Machinability Study of AISI Low Alloy Steels Using Classification and Machine Learning Approaches

Abstract

Machinability is one of the most important properties of steel. It shows how easily a metal piece can be processed with a cutting tool. In this paper, we use statistical and advanced machine learning methods to investigate the influence of different mechanical properties on the machinability of low alloy steels of type AISI 4000 and AISI 5000. Based on dataset of 375 different steels, classification and regression models were constructed and analyzed using random forest, and CART Ensembles and bagging. The obtained classification and regression models achieved an accuracy of about 99.5% and 94.7%, respectively. The results of the built models are comparable in accuracy with the experimental ones and provide an alternative and economical way to determine the machinability index of low alloy steels with known predictors.