Education, Science, Technology, Innovation and Life
Open Access
Sign In

Identification of time-frequency maps of bearing faults based on hyperparameter optimization SSA-GoogleNet

Download as PDF

DOI: 10.23977/jemm.2024.090105 | Downloads: 10 | Views: 226


Ziqiang Luo 1


1 School of Mechanical and Automation Engineering, Wuhan University of Science and Technology, Wuhan, 430065, China

Corresponding Author

Ziqiang Luo


In order to solve the problem of poor noise immunity of neural network in bearing fault detection and to meet the high demand for adaptive extraction ability of network features in the industrial field, this paper proposes a hyper-parameter optimization method to eliminate the error of human-set parameters. Aiming at the non-smoothness and non-linear characteristics of the bearing fault signal, the two-dimensional wavelet transform (2DWT) is used to extract the feature values of the vibration signal of the bearing fault, and it is proposed to use the CNN convolutional neural network (CNN) to train the fault model. Convolutional Neural Network (CNN) is proposed to be used for fault model training. Firstly, the 2D wavelet transform is applied to the bearing vibration signal to extract the time-frequency map of the vibration signal; secondly, the extracted time-frequency map is used as the training object of the convolutional neural network to train the model, and then multiple convolutional neural network frameworks are used for noise immunity test by adding Gaussian white noise to the original signal to construct the test framework, and then the framework with the best noise immunity is used as the base network; and then the CNN is used as the base network. Sparrow Search Algorithm (SSA), Grey Wolf Optimizer (GWO), and Whale Optimization Algorithm (GWO) to find the best parameters. Validated by the public dataset of Western Reserve University, the method can effectively improve the fault identification accuracy of the model and can get rid of the problem of poor network robustness.


2D wavelet transform, convolutional neural network, sparrow optimization algorithm, hyperparameter optimization


Ziqiang Luo, Identification of time-frequency maps of bearing faults based on hyperparameter optimization SSA-GoogleNet. Journal of Engineering Mechanics and Machinery (2024) Vol. 9: 24-32. DOI:


[1] Shen J, Xu F. Method of fault feature selection and fusion based on poll mode and optimized weighted KPCA for bearings. Measurement, 2022(194):194. DOI: 10.1016/j.measurement.2022.110950.
[2] Lin Li, Xining Zhang, Shuyu Liu, et al., "Bearing Fault Diagnosis Method Based on Two-Dimensional Empirical Wavelet Texture Domain Feature Adaptive Extraction," Journal of Xi'an Jiaotong University, vol. 55, no. 12, pp. 79-86, 2021.  
[3] Gang Yang, Limu Qin, Kun Lv, et al., "Bearing Fault Diagnosis Based on Continuous Cross Wavelet Coherence Analysis and Adaptive CYCBD," Vibration and Shock, vol. 42, no. 21, pp. 17-28, 2023. DOI: 10.13465/j.cnki.jvs. 2023.21.003. 
[4] Cong Wu, Mengnan Li, Kun Li, "Elevator Bearing Fault Detection Based on PBO and CNN," Information Technology, vol. 47, no. 4, pp. 73-78, 2023. DOI: 10.13274/j.cnki.hdzj.2023.04.014.
[5] Siyan He, Ya Liu, Xincheng Tian, "Bearing Fault Diagnosis Research Based on Wavelet Packet-AR Spectrum and Deep Learning," Computer Applications Research, vol. 36, no. 6, pp. 1758-1761, 1766, 2019. 
[6] Yuetian Wang, Sichao Fu, Qinmu Peng, et al., "Graph Convolutional Neural Network for Multi-Perspective Information Interaction in Semi-Supervised Scenes," Journal of Software, 2023-12-01. DOI: 10.13328/j.cnki.jos.007007.  
[7] Junqing Li, Jing Liu, "Motor Bearing Fault Diagnosis Method Combining Convolutional Neural Network and Transfer Learning," Journal of North China Electric Power University (Natural Science Edition), vol. 50, no. 1, pp. 76-83, 91, 2023. DOI: 10.3969/j.ISSN.1007-2691.2023.01.09.
[8] Qinye He, Hanyu Lu, Yongyi Yuan, et al., "Research on Brightness Temperature Downscaling Method Based on Two-Dimensional Wavelet Transform," Journal of Guangxi University (Natural Science Edition), vol. 47, no. 3, pp. 746-755, 2022. 
[9] Yanyan Bao, Guangze Yang, Wei Chen, et al., "Voiceprint Recognition of 750kV Transformer and Bushing Discharge Mixed Signal Based on SBSS and CNN," Journal of Southwest Jiaotong University, 2023-12-13.  
[10] S. Mirjalili, "Dragonfly algorithm: A new optimization algorithm for solving discrete and continuous optimization problems," Neural Computing and Applications, vol. 27, no. 4, pp. 1053-1073, 2016.

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.