基于域自适应对抗生成样本的金属损伤导波智能迁移识别方法

PDF(4820 KB)

振动与冲击 ›› 2025, Vol. 44 ›› Issue (3) : 191-201.

故障诊断分析

基于域自适应对抗生成样本的金属损伤导波智能迁移识别方法

王莉1，2，刘国强2，杨宇2，张超1，裘进浩*1

作者信息 +

Intelligent transfer recognition method for metal damage with guided wave based on domain adaptive adversarial generated samples

WANG Li1,2, LIU Guoqiang2, YANG Yu2, ZHANG Chao1, QIU Jinhao*1

Author information +

文章历史 +

摘要

针对工程场景下缺乏大量标注完备的真实损伤样本，而难以学习到可用的智能诊断模型的难题，本文提出了一种基于域自适应对抗生成样本的金属损伤导波智能迁移诊断方法。首先，采用有限元仿真得到了大量标签丰富的模拟损伤导波监测数据，然后采用生成对抗神经网络(Wasserstein Generative Adversarial Networks with Gradient penalty, WGAN-GP)实现了模拟损伤监测样本至真实损伤的域自适应对抗样本的生成，最后构建了基于对抗生成样本的损伤智能诊断模型，实现了对未知标签真实损伤监测样本的高可靠分类诊断。金属开孔结构疲劳裂纹导波监测试验验证结果表明：所提方法可实现模拟损伤导波识别知识至疲劳损伤的跨域迁移，且在无真实损伤标注样本时也可实现对裂纹损伤的高精度智能识别。

Abstract

This paper proposes an intelligent migration detection method for fatigue crack using guided wave based on adversarial synthetic samples with domain adaptation, to solve the problem that insufficient labeled data in engineering application limits to train a reliable intelligent damage detection model. Firstly, abundant labeled guided wave signals of numerical damages are obtained by the finite element simulation. Then, synthetic samples similar to the experiment guided wave signals and consistent with the labels of simulation guided wave signal are generated by Wasserstein generative adversarial network with gradient penalty (WGAN-GP). Finally, the convolutional neural network (CNN) as the damage classifier is trained with the synthetic samples and the unlabeled experimental samples will be accurately detected. The validity of the proposed method is assessed by detecting fatigue cracks on metal center-hole specimens. The results show that the proposed method can achieve domain knowledge transfer from the simulated damage to the experimental damage, and achieve a reliable fatigue crack detection performance even no available labeled damaged data in the experiment domain.

导出引用

王莉1, 2, 刘国强2, 杨宇2, 张超1, 裘进浩1. 基于域自适应对抗生成样本的金属损伤导波智能迁移识别方法[J]. 振动与冲击, 2025, 44(3): 191-201

WANG Li1, 2, LIU Guoqiang2, YANG Yu2, ZHANG Chao1, QIU Jinhao1. Intelligent transfer recognition method for metal damage with guided wave based on domain adaptive adversarial generated samples[J]. Journal of Vibration and Shock, 2025, 44(3): 191-201

参考文献

[1] 孙侠生, 肖迎春. 飞机结构健康监测技术的机遇与挑战[J]. 航空学报, 2014, 35(2): 3199-3212.
SUN X S, XIAO Y C. Opportunities and challenges of aircraft structural health monitoring[J]. Acta Aeronautica et Astronautica Sinica, 2014, 35(2): 3199-3212.
[2] WANG L, ZHANG C, TAO C C, et al. Prediction of multiple fatigue crack growth based on modified Paris model with particle filtering framework[J]. Mechanical Systems and Signal Processing, 2023, 190, 110124.
[3] 王莉, 刘国强, 肖迎春. 基于代理模型的复合材料加筋壁板分层损伤定量监测方法[J]. 复合材料学报, 2020, 37(2): 302-308.
WANG L, LIU G Q, XIAO Y C. Quantitative monitoring method for delamination damage of stiffened composite panel based on surrogate model[J]．Acta Materiae Compositae Sinica, 2020, 37(2): 302-308．
[4] 刘国强, 肖迎春, 李明. 复合材料胶接损伤的指数监测方法[J]. 航空学报, 2012, 33(7): 1275-1280.
LIU G Q, XIAO Y C, LI M. An index method of monitoring composites debonding[J]. Acta Aeronautica et Astronautica Sinica, 2012, 33(7): 1275-1280 (in Chinese).
[5] WANG L, LIU G Q, ZHANG C, et al. FEM simulation-based adversarial domain adaptation for fatigue crack detection using Lamb wave[J]. Sensors, 2023, 23(4), 1943.
[6] 田翔. 基于Lamb波和二叉树支持向量机的航空结构裂纹监测[D]. 南京: 南京航空航天大学, 2012: 25-40.
TIAN X. Crack monitoring on aerospace structure based on Lamb waves and binary tree support vector machines[D]. Nanjing: Nanjing University of Aeronautics and Astronautics, 2012: 25-40 (in Chinese).
[7] VITOLA J, TIBADUIZA D, ANAYA M, et al. Structural damage detection and classification based on machine learning algorithms[C]//8th European Workshop On Structural Health Monitoring. Spain: Bilbao, 2016: 1-10.
[8] SU Z Q, YE L. Digital Damage Fingerprints (DDF) and its application in quantitative damage identification[J]. Composite Structures, 2005, 67: 197-204.
[9] 杨宇, 周雨熙, 王莉. 一种集成多个机器学习模型的复合材料结构损伤识别方法[J]. 数据采集与处理, 2020, 35(2): 278-287.
YANG Y, ZHOU Y X, WANG L. Integrated method of multiple machine learning models for damage recognition of composite structures[J]. Journal of Data Acquisition and Processing, 2020, 35(2): 278-287 (in Chinese).
[10] 陈素文, 李国强. 人工神经网络在结构损伤识别中的应用[J]. 振动、测试与诊断, 2001, 2(2): 116-124.
CHEN S W, LI G Q. Application of artificial neural networks to damage identification of structures[J]. Journal of Vibration, Measurement & Diagnosis, 2001, 2(2): 116-124 (in Chinese).
[11] CHEN J, WU W Y, REN Y Q, et al. Fatigue crack evaluation with the guided wave–convolutional neural network ensemble and differential wavelet spectrogram[J]. Sensors, 2022, 22, 307.
[12] WU J, XU X B, LIU C, et al. Lamb wave-based damage detection of composite structures using deep convolutional neural network and continuous wavelet transform[J]. Composites Structures, 2021, 276, 114590.
[13] LEE H, LIM H J, SKINNER T, et al. Automated fatigue damage detection and classification technique for composite structures using Lamb waves and deep autoencoder[J]. Mechanical Systems and Signal Processing, 2022, 163, 108148.
[14] 杨宇, 王彬文, 吕帅帅, et al. 一种基于深度学习的复合材料结构损伤导波监测方法[J]. 航空科学技术, 2020, 31(07): 102-108.
YANG Y, WANG B W, LV S S. A Deep-Learning-Based Method For Damage Identification of Composite Laminates[J]. Aeronautical Science & Technology, 2020, 31(07): 102-108 (in Chinese).
[15]TENG S, CHEN X D, CHEN G F, et al. Structural damage detection based on transfer learning strategy using digital twins of bridges[J]. Mechanical Systems and Signal Processing, 2023, 191, 110160.
[16]马波, 蔡伟东, 赵大力. 基于GAN样本生成技术的智能诊断方法[J]. 振动与冲击, 2020, 39(18): 153-160.
MA B, CAI W D, ZHAO D L. Intelligent diagnosis method based on GAN sample generation technology[J]. Journal of Vibration and Shock, 2020, 39(18): 153-160 (in Chinese).
[17]王育鹏, 吕帅帅, 杨宇. 基于域自适应的复合材料结构损伤识别方法[J]. 航空学报, 2022, 43(6): 526752.
WANG Y P, LV S S, YANG Y. Damage recognition of composite structures based on domain adaptive model[J]. Acta Aeronautica et Astronautica Sinica, 2022, 43(6): 526752 (in Chinese).
[18]SHAO W H, SUN H, WANG Y S, et al. A multi-level damage classification technique of aircraft plate structures using Lamb wave-based deep transfer learning network[J]. Smart Materials and Structures, 2022, 31, 075019.
[19]PARZIALE M, LOMAZZI L, RASTIN Z, et al. Unsupervised damage localization in composite plates using lamb waves and conditional generative adversarial networks[J]. Procedia Structural Integrity, 2024, 52, 551-559.
[20]ZHANG B., HONG X B, LIU Y. Multi-task deep transfer learning method for guided wave based integrated health monitoring using piezoelectric transducers[J]. IEEE Sensors Journal, 2020, 3009194.
[21]ZHANG B, HONG X B, LIU Y. Distribution adaptation deep transfer learning method for cross-structure health monitoring using guided waves[J]. Structural Health Monitoring, 2021, 00(0): 1-19.
[22]SAWANT S, SETHI A, BANERJEE S, et al. Unsupervised learning framework for temperature compensated damage identification and localization in ultrasonic guided wave SHM with transfer learning[J]. Ultrasonics, 2023, 130: 106931.
[23]GULRAJANI I, AHMED F, ARJOVSKY M, et al. Improved training of Wasserstein GANs[J]. arXiv 2017, arXiv: 1704.00028v3.
[24]HE J J, GUAN X F, PENG T S, et al. A multi-feature integration method for fatigue crack detection and crack length estimation in riveted lap joints using Lamb waves[J]. Smart Materials and Structures, 2013, 22, 105007.
[25]PANDEY P, RAI A, MIRTA M. Explainable 1-D convolutional neural network for damage detection using Lamb wave[J]. Mechanical Systems and Signal Processing, 2022, 164: 108220.
[26]方岱宁, 刘彬. 力电耦合物理力学计算方法[M]. 北京: 高等教育出版社, 2012.
FANG D N, LIU B. Electromechanical Coupling Computational Methods of Physical Mechanics[M]. Beijing: Higher Education Press, 2012 (in Chinese).
[27]LIU G Q, XIAO Y C, ZHANG H, et al. Baseline Signal Reconstruction for Temperature Compensation in Lamb Wave-Based Damage Detection[J]. Sensors, 2016, 16, 1273.
[28]杨宇, 王彬文, 曹雪洋, et al. 导波受载荷影响补偿的深度学习神经网络方法[J]. 振动、测试与诊断, 2022, 42(2): 812-819.
YANG Y, WANG B W, CAO X Y. Compensation method for load effect on guided wave based on deep learning neural network[J]. Journal of Vibration, Measurement & Diagnosis, 2022, 42(2): 812-819 (in Chinese).
[29]LUO J, HUANG J Y, LI H M. A case study of conditional deep convolutional generative adversarial networks in machine fault diagnosis[J]. Journal of Intelligent Manufacturing, 2011, 32: 407-425.
[30]YANG L, FAN W T, BONGUILA N. Robust unsupervised image categorization based on variational autoencoder with disentangled latent representations[J]. Knowledge-Based Systems, 2022, 246: 108671.
[31]MAATERN L, HINTON G. Visualizing data using t-SNE[J]. Journal of Machine Learning Research, 2008, 9: 2579-2605.