Abstract
To tackle real-time localization and force identification challenges induced by complex stiffness distributions, this study proposes a low-speed impact monitoring method for aerospace composite panels. By integrating distributed fiber optic sensing with physics-informed deep learning, a two-stage strategy simultaneously identifies impact location and force. This study collected dynamic strain responses from 165 impact events using 48 fiber Bragg grating sensors. The framework utilizes a Convolutional Neural Network (CNN) for initial detection and regional classification, which subsequently activates a physics-informed submodel. By embedding linear force-strain priors and geometric constraints, this approach overcomes data scarcity issues common in pure data-driven models. The results validate that this physics-data fusion offers superior robustness and accuracy for aerospace structural health monitoring.
| Original language | English |
|---|---|
| Article number | 113207 |
| Journal | Aerospace Science and Technology |
| Volume | 178 |
| DOIs | |
| State | Published - Nov 2026 |
Keywords
- Composite-reinforced plate
- Convolutional neural network
- Low-speed impact
- Physical information network
- Structural health monitoring
Fingerprint
Dive into the research topics of 'Research on low-velocity impact monitoring method for composite stiffened plates based on FBG and PINNs'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver