[2C01-07] Data Science, Fuel Research with Machine Learning

[2C02] New Developments in Nuclear Fuel Research through Integration with Data Science

(2)Design of machine learning models for highly accurate prediction of thermal conductivity

*Koki Takeichi¹, Ken Kurosaki^1,4, Masaya Kumagai^1,3, Yuji Ohishi² (1. Kyoto Univ., 2. Osaka Univ., 3. SAKURA internet Inc., 4. Univ. of Fukui)

Keywords:Materials Informatics, Machine learning, Uranium compounds, Nuclear fuel, Thermal conductivity

In recent years, Materials Informatics, which uses information science in materials science, has become mainstream in the field of materials research for magnetic and thermoelectric materials, reducing the time and cost required for materials development. However, MI research has not progressed sufficiently in the field of nuclear materials research. In this study, a machine learning model with structural information of materials is used for the development of nuclear fuels in the nuclear field. Specifically, we will construct a machine learning model for predicting thermal conductivity that includes information on the crystal structure of materials for a method to comprehensively search for uranium compounds with high thermal conductivity using machine learning.