BAT-NO: A Boundary-Condition-Aware Transformer Neural Operator for Crashworthiness Prediction of Vehicle Components

2026-10-02

AI Summary

This study addresses the prohibitive computational cost of high-fidelity crash simulations and the limited boundary generalization of conventional surrogate models by proposing a boundary-condition-aware Transformer neural operator. Methodologically, it introduces a local-global hybrid mechanism to propagate boundary information, coupled with sliced attention to model interactions across physical domains. By integrating recursive mesh processing with a latent mesh Fourier operator, the framework enables autoregressive prediction of displacement fields and crash responses under varying geometries and boundary conditions. Experimental results demonstrate that the proposed model reduces errors in the most complex scenarios by 32.6% compared to the second-best baseline, achieving a test-set displacement error of only 0.267 mm and median errors below 3% for critical crash metrics, thereby significantly enhancing generalization across the design space.

Abstract

High-fidelity finite-element simulations provide accurate crashworthiness predictions, but their cost limits iterative design exploration. Deep learning surrogates can reduce this cost, but many component-level models are developed under a single prescribed boundary condition, limiting generalisation to boundary variations. This work proposes a Boundary-Condition-Aware Transformer Neural Operator (BAT-NO) for autoregressive prediction of transient displacement fields and scalar crashworthiness responses under variations in geometry and boundary conditions. A B-pillar simulation framework evaluates generalisation across variations in geometry, impact position and velocity, and support stiffness. BAT-NO combines recurrent mesh processing with latent-grid Fourier operator processing. Boundary-condition information is transferred to the latent grid through a hybrid local--global mechanism. Slice-based attention models interactions among physically related regions, while direct boundary-to-grid projection preserves local spatial structure. Across the validation sets for the shape-only, shape-and-loading, and shape-loading-boundary cases, BAT-NO achieves the lowest mean final-step mean nodal Euclidean displacement error among the evaluated baselines. In the most challenging case, it reduces the mean error by 32.6% relative to the second-best model. Hyperparameter tuning reduces the validation error from 0.451 to 0.269 mm, with a comparable error of 0.267 mm on 300 unseen test simulations sampled within the investigated design space. An attention-based scalar decoder jointly predicts six response trajectories with a mean relative error of 2.46%. Most derived crashworthiness indicators have median errors below 3%. These results show that explicit local and global boundary-condition representations improve crashworthiness prediction over expanded component-level design spaces.

Problem

• crashworthiness prediction
• boundary conditions
• neural operator
• surrogate model
• finite-element simulation

Innovation

• Transformer Neural Operator
• Boundary-Condition-Aware
• Crashworthiness Prediction
• Fourier Operator
• Autoregressive Prediction