Connecting Materials Science and Continuum Theory for Smart Materials via Machine Learning
The seminar will be presented by Dr.-Ing. habil. Adrian Ehrenhofer, Institute of Solid Mechanics, Technische Universität Dresden.
Abstract:
The specific pairing of materials and the layering setup determine which kind of functionality can be realized in a smart composite structure. Designing active-passive systems therefore connects the fields of materials science – with a focus on the Processing-Structure-Property-Performance relationship for predicting a material’s capabilities – to the field of continuum theory, which focuses on the geometry and functionality of a material in a specific setup.
The end-to-end engineering framework combines machine learning methods with continuum approaches, starting from abstract representations of synthesis and processing descriptors and ending with the actuation performance of a smart material in a specific composite functionality.
The methods combined range from (i) latent-space representations via embeddings, as used in Large Language Models, and (ii) classical feed-forward neural networks to (iii) continuum-based multi-field modelling with the Stimulus-Expansion Model and (iv) its integration into structural mechanics approaches.
The practical applicability of the approach will be illustrated through the case study of an active hydrogel-coated mesh that opens and closes in response to the level of hydration.
Short Bio:
Dr.-Ing. habil. Adrian Ehrenhofer is a Research Group Leader at the Institute of Solid Mechanics at Technische Universität Dresden. His work connects multi-field modelling of smart structures, data-driven material discovery using classical machine learning and generative artificial intelligence, and multi-field biomechanics.
From September to October 2026, he will be visiting Prof. Erasmo Carrera’s research group at DIMEAS.
(941 KB)