Optimized Machine Learning Algorithms to predict wear behavior of Tribo- Informatics

Abstract

Wear rate prediction is most important in industrial applications. Machine learning (ML) has made an admirable contribution to the field of tribology. Standard ML models are extremely dependent on the parameter values; hence, tuning plays a crucial role in enhancing predictive performance. ML models largely work empirically, based on the data availability and application domain, the parameter tuning process effectively attains the desired accuracy of the models. The main aim of this study is to develop optimized ML models which render better accuracy than the previous study by using a grid search hyperparameter optimization technique. Five ML models namely Random Forest (RF), Support Vector Machine (SVM), K- Nearest Neighbor (KNN), Gaussian Process Regression (GPR), and Linear Regression (LR) are designed by tuning the parameters which lead to the optimization of models concerning the prediction accuracy.