基于声发射信号时频图深度学习的桥梁钢桁架焊接节点损伤程度识别

李丹1 2,沈鹏1,贺文宇1,向抒林3

振动与冲击 ›› 2024, Vol. 43 ›› Issue (1) : 107-115.

PDF(3591 KB)
PDF(3591 KB)
振动与冲击 ›› 2024, Vol. 43 ›› Issue (1) : 107-115.
论文

基于声发射信号时频图深度学习的桥梁钢桁架焊接节点损伤程度识别

  • 李丹1 2,沈鹏1,贺文宇1,向抒林3
作者信息 +

Identification of damage degree of welded joints in bridge steel trusses based on deep learning of time-frequency maps of AE signals

  • LI Dan1,2, SHEN Peng1, HE Wenyu1, XIANG Shulin3
Author information +
文章历史 +

摘要

针对桥梁钢桁架疲劳损伤识别难度大、精度低的现状,提出基于声发射信号时频分析与深度学习的钢桁架焊接节点损伤程度识别方法。对桁架节点在桥梁运营状态下产生的声发射信号进行小波变换,表征不同损伤程度信号的时频能量分布模式,然后建立卷积神经网络(convolutional neural network,CNN)模型对时频图进行损伤特征提取,并通过迁移学习思想提升模型的训练效率和学习能力,从而实现桁架焊接节点严重损伤、轻微损伤和噪声工况的准确识别。进一步对模型各卷积层激活区域进行可视化分析,解剖模型的损伤特征学习过程及分类逻辑。某悬索桥中央纵向腹板钢桁架焊接节点现场试验结果表明:相较于利用时域波形进行特征学习的一维卷积神经网络模型,时频图包含了更丰富的损伤信息,所建立的二维卷积神经网络模型对钢桁架焊接节点三种损伤程度的识别准确率超过94%,具有更强鲁棒性和实际应用价值。

Abstract

In view of the difficulty in accurate detection of fatigue damages in bridge steel trusses, this study proposes a method to identify the damage degree of welded truss joints based on time-frequency analysis and deep learning of acoustic emission signals. The acoustic emission signals generated by the truss joints during operation are firstly analyzed by wavelet transform to characterize their energy distribution patterns in the time-frequency domain for different damage degrees. After that, a convolutional neural network (CNN) model is established to extract damage features from the time-frequency diagrams. The training efficiency and learning ability of the model are improved through transfer learning. Accurate identification of severe damage, minor damage, and intact cases of the truss joints can then be realized. Further, the activation areas in each convolution layer of the model are visualized to reveal the damage feature learning process and classification logic. Filed test was carried out on the central longitudinal steel truss web of a suspension bridge. The results showed that compared with the one-dimensional CNN model using time-domain waveforms of acoustic emission signals for feature learning, the two-dimensional CNN, taking time-frequency diagrams that contained more abundant damage information as the input, achieved an accuracy of more than 94% in identifying the three damage degrees of the truss joints. It behaved with higher robustness and potential for practical applications.

关键词

钢桁架 / 焊接节点 / 损伤程度 / 声发射 / 时频分析 / 深度学习

Key words

Steel trusses / welded joints / damage degree / acoustic emission / time-frequency analysis / deep learning

引用本文

导出引用
李丹1 2,沈鹏1,贺文宇1,向抒林3. 基于声发射信号时频图深度学习的桥梁钢桁架焊接节点损伤程度识别[J]. 振动与冲击, 2024, 43(1): 107-115
LI Dan1,2, SHEN Peng1, HE Wenyu1, XIANG Shulin3. Identification of damage degree of welded joints in bridge steel trusses based on deep learning of time-frequency maps of AE signals[J]. Journal of Vibration and Shock, 2024, 43(1): 107-115

参考文献

[1] 张清华, 劳武略, 崔闯, 等. 钢结构桥梁疲劳2020年度研究进展[J]. 土木与环境工程学报(中英文), 2021, 43(S1): 79-90. ZHANG Qinghua, LAO Wulue, CUI Chuang, et al. State-of-the-art review of fatigue of steel bridge in 2020[J]. Journal of Civil and Environmental Engineering (Chinese and English), 2021, 43(S1): 79-90. [2] BAO Y Q, CHEN Z C, WEI S Y, et al. The state of the art of data science and engineering in structural health monitoring[J]. Engineering, 2019, 5(2): 234-242. [3] 唐蕾, 黄天立, 万熹. 基于变分模态分解和同步提取变换识别时变结构瞬时频率[J]. 振动与冲击, 2022, 41(6): 197-205. TANG Lei, HUANG Tianli, WAN Xi. Instantaneous frequency identification of time-varying structures using variational mode decomposition and synchroextracting transform[J]. Journal of Vibration and Shock, 2022, 41(6): 197-205. [4] 贺文宇, 丁绪聪, 任伟新. 环境激励下移动车辆对桥梁模态参数识别的影响研究[J]. 振动与冲击, 2021, 40(03): 48-53. He Wenyu, DING Xucong, REN Weixin. Effects of moving vehicle on bridge modal parametric identification under ambient excitation[J]. Journal of Vibration and Shock, 2021, 40(03): 48-53. [5] LI S L, KANG Z Z, Wu G M, et al. Acoustic emission-based transition monitoring of mechanical mechanism for bolted shear connection in GFRP–UHPC hybrid beams[J]. Measurement, 2022, 198: 111358. [6] MCCRORY J P, VINOGRADOV A, PEARSON M R, et al. Acoustic Emission Monitoring of Metals[M]. Acoustic Emission Testing, 2022: 529-565. [7] SILVERSIDES I, MASLOUHI A, LAPLANTE G. Acoustic emission monitoring of interlaminar delamination onset in carbon fibre composites[J]. Structural Health Monitoring, 2013, 12(2): 126-140. [8] DROUBI M G, FAISAL N H, ORR F, et al. Acoustic emission method for defect detection and identification in carbon steel welded joints[J]. Journal of Constructional Steel Research, 2017, 134: 28-37. [9] MEGID W A, CHAINEY M A, LEBRUN P, et al. Monitoring fatigue cracks on eyebars of steel bridges using acoustic emission: A case study[J]. Engineering Fracture Mechanics, 2019, 211: 198-208. [10] LI D, NIE J H, REN W X, et al. A novel acoustic emission source location method for crack monitoring of orthotropic steel plates[J]. Engineering Structures, 2022, 253: 113717. [11] RAJEEV P, WIJESUNDARA K K. Energy-based damage index for concentrically braced steel structure using continuous wavelet transform[J]. Journal of Constructional Steel Research, 2014, 103: 241-250. [12] 孙增寿, 范科举. 基于提升小波熵指标的梁板组合桥损伤识别研究[J]. 振动与冲击, 2012, 31(11): 114-117. SUN Zenshou, FAN Keju. Damage detection for a gird-slab combined bridge based on lifting wavelet entropy indexes[J]. Journal of Vibration and Shock, 2012, 31(11): 114-117. [13] KUANG K S C, Li D, KOH C G. Acoustic emission source location and noise cancellation for crack detection in rail head[J]. Smart Struct Syst, 2016, 18(5): 1063-1085. [14] ZHAO R, YAN R Q, CHEN Z H, et al. Deep learning and its applications to machine health monitoring[J]. Mechanical Systems and Signal Processing, 2019, 115: 213-237. [15] KHAN S, YAIRI T. A review on the application of deep learning in system health management[J]. Mechanical Systems and Signal Processing, 2018, 107: 241-265. [16] HESSER D F, MOSTAFAVI S, KOCUR G K, et al. Identification of acoustic emission sources for structural health monitoring applications based on convolutional neural networks and deep transfer learning[J]. Neurocomputing, 2021, 453: 1-12. [17] BAO Y Q, TANG Z Y, LI H, et al. Computer vision and deep learning–based data anomaly detection method for structural health monitoring[J]. Structural Health Monitoring, 2018, 18(2): 401-421. [18] LI D, WANG Y, YAN W J, et al. Acoustic emission wave classification for rail crack monitoring based on synchrosqueezed wavelet transform and multi-branch convolutional neural network[J]. Structural Health Monitoring, 2020, 20(4): 1563-1582. [19] GUO J H, LIU C, CAO J F, et al. Damage identification of wind turbine blades with deep convolutional neural networks[J]. Renewable Energy, 2021, 174: 122-133. [20] 骆勇鹏, 王林堃, 郭旭, 等. 利用单传感器数据基于GAF-CNN的结构损伤识别[J]. 振动、测试与诊断, 2022, 42(01): 169-176+202-203. LUO Yongpeng, WANG Linkun, GUO Xu. et al. Structural damage identification using single sensor data based on GAF-CNN[J]. Journal of Vibration Measurement and Diagnosis, 2022, 42(01): 169-176+202-203. [21] RIOUL O, DUHAMEL P. Fast algorithms for discrete and continuous wavelet transforms[J]. IEEE Transactions on Information Theory, 1992, 38(2): 569-586. [22] 闫维明, 马裕超, 何浩祥, 等. 双线性时频分布交叉项提取及损伤识别应用[J]. 振动、测试与诊断, 2014, 34(6): 1014-1021. YAN Weiming, MA Yuchao, HE Haoxiang, et al. Cross term extraction of bilinear time-frequency distribution and its application to damage identification[J]. Journal of Vibration Measurement and Diagnosis, 2014, 34(6): 1014-1021. [23] ZITTO M E, PIOTRKOWSKI R, GALLEGO A, et al. Damage assessed by wavelet scale bands and b-value in dynamical tests of a reinforced concrete slab monitored with acoustic emission[J]. Mechanical Systems and Signal Processing, 2015, 60: 75-89. [24] LI D, KUANG K S C, KOH C G. Fatigue crack sizing in rail steel using crack closure-induced acoustic emission waves[J]. Measurement Science and Technology, 2017, 28(6): 065601. [25] HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. science, 2006, 313(5786): 504-507. [26] 张义清, 谭继文, 孟庆文,等. 基于迁移学习的钢丝绳断丝定量检测方法[J]. 振动与冲击, 2022, 41(12): 261-266. ZHANG Yiqin, TAN Jiwen, MENG Qingwen, et al. A quantitative testing method for broken wires in steel wire ropes based on transfer learning[J]. Journal of Vibration and Shock, 2022, 41(12): 261-266. [27] KINGMA D P, BA J. Adam: A Method for Stochastic Optimization[J]. Computer Science, 2014. [28] SUN Y, DING S, ZHANG Z, et al. An improved grid search algorithm to optimize SVR for prediction[J]. Soft Computing, 2021, 25: 5633-5644. [29] WON J, PARK J W, JANG S, et al. Automated structural damage identification using data normalization and 1-dimensional convolutional neural network[J]. Applied Sciences, 2021, 11(6): 2610-2623.

PDF(3591 KB)

Accesses

Citation

Detail

段落导航
相关文章

/