GA-VINO: A Geometry-Aware Variational Physics-Informed Neural Operator for Mindlin–Reissner Plates
Plate and shell structures are widely used in engineering fields. Rapid response prediction for such structures under complex geometries, heterogeneous materials, and varying loads is important for engineering design, but conventional numerical methods usually require repeated modeling and solution when the physical configuration changes. To address this issue, this study proposes a geometry-aware variational physics-informed neural operator (GA-VINO) for Mindlin–Reissner plates. GA-VINO represents the plate geometry using boundary point clouds and incorporates a material encoder, a load encoder, and a scalar-parameter branch to handle spatially random material fields, spatially varying loads, and sample-level uniform parameters. Through multi-branch point cloud encoding and cross-attention, GA-VINO fuses geometric, material, loading, and query point information, and predicts the transverse deflection and rotations at arbitrary query locations. Unlike conventional data-driven neural operators, GA-VINO requires no labeled solution data during training. Instead, it minimizes a variational physics-informed loss constructed from the discretized total potential energy of the Mindlin–Reissner plate. Compared with grid-based neural operators, GA-VINO directly processes irregular point clouds and allows different physical fields to be discretized on different point sets, avoiding forced interpolation onto a common grid. Numerical results demonstrate that GA-VINO achieves accurate predictions across varying geometries, material properties, and loading conditions, with lower H 1 seminorm and energy-norm errors and more accurate predictions of bending moments and transverse shear forces than data-driven baselines. Furthermore, GA-VINO enables millisecond-level full-field inference and demonstrates useful generalization capability under moderate geometric distribution shifts, although its prediction accuracy decreases under stronger geometric extrapolation.
Authors
- Timon Rabczuk (ORCID: https://orcid.org/0000-0002-7150-296X)
- Xiaoying Zhuang (ORCID: https://orcid.org/0000-0001-6562-2618)
- Daobo Sun (ORCID: https://orcid.org/0009-0000-5197-9579)
- Y Jin (ORCID: https://orcid.org/0000-0002-2769-7376)
- Yizheng Wang (ORCID: https://orcid.org/0000-0002-3899-7008)
- Siqi Wang (ORCID: https://orcid.org/0009-0009-2104-7784)
- Yilong Zhang
Institutions
- Leibniz University Hannover (DE)
- Xiamen University (CN)
- Fudan University (CN)
- Bauhaus-Universität Weimar (DE)
- Tsinghua University (CN)
Publication Details
- Journal
- Computer Methods in Applied Mechanics and Engineering
- Published
- 2026-11-21
- DOI
- https://doi.org/10.1016/j.cma.2026.119609
- Primary Topic
- Model Reduction and Neural Networks
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Fudan University