Abstract
Current research on digital twin (DT) technology for bridge health monitoring primarily focuses on virtual model construction, data sharing, and visualisation, while often lacking integration with simulation model updating and intelligent prediction. To address this gap, this study presents a DT-based framework that enables condition-aware synchronisation of structural monitoring data and finite element models, ensuring relevant information is delivered when most needed. The system centres on the analysis of structural physical performance and timely visualisation under dynamic loading conditions. A high-fidelity digital model of a bridge was developed using precise 3D modelling and rendering techniques to accurately represent the physical structure. To improve simulation accuracy, a genetic algorithm (GA)-based finite element model updating (FEMU) method was employed to refine key mechanical parameters. Numerical simulations were performed using ANSYS, and a Gaussian-weighted K-nearest neighbours (GW-KNN) method was introduced to achieve mesh reduction, enhancing computational efficiency and improving physical field resolution. Furthermore, surrogate modelling was implemented using radial basis function (RBF) networks combined with neural network algorithms to approximate the bridge’s structural response under dynamic loads. This approach significantly increased training speed and reduced computational overhead. The surrogate model was thoroughly validated, demonstrating its reliability and accuracy for right-time structural simulation during dynamic load testing.
| Original language | English |
|---|---|
| Journal | Structure and Infrastructure Engineering |
| DOIs | |
| State | Accepted/In press - 2025 |
Keywords
- Digital twin
- finite element analysis
- genetic algorithms
- model updating
- steel bridges
- surrogate models
- timely simulation
Fingerprint
Dive into the research topics of 'A digital twin-based system for timely simulation of bridge structural performance with finite element model updating'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver