The mixed Deep Energy Method for resolving concentration features in finite strain hyperelasticity

• Published in: Journal of computational physics Vol. 451; p. 110839
• Main Authors: Fuhg, Jan N., Bouklas, Nikolaos
• Format: Journal Article
• Language: English
• Published: Cambridge Elsevier Inc 15.02.2022; Elsevier Science Ltd
• Subjects: Approximation; Artificial neural networks; Boundary conditions; Computational physics; Concentrated loads; Deep energy method; Domains; Energy methods; Finite element method; Finite strain; Hyperelasticity; Inverse problems; Mixed deep energy method; Neural networks; Notches; Numerical integration; Parameter estimation; Physics-informed neural networks; Solid mechanics; Strain; Mixed deep energy method; Solid mechanics; Finite strain; Physics-informed neural networks; Deep energy method; Hyperelasticity

•Using machine learning to solve PDEs has seen increased interest in solid mechanics.•Current methods fail to reliably resolve concentration features in hyperelasticity.•Mixed Deep Energy Method (mDEM) is introduced that is able to resolve concentrations.•mDEM utilizes stress as an additional output...