Research on structural damage identification based on multi-head self-attention and convolutional neural networks

PDF(2122 KB)

Journal of Vibration and Shock ›› 2022, Vol. 41 ›› Issue (24) : 60-71.

ZHANG Jianfei，HUANG Chaodong，WANG Zifan

Author information +

History +

Abstract

In order to improve the performance of the CNN model for structural damage identification, a CNN model based on multi-head self-attention was proposed, which takes structural vibration acceleration signal as input. The acceleration signal of structural vibration was firstly fed into the model and processed with one-dimensional CNN to extract local features. The multi-head self-attention was then utilized to attend to important information in different positions and different representation subspaces of the input data to learn global features. All the learned features were finally used for structural damage pattern recognition. The results of numerical and shaking table tests of cantilever beam show that compared with CNN model, CNN-LSTM joint model and CNN-BiLSTM joint model, the CNN model based on multi-head self-attention has lower complexicy, easier training, higher damage detecting accuracy, stronger anti-noise capacity and better ability to distinguish damage modes with similar characteristics.
Keywords: deep learning; multi-head self-attention; convolutional neural networks; structural damage identification

Key words

deep learning / multi-head self-attention / convolutional neural networks / structural damage identification

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

ZHANG Jianfei，HUANG Chaodong，WANG Zifan. Research on structural damage identification based on multi-head self-attention and convolutional neural networks[J]. Journal of Vibration and Shock, 2022, 41(24): 60-71

References

[1] 朱宏平, 余璟, 张俊兵. 结构损伤动力检测与健康监测研究现状与展望[J]. 工程力学, 2011, 28(2):1-11, 17.
ZHU Hong-ping, YU Jing, ZHANG Jun-bing. A summary review and advantages of vibration-based damage identification methods in structural heath monitoring [J]. Engineering Mechanics, 2011, 28(2):1-11, 17.
[2] 杨佑发, 熊丽, 陈远. 基于神经网络的框架结构损伤多重分步识别[J]. 建筑科学与工程学报, 2011, 28(1): 106-111.
YANG You-fa, XIONG Li, CHEN Yuan. Multi-stage damage identification for frame structures based on neural network[J]. Journal of Architecture and Civil Engineering, 2011, 28(1): 106-111.
[3] 张刚刚, 王春生, 徐岳. 基于径向基函数神经网络的斜拉桥损伤识别[J]. 长安大学学报(自然科学版), 2006, 25(1): 49-53.
ZHANG Gang-gang, WANG Chun-sheng, XU Yue. Damage detection of cable-stayed bridge based on RBF neural networks[J]. Journal of Chang'an University(Natural Science Edition), 2006, 25(1): 49-53.
[4] 刘寒冰, 焦峪波, 程永春等. 基于模态曲率理论及神经网络的简支梁桥损伤识别[J]. 吉林大学学报(工学版), 2011, 41(4): 963-967
LIU Han-bin, JIAO Yu-bo, CHENG Yong-chun, et al. Damage identification for simply supported beam bridge based on modal curvature theory and neural network[J]. Journal of Jilin University (Engineering and Technology Edition), 2011, 41(4): 963-967.
[5] Avci O, Abdeljaber O, Kiranyaz S, et al. A review of vibration-based damage detection in civil structures: From traditional methods to Machine Learning and Deep Learning applications[J]. Mechanical Systems and Signal Processing, 2021, 147: 107077.
[6] 周飞燕, 金林鹏, 董军. 卷积神经网络研究综述[J]. 计算机学报, 2017, 40(7): 1-23.
ZHOU Fei-yan, JIN Lin-peng, DONG Jun. Review of convolutional neural network [J]. Chinese Journal of Computers, 2017, 40(7): 1-23.
[7] 罗雨舟, 向天宇, 郝柳青. 卷积神经网络在结构损伤检测中的应用[J]. 土木工程与管理学报, 2020, 37(3): 155-161, 173.
LUO Yu-zhou, XIANG Tian-yu, HAO Liu-qing. Application of convolutional neural network in structural damage diagnosis [J]. Journal of Civil Engineering and Management, 2020, 37(3): 155-161, 173.
[8] 李雪松, 马宏伟, 林逸洲. 基于卷积神经网络的结构损伤识别[J]. 振动与冲击, 2019, 38(1): 159-167.
LI Xue-song, MA Hong-wei, LIN Yi-zhou. Structural damage identification based on convolution neural network [J]. Journal of Vibration and Shock, 38(1): 159-167.
[9] 李书进, 赵源, 孔凡等. 卷积神经网络在结构损伤诊断中的应用[J]. 建筑科学与工程学报, 2020, 37(6): 29-37.
LI Shu-jin, ZHAO Yuan, KONG Fan, et al. Application of convolutional neural networks in structural damage identification [J]. Journal of Architecture and Civil Engineering, 2020, 37(6): 29-37.)
[10] Khodabandehlou H, Pekcan G, Sami Fadali M. Vibration-based structural condition assessment using convolution neural networks [J]. tructural Control & Health Monitoring, 2019, 26, e2308.
[11] Lin Y, Nie Z. Structural damage detection with automatic feature-extraction through deep learning [J]. Computer-Aided Civil and Infrastructure Engineering, 2017, 32(12): 1025-1046.
[12] Liu T, Xu H, Ragulskis M, et al. A data-driven damage identification framework based on transmissibility function datasets and one-dimensional convolutional neural networks: verification on a structural health monitoring benchmark structure [J]. Sensors, 2020, 20, 1059.
[13] 杨建喜, 张利凯, 李韧等. 联合卷积与长短记忆神经网络的桥梁结构损伤识别研究[J]. 铁道科学与工程学报, 2020, 17(8): 1893-1902.
YANG Jian-xi, ZHANG Li-kai, Li Ren, et al. Research on bridge structural damage detection based on convolutional and long short-term memory neural networks [J]. Journal of Railway Science and Engineering, 2020, 17(8): 1893-1902.
[14] Yang J, Yang F, Zhou Y, et al. A data-driven structural damage detection framework based on parallel convolutional neural network and bidirectional gated recurrent unit [J]. Information Sciences, 2021, 556: 103-117.
[15] Vaswani A, Shazeer N, Parma N, et al. Attention Is All You Need[C]// Ulrike von Luxburg, Isabelle Guyon. Proceedings of the 31st International Conference on Neural Information Processing Systems. NY, USA: Curran Associates Inc., 2017. 6000-6010.
[16] 曹晖, 林秀萍. 结构损伤识别中噪声的模拟[J]. 振动与冲击, 2010, 29(5): 106-109.
CAO Hui, LIN Xiu-ping. Noise simulation in structural damage identification [J]. Journal of Vibration and Shock, 29(5): 106-109.