An Approach of Data Science for the Prediction of Wear Behaviour of Hypereutectoid Steel

The data science has developed into an effective approach for predicting wear behaviour of solid materials. This contemporary work has mainly focused on the prediction of wear rate for the hypereutectoid steel which are tested under different operating conditions (sliding speed, normal pressure and sliding distance). To train and forecast the wear rate of the models, supervised machine learning methods such as random forest, Gaussian process regression, and support vector machine were used. Analysis of variance yielded 63.74%, 28.52%, and 1.08% for the contributions of normal pressure, sliding speed, and sliding distance, respectively. In order of prediction accuracy, the three machine learning algorithms used are: RF, GPR, and SVM. RF outperformed all other created models in terms of R2 (training and test), mean absolute error, and root mean squared error. At greater loads and speeds, the worn surface reveals that sub-surfaces fractured and broke, forming a plate-like structure of wear debris. The discoveries might speed up the creation of new functional hypereutectoid steel by allowing for the manufacture of hypereutectoid steel with controlled wear properties. © 2022, The Author(s), under exclusive licence to Springer Nature Switzerland AG.