Ira A. Fulton College of Engineering and Technology
First Faculty Advisor
Dr. David T. Fullwood
First Faculty Reader
Dr. Michael P Miles
Dr Brian Jensen
Magnesium, Machine Learning, Twin Modeling, Materials Science, Decision Trees, EBSD
Machine learning is being adopted in various areas of materials science to both create predictive models and to uncover correlations which reveal underlying physics. However, these two aims are often at odds with each other since the resultant predictive models generally become so complex that they can essentially be described as a black box, making them difficult to understand. In this study, complex relationships between microstructure and twin formation in AZ31 magnesium are investigated as a function of increasing strain. Supervised machine learning is employed, in the form of J-48 decision trees. In one approach, strain is incorporated as an implicit attribute in a single predictive model; in a second method, separate decision trees are formed for each strain level, and the structure of the trees is compared to understand the influence of strain on the twin activity. A comparison of the methods shows that the second better uncovers the underlying physics of twin formation as a function of strain. The correlations revealed by the second method are found to exhibit similarities with parameters used in conventional modeling techniques, leading to the conclusion that machine learning has potential to assist in future microstructural modeling.
BYU ScholarsArchive Citation
Orme, Andrew D., "Evolution of Mg AZ31 Twin Activation with Strain: A Machine Learning Study" (2018). Undergraduate Honors Theses. 17.