Bearing Fault Diagnosis under Variable Rotational Speeds Using Stockwell Transform-Based Vibration Imaging and Transfer Learning
Open accessThe DOI matches its Crossref record; 40 of the 41 references it cites were found in a registry or an index.
Abstract
In this paper, discrete orthonormal Stockwell transform (DOST)-based vibration imaging is proposed as a preprocessing step for supporting load and rotational speed invariant scenarios for signals of various health conditions. For any health condition, features can easily be extracted from its generated health pattern. To automate the feature selection process, a convolutional neural network (CNN)-based transfer learning (TL) approach for diagnosis has also been introduced. Transfer learning allows an established model to use feature knowledge obtained under one set of working conditions through hidden layers to diagnose faults that occur under other working conditions. The network learns from the massive source dataset, and that knowledge is applied to the target data to identify faults. Using the bearing dataset of Case Western Reserve University, the proposed approach yields an average 99.8% classification accuracy and, specifically, 99.99% for healthy condition (HC), 99.95% for inner race fault (IRF), 99.96% for ball fault (BF), 99.68% for outer race fault for 12 o’clock sensor position (ORF@12), 99.93% for outer race fault for 3 o’clock sensor position (ORF@3), and 99.89% for outer race fault for 6 o’clock sensor position (ORF@6). In this paper, the proposed approach is compared with conventional artificial neural networks (ANNs), support vector machines (SVMs), hierarchical CNNs, and deep autoencoders. The proposed approach outperforms these conventional methods in the accuracy under all working conditions.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Compared with the diagnosis model just trained with a small number of target-domain samples, the fine-tuned diagnosis model presented the faster convergence rate and higher diagnosis accuracy.”
From Applications of machine learning to machine fault diagnosis: A review and roadmap (2020) · cites this paper as [37]
Citations
Cites
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805 · cited 1×
“This network makes optimal use of the indigenous connections (instead of fully-connected layers), weight distribution, and spatial or progressive subsampling to achieve invariance of shifting, scaling, and distortion in their inputs [39,40].”
BackgroundA review on machinery diagnostics and prognostics implementing condition-based maintenance
Jardine, A. K. S., Lin, D., & Banjevic, D. (2006) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2005.09.012 · Cited by 4,816
Localization of the complex spectrum: the S transform
Stockwell, R. G., Mansinha, L., & Lowe, R. P. (1996) · IEEE Transactions on Signal Processing · doi:10.1109/78.492555 · Cited by 3,486 · cited 1×
“If a 1D function of time t for a constant frequency f is represented as V s ( τ , f ) , then the continuous [30,35] ST of a function h ( t ) is: Equation (2) demonstrates the change of the amplitude and phase for this frequency over time.”
BackgroundA review on electrical motors energy use and energy savings
Saidur, R. (2010) · Renewable and Sustainable Energy Reviews · doi:10.1016/j.rser.2009.10.018 · Cited by 499 · cited 1×
“In electromechanical engineering, motion is usually determined by mechanical device structures (e.g., rotating machines or induction motors), which leads to satisfactory records of nearly 70% of the gross energy ingestion in modern manufacturing economics [1,2].”
BackgroundCao, P., Zhang, S., & Tang, J. (2018) · IEEE Access · doi:10.1109/access.2018.2837621 · Cited by 460 · cited 3×
“Thus, the last layer of the target network is trained, providing an accuracy of output based on the learning from the source network [34].”
BackgroundMethodA novel method for intelligent fault diagnosis of rolling bearings using ensemble deep auto-encoders
Shao, H., Jiang, H., Lin, Y., & Li, X. (2018) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2017.09.026 · Cited by 460 · cited 3×
“Examples include deep autoencoders [27,28], ANNs (artificial neural networks), and hierarchical convolution network [29].”
BackgroundMethodLu, C., Wang, Z., & Zhou, B. (2017) · Advanced Engineering Informatics · doi:10.1016/j.aei.2017.02.005 · Cited by 434 · cited 2×
“Examples include deep autoencoders [27,28], ANNs (artificial neural networks), and hierarchical convolution network [29].”
MethodA basis for efficient representation of the S-transform
Stockwell, R. G. (2007) · Digital Signal Processing · doi:10.1016/j.dsp.2006.04.006 · Cited by 417 · cited 2×
“To address this issue, the Stockwell transform (S-transform) is employed in this paper [30].”
BackgroundMethodTransfer learning with neural networks for bearing fault diagnosis in changing working conditions
Zhang, R., Tao, H., Wu, L., & Guan, Y. (2017) · IEEE Access · doi:10.1109/access.2017.2720965 · Cited by 378 · cited 6×
“To validate the performance of the proposed method, some recent deep learning approaches, including raw signal-based TL approach [33], hierarchical CNN-based approach [29], and ensemble deep autoencoders approach [27], are compared with the proposed method.”
BackgroundMethodInfluentialSoualhi, A., Clerc, G., & Razik, H. (2013) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2012.2230598 · Cited by 264 · cited 1×
“Multiple signature analysis of vibrations and motor currents has also been researched to improve reliability [19].”
BackgroundOne‐dimensional convolutional neural networks for spectroscopic signal regression
Malek, S., Melgani, F., & Bazi, Y. (2018) · Journal of Chemometrics · doi:10.1002/cem.2977 · Cited by 248 · cited 2×
“In [32], U. Battisti et al. assumed that one could find such a basis E wp of L 2 ([0,1]), where w will be dependent on choice.”
BackgroundMethodFailure identification and analysis for high-voltage induction motors in the petrochemical industry
Thorsen, O. V., & Dalva, M. (1999) · IEEE Transactions on Industry Applications · doi:10.1109/28.777188 · Cited by 228 · cited 1×
“Bearings play an important role in condition monitoring because, in more than 50% of the cases involving induction motor failures, they are the root cause of the failure [2,8].”
BackgroundLi, S., Liu, G., Tang, X., Lu, J., & Hu, J. (2017) · Sensors · doi:10.3390/s17081729 · Cited by 211
A Hybrid Feature Model and Deep-Learning-Based Bearing Fault Diagnosis
Sohaib, M., Kim, C.-H., & Kim, J.-M. (2017) · Sensors · doi:10.3390/s17122876 · Cited by 200
Self-citationSaimurugan, M., Ramachandran, K. I., Sugumaran, V., & Sakthivel, N. R. (2011) · Expert Systems with Applications · doi:10.1016/j.eswa.2010.09.042 · Cited by 194 · cited 1×
“Existing signal processing techniques have employed feature analysis approaches to detect statistical features for further classification employing classical machine learning algorithms, such as SVM (support vector machine) and PSVM (proximal support vector machine) [25,26].”
MethodHuo, Z., Zhang, Y., Francq, P., Shu, L., & Huang, J. (2017) · IEEE Access · doi:10.1109/access.2017.2661967 · Cited by 187 · cited 1×
“Adjusting the window size is often a solution to gather important information from vibration signals [22].”
BackgroundZhao, S., Liang, L., Xu, G., Wang, J., & Zhang, W. (2013) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2013.04.006 · Cited by 174 · cited 1×
“These fault diagnosis studies have shown that diagnosis of bearing wear can reduce maintenance expenses and enhance machine reliability [6,12–16].”
BackgroundEffect of number of features on classification of roller bearing faults using SVM and PSVM
Sugumaran, V., & Ramachandran, K. I. (2010) · Expert Systems with Applications · doi:10.1016/j.eswa.2010.09.072 · Cited by 157 · cited 1×
“Existing signal processing techniques have employed feature analysis approaches to detect statistical features for further classification employing classical machine learning algorithms, such as SVM (support vector machine) and PSVM (proximal support vector machine) [25,26].”
MethodTransfer representation learning for medical image analysis
Shie, C., Chuang, C.-H., Chou, C., Wu, M.-H., & Chang, E. Y. (2015) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2015.7318461 · Cited by 146 · cited 2×
“In practice, TL creates a robust fault diagnosis process [38].”
BackgroundMethodFault Diagnosis from Raw Sensor Data Using Deep Neural Networks Considering Temporal Coherence
Zhang, R., Peng, Z., Wu, L., Yao, B., & Guan, Y. (2017) · Sensors · doi:10.3390/s17030549 · Cited by 132
Dynamic modelling of wear evolution in rolling bearings
El-Thalji, I., & Jantunen, E. (2014) · Tribology International · doi:10.1016/j.triboint.2014.11.021 · Cited by 116 · cited 2×
“To perform the orthogonal transform, this method uses the discrete Fourier transform (DFT) [3].”
BackgroundMethodFast Discrete Orthonormal Stockwell Transform
Wang, Y., & Orchard, J. (2009) · SIAM Journal on Scientific Computing · doi:10.1137/080737113 · Cited by 113 · cited 1×
“Later, in [37], Y. Wang et al. proposed a fast algorithm for the ST with a time complexity of O ( N log N ) .”
BackgroundBall bearing damage detection using traditional signal processing algorithms
Bediaga, I., Mendizabal, X., Arnáiz, A., & Muñoa, J. (2013) · IEEE Instrumentation & Measurement Magazine · doi:10.1109/mim.2013.6495676 · Cited by 90 · cited 1×
“In real time, bearing fault diagnosis is performed through the collected data (i.e., vibration acceleration signals, acoustic emission signals, and motor currents), which has been an important area of study over the last few decades [6,9–11].”
BackgroundKim, B.-D., Lee, S. H., Lee, M. G., Ni, J., Song, J., & Lee, C. W. (2007) · Journal of Materials Processing Technology · doi:10.1016/j.jmatprotec.2006.11.222 · Cited by 87 · cited 1×
“Time—frequency signal analysis approaches have been employed to solve these issues, such as fast Fourier transform (FFT) and short-term-Fourier transform (STFT), which convert time-domain signals into frequency-domain signals for further analysis [20].”
MethodTahir, M. M., Khan, A. Q., Iqbal, N., Hussain, A., & Badshah, S. (2016) · IEEE Access · doi:10.1109/access.2016.2608505 · Cited by 80 · cited 1×
“Another powerful signal processing approach that has been considered is wavelet analysis [21].”
BackgroundTra, V., Kim, J., Khan, S. A., & Kim, J.-M. (2017) · Sensors · doi:10.3390/s17122834 · Cited by 76 · cited 1×
“This network makes optimal use of the indigenous connections (instead of fully-connected layers), weight distribution, and spatial or progressive subsampling to achieve invariance of shifting, scaling, and distortion in their inputs [39,40].”
BackgroundSelf-citationDiagnosis and Prognosis for Complicated Industrial Systems—Part I
Yin, S., Ding, S. X., & Zhou, D. (2016) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2016.2522944 · Cited by 72
Qin, X., Li, Q., Dong, X., & Lv, S. (2017) · Shock and Vibration · doi:10.1155/2017/2623081 · Cited by 64
Window-dependent bases for efficient representations of the Stockwell transform
Battisti, U., & Riba, L. (2015) · Applied and Computational Harmonic Analysis · doi:10.1016/j.acha.2015.02.002 · Cited by 61 · cited 1×
“, in [36], extended that work and provided an adapted basis to decompose the ST with a general admissible window.”
BackgroundKhan, S. A., & Kim, J.-M. (2016) · Shock and Vibration · doi:10.1155/2016/8729572 · Cited by 61 · cited 2×
“Bearings play an important role in condition monitoring because, in more than 50% of the cases involving induction motor failures, they are the root cause of the failure [2,8].”
BackgroundSelf-citationSohaib, M., & Kim, J.-M. (2018) · Shock and Vibration · doi:10.1155/2018/2919637 · Cited by 52 · cited 1×
“Examples include deep autoencoders [27,28], ANNs (artificial neural networks), and hierarchical convolution network [29].”
MethodSelf-citationAn energy kurtosis demodulation technique for signal denoising and bearing fault detection
Wang, W., & Lee, H. (2013) · Measurement Science and Technology · doi:10.1088/0957-0233/24/2/025601 · Cited by 51 · cited 1×
“Enveloping with energy kurtosis [23] is another approach.”
BackgroundHuang, W., Sun, H., & Wang, W. (2017) · Sensors · doi:10.3390/s17061279 · Cited by 44 · cited 1×
“These fault diagnosis studies have shown that diagnosis of bearing wear can reduce maintenance expenses and enhance machine reliability [6,12–16].”
BackgroundA Signal Theoretic Approach for Envelope Analysis of Real-Valued Signals
Yang, Y. (2017) · IEEE Access · doi:10.1109/access.2017.2688467 · Cited by 43 · cited 1×
“However, detecting the envelopes of real signals does not always rely on analytic signals [24].”
BackgroundNon-Mutually Exclusive Deep Neural Network Classifier for Combined Modes of Bearing Fault Diagnosis
Duong, B. P., & Kim, J.-M. (2018) · Sensors · doi:10.3390/s18041129 · Cited by 33 · cited 3×
“These flaws create shocks and stimulate high-occurrence reverberations of the bearings and machine assembly due to the recurring effect on spinning parts [17].”
BackgroundMethodSelf-citationTra, V., Kim, J., Khan, S. A., & Kim, J.-M. (2017) · The Journal of the Acoustical Society of America · doi:10.1121/1.4991329 · Cited by 32
Self-citationDiagnosis and Prognosis for Complicated Industrial Systems—Part II
Yin, S., Ding, S. X., & Zhou, D. (2016) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2016.2538745 · Cited by 27
Jiang, X., Wu, L., & Ge, M. (2017) · Entropy · doi:10.3390/e19050231 · Cited by 25
Speed Invariant Bearing Fault Characterization Using Convolutional Neural Networks
Appana, D. K., Ahmad, W., & Kim, J.-M. (2017) · Lecture notes in computer science · doi:10.1007/978-3-319-69456-6_16 · Cited by 24 · cited 1×
“Conventional, straightforward neural networks make the feature selection process much easier through their convoluted encoding layers compared to traditional methods [17,31,32].”
MethodSelf-citationKhan, S. A., & Kim, J.-M. (2016) · The Journal of the Acoustical Society of America · doi:10.1121/1.4945818 · Cited by 23 · cited 1×
“Bearing fault diagnosis studies have extensively investigated vibration signals and motor current analysis [17,18].”
BackgroundSelf-citation
Cited by
Applications of machine learning to machine fault diagnosis: A review and roadmap
Lei, Y., Yang, B., Jiang, X., Jia, F., Li, N., & Nandi, A. K. (2020) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2019.106587 · Cited by 2,997
Li, W., Huang, R., Li, J., Liao, Y., Chen, Z., He, G., Yan, R., & Gryllias, K. (2022) · Mechanical Systems and Signal Processing · doi:10.1016/j.ymssp.2021.108487 · Cited by 816
Artificial Intelligence in Advanced Manufacturing: Current Status and Future Outlook
Arinez, J. F., Chang, Q., Gao, R. X., Xu, C., & Zhang, J. (2020) · Journal of Manufacturing Science and Engineering · doi:10.1115/1.4047855 · Cited by 584
“Reported research on this topic this fi eld can be characterized into three categories: (1) model transfer among different working conditions [158,159], (2) model transfer among different machines [160], and (3) model transfer among different fault locations or types [161].”
Deep Transfer Learning for Bearing Fault Diagnosis: A Systematic Review Since 2016
Chen, X., Yang, R., Xue, Y., Huang, M., Ferrero, R., & Wang, Z. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3244237 · Cited by 516
“Besides replacing the output layer, Kim et al. [108] also proposed a selective parameter freezing (SPF) strategy to freeze output-sensitive parameters while tuning the model.”
MethodA systematic review of deep transfer learning for machinery fault diagnosis
Li, C., Zhang, S., Qin, Y., & Estupinan, E. (2020) · Neurocomputing · doi:10.1016/j.neucom.2020.04.045 · Cited by 472
Metadata
- Authors
- Hasan, M. J., & Kim, J.-M.
- Affiliations
- University of Ulsan (KR)
- Corresponding author
- Kim, Jong-Myon
- Venue
- Applied Sciences
- Publisher
- MDPI AG
- Published
- 2018-11-22
- Volume
- 8
- Issue
- 12
- Pages
- 2357
- ISSN
- 2076-3417 (ISSN-L)
- OpenAlex
- W2900791983
- Cited by
- 105
- References
- 41
- Citations by year
- 2026: 3 · 2025: 7 · 2024: 13 · 2023: 15 · 2022: 18 · 2021: 22 · 2020: 17 · 2019: 10
- FWCI
- 7.58
- Type
- journal article
- Language
- en
- Open access
- gold
- dx.doi.org
- Licence
- Locations
- Applied Sciences (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Open Access Institutional Repository at Robert Gordon University (Robert Gordon University) (submittedVersion, cc-by)
- Funders
- Korea Institute of Energy Technology Evaluation and Planning (20181510102160)
- Topics
- Physical Sciences › Engineering › Control and Systems Engineering › Machine Fault Diagnosis Techniques
- Physical Sciences › Engineering › Control and Systems Engineering › Fault Detection and Control Systems
- Physical Sciences › Engineering › Mechanical Engineering › Gear and Bearing Dynamics Analysis
- Keywords
- bearing fault diagnosis, variable rotational speed, S-transform, vibration imaging, transfer learning, convolutional neural networks, feature selection, inner race fault, ball fault, outer race fault, Case Western Reserve University bearing dataset
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.