Physics-based machine learning for computational failure mechanics

Einladung zum Vortrag

--------------------------------------

--------------------------------------

This presentation introduces a physics-based machine learning framework for modeling brittle and ductile fracture, as well as cyclic damage and fatigue. Physical principles, such as governing equations and fundamental constraints, are embedded directly into the neural network architecture. As a result, the proposed framework overcomes key limitations of classical data-driven models, which depend heavily on large datasets and provide no guarantees of physical consistency.