基于卷积神经网络的稳定图自动分析方法

苏亮,宋明亮,董石麟

振动与冲击 ›› 2018, Vol. 37 ›› Issue (18) : 59-66.

PDF(1982 KB)
PDF(1982 KB)
振动与冲击 ›› 2018, Vol. 37 ›› Issue (18) : 59-66.
论文

基于卷积神经网络的稳定图自动分析方法

  • 苏亮,宋明亮,董石麟
作者信息 +

Automatic analysis of stabilization diagrams using a convolutional neural network

  • SU Liang,SONG Mingliang,DONG Shilin
Author information +
文章历史 +

摘要

提出一种基于卷积神经网络(Convolutional Neural Network, CNN)的稳定图自动分析方法。在获得不同结构的稳定图之后,按照各自的频率识别精度要求,将稳定图均分成若干个频带,得到单一模态稳定图作为CNN训练样本。然后,通过平移、改变稳定点标记等技术手段对样本进行扩充,再将预处理好的训练样本代入CNN,通过跟踪损失函数在训练过程中变化规律,对如学习率等CNN参数进行调优,最终得到可自动判别稳定图中虚假模态的CNN。以3自由度弹簧质量数值模型、7自由度弹簧质量数值模型、以及一座钢筋混凝土框架结构大楼、瑞士Z24桥加速度实测数据验证了所搭建CNN模型的有效性。训练和预测结果表明,搭建的CNN亦可用于其他一般结构的稳定图自动分析,具有一定的通用性。在无需人为提取任何特征参数,也无需设定任何阈值的情况下,即可自动且准确、快速地剔除稳定图上的虚假模态。

Abstract

A convolutional neural network (CNN) methodology was proposed to automatically interpret stabilization diagrams.Once the stabilization diagrams of different structures had been obtained, they were equally distributed into several frequency bands according to the accuracy requirements of each frequency identification, which were called single mode stabilization diagrams.These frequency bands samples were used as learning samples of the CNN.After that, these learning samples were expanded with some technical methods such as translating and changing the label of stable poles on the stabilization diagram.Then, the preprocessed learning samples were substituted into the constructed CNN.The parameters of CNN, such as learning ratio, were tuned by tracking the changing rule of losing function during the learning process.Finally, a CNN which can automatically eliminate the spurious modes on the stabilization diagram was obtained.The constructed CNN was verified by a 3 degree of freedom(DOF), a 7DOF spring-mass model as well as the accelerometer data of a reinforced concrete frame structure and the Swiss Z24 bridge.The robust learning and prediction results prove that the constructed CNN is effective for analyzing any stabilization diagram of different structures.It can automatically and accurately eliminate the spurious modes on the stabilization diagram immediately without  extracting any characteristic parameters or setting any thresholds of them.
 

关键词

自动识别 / 卷积神经网络 / 稳定图 / 虚假模态 / 模态参数 / 深度学习

Key words

 automatic identification / convolutional neural network / stabilization diagram / spurious modes / modal parameter / deep learning

引用本文

导出引用
苏亮,宋明亮,董石麟. 基于卷积神经网络的稳定图自动分析方法[J]. 振动与冲击, 2018, 37(18): 59-66
SU Liang,SONG Mingliang,DONG Shilin. Automatic analysis of stabilization diagrams using a convolutional neural network[J]. Journal of Vibration and Shock, 2018, 37(18): 59-66

参考文献

[1] 吴智深, 张建. 结构健康监测先进技术及理论[M]. 北京: 科学出版社, 2015.
[2] H. Van der Auweraer, “Structural Dynamics Modeling using Modal Analysis: Applications, Trends and Challenges,” Proceedings of the 2001 IEEE Instrumentation and Measurement Technology Conference, Budapest, Hungary, 2001.
[3] Rainieri C, Fabbrocino G, Cosenza E. Automated Operational Modal Analysis as Structural Health Monitoring Tool: Theoretical and Applicative Aspects[J]. Key Engineering Materials, 2007, 347(347):479-484.
[4] Phillips A W, Allemang R J. Additional mechanisms for providing clear stabilization (consistency) diagrams[C]// Proceedings, International Conference on Noise and Vibration Engineering (ISMA). 2008: 15.
[5] Auweraer H V D, Peeters B. Discriminating physical poles from mathematical poles in high order systems: use and automation of the stabilization diagram[C]// Instrumentation and Measurement Technology Conference, 2004. Imtc 04. Proceedings of the, IEEE. IEEE Xplore, 2004:2193-2198 Vol.3.
[6] S. Chauhan, D. Tcherniak, Clustering approaches to automatic modal parameter estimation, in: Proceedings of the 27th SEM International Modal Analysis Conference, Orlando, FL, USA, 2009.
[7] Deraemaeker A, Reynders E, De Roeck G, et al. Vibration-based structural health monitoring using output-only measurements under changing environment [J]. Mechanical systems and signal processing, 2008, 22(1): 34-56.
[8] 刘进明, 应怀樵, 章关永. OMA模态参数的优化及盲分析技术探讨[J]. 振动、测试与诊断, 2012, 32(6):1016-1020.
Liu Jin-ming, Ying Huai-qiao, Zhang Guan-yong. Optimization of OMA modal parameter and technology study of blind analysis [J].  Journal of Vibration Measurement & Diagnosis, 2012, 32(6):1016-1020.
[9] 徐琪泽, 吴金志, 张毅刚. 基于振型相关性的结构模态参数频域自动识别[J]. 建筑结构, 2015(5):39-43.
Xu Qi-ze, Wu Jin-Zhi, Zhang Yi-gang. An automated frequency domain modal identification method based on modal assurance criterion [J]. Building Structure: 2015(5):39-43.
[10] Magalhaes F, Cunha A, Caetano E. Online automatic identification of the modal parameters of a long span arch bridge [J]. Mechanical Systems and Signal Processing, 2009, 23(2): 316-329.
[11] 孙国富. 基于模糊聚类的模态参数自动识别 [J]. 振动与冲击, 2010, 29(9): 86-88.
Sun Guo-fu. Automatic modal parameters identification based on fuzzy clustering [J]. Journal of vibration and shock, 2010, 29(9): 86-88.
[12] 汤宝平, 章国稳, 陈卓. 基于谱系聚类的随机子空间模态参数自动识别[J]. 振动与冲击, 2012, 31(10): 92-96.
Tang Bao-ping, Zhang Guo-wen, Chen Zhuo. Automatic stochastic subspace identification of modal parameters based on hierarchical clustering method [J]. Journal of vibration and shock, 2012, 31(10): 92-96.
 [13] 苏亮, 宋明亮,董石麟,等. 循环遗传聚类法稳定图自动分析[J]. 浙江大学学报工学版, 2017, 51(3):24-33.
Su Liang, Song Ming-liang, Dong Shi-lin, et al. Automatic analysis of stabilization diagram using iterative genetic-fuzzy clustering method [J]. Journal of Zhejiang University: Engineering Science, 2017, 51(3):24-33.
[14] 孙鑫晖, 张令弥. 宽频带模态识别算法中极点的自动选取[J]. 地震工程与工程振动学报, 2009, 29(1): 130-134.
Sun Xin-hui, Zhang ling-mi. Automatic pole selection for broad band modal identification [J]. Journal of Earthquake Engineering and Engineering Vibration, 2009, 29(1): 130-134.
[15] Hinton G E, Osindero S, Teh Y W. A fast learning algorithm for deep belief nets.[J]. Neural Computation, 2006, 18(7): 1527-1554.
[16] 李玉鑑 张婷. 深度学习导论及案例分析[M]. 北京: 机械工业出版社, 2016.
[17] 李彦冬, 郝宗波, 雷航. 卷积神经网络研究综述[J]. 计算机应用, 2016, 36(9):2508-2515.
Li Yan-dong, Hao Zong-bo, Lei Hang. Survey of convolutional neural network [J]. Journal of Computer Applications, 2016, 36(9):2508-2515.
[18] Hubel D H, Weisel T N. Receptive fields, binocular interaction and functional architecture in cat's visual cortex [J]. Journal of Physiology: 1962, 160, 106-154.
[19] Zeiler M D, Fergus R. Stochastic Pooling for Regularization of Deep Convolutional Neural Networks[J]. Eprint Arxiv, 2013: 1301,3557-3566.
[20] 贾世杰, 杨东坡, 刘金环. 基于卷积神经网络的商品图像精细分类[J]. 山东科技大学学报(自然科学版), 2014, 33(6):91-96.
Jia Shi-jie, Yang Dong-po, Liu Jin-huan. Product Image Fine-grained Classification Based on Convolutional Neural Network [J]. JournaI of Shandong University of Science and Technology, 2014, 33(6):91-96.
[21] Hinton G E, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. Computer Science, 2012, 3(4): 212-223.
[22] 章国稳, 汤宝平, 潘飞. 特征系统实现算法的虚假模态剔除方法[J]. 重庆大学学报, 2012, 35(3):20-25.
Zhang Guo-wen, Tang Bao-ping, Pan Fei. The method for removing the spurious modes of eigensystem realization algorithm [J]. Journal of Chongqing University(Natural Science Edition), 2012, 35(3):20-25.
[23] 章国稳, 汤宝平, 孟利波. 基于特征值分解的随机子空间算法研究[J]. 振动与冲击, 2012, 31(7):74-78.
Zhang Guo-wen, Tang Bao-ping, Meng Li-bo. Improved stochastic subspace identification algorithm based on eigendecomposition [J]. Journal of Vibration and Shock, 2012, 31(7):74-78.
[24] Brincker R, Ventura C. Introduction to operational modal analysis[M]. Chichester: John Wiley & Sons Ltd, 2015.
[25] MAECK J, DE ROECK G. Description of Z24 benchmark [J]. Mechanical Systems and Signal Processing, 2003, 17(1): 127-131.
[26] Mohanty P, Reynolds P, Pavic A. Automated interpretation of stability plots for analysis of a non-stationary structure[C]//25th International Modal Analysis Conference (IMAC XXV). Orlando, Florida USA. Society for Experimental Mechanics, 2007.
[27] Reynders E, De Roeck G. Reference-based combined deterministic–stochastic subspace identification for experimental and operational modal analysis [J]. Mechanical Systems and Signal Processing, 2008, 22(3): 617-637.

PDF(1982 KB)

Accesses

Citation

Detail

段落导航
相关文章

/