Adaptive Distance-Aware Trunk Deep Operator Learning for Long-Span Roadway Bridges

2026-06-18 · Source: Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Structural Engineering · Depth: Expert, quick

Summary

An adaptive-trunk DeepONet (AD-DeepONet) framework is proposed for predicting localized structural responses in long-span roadway bridges. This addresses the high computational cost of traditional Finite Element (FE) analysis, critical for influence surface generation and structural digital twins. The framework dynamically constructs a load-dependent learning domain using a K-Nearest Neighbors (KNN) strategy. It enhances the trunk network with distance-aware features and integrates a physics-based full-field reconstruction via a stiffness-informed Schur complement formulation. For scalable training, response data is generated using a reduced-order equivalent shell model. Validated on a benchmark bridge and the real-world Mussafah Bridge, the method achieves FEM-level accuracy, with relative errors below 5%. It reduces total response evaluation time by approximately 60x, and inference is up to four orders of magnitude faster than FEM.

Key takeaway

For structural engineers developing bridge analysis tools or digital twins, this AD-DeepONet framework offers a significant leap in computational efficiency. You can achieve FEM-level accuracy with response evaluation times reduced by 60x. This enables rapid generation of influence lines and full-field responses under various vehicular loads. Consider integrating this adaptive operator learning approach to accelerate large-scale bridge analysis and enhance your digital twin applications' practicality.

Key insights

AD-DeepONet accurately predicts localized bridge responses 60x faster than FEM by adaptively focusing on influence zones and using physics-informed reconstruction.

Principles

• Focus learning on structural influence zones.
• Encode geometric load-node relationships.
• Integrate physics for full-field reconstruction.

Method

Dynamically construct a load-dependent learning domain via KNN. Enhance the trunk network with distance-aware features. Reconstruct full-field responses using a stiffness-informed Schur complement formulation. Train with a reduced-order shell model.

In practice

• Generate influence lines for bridges rapidly.
• Develop efficient structural digital twins.
• Analyze large-scale bridges under arbitrary loads.

Topics

• DeepONet
• Bridge Engineering
• Structural Digital Twins
• Finite Element Analysis
• Operator Learning
• Reduced Order Modeling

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• 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.
• eCNNTO: A Highly Generalizable ConvNet for Accelerating Topology Optimization — eCNNTO employs CNNs and a novel training strategy to accelerate topology optimization, achieving high generalizability and significant iteration reduction.
• MR-GVNO: A Geometry-Aware Variational Physics-Informed Neural Operator for Mindlin-Reissner Plates on Irregular Domains — MR-GVNO uses a geometry-aware neural operator and physics-informed loss for rapid, label-free plate response prediction on irregular domains.
• Adaptive Domain Decomposition Physics-Informed Neural Networks for Traffic State Estimation with Sparse Sensor Data — ADD-PINN improves traffic state estimation from sparse sensor data by adaptively decomposing domains based on residual errors.
• Bridging Data Gaps in Structural Fragility Modeling through Transfer Learning: Methodology and Case Studies — Transfer learning effectively bridges data gaps in structural fragility modeling under domain shift and scarcity.
• Neural Operator-Based Surrogate Model for CFD:Helical Coil Steam Generator in Small Modular Reactor — Neural operator-based surrogates, combined with ROMs and multi-scale techniques, enable real-time CFD for SMR digital twins.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.