A Bearing Fault Diagnosis Method Based on Fusion of CNN-BiLSTM-Transformer and Cross-Attention

Xun Yuan; Wei Wu

doi:10.54097/594q8441

Authors

Xun Yuan
Wei Wu

DOI:

https://doi.org/10.54097/594q8441

Keywords:

Rolling Bearing; Convolutional Neural Network (CNN); Bidirectional Long Short-Term Memory (BiLSTM); Cross-Attention; Transformer.

Abstract

To address the limitations of traditional diagnostic models in achieving high accuracy in rolling bearing fault diagnosis and their inadequate extraction of vibration signal features across both frequency and time domains, this paper introduces a fault diagnosis approach that integrates multiple neural networks with a cross-attention mechanism. Initially, one-dimensional fault signals undergo processing through the Fast Fourier Transform (FFT) and Variational Mode Decomposition (VMD) techniques to obtain spectral information and time-domain features, respectively. The combined application of FFT and VMD facilitates the efficient extraction of multi-scale features from the signals. Subsequently, a CNN-Transformer network is utilized to capture the spatial characteristics of the pre-processed multi-scale fault signal features. This is followed by the deployment of a Bidirectional Long Short-Term Memory (BiLSTM) network to extract sequential information, with a particular emphasis on pivotal temporal features. Additionally, the integration of the Transformer network further enhances feature extraction and fusion capabilities. Ultimately, a cross-attention mechanism is implemented to seamlessly integrate time-domain and frequency-domain features, thereby bolstering the model's performance and generalization capacity for fault classification tasks. Experimental findings validate the proposed model's diagnostic effectiveness, achieving an accuracy and recall rate exceeding 99% in rolling bearing fault diagnosis.

Downloads

Download data is not yet available.

References

[1] Guo Y ,Zhou J ,Dong Z , et al.Research on bearing fault diagnosis based on novel MRSVD-CWT and improved CNN-LSTM[J].Measurement Science and Technology,2024,35(9):

[2] Kang J ,Zhu X ,Shen L , et al.Fault diagnosis of a wave energy converter gearbox based on an Adam optimized CNN-LSTM algorithm[J].Renewable Energy,2024,231121022-121022.

[3] Xi Tao, Yang Weizhen. Research on Gearbox Fault Diagnosis Based on Optimized VMD and CNN [J]. Mechanical Science and Technology: 1-11

[4] Mao Chenglong, Zhang Mei, Zhang Jie. Rolling Bearing Fault Diagnosis Based on MSSA-VMD and CNN-BiLSTM [J]. Journal of Chongqing Technology and Business University (Natural Science Edition): 1-10

[5] Wang Minjuan, Jia Qian, Wang Youming, Ding Wenke. Rolling Bearing Fault Diagnosis Method Based on IMSE and Parameter-Optimized VMD [J]. Journal of Xi'an University of Posts and Telecommunications: 1-10.

[6] Cao Jingsheng, Yu Yang, Wang Qi, Dong Yining. Intelligent Fault Diagnosis of Motor Bearings Based on Optimized VMD-CNN-BiLSTM [J]. Modern Electronics Technique, 2024, 47(12): 115-121.

[7] Xie Fengyun, Wang Gan, Shang Jiandong, Fan Qiuyang, Zhu Haiyan. Gearbox Fault Diagnosis Based on Adaptive Variational Mode Decomposition [J]. Journal of Propulsion Technology: 1-11.

[8] Meng Jingyu, Yang Liming, Zhang Cheng, Wu Boyang, Xu Guoping, Yu Jian. Feature Extraction and Diagnosis of Wind Turbine Gearbox Faults Based on Zoom-FFT-CEEMD and Wavelet Packet Denoising [J]. Micro Motors, 2024, 52(04): 28-32=37.

[9] Gao Hongwei, Li Xincheng, He Xiaoning, Liu Chang'an, Wang Hanchi, Li Wencheng, Zhang Baohui. Research on Wind Turbine Gearbox Fault Diagnosis Method Based on Hilbert Transform [J]. Machine Tool & Hydraulics, 2024, 52(09): 215-220.

[10] Wei Hangxin, Cheng Huan, Wu Wei, Wang Xiaorong. Rolling Bearing Fault Diagnosis Based on VMD Visualization and Deep Learning [J]. Mechanical Design and Manufacture: 1-6.

[11] Zhu Junjie, Zhang Qinghua, Zhu Guanhua, Su Naiquan. Rolling Bearing Fault Diagnosis Based on EEMD and CNN-SVM [J]. Machine Tool & Hydraulics: 1-9.

[12] Zhou Ruifeng. Research and Implementation of Intelligent Fault Diagnosis Methods for Rolling Bearings [D]. Dalian University of Technology, 2009.

[13] Xu Jingwen, Yang Ping, Yin Xiaojun. Gear Fault Diagnosis Based on EMD Decomposition and Levy-SSA-BP Neural Network [J]. Mechanical Transmission, 2024, 48(05): 152-157.

[14] Long Xiafei, He Zhicheng, Zhou Ling, Liu Weiqiang, Liang Kai. Research on Wind Turbine Gearbox Fault Diagnosis Based on KOA-CNN-BiLSTM-AM [J]. Machine Tool & Hydraulics: 1-10.

[15] Wang Jinhua, Liu Zhengqi, Cao Jie, Liu Yunqiang, Chen Li. Gearbox Fault Diagnosis Based on R Vine Copula-DBN [J]. Journal of Beijing University of Aeronautics and Astronautics: 1-15.

[16] Wang Chengbing. Gearbox Fault Diagnosis Based on Multi-Sensor Information Fusion and Dual-Stream CNN [J]. Construction Machinery, 2024, 55(04): 69-78+10-11.

[17] Xu M ,Yu Q ,Chen S , et al.Rolling Bearing Fault Diagnosis Based on CNN-LSTM with FFT and SVD [J]. Information, 2024, 15(7):399-399.

[18] Yu T ,Li C .Fault diagnosis method of rolling bearing based on MSSA-VMD algorithm[J].Journal of Physics: Conference Series, 2024,2752(1):

[19] Cui K ,Liu M ,Meng Y .A new fault diagnosis of rolling bearing on FFT image coding and L-CNN[J].Measurement Science and Technology,2024,35(7):

[20] Zhenzhen J ,Deqiang H ,Zexian W .Intelligent fault diagnosis of train axle box bearing based on parameter optimization VMD and improved DBN[J].Engineering Applications of Artificial Intelligence,2022,110.