Mask-Morph Graph U-Net: A Generalisable Mesh-Based Surrogate for Crashworthiness Field Prediction under Large Geometric Variation

2026-05-13

AI Summary

Traditional nonlinear finite element crash simulations are computationally expensive and ill-suited for iterative optimization, while existing graph neural network surrogates suffer from limited generalization under large geometric variations. This work proposes Mask-Morph Graph U-Net, which employs barycentric parameterization to adaptively align a fixed coarse graph hierarchy with input meshes and introduces edge-specific downsampling and upsampling layers to enhance spatial correspondence. By combining node-mask-supervised pretraining with a parameter-efficient fine-tuning strategy that freezes high-parameter edge layers, the model achieves high accuracy while significantly improving out-of-distribution and cross-component generalization as well as data efficiency. Experiments demonstrate that the proposed method consistently outperforms baselines across in-distribution, out-of-distribution, and cross-component tasks, yielding lower test errors and reduced train-test gaps.

Abstract

Nonlinear finite element crash simulations are accurate but computationally expensive, limiting their use in iterative design optimisation. Machine-learning surrogate models based on graph neural networks (GNNs) offer a faster alternative. Message-passing GNNs are widely used for mesh simulation, and their shared node and edge update functions are relatively generalisable across varying graph structures. By contrast, non-shareable edge-specific aggregation layers can capture nonlinear relationships more accurately but usually require fixed graph connectivity, which limits generalisability. This paper presents Mask-Morph Graph U-Net (MMGUNet), a practical approach to addressing the limitation of hierarchical Graph U-Net architectures that use edge-specific downsampling and upsampling layers. Fixed coarse graph connectivity is required for edge-specific layers. To retain this while improving spatial correspondence, the proposed method morphs the coarsened graph hierarchy to each input mesh using feature-aligned barycentric parameterisation before constructing cross-graph edges. It further applies node masking during supervised pretraining, followed by parameter-efficient fine-tuning in which high-parameter edge-specific layers are frozen. The proposed approach is evaluated in in-distribution, out-of-distribution, and cross-component transfer settings using mean Euclidean distance and maximum intrusion percentage error. Results show that coarse-graph morphing improves test accuracy relative to a fixed-coarse-graph baseline, while masked supervised pretraining reduces the train-test discrepancy and improves data efficiency during transfer. The proposed model also achieves lower prediction error compared with external baselines. These results demonstrate a practical route toward reusable, data-efficient mesh-based surrogate modelling for crashworthiness design exploration.

Problem

• crashworthiness
• graph neural networks
• mesh-based surrogate
• geometric variation
• generalisability

Innovation

• Graph U-Net
• mesh morphing
• edge-specific aggregation
• masked pretraining
• crashworthiness surrogate