Separability Index-Based Feature Selection and a Two-Tier Classifier for Improving Diagnostic Performance in Bearings

  • Viet Tra
  • Bach Phi Duong
  • Jong-Myon KimEmail author
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 10884)


Faults in bearings have traditionally been diagnosed by identifying the characteristic defect frequencies in the envelope power spectrum (EPS). Although EPS-based methods are effective for constant operating conditions, it makes difficult to detect a fault when the machine operates at variable shaft speeds. To address this issue, we propose a method that performs as follows: (a) a feature extraction technique is used to extract as many information as possible from different sub-bands using a non-stationary signal analysis method (e.g., discrete wavelet pack transform (DWPT)), (b) a novel feature selection method is then applied to select the most informative features from the extracted high-dimensional feature pool, and (c) the selected features are further fed to a two tiers classifiers to identify the condition of a bearing. This two tiers classifier includes the combination of support vector machine groups in the first tier and the combination multilayer perceptron in the second tier. The proposed method is tested for a bearing fault diagnosis application using acoustic emission (AE) signal under various conditions.


Bearing fault diagnosis Support vector machine Multilayer perceptron Feature selection 



This work was supported by the Korea Institute of Energy Technology Evaluation and Planning (KETEP) and the Ministry of Trade, Industry & Energy (MOTIE) of the Republic of Korea (Nos. 20162220100050, 20161120100350, 20172510102130). It was also funded in part by The Leading Human Resource Training Program of Regional Neo industry through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and future Planning (NRF-2016H1D5A1910564), and in part by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2016R1D1A3B03931927).

The authors declare that there is no conflict of interest regarding the publication of this manuscript.


  1. 1.
    Tra, V., Kim, J., Khan, S.A., Kim, J.-M.: Bearing fault diagnosis under variable speed using convolutional neural networks and the stochastic diagonal levenberg-marquardt algorithm. Sensors 17, 2834 (2017)CrossRefGoogle Scholar
  2. 2.
    Tra, V., Kim, J., Khan, S.A., Kim, J.-M.: Incipient fault diagnosis in bearings under variable speed conditions using multiresolution analysis and a weighted committee machine. J. Acoust. Soc. Am. 142, EL35–EL41 (2017)CrossRefGoogle Scholar
  3. 3.
    Rauber, T.W., de Assis Boldt, F., Varejão, F.M.: Heterogeneous feature models and feature selection applied to bearing fault diagnosis. IEEE Trans. Industr. Electron. 62, 637–646 (2015)CrossRefGoogle Scholar
  4. 4.
    Kang, M., Islam, M.R., Kim, J., Kim, J.-M., Pecht, M.: A hybrid feature selection scheme for reducing diagnostic performance deterioration caused by outliers in data-driven diagnostics. IEEE Trans. Industr. Electron. 63, 3299–3310 (2016)CrossRefGoogle Scholar
  5. 5.
    Ma, H., Xiong, Y., Fang, H., Gu, L.: Fault diagnosis of bearing based on fuzzy support vector machine. In: Prognostics and System Health Management Conference (PHM), pp. 1–4. IEEE (2015)Google Scholar

Copyright information

© Springer International Publishing AG, part of Springer Nature 2018

Authors and Affiliations

  1. 1.School of IT ConvergenceUniversity of UlsanUlsanSouth Korea

Personalised recommendations