MR-GVNO: A Geometry-Aware Variational Physics-Informed Neural Operator for Mindlin-Reissner Plates on Irregular Domains

2026-06-15 · Source: Artificial Intelligence · Field: Science & Research — Engineering & Applied Sciences, Artificial Intelligence & Machine Learning · Depth: Expert, quick

Summary

MR-GVNO, a geometry-aware variational neural operator, addresses rapid response prediction for Mindlin-Reissner plate problems on irregular domains. This method overcomes the high computational costs of conventional finite element methods by processing boundary point clouds to represent complex geometries. It employs separate encoders for spatially varying material fields, pressure loads, and scalar physical parameters, integrating these inputs via a cross-attention mechanism. Trained without labeled solution data using a variational physics-informed loss derived from total potential energy, MR-GVNO predicts transverse deflections and rotations at arbitrary locations. Numerical experiments on single-hole, double-hole, and L-shaped plates confirm accurate predictions under homogeneous and heterogeneous materials and uniform/random loads, achieving millisecond-level full-field inference and favorable cross-geometry generalization.

Key takeaway

For structural engineers or computational scientists developing simulation tools, MR-GVNO offers a significant advancement for Mindlin-Reissner plate analysis. Its ability to directly process irregular geometries via point clouds and achieve millisecond-level, full-field inference without labeled training data can drastically reduce computational overhead. You should consider integrating neural operator approaches like MR-GVNO to accelerate design iterations and explore complex structural behaviors more efficiently.

Key insights

MR-GVNO uses a geometry-aware neural operator and physics-informed loss for rapid, label-free plate response prediction on irregular domains.

Principles

• Represent irregular geometries with boundary point clouds.
• Integrate diverse physical inputs via cross-attention.
• Train without labeled data using variational physics-informed loss.

Method

MR-GVNO uses boundary point clouds for geometry, separate encoders for material/load fields, cross-attention for input integration, and a variational physics-informed loss from total potential energy to predict deflections and rotations.

In practice

• Predict plate responses on irregular domains.
• Handle heterogeneous materials and varying loads.
• Achieve millisecond-level full-field inference.

Topics

• Neural Operators
• Mindlin-Reissner Plates
• Physics-Informed Neural Networks
• Irregular Domains
• Computational Mechanics
• Finite Element Methods

Best for: AI Scientist, Research Scientist

Related on AIssential

• On the training of physics-informed neural operators for solving parametric partial differential equations — Physics-informed neural operators (PINOs) can match or outperform data-driven methods with robust training.
• Mesh Graph Neural Network Framework for Accelerating Finite Element Simulation for Arbitrary Geometries — Mesh Graph Networks effectively generalize finite element analysis predictions across arbitrary structural geometries by encoding relative features.
• Neural operators for free-boundary problems — Topological conjugacy enables neural operators to model complex free-boundary problems effectively.
• On Approximating the Dynamic Response of Synchronous Generators via Operator Learning: A Step Towards Building Deep Operator-based Power Grid Simulators — Operator Learning, specifically DeepONets, can accurately model synchronous generator dynamics for power grid simulation.
• Unified generalization analysis for physics informed neural networks — Koopman-based analysis can unify generalization bounds for PINNs and VPINNs, even with nonlinear differential operators.
• Reformulating Neural Operators in $d+1$ Dimensions for Embedding Evolution — Neural operators redefined in d+1 dimensions using Schrödingerisation better capture system evolution and achieve superior performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.