534
J. Kim and J.-M. Kim
of DNN, which indicates the degree of importance for each input sample. The result
shows the low frequency magnitudes are important to classify the AE signals from
healthy-conditioned bearing, and bearing characteristic frequencies are important to
classify the AE signals from bearing failures. In the future works, the explainable
method for DNN should be researched for various experiment for verification.
Acknowledgements 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 (No. 20181510102160, No. 20192510102510).
References
1. N. Subhasis, T. Hamid, L. Xiaodong, Condition monitoring and fault diagnosis of electrical
motors—a review. IEEE Trans. Energy Convers. 20(4), 719–729 (2005)
2. M, Elforjani, D. Mba, Accelerated natural fault diagnosis in slow speed bearings with acoustic
emission. Eng. Fract. Mech. 77(1), 112–127 (2010)
3. K. Wasmer, F. Saeidi, B. Meylan, T.L. Quan, S.A. Shevchik, When AE (Acoustic Emission)
meets AI (Artificial Intelligence) II. EWGAE (2018)
4. M.M. Manjurul Islam, J.-M. Kim, Automated bearing fault diagnosis scheme using 2D representation of wavelet packet transform and deep convolutional neural network. Comput. Ind.
106(1), 142–153 (2019)
5. R.R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, D. Batra, Grad-CAM: visual
explanations from deep networks via gradient-based localization. ICCV (2017)
6. M. Kang, J. Kim, L.M. Wills, J.-M. Kim, Time-varying and multiresolution envelope analysis
and discriminative feature analysis for bearing fault diagnosis. IEEE Trans. Ind. Electron. 62(12),
7749–7761 (2015)
7. T.N. Sainath, A.R. Mohamed, B. Kingsbury, B. Ramabhadran, Deep convolutional neural
networks for LVCSR. ICASSP (2013)
8. Z. Zhang, S. Mert, Generalized cross entropy loss for training deep neural networks with noisy
labels. NeurIPS (2018)
J. Kim and J.-M. Kim
of DNN, which indicates the degree of importance for each input sample. The result
shows the low frequency magnitudes are important to classify the AE signals from
healthy-conditioned bearing, and bearing characteristic frequencies are important to
classify the AE signals from bearing failures. In the future works, the explainable
method for DNN should be researched for various experiment for verification.
Acknowledgements 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 (No. 20181510102160, No. 20192510102510).
References
1. N. Subhasis, T. Hamid, L. Xiaodong, Condition monitoring and fault diagnosis of electrical
motors—a review. IEEE Trans. Energy Convers. 20(4), 719–729 (2005)
2. M, Elforjani, D. Mba, Accelerated natural fault diagnosis in slow speed bearings with acoustic
emission. Eng. Fract. Mech. 77(1), 112–127 (2010)
3. K. Wasmer, F. Saeidi, B. Meylan, T.L. Quan, S.A. Shevchik, When AE (Acoustic Emission)
meets AI (Artificial Intelligence) II. EWGAE (2018)
4. M.M. Manjurul Islam, J.-M. Kim, Automated bearing fault diagnosis scheme using 2D representation of wavelet packet transform and deep convolutional neural network. Comput. Ind.
106(1), 142–153 (2019)
5. R.R. Selvaraju, M. Cogswell, A. Das, R. Vedantam, D. Parikh, D. Batra, Grad-CAM: visual
explanations from deep networks via gradient-based localization. ICCV (2017)
6. M. Kang, J. Kim, L.M. Wills, J.-M. Kim, Time-varying and multiresolution envelope analysis
and discriminative feature analysis for bearing fault diagnosis. IEEE Trans. Ind. Electron. 62(12),
7749–7761 (2015)
7. T.N. Sainath, A.R. Mohamed, B. Kingsbury, B. Ramabhadran, Deep convolutional neural
networks for LVCSR. ICASSP (2013)
8. Z. Zhang, S. Mert, Generalized cross entropy loss for training deep neural networks with noisy
labels. NeurIPS (2018)