Recently, a research team led by Professor Figueiredo of the Department of Metallurgy and Materials Engineering at the Federal University of Minas Gerais, Brazil, in collaboration with Professor Kim Hyung-seop of the Graduate School of Steel and Environmental Materials Technology and the Department of Materials Engineering, and Lee Jeong-ah, a doctoral student in the Department of Materials Engineering, developed an optimal artificial intelligence model to predict the yield strength of various metals, effectively resolving existing cost and time constraints. This study was published online. Materials JournalAn international journal in the field of metallurgy and materials engineering.
Yield strength is the point at which a material, such as a metal, begins to deform under external stress. In materials engineering, accurately predicting yield strength is important for developing high-performance materials and enhancing structural stability. However, predicting this property requires considering numerous variables, such as particle size and the type of impurities in the material, and typically requires extensive experiments over a long period of time to collect data.
To solve this problem, the Hall-Petch equation, which establishes the relationship between the yield strength of the material and the grain size, is commonly used. However, it has limitations in accurately predicting the yield strength of a new material when considering specific properties and various environmental conditions such as temperature and strain rate.
In this study, the team combined physical theory with artificial intelligence (AI) techniques to increase accuracy while reducing the cost and time required to predict yield strength. They developed a machine learning model that applies the “grain boundary sliding” mechanism that describes how particles within a material move relative to each other, and a machine learning algorithm that predicts yield strength.
First, the team used a black-box model to analyze the effect of various material properties on yield strength. Then, a white-box model with clear inputs and outputs was developed to improve the accuracy of yield strength prediction.
The team validated the model using a variety of iron-based alloys that were not included in the training data for the yield strength prediction model. The results showed that the model was highly accurate, with a mean absolute error of 7.79 MPa when compared to the actual yield strength, even when predicted with untrained data.
Professor Hyung-Seop Kim of POSTECH expressed his ambition, saying, “We have developed a general-purpose AI model that can accurately predict the yield strength of various metals.” He added, “We will continue to actively utilize AI technology to make great progress in materials engineering research.”
This study was conducted with the support of the Nanomaterial Technology Development Project of the National Research Foundation of Korea, funded by the Ministry of Science and ICT.