基于特征交叉注意力机制融合的轴承故障诊断方法

PDF(2807 KB)

振动与冲击 ›› 2025, Vol. 44 ›› Issue (12) : 228-237.

故障诊断分析

基于特征交叉注意力机制融合的轴承故障诊断方法

赵国超1, 2, 刘崇德*1, 宋宇宁3, 金鑫1, 2, 李伟华1

作者信息 +

Bearing fault diagnosis method based on fusion of time-frequency features by cross-attention mechanism

ZHAO Guochao1,2,LIU Chongde*1,SONG Yuning3,JIN Xin1,2,LI Weihua1

Author information +

文章历史 +

摘要

为了解决轴承振动信号特征提取不充分导致故障诊断准确率低的问题，本文提出一种基于特征交叉注意力机制融合的轴承故障诊断方法，建立CNN-BiTCN-CATTM诊断模型。采用变分模态分解和快速傅立叶变换对原始信号进行重构，分别使用双向时间卷积网络（bidirectional temporal convolutional network，BiTCN）和卷积神经网络（convolutional neural network，CNN）提取时频特征，通过交叉注意力机制（cross-attention mechanism，CATTM）融合时频特征的能力，充分提取原始信号故障特征，利用全连接层实现滚动轴承故障类型的精确诊断。试验研究表明，在含信噪比为9.32、标准差为2.98的高斯白噪声的环境下，使用CNN-BiTCN-CATTM模型轴承故障分类准确率为99.88%，相较于使用CNN、BiTCN和结合自注意力机制的卷积神经网络（CNN with self-attention mechanism，CNN-SATTM）诊断轴承故障，准确率分别提升约22.79%、4.85%和4.19%。在引入信噪比为3.31、标准差为5.96的高斯白噪声时，本模型仍然可以达到96.12%的诊断准确率。CNN-BiTCN-CATTM模型能够深入提取轴承信号中的故障特征，有效提高故障分类准确性。

Abstract

In order to solve the problem of low fault diagnosis accuracy, this paper proposes a bearing fault diagnosis method based on feature cross-attention mechanism fusion and develops the CNN-BiTCN-CATTM model. The original signal is reconstructed using variational mode decomposition and fast fourier transform, while bidirectional temporal convolutional networks (BiTCN) and convolutional neural networks (CNN) are used to extract time-frequency features. The cross-attention mechanism (CATTM) is applied to fuse these features, fully capturing fault characteristics from the original signal. Experiments show that in an environment with Gaussian white noise (SNR = 9.32, standard deviation = 2.98), the CNN-BiTCN-CATTM model achieves a bearing fault classification accuracy of 99.88%, which is about 22.79%, 4.85%, and 4.19% higher than using CNN, BiTCN, and CNN-SATTM, respectively. Even with Gaussian white noise (SNR = 3.31, standard deviation = 5.96), the model still achieves a diagnostic accuracy of 96.12%. The CNN-BiTCN-CATTM model effectively extracts deep fault features and significantly improves fault classification accuracy.

导出引用

赵国超1, 2, 刘崇德1, 宋宇宁3, 金鑫1, 2, 李伟华1. 基于特征交叉注意力机制融合的轴承故障诊断方法[J]. 振动与冲击, 2025, 44(12): 228-237

ZHAO Guochao1, 2, LIU Chongde1, SONG Yuning3, JIN Xin1, 2, LI Weihua1. Bearing fault diagnosis method based on fusion of time-frequency features by cross-attention mechanism[J]. Journal of Vibration and Shock, 2025, 44(12): 228-237

参考文献

[1] Jiang H, Li C, Li H. An improved EEMD with multiwavelet packet for rotating machinery multi-fault diagnosis[J]. Mechanical Systems and Signal Processing, 2013, 36(2): 225- 239.
[2] Lei Y, Lin J, He Z, et al. A review on empirical mode decomposition in fault diagnosis of rotating machinery[J]. Mechanical systems and signal processing, 2013, 35(1-2): 108-126.
[3] 杜小磊,陈志刚,许旭,等.基于小波卷积自编码器和LSTM网络的轴承故障诊断研究[J].机电工程,2019,36(07):663-668.
DU Xiao-lei, CHEN Zhi-gang, XU Xu, et al. Fault diagnosis of bearing based on wavelet convolutional auto- encoder and LSTM network[J].Journal of Mechanical & Electrical Engineering, 2019,36(07):663-668.
[4] Shao H D, Jiang H K, Zhao K, et al. A novel tracking deep wavelet auto-encoder method for intelligent fault diagnosis of electric locomotive bearings[J]. Mechanical Systems and Signal Processing, 2018, 110: 193-209.
[5] 雷亚国,贾峰,孔德同,等.大数据下机械智能故障诊断的机遇与挑战[J].机械工程学报,2018,54(05):94-104.
LEI Ya-guo, JIA Feng, KONG De-tong, et al. Opportunities and challenges of machinery intelligent fault diagnosis in big data era[J].Journal of Mechanical Engineering, 2018,54(05): 94-104.
[6] Sadoughi M, Hu C. Physics-based convolutional neural network for fault diagnosis of rolling element bearings[J]. IEEE Sensors Journal, 2019, 19(11): 4181-4192.
[7] Hasan M J, Islam M M M, Kim J M. Acoustic spectral imaging and transfer learning for reliable bearing fault diagnosis under variable speed conditions[J]. Measurement, 2019, 138: 620 -631.
[8] Gao Y, Kim C H, Kim J M. A novel hybrid deep learning method for fault diagnosis of rotating machinery based on extended WDCNN and long short-term memory[J]. Sensors, 2021, 21(19): 6614.
[9] 郭盼盼,张文斌,崔奔,等.基于增强深度卷积神经网络的滚动轴承多工况故障诊断方法[J/OL].振动工程学报,1-14 [2024-10-04].https://link.cnki.net/urlid/32.1349.TB.20230830.1129.002.
GUO Pan-pan, ZHANG Wen-bin, CUI Ben, et al. Multi- condition fault diagnosis method of rolling bearing based on enhanced deep convolutional neuralnetwork[J/OL]. Journal of Vibration Engineering, 1-14[2024-10-04].
[10] Song B, Liu Y, Fang J, et al. An optimized CNN-BiLSTM network for bearing fault diagnosis under multiple working conditions with limited training samples[J]. Neurocomputing, 2024, 574: 127284.
[11] Zhang Q, Wei X, Wang Y, et al. Convolutional neural network with attention mechanism and visual vibration signal analysis for bearing fault diagnosis[J]. Sensors, 2024, 24(6): 1831.
[12] Wang Y, Li D, Li L, et al. A novel deep learning framework for rolling bearing fault diagnosis enhancement using VAE- augmented CNN model[J]. Heliyon, 2024, 10(15):91-100.
[13] 张洪亮,余其源,秦超群,等.基于信息融合及双连接注意力残差网络的轴承故障诊断[J].振动与冲击,2023,42(20): 114-123.
ZHANG Hong-liang, YU Qi-yuan, QIN Chao-qun, et al. Bearing fault diagnosis based on double-connected attention residual network and information fusion[J].Journal of Vibration and Shock,2023,42(20):114-123.
[14] 赵小强,柴靖轩.改进卷积胶囊网络的滚动轴承故障诊断方法[J].振动工程学报,2024,37(05):885-895.
ZHAO Xiao-qiang, CHAI Jing-xuan.Improved convolutional capsule network method for rolling bearing fault diagnosis [J]. Journal of Vibration Engineering,2024,37(05):885-895.
[15] 王亚萍,高圣延,王新生,等.多源域下M A A N与M K-MM D的滚动轴承跨工况复合故障诊断[J/OL].振动工程学报,1-12[2024-10-04].https://link.cnki.net/urlid/32.1349.TB.20240826.1700.006.
WANG Ya-ping, GAO Sheng-yan, WANG Xing-sheng, et al. Composite fault diagnosis of roll bearings across operating conditions with MAAN and MK-MMD in multi-source domains[J/OL].Journal of Vibration Engineering,1-12 [2024- 10-04].
[16] Dong Z, Zhao D, Cui L. An intelligent bearing fault diagnosis framework: one-dimensional improved self-attention- enhanced CNN and empirical wavelet transform[J]. Nonlinear Dynamics, 2024, 112(8): 6439-6459.
[17] Zhou H, Liu R, Li Y, et al. A rolling bearing fault diagnosis method based on a convolutional neural network with frequency attention mechanism[J]. Structural Health Monitoring, 2024, 23(4): 2475-2495.
[18] Keshun Y, Puzhou W, Yingkui G. Towards efficient and interpretative rolling bearing fault diagnosis via quadratic neural network with BI-LSTM[J]. IEEE Internet of Things Journal, 2024.
[19] 尹爱军,吕明阳,杨敏英,等.RepVGG与CapsNet融合的轴承故障诊断[J].振动与冲击,2024,43(14):301-307.
YIN Ai-jun, LÜ Ming-yang, YANG Min-ying, et al. Bearing fault diagnosis based on the combined use of RepVGG and CapsNet[J].Journal of Vibration and Shock,2024,43(14): 301- 307.
[20] 宁方立,王珂,郝明阳.融合CNN和ViT的声信号轴承故障诊断方法[J].振动与冲击,2024,43(03):158-163+170.
NING Fang-li, WANG Ke, HAO Ming-yang. Fault diagnosis method for bearing based on fusing CNN and ViT[J].Journal of Vibration and Shock,2024,43(03):158-163+170.
[21] 王敏,邓艾东,马天霆,等.基于双注意力机制的MSCN -BiGRU的滚动轴承故障诊断方法[J].振动与冲击,2024,43(06):84-92+103.
WANG Min, DENG Ai-dong, MA Tian-ting, et al. Rolling bearing fault diagnosis method based on a multi-scale and improved gated recurrent neural network with dual attention [J].Journal of Vibration and Shock,2024,43 (06):84- 92+103.
[22] Sprangers, Olivier, Sebastian Schelter, and Maarten de Rijke. Parameter-efficient deep probabilistic forecasting[J]. International Journal of Forecasting, 2023, 39(1): 332-345.
[23] Jian L, Xiong S, Yan H, et al. Rethinking cross-attention for infrared and visible image fusion[J]. IEEE transactions on intelligent transportation systems,2024, 25(11): 16300-16311.
[24] 李辉,徐伟烝.噪声干扰下的CCSD-CNN轴承故障诊断方法[J].轴承,2023,(10):93-100.
LI Hui, XU Wei-zheng. Fault diagnosis method for bearings based on CCSD-CNN under noise interference[J]. Bearing, 2023,(10): 93-100.
[25] Smith W A, Randall R B. Rolling element bearing diagnostics using the Case Western Reserve University data: A benchmark study[J]. Mechanical systems and signal processing, 2015, 64: 100-131.
[26] 尹久,张杰.基于HMFDE和t-SNE的旋转机械故障诊断方法[J].机电工程,2024,41(06):1058-1067.
YIN Jiu, ZHANG Jie. Fault diagnosis method of rotating machinery based on HMFDE and t-SNE[J].Journal of Mechanical & Electrical Engineering,2024,41(06):1058-1067.
[27] 殷秀丽,谢丽蓉,杨欢,等.特征选择与t-SNE结合的滚动轴承故障诊断[J].机械科学与技术,2023,42(11):1784-1793.
YIN Xiu-li, XIE Li-rong, YANG Huan, et al. Rolling bearing fault diagnosis combined feature selection with t-distributed stochastic neighbor embedding[J].Mechanical Science and Technology for Aerospace Engineering,2023,42(11):1784- 1793.