基于卷积神经网络的结构损伤识别

PDF(3999 KB)

振动与冲击 ›› 2019, Vol. 38 ›› Issue (1) : 159-167.

论文

基于卷积神经网络的结构损伤识别

李雪松 1 ，马宏伟 1,2，林逸洲 3

作者信息 +

Structural damage identification based on convolution neural network

LI Xuesong1, MA Hongwei1,2, LIN Yizhou3

Author information +

文章历史 +

摘要

为解决结构的健康监测问题，找到合适的结构损伤识别特征，本文使用卷积神经网络提取结构特征来识别损伤，并通过IASC-ASCE SHM Benchmark第一阶段模拟数据验证其有效性，同时与小波包频带能量特征、前五阶本征模态函数能量特征做同分类器准确率对比，证明了卷积神经网络在自动提取特征方面的优势。在分析卷积神经网络自动提取特征的鲁棒性时，发现单一噪声数据训练的特征抗噪能力有一定局限性，为了获得更好的特征抗噪能力，本文提出混合噪声训练模式，验证了含噪声0%-50%的样本数据，均取得良好识别结果。同时在进行卷积核特征可视化工作中发现，混噪模式训练的卷积核能够识别更多阶次的频率信息。

Abstract

Here, a convolution neural network was used to extract structural features, identify damage and solve problems of structural damage identification.The effectiveness of this method was verified with IASC-ASCE SHM Benchmark Phase 1 simulation data.Then, comparing the same classifier accuracies for energy characteristics of the convolution neural network, the wavelet packet and the first 5 IMFs obtained by EMD, advantages of the convolution neural network in automatically extracting features were proved.In analyzing the robustness of features’ automatic extraction of the convolution neural network, it was found that the characteristic anti-noise ability of a single noise data training mode is limited.In order to acquire the better characteristic anti-noise ability, a mixed noise training mode was proposed.The validity of this training mode was verified using the sample data with noise of 0%—50% to obtain good recognition results.At the same time, it was found in visualization of the convolution’s kernel features that the convolution kernel of the mixed noise training mode can identify more orders of frequency information.

导出引用

李雪松 1,马宏伟 1,2，林逸洲 3. 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 Xuesong1, MA Hongwei1,2, LIN Yizhou3. Structural damage identification based on convolution neural network[J]. Journal of Vibration and Shock, 2019, 38(1): 159-167

参考文献

[1] 马宏伟, 聂振华. 桥梁安全监测最新研究进展与思考[J]. 力学与实践, 2015, 37(2):161-170.
MA Hongwei, NIE Zhenhua. RECENT ADVANCES AND REVIEW OF BRIDGE SAFETY MONITORING[J]. Mechanics in Engineering, 2015, 37(2):161-170.
[2] 赵俊, 程良彦, 马宏伟. 基于曲率模态的拱板结构损伤识别[J]. 暨南大学学报(自然科学与医学版), 2008, 29(5):470-477.
Zhaojun, Chengliangyan, Ma Hongwei. Identification of Arch Structure Damage Based on Curvature Modes[ J] .Journal of Jinan University(Natural Science and Medicine), 2008, 29(5): 470-477.
[3] 聂振华, 程良彦, 马宏伟. 基于结构动力特性的损伤检测可视化方法[J]. 振动与冲击, 2011, 30(12):7-13.
Nie Z H, Cheng L Y, Hong-Wei M A. Visualization method for structural damage detection based on its dynamic characteristics[J]. Journal of Vibration & Shock, 2011, 30(12):7-13.
[4] 王志华, 张向东, 马宏伟. 基于应变曲率法的悬臂梁多位置损伤识别实验研究[C]// 全国结构工程学术会议. 2003.
[5] Peeters B,De roeck G. One‐year Monitoring of the Z24‐bridge: Environmental Effects Versus Damage Events[J]. Earthquake Engineering & Structural Dynamics, 2015, 30(2): 149-171.
[6] 朱劲松, 孙雅丹. 基于小波包能量的桥梁损伤识别指标[J]. 振动、测试与诊断, 2015, 35(4):715-721.
Zhu J, Sun Y. Wavelet packet energy based damage detection index for bridge[J]. Zhendong Ceshi Yu Zhenduan/journal of Vibration Measurement & Diagnosis, 2015, 35(4):715-721.
[7] 梁永涛. 移动荷载作用下基于HHT和相空间重构法的桥梁损伤识别[D]. 暨南大学, 2014.
[8] 瞿益丹. 基于HHT和SVM的流动轴承故障振动信号的诊断研究[D]. 中南大学, 2012.
[9] 肖书敏, 闫云聚, 姜波澜. 基于小波神经网络方法的桥梁结构损伤识别研究[J]. 应用数学和力学, 2016, 37(2):149-159.
Xiao S M, Yan Y J, Jiang B L. Damage Identification for Bridge Structures Based on the Wavelet Neural Network Method[J]. Applied Mathematics & Mechanics, 2016.
[10] 管德清, 廖俊文, 吴兆. 基于应变模态小波神经网络的结构损伤识别方法[J]. 吉首大学学报(自科版), 2015, 36(3):45-51.
Guan D Q. Structure Damage Identification Method by the Wavelet-Neural Network Analysis of Strain Mode[J]. 2015.
[11] Hinton GE,Osindero S,Teh YW. A Fast Learning Algorithm for Deep Belief Nets[J]. Neural Computation, 2006, 18(7): 1527-1554.
[12] 孙志军, 薛磊, 许阳明,等. 深度学习研究综述[J]. 计算机应用研究, 2012, 29(8):2806-2810.
Sun Z J, Xue L, Yang-Ming X U, et al. Overview of deep learning[J]. Application Research of Computers, 2012.
[13] Abdel-hamid O,Mohamed AR,Jiang H, et al. Applying Convolutional Neural Networks Concepts to Hybrid Nn-hmm Model for Speech Recognition[C]//Ieee International Conference on Acoustics, Speech and Signal Processing, [S.l.]: [s.n.], 2012: 4277-4280.
[14] Huang NE. Huang, N.e., Et Al.: the Empirical Mode Decomposition and the Hilbert Spectrum for Nonlinear and Non-stationary Time Series Analysis. Proc. R. Soc. Lond. a 454, 903-995[J]. Proceedings of the Royal Society a Mathematical Physical & Engineering Sciences, 1998, 454(1971): 903-995.
[15] Hubel DH,Wiesel TN. Receptive Fields, Binocular Interaction and Functional Architecture in the Cat's Visual Cortex.[J]. Journal of Physiology, 1962, 160(1): 106.
[16]LeCun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86(11), 2278-2324.
[17] 焦李成. 神经网络系统理论[M].西安电子科技大学出版社,1990.
[18] 何鹏程. 改进的卷积神经网络模型及其应用研究 [D].大连理工大学, 2015.
[19] 张兢, 路彦和. 基于小波包频带能量检测技术的故障诊断[J]. 微计算机信息, 2006, 22(4):202-204.
Chongqing, Zhang, Lu J, et al. On the Fault Diagnosis based on the Detecting Technology of Wavelet Packet Frequency Band Energy[J]. 2006.
[20] Andrychowicz, M., Denil, M., Gomez, S., Hoffman, M. W., Pfau, D., Schaul, T., & De Freitas, N. (2016). Learning to learn by gradient descent by gradient descent. In Advances in Neural Information Processing Systems (pp. 3981-3989).
[21] Ioffe, S., & Szegedy, C. (2015). Batch normalization: Accelerating deep network training by reducing internal covariate shift. arXiv preprint arXiv:1502.03167.
[22]Clevert, D. A., Unterthiner, T., & Hochreiter, S. (2015). Fast and accurate deep network learning by exponential linear units (elus). arXiv preprint arXiv:1511.07289.
[23] Gimpel K, Smith N. Softmax-margin CRFs: Training log-linear models with cost functions[C]// Human Language Technologies: Conference of the North American Chapter of the Association of Computational Linguistics, Proceedings, June 2-4, 2010, Los Angeles, California, USA. DBLP, 2010:733-736.
[24] Johnson EA,Lam HF,Katafygiotis LS, et al. Phase I Iasc-asce Structural Health Monitoring Benchmark Problem Using Simulated Data[J]. Journal of Engineering Mechanics, 2004, 130(1): 3-15.
[25] Lee VW,Kim C,Chhugani J, et al. Debunking the 100x Gpu Vs. Cpu Myth:an Evaluation of Throughput Computing on Cpu and Gpu[J]. Acm Sigarch Computer Architecture News, 2010, 38(3): 451-460.