Differentiable finite element method with Galerkin discretization for fast and accurate inverse analysis of multidimensional heterogeneous engineering structures

• Published in: Computer methods in applied mechanics and engineering Vol. 437; p. 117755
• Main Authors: Wang, Xi, Yin, Zhen-Yu, Wu, Wei, Zhu, He-Hua
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.03.2025
• Subjects: Differentiable Finite Element Method (DFEM); Heterogeneous engineering structures; Inverse analysis; Physics-Encoded Numerical Network (PENN); Physics-Informed Neural Network (PINN); Inverse analysis; Physics-Encoded Numerical Network (PENN); Physics-Informed Neural Network (PINN); Heterogeneous engineering structures; Differentiable Finite Element Method (DFEM)

•Galerkin discretization is utilized to build a novel differentiable finite element method (DFEM) that encodes physics and significantly reduces training cost.•DFEM embeds the weak-form physics, boundary/initial conditions, and data constraints into the network architecture.•Both the accuracy and ef...