Projet de fin d'étude : Modeling diffusion in a network of varied connectivities at the polycrystal scale: Deep Learning solution of an inverse problem for assigning diffusion coefficients on a Nickel microstructure
Etudiant : ES-SAYEH RABIE
Filière : Master Big Data Analytics & Smart Systems (BDSaS)
Encadrant : Pr. ZINEDINE AHMED
Annèe : 2024
Résumé : The project aims to design nickel microstructures less sensitive to hydrogen embrittlement using numerical simulations and machine learning. The impact of hydrogen on the durability of metals is a major concern in many sectors, motivating experimental and numerical studies on hydrogen diffusion and trapping in interaction with crystal defects. The work seeks to link hydrogen diffusion in the material to microstructural parameters such as the distribution of grain boundaries and triple junctions. The tools developed include finite element simulations to predict hydrogen diffusion and machine learning techniques, including deep neural networks, to generate optimized microstructures with controlled diffusion properties.