Bearing condition identification method based on enhanced synchrosqueezing stockwell transform and improved ensemble deep extreme learning machine

被引：0
|
作者：
Du X.-L. [1 ]
Xiao L. [2 ]
Zhou Q.-H. [3 ]
Chen Z.-G. [3 ]
机构：
[1] College of Mechanical and Electrical Engineering, University of Electronic Science and Technology of China, Chengdu
[2] Advanced Technology Research Institute, Beijing Institute of Technology, Jinan
[3] School of Mechanical-Electronic and Vehicle Engineering, Beijing University of Civil Engineering and Architecture, Beijing
来源：
Dianji yu Kongzhi Xuebao/Electric Machines and Control | 2022年 / 26卷 / 11期
关键词：
condition identification; deep learning; ensemble learning; extreme learning machine; motor bearing; synchrosqueezing stockwell transform;
D O I：
10.15938/j.emc.2022.11.015
中图分类号：
学科分类号：
摘要：
In terms of the drawbacks of low time-frequency concentration and tardiness convergence rate of deep network in intelligent working conditions identification of motor bearings based on “time frequency spectrum diagram—deep network”, an enhanced synchrosqueezing stockwell transform (ESST) with improved ensemble deep extreme learning machine (IEDELM) was proposed. The ESST-IEDELM method firstly performed ESST transformation on the collected motor bearing vibration signals to obtain the time—frequency spectrograms and then were reshaped into vectors. Secondly, nine deep extreme learning machines were constructed by employing nine activation functions, and self-organizing strategy and convolution strategy were introduced to promote the robustness of deep extreme learning machine. Finally, the time-frequency vectors were fed into IEDELM for automatic feature learning and the recognition result was given using the ensemble averaging method. The experimental results indicate that the ESST—IEDELM method reaches condition identification accuracy of 99. 87%, and the standard deviation is only 0. 12 and is superier than other methods in the feature extraction of motor bearing vibration signal and the accuracy of condition identification. Therefore, the ESST-IEDELM model can be applied to the engineering application of conditions identification of motor bearings. © 2022 Editorial Department of Electric Machines and Control. All rights reserved.
引用
收藏
页码：141 / 150
页数：9
相关论文
共 19 条
  • [11] ASGHAR M Z, ABBAS M, ZEESHAN K, Et al., Assessment of deep learning methodology for self-organizing 5G networks[ J], Applied Sciences, 9, 15, (2019)
  • [12] WEI J, LIU H, YAN G, Et al., Robotic grasping recognition using multi-modal deep extreme learning machine [ J], Multidimensional Systems and Signal Processing, 28, 3, (2016)
  • [13] (2019)
  • [14] LU C, WANG Z Y, QIN W L, Et al., Fault diagnosis of rotary machinery components using a stacked denoising autoencoder-based health state identification, Signal Processing, 130, C, (2016)
  • [15] VINCENT P, LAROCHELLE H, LAJOIE I, Et al., Stacked denoising autoencoders, Journal of Machine Learning Research, 11, 12, (2010)
  • [16] CHEN M M, XU Z X, WEINBERGER K, Et al., Marginalized denoising autoencoders for domain adaptation[J], Proceeding of the Twenty-ninth International Conference on Machine Learning, 5, (2012)
  • [17] LI Z R, KANG Y, LV W J, Et al., High-emitter identification model establishment using weighted extreme learning machine and active sampling [J], Neurocomputing, 30, 2, (2021)
  • [18] SHAO H D, JIANG H K, ZHAO K W, Et al., A novel tracking deep wavelet auto-encoder method for intelligent fault diagnosis of electric locomotive bearings [J], Mechanical Systems and Signal Processing, 110, 9, (2018)
  • [19] TAO Shasha, GUO Shunsheng, Fault diagnosis of bearing based on deep wavelet automatic encoder and extreme learning machine [ J ], Science Technology and Engineering, 20, 29, (2020)
12