State-of-the-Art Review on the Aspects of Martensitic Alloys Studied via Machine Learning

Abstract

Though the martensitic transformation has been a commonly investigated topic in the field of experimental and computational materials science, the understanding of this mechanism in a variety of alloys is yet far from complete. In this era of Industry 4.0, there have been ongoing trends on employing machine learning (ML) techniques for the study of the martensitic alloys, and such data-driven approaches are expected to unravel a great amount of information about the process-structure-property behaviour relationship in this class of materials. However, with the availability of a large variety of datasets and with an option to use different ML models, a bulk amount of information has already been generated with regard to martensitic alloys. The discovery and design of shape memory alloys can be accelerated if the multi-principal element functional alloys and martensitic transformation phenomenon are studied extensively using machine learning techniques. Thus, it is necessary to highlight the major categories or aspects of these alloys that have been predicted with ML. The present work performs a state-of-the-art review on the machine learning models developed for the quantification of aspects such as martensitic start temperature (Ms), materials properties, microstructure, mechanisms etc., on the alloys.