Professor Hyoung Seop Kim, along with Jeong Ah Lee, a PhD candidate, both from the Graduate Institute of Ferrous & Eco Materials Technology and the Department of Materials Science and Engineering, Pohang University of Science and Technology (POSTECH), collaborated with Professor Figueiredo from Universidade Federal de Minas Gerais’s Department of Metallurgical and Materials Engineering in Brazil to create an advanced artificial intelligence model. This model accurately predicts the yield strength of different metals, effectively overcoming conventional time and cost constraints.
The yield strength indicates when a material, like a metal, begins to change shape due to external stress. In materials engineering, accurately predicting yield strength is vital for creating high-performance materials and improving structural stability.
However, predicting this property involves considering many factors like grain size and types of impurities in the material and typically requires extensive experimentation over long periods to collect data.
To tackle this challenge, the Hall-Petch equation, which establishes the connection between a material’s yield strength and its grain size, is commonly utilized. However, it has limitations in accurately predicting the yield strength of new materials, taking into account their specific characteristics and various environmental conditions such as temperature and strain rate.
This research project involved integrating physical principles with AI methods to increase prediction accuracy and reduce the time and cost required for predicting yield strength. A machine learning model was created alongside a machine learning algorithm to utilize “grain boundary sliding” mechanisms in predicting yield strength by understanding how particles move within a material.
Initially, a black-box model was utilized by the team to assess the influence of different material properties on yield strength. Subsequently, a white-box model was developed with clearly defined inputs and outputs to improve the precision of yield strength predictions.
The model’s validation involved testing various iron-based alloys that were not included in the training data for the yield strength prediction model. The findings indicated that the model exhibited high accuracy, with an average absolute error of 7.79 MPa when compared to the actual yield strength, even when making predictions on untrained data.
Professor Hyoung Seop Kim of POSTECH expressed his aspirations, saying, “We have developed a general-purpose AI model that can accurately predict the yield strength of different types of metals and under various experimental conditions.” He added, “We will continue to actively utilize AI technology to make significant advances in materials engineering research.”
Journal reference:
- Jeong Ah Lee, Roberto B. Figueiredo, Hyojin Park, Jae Hoon Kim, Hyoung Seop Kim. Unveiling yield strength of metallic materials using physics-enhanced machine learning under diverse experimental conditions. Acta Materialia, 2024; DOI: 10.1016/j.actamat.2024.120046