- Cet évènement est passé
Séminaire – Lori Graham-Brady ‘ML-enhanced approaches for modeling complex mechanics’
mercredi 14 février à 9 h 00 – 10 h 15
Dans le cadre du projet ANR-FNS CLIP-, le GeM accueillera Lori Graham-Brady, chercheuse à l’université John Hopkins qui donnera un séminaire le mercredi 14 février de 9h à 10h15 à Centrale Nantes (Amphi B8).
Résumé :
ML-enhanced approaches for modeling complex mechanics
The simultaneous growth of effective machine-learning-enabled modeling and high-throughput fabrication, characterization and testing of materials has ushered in a new paradigm for applying AI to materials design (AI-M). While tremendous progress has been made in materials design for several disciplines, such as pharmaceuticals, there remains much to be done the realm of materials in extreme environments. Damage, fracture and fragmentation are multi-scale phenomena that require significant computational effort to represent in fully physics-based models. This talk describes recent advances for machine learning (ML) models for elastic, elastoplastic and spall behavior. The results show a surprising level of robustness in the model predictions and an ability to transfer the model to material microstructures outside of the training data.