The objective of this internship is to develop a learning metamodel (AI) that can be integrated into MAIA. Firstly, it will be necessary to understand which factors have the most impact on thermal conductivity via a sensitivity analysis. Then, using experimental designs and the MEROPE tool, microstructures of the fuel will be generated and thermal calculations will be carried out to determine the corresponding thermal conductivity. These results will serve as a learning basis for the construction of a metamodel by machine learning. A comparison of different types of metamodels can be carried out in order to select the most suitable (neural network, random forests, etc.).