.png){kind=icon}
返回
Physics-Augmented Spatial-Temporal graph convolutional network for damage localization using Ultrasonic guided waves
.png){kind=icon}
PRE
AI
DOI：10.1016/j.ymssp.2024.111738.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Structural health monitoring technology based on ultrasonic guided waves (UGW) is a potential application for damage detection in plate-like composite structures. This paper introduces a novel physics-augmented spatial-temporal graph convolutional network (P-STGCN) for the damage detection task in composite plate. Firstly, the graph topology containing edge weight attributes is constructed based on the spatial correlation of the sensors and the UGW scattering principle, which strengthens the expression of the data spatial information of the data. Then, a spatial feature extractor and a temporal feature extractor are constructed based on graph convolutional network and convolutional neural network, respectively. Further, the spatial-temporal fusion method is proposed for the first time in the field of UGW damage detection, which enables the perception of multi-level damage information. In addition, a sample compensation attention module based on wavefield scattering energy is designed to compensate for the samples' spatial distribution. Finally, a dataset was constructed in a composite plate containing only four sensors to evaluate the model's performance. The results show that the average localization error is 10.36 mm. Compared with other deep learning localization methods, the P-STGCN model is superior in both localization accuracy and computational resources.
Keyword:
Composite structures
Ultrasonic guided wave
Damage localization
Graph convolutional networks
Convolutional neural networks
期刊
IF:
8.9
论文数:
1.3W
被引数:
6.6W
机构
引用论文
Optimization of multi-directional fiber architecture for resistance and ductility of bolted FRP profile joints.png)
COMPOSITE STRUCTURES
COMPOSITE STRUCTURESIF7.1
PREAI
分享收藏A novel machine learning model to design historical-independent health indicators for composite structures一种新颖的机器学习模型，用于设计复合结构的历史独立健康指标
OAAI
OAAI分享收藏Damage localization for composite structure using guided wave signals with Gramian angular field image coding and convolutional neural networksCOMPOSITE STRUCTURES
COMPOSITE STRUCTURESIF7.1
PREAI
分享收藏Full waveform inversion guided wave tomography with a recurrent neural network基于递归神经网络的全波形反演导波层析成像ULTRASONICS
ULTRASONICSIF4.1
PREAI
分享收藏Damage Detection and Characterization in Composites Using a Geometric Modification of the RAPID AlgorithmIEEE SENSORS JOURNALOAAI
IEEE SENSORS JOURNALIF4.5
OAAI分享收藏
查看更多内容