Work

A deep neural network surrogate for fast mechanical parameter identification using the ring tensile test

Year: 2026
Type: article
Abstract: This work introduces a deep neural network (DNN) surrogate model designed to accelerate mechanical parameter identification using the ring tensile test. The surrogate is trained on finite element simulations of the ring tensile test using an anisotropic constitutive model, serving as an efficient approximation of the forward problem. Its main application is the batch characterization of small aortic tissue samples, where specimen availability and computational cost limit traditional inverse analyses. Validation against experimental data from thoracic aorta samples of Wistar rats shows strong predictive capability, with initial correlation coefficients of 0.73 or higher across all identified parameters. The DNN surrogate provides accurate initial estimates that can be refined through finite element-based inverse analysis, allowing to include the residual stress effects in vessel mechanics. While training requires a substantial computational cost, the framework enables rapid, high-throughput characterization and can be generalized to other non-standard mechanical tests involving complex stress-strain states. • DNN surrogate accelerates arterial parameter identification from ring tensile tests. • The framework combines FEM simulations with deep learning for fast optimization. • Achieves accurate prediction of hyperelastic model parameters in arteries. • DNN assists sensitivity analysis and initial guess refinement in inverse problems. • Enables rapid, high-throughput mechanical characterization beyond standard methods.
Language: English
Open Access status: gold
Estimated APC paid: $3,650
OpenAlex ID: W7128742072
FWCI: 1.789
Cites: 47
Cited by: 1
Related to: