DeepFEAv2: Deep Learning for Transient Finite Element Analysis Beyond Structured Meshes

DeepFEAv2 extends the original DeepFEA framework to handle unstructured finite element meshes, overcoming the limitation of previous approaches restricted to structured grids. The framework jointly predicts Node-based and Element-based Outputs (NEO) over time for transient mechanical simulations. By leveraging neural network surrogate models, this method significantly reduces the high computational cost associated with traditional finite element analysis while maintaining predictive accuracy for complex geometries.