基于多通道一维卷积神经网络的建筑结构损伤识别

PDF(1538 KB)

振动与冲击 ›› 2024, Vol. 43 ›› Issue (24) : 216-224.

论文

基于多通道一维卷积神经网络的建筑结构损伤识别

熊青松1,2,3，熊海贝1,2，袁程4，陈琳1,2，孔庆钊1, 2

作者信息 +

Damage identification of building structures based on multi-channel one-dimensional convolutional neural networks

XIONG Qingsong1,2,3,XIONG Haibei1,2,YUAN Cheng4,CHEN Lin1,2,KONG Qingzhao1,2

Author information +

文章历史 +

摘要

面向建筑结构实际工程中监测数据获取范式以及现阶段相关数据驱动算法的局限性，提出一种基于多通道一维卷积神经网络（1D convolutional neural network,,1D CNN）的建筑结构损伤识别算法。该方法通过对建筑各楼层测点振动数据进行融合特征提取，利用多通道1D CNN隐式提取测点间拓扑关系，抽取信号高维损伤特征，从而实现建筑结构损伤的有效识别。基于IASC-ASCE benchmark模型，验证所提方法的有效性。并与传统单通道1D CNN进行对比分析，结果表明其在识别精度上实现明显提升，预测精度达0.989。同时，模型训练过程中引入监视机制，大幅提升模型训练效率，对比分析表明所提出多通道模型架构训练收敛更快，结构损伤特征抽取更为稳定。

Abstract

In addressing the paradigm of acquiring monitoring data in practical engineering for building structures and the limitations of current data-driven algorithms, a building structure damage identification algorithm based on a multi-channel one-dimensional convolutional neural network (1D CNN) is proposed. This method merges feature extraction from vibration data of various building floors, implicitly utilizes multi-channel 1D CNN to extract topological relationships between monitoring points, and extracts high-dimensional damage features from signals to achieve effective identification of building structural damage. Validity of the proposed approach is verified using the IASC-ASCE benchmark model. Comparative analysis with traditional single-channel 1D CNN demonstrates a significant improvement in identification accuracy, with a predictive accuracy reaching 0.989. Furthermore, the introduction of monitoring mechanisms during model training greatly enhances training efficiency. Comparative analysis indicates that the proposed multi-channel model architecture converges faster during training and exhibits more stable extraction of structural damage features.

导出引用

熊青松1, 2, 3, 熊海贝1, 2, 袁程4, 陈琳1, 2, 孔庆钊1, 2. 基于多通道一维卷积神经网络的建筑结构损伤识别[J]. 振动与冲击, 2024, 43(24): 216-224

XIONG Qingsong1, 2, 3, XIONG Haibei1, 2, YUAN Cheng4, CHEN Lin1, 2, KONG Qingzhao1, 2. Damage identification of building structures based on multi-channel one-dimensional convolutional neural networks[J]. Journal of Vibration and Shock, 2024, 43(24): 216-224

参考文献

[1] 徐阳, 金晓威, 李惠. 土木工程智能科学与技术研究现状及展望[J]. 建筑结构学报, 2022, 43(09): 23-35.
XU Yang, JIN Xiao-wei, LI Hui. State-of-the-art and prospect of intelligent science and technology in civil engineering[J]. Journal of Building Structures, 2022, 43(09): 23-35.
[2] 鲍跃全,李惠.人工智能时代的土木工程[J].土木工程学报,2019,52(05):1-11.
BAO Yue-quan, LI Hui. Artificial Intelligence for civil engineering[J]. China Civil Engineering Journal. 2019,52(05):1-11.
[3] Johnson EA, Lam H-F, Katafygiotis LS, Beck JL. Phase I IASC-ASCE structural health monitoring benchmark problem using simulated data. Journal of engineering mechanics. 2004;130(1):3-15.
[4] Dang HV, Raza M, Nguyen TV, et al. Deep learning-based detection of structural damage using time-series data[J]. Structure and Infrastructure Engineering. 2021,17(11):1474-1493.
[5] Xiong Q, Yuan C, He B, et al. GTRF: A general deep learning framework for tuples recognition towards supervised, semi-supervised and unsupervised paradigms[J]. Engineering Applications of Artificial Intelligence. 2023,124:106500.
[6] Sun, H, Li S, Yu. A deep learning-based bridge damage detection and localization method [J]. Mechanical Systems and Signal Processing, 2023, 193:110277.
[7] Wang H, Barone G, Smith A. A novel multi-level data fusion and anomaly detection approach for infrastructure damage identification and localization [J]. Engineering Structures. 2023, 292:116473.
[8] Toufigh V, Ranjbar I. Unsupervised deep learning framework for ultrasonic-based distributed damage detection in concrete: integration of a deep auto-encoder and Isolation Forest for anomaly detection. Structural Health Monitoring [J]. 2023,0(0).
[9] Wang X, Du Y, Zhou X, et al. Data anomaly detection through semi-supervised learning aided by customized data augmentation techniques[J]. Structural Control and Health Monitoring. 2023, 22;2023.
[10] 阳洋, 王者伟, 凌园,等. 基于统计矩理论和贝叶斯方法的框架结构损伤识别方法研究[J]. 建筑结构学报,2023, 44(02):217-226.
YANG Yang, WANG Zhe-wei, LING Yuan, et al. Damage identification method of frame structure based on statistical moment theory and Bayesian method[J]. Journal of Building Structures, 2023, 44(02): 217-226.
[11] 阳洋, 李建雷, 梁晋秋,等. 基于统计矩理论的无模型结构损伤识别方法研究 [J]. 建筑结构学报, 2019, 40(09):196-204.
YANG Yang, LI Jian-lei, LIANG Jin-qiu, et al. Model-free structural damage identification method based on statistical moment theory [J]. Journal of Building Structures, 2019, 40(09): 196-204.
[12] 杨秋伟，孙斌祥．结构损伤识别的改进柔度灵敏度方法研究[J]．振动与冲击，2011，30 (5) : 27-31．
YANG Qiu-wei, SUN Bin-xiang. An improved flexibility sensitivity method for structural damage detection [J]. Journal of Vibration and Shock,2011, 30( 5) : 27-31.
[13] Ji X, Zhuang Y, Miao Z, Cheng Y. Vision‐based seismic damage detection and residual capacity assessment for an RC shaking table test structure. Earthquake Engineering & Structural Dynamics[J]. 2023, 52(3):806-27.
[14] Tavasoli S, Pan X, Yang TY. Real-time autonomous indoor navigation and vision-based damage assessment of reinforced concrete structures using low-cost nano aerial vehicles[J]. Journal of Building Engineering. 2023, 68:106193.
[15] 刘纲,顾嘉伟,李孟珠,等.基于数字图像相关的古建筑彩绘梁损伤识别方法[J].土木与环境工程学报, 2022, 44(02):148-157.
LIU Gang, GU Jia-wei, LI Meng-zhu, et al. Damage identification method of painted beams based on digital image correlation[J]. Journal of Civil and Environmental Engineering, 2021, 41(04):145-156.
[16] 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(0):107077.
[17] 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(01):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,2019,38(01):159-167.
[18] 熊青松, 熊海贝, 孔庆钊,等. 基于振动信号深度表征学习的高层建筑结构振动识别[J]. 习的高层建筑结构振动识别[J/OL]．(2022-11-10)[2024-03-28].https://kns.cnki.net/kcms/detail/11.2595.O3.20221109.1733.008.html.
[19] Kiranyaz S, Avci O, Abdeljaber O, et al. 1D convolutional neural networks and applications: A survey[J]. Mechanical systems and signal processing. 2021,151:107398.
[20] Abdeljaber O, Avci O, Kiranyaz S, Gabbouj M, et al. Real-time vibration-based structural damage detection using one-dimensional convolutional neural networks[J]. Journal of Sound and Vibration. 2017,388:154-70.
[21] 骆勇鹏,王林堃,廖飞宇,等.基于一维卷积神经网络的结构损伤识别[J].地震工程与工程振动,2021,41(04):145-156.
LUO Yong-peng, WANG Lin-kun, LIAO Fei-yu, et al. Vibration-based structural damage identification by 1-dimensional convolutional neural network[J]. Earthquake Engineering and Engineering Dynamics. 2021,41(04):145-156.
[22] Hinton G.E., Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors [J]. arXiv preprint arXiv.2012,1207.0580.
[23] 孙柏涛, 李洋. 中国震害预测工作的沿革与发展[J]. 工程力学, 2021, 38(1): 1-7, 51.
SUN Bai-tao, LI Yang. Evolution and development of earthquake disaster prediction in China [J]. Engineering Mechanics, 2021, 38(1): 1-7, 51.
[24] Gao Y, Zhai P, Mosalam KM. Balanced semisupervised generative adversarial network for damage assessment from low-data imbalanced-class regime [J]. Computer‐Aided Civil and Infrastructure Engineering. 2021, 36(9):1094-113.
[25] Yao Y, Rosasco L, Caponnetto A. On early stopping in gradient descent learning [J]. Constructive Approximation. 2007,26:289-315.
[26] Kingma DP, Ba J. Adam: A method for stochastic optimization[J]. arXiv preprint.2014:1412.6980.
[27] Manavski A, Valle G. CUDA compatible GPU cards as efficient hardware accelerators for Smith-Waterman sequence alignment [J]. BMC Bioinformatics, 2008, 9(2): S10.