature significance, Variance Inflation Factors (VIFs) analyzed multicollinearity, identified influential elements for wear rate prediction and aiding in the development of novel Lightweight High Entropy Alloys (LHEAs) coating compositions. For phase prediction, four ML models including RF, GNB, ANN and Logistic regression were evaluated. Results demonstrated that XGBoost achieved the highest predictive effectiveness with an R2 of 0.98 and the lowest error values, validated against experimental data. In phase prediction, the RF model exhibited the best accuracy of 98.5% for novel LHEA coatings. These findings highlight the potential of ML techniques in facilitating material design and coating optimization.

Topics & Concepts