Designing a lightweight 1D convolutional neural network with Bayesian optimization for wheel flat detection using carbody accelerations

References

• Leiste M, Hecht M, Jobstfinke D, Deghela U. Roadmap zur Digitalisierung der Wagentechnischen Untersuchung. Chair of Rail Vehicles. Intern report 16/2018. Technische Universität Berlin; 2018.
   Google Scholar
• GCU. General contract of use for wagons – GCU. Edition dated. 1 January 2018.
   Google Scholar
• Bernal E, Spiryagin M, Cole C. Onboard condition monitoring sensors, systems and techniques for freight railway vehicles: a review.IEEE. Sens J. 2019;19:4–24.
   Web of Science ®Google Scholar
• Schaeffler. Axlebox generator for railway applications.2018. [cited 2020 Jan28]. Available from: https://www.schaeffler.com/remotemedien/media/_shared_media/08_media_library/01_publications/schaeffler_2/brochure/downloads_1/org_de_en.pdf
   Google Scholar
• Wheelwright HE, Vincent D. Track defect and wheel damage: detection and location. Perpetuum’s white paper. 2019 [cited 2020 Jan 28]. Available from: https://perpetuum.com/download/track-defect-and-wheel-damage-detection-and-location/?wpdmdl=1480&refresh=5e302ed474a071580216020
   Google Scholar
• DB Cargo. Mit intelligenten und leisen Güterwagen für Wachstum im Schienengüterverkehr. Deutsche Bahn. Feb 2018. Available from: https://www.deutschebahn.com/de/presse/pressestart_zentrales_uebersicht/DB-Cargo–Mit-intelligenten-und-leisen-G%C3%BCterwagen-f%C3%BCr-Wachstum-im-Schieneng%C3%BCterverkehr-1440210
   Google Scholar
• Harms C. Im minutentakt: wie VTG die Güter aus dem Schwarzen Loch holt. Allianz pro Schiene.2017.[cited 2020 Jan 28]. Available from: https://www.allianz-pro-schiene.de/themen/aktuell/interview-hanno-schell-digitalisierung-gueterwagen-vtg/
   Google Scholar
• Shi D, Marin-Perianu R, Marin-Perianu M, Vincent D, Ulianov C, Sutthaweekul R, Hamer A, Tian G. Wireless data communication concept for cargo condition monitoring system. Technical report. Deliverable D2.3. EU Shift2Rail project; Grand Nr:730863-S2-OC-IP5-03-2015; 2018
   Google Scholar
• Alemi A, Corman F, Lodewijks G. Condition monitoring approaches for the detection of railway wheel defects. Proc Inst Mech Eng F J Rail Rapid Transit. 2016;231:961–981.
   Web of Science ®Google Scholar
• Jiang H, Lin J. Fault diagnosis of wheel flat using empirical mode decomposition-Hilbert envelope spectrum. Math Prob Eng. 2018;2018:1–16.
   Web of Science ®Google Scholar
• Zhao R, Shi H. Research on wheel-flat recognition algorithm for high-speed train based on high-order spectrum feature extraction. J Mech Eng. 2017;53:102.
   Google Scholar
• Li Y, Zuo MJ, Lin J. Fault detection method for railway wheel flat using an adaptive multiscale morphological filter. Mech Syst Signal Process. 2017;84:642–658.
   Web of Science ®Google Scholar
• Liang B, Iwnicki SD, Zhao Y, et al. Railway wheel-flat and rail surface defect modelling and analysis by time–frequency techniques. Veh Syst Dyn. 2013;51:1403–1421.
   Web of Science ®Google Scholar
• Gericke C Methoden zur on-board-Diagnose von Radlaufflächenschäden: ein Beitrag zur zustandsorientierten Instandhaltung von Schienenfahrzeugen. Dissertation; Technische Universität Berlin. 2013
   Google Scholar
• Bosso N, Gugliotta A, Zampieri N. Wheel flat detection algorithm for onboard diagnostic. Measurement. 2018;123:193–202.
   Web of Science ®Google Scholar
• Bernal E, Spiryagin M, Cole C. Wheel flat detectability for Y25 railway freight wagon using vehicle component acceleration signals. Veh Syst Dyn. 2019;1–21. DOI:10.1080/00423114.2019.1657155
   Web of Science ®Google Scholar
• Shi D, Bruni S, Stefano A, Hyde P, Regazzi D, Vincent D. Models for reliability, statistical information, real time health status of the rolling stock, prognostic analysis, and economic data. Technical report. Deliverable D4.2. EU Shift2Rail project. Grand no: 730863-S2R-OC-IP5-03-2015; 2018
   Google Scholar
• Jangamreddy N A survey on specialised hardware for machine learning. 2019. DOI: 10.13140/RG.2.2.20697.26725
   Google Scholar
• Cheng Y, Wang D, Zhou P, Zhang T. A survey of model compression and acceleration for deep neural networks. IEEE signal processing magazine; 2019. Special issue on deep learning for image understanding (arxiv extended version).
   Google Scholar
• Landola F, Han S, Moskewicz MW, Ashraf K, Dally WJ, Keutzer K. Squeezenet: AlexNet-level accuracy with 50x fewer parameters and <0.5mb model size. arxiv.org. 2016. p. arXiv:1602.07360v4.
   Google Scholar
• Chollet F. Xception: deep learning with depthwise separable convolutions. arxiv.org. 2016. p. arXiv:1610.02357v3.
   Google Scholar
• Zhang X, Zhou X, Lin M, et al. Shufflenet: an extremely efficient convolutional neural network for mobile devices. In Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition. Jun19-21; Salt Lake City; 2018.
   Google Scholar
• Howard AG, Zhu M, Chen B, Kalenichenko D, Wang W, Weyand T, Andreetto M, Adam H. Mobilenets: efficient convolutional neural networks for mobile vision applications. arxiv.org. 2017: p. arXiv:1704.04861.
   Google Scholar
• Ma N, Zhang X, Zheng H, Zheng H, Sun J. Shufflenet v2: practical guidelines for efficient CNN architecture design. 2018. p. arXiv:1807.11164v1.
   Google Scholar
• Sandler M, Howard A, Zhu M, Zhmoginov A, Chen L. Mobilenetv2: inverted residuals and linear bottlenecks. arxiv.org. 2018: p. arXiv:1801.04381.
   Google Scholar
• Howard A, Sandler M, Chu G, Chen L, Chen B, Tan M, Wang W, Zhu Y, Pang R, Vasudevan V, Le Q, Adam H. Searching for MobileNetV3. 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Oct 27-Nov 02; Seoul; 2019.
   Google Scholar
• Passalis N, Tefas A. Training lightweight deep convolutional neural networks using bag-of-features pooling. IEEE Trans Neural Netw Learn Syst. 2019;30:1705–1715.
   PubMed Web of Science ®Google Scholar
• Lu Z, Qin S, Li L, et al. One-shot learning hand gesture recognition based on lightweight 3D convolutional neural networks for portable applications on mobile systems. IEEE Access. 2019;7:131732–131748.
   Web of Science ®Google Scholar
• Yu R, Xu X, Shen Y. Rhnet: lightweight dilated convolutional networks for dense objects counting. Chin Contr Conf. 2019;2019:8455–8459.
   Google Scholar
• Cui Y, Shi Y, Sun X, Yin W. S-net: a lightweight convolutional neural network for n-dimensional signals. 2018 IEEE International Conference on Multimedia & Expo Workshops (ICMEW). Jul 23–27; San Diego; 2018; 1–4.
   Google Scholar
• Li D, Zeng S, Zhou Y, Pan B, Yang L. Design of high performance convolutional neural network for lightweight platform. 2019 IEEE Fourth International Conference on Data Science in Cyberspace (DSC). Jun 23–25; Hangzhou; 2019; 1–8.
   Google Scholar
• Xiang H, Chen L, Xu W. DrfNet: a lightweight and high accuracy network for resource-limited implementation. 2017 IEEE 12th International Conference on ASIC (ASICON). Oct 25-27; Guiyang; 2017; 1086–1089.
   Google Scholar
• Zhang X, Zou Y, Wang W. Ld-CNN: a lightweight dilated convolutional neural network for environmental sound classification. 2018 24th International Conference on Pattern Recognition (ICPR). Aug 20–24; Berijing; 2018; 373–378.
   Google Scholar
• Liu H, Yao D, Yang J, et al. Lightweight convolutional neural network and its application in rolling bearing fault diagnosis under variable working conditions. Sensors (Basel). 2019 6;19(22):4827. .
   Web of Science ®Google Scholar
• Wang B, Lei Y, Li N, et al. Deep separable convolutional network for remaining useful life prediction of machinery. Mech Syst Signal Process. 2019;134:106330.
   Web of Science ®Google Scholar
• Ma S, Liu W, Cai W, et al. Lightweight deep residual CNN for fault diagnosis of rotating machinery based on depthwise separable convolutions. IEEE Access. 2019;7:57023–57036.
   Web of Science ®Google Scholar
• Ma S, Cai W, Liu W, et al. A lighted deep convolutional neural network based fault diagnosis of rotating machinery. Sensors (Basel). 2019;19(10):2381.
   Web of Science ®Google Scholar
• Iwnicki SD, Stichel S, Orlova A, et al. Dynamics of railway freight vehicles. Veh Syst Dyn. 2015;53:995–1033.
   Web of Science ®Google Scholar
• Velasco-Montero D, Fernández-Berni J, Carmona-Galan R, Rodriguez-Vazquez A. Performance analysis of real-time DNN inference on Raspberry Pi. Real Time Image Video Process. Apr 16; Florida; 2018.
   Google Scholar
• Shi D, Ulianov C, Sutthaweekul R, Hamer A, Cheng L, Vincent D. Energy concept for cargo condition monitoring. Technical report; Deliverable D2.2; EU Shift2Rail project; Grand no: 730863 - S2R-OC-IP5-03-2015; 2019.
   Google Scholar
• Frigo M, Johnson S.G. FFT benchmark methodology. 2020 [Cited 2020 Feb 03]. Available from: http://www.fftw.org/speed/method.html
   Google Scholar
• Munoz A, Ertle R, Unser M. Continuous wavelet transform with arbitrary scales and O(n) complexity. Signal Process. 2002;82:749–757.
   Web of Science ®Google Scholar
• Sundararajan D Fundamentals of the discrete Haar wavelet transform. 2011[Cited 2020 Feb 303]. Available from: https://www.dsprelated.com/Documents/d_sundararajan_lpaper.pdf
   Google Scholar
• Wang YH, Yeh CH, Young HWV, et al. On the computational complexity of the empirical mode decomposition algorithm. Phys A Stat Mech Appli. 2014;400:159–167.
   Web of Science ®Google Scholar
• Louppe G Understanding random forests: from theory to practice. PhD dissertation. arXiv:1407.7502v3. 2015.
   Google Scholar
• Claesen M, Smet FD, Suykens JAK, et al. Fast prediction with SVM models containing RBF kernels. arxiv.org. 2014. p. arXiv:1403.0736.
   Google Scholar
• Zhang F, Webb GI A comparative study of semi-naive Bayes methods in classification learning. Proceedings of the Fourth Australasian Data Mining Workshop; Dec 05–06; Sydney; 2005; 141–156.
   Google Scholar
• Rasmussen C, Williams C. Gaussian processes for machine learning. MIT Press. 2006 [Cited 2020 Feb 03]. ISBN 026218253X. Available from: http://www.gaussianprocess.org/gpml/chapters/RW.pdf
   Google Scholar
• LeCun Y, Boser B, Denker J.S, Henderson D, Howard R.E, Hubbard W, Jackel L.D. Handwritten digit recognition with a back-propagation network. Proc Adv Neural Inf Process Syst. 1990; 02: 396–404.
   Google Scholar
• Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift. arxiv.org. 2015. p. arXiv:1502.03167.
   Google Scholar
• Hu J, Shen L, Albanie S, et al. Squeeze-and-excitation networks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition; Jun 19–21; Salt Lake City; 2018; 7132–7141.
   Google Scholar
• He K, Zhang X, Ren S, Sun J. Deep residual learning for image recognition. Proc IEEE Conf Comput Vis Pattern Recognit(CVPR); Jun 27–30; Las Vegas; 2016;770–778.
   Google Scholar
• Veit A, Wilber M, Belongie S. Residual networks behave like ensembles of relatively shallow networks. arxiv.org. 2016. p. arXiv:1605.06431.
   Google Scholar
• Maas A.L, Hannun A.Y, Andrew Y. Rectifier nonlinearities improve neural network acoustic models. In: Proc. 30th International Conference on Machine Learning (ICML); Jun 16–21; Atlanta; 2013. p. 30.
   Google Scholar
• Djork-Arné C, Unterthiner T, Hochreiter S. Fast and accurate deep network learning by exponential linear units (ELUS). arxiv.org. 2016. p. arXiv:1511.07289.
   Google Scholar
• Ramach P, Zoph B, Le Q. Searching for activation functions. arxiv.org. 2017. p. arXiv:1710.05941.
   Google Scholar
• Misra D. Mish: a self-regularized non-monotonic neural activation function. arxiv.org. 2019. p. arXiv:1908.08681.
   Google Scholar
• Martinez-Cantin R. Funneled Bayesian optimization for design, tuning and control of autonomous systems. IEEE Trans Cybern. 2019;49:1489–1500.
   PubMed Web of Science ®Google Scholar
• Kingma D, Ba J Adam: a method for stochastic optimisation. 3rd International Conference on Learning Representations (ICLR); San Diego; 2015; 1–13
   Google Scholar
• Lin M, Chen Q, Yan S. Network in network. arxiv.org. 2014. p. arXiv:1312.4400.
   Google Scholar
• Kou L, Qin Y, Zhao X, et al. Integrating synthetic minority oversampling and gradient boosting decision tree for bogie fault diagnosis in rail vehicles. Proc Inst Mech Eng F J Rail Rapid Transit. 2018;233(3):312–325.
   Web of Science ®Google Scholar
• Ren Z. An investigation on wheel/rail impact dynamics with a three-dimensional flat model. Veh Syst Dyn. 2018;57(3):369–388.
   Web of Science ®Google Scholar
• Wang X, Makis V, Yang M. A wavelet approach to fault diagnosis of a gearbox under varying load conditions. J Sound Vib. 2010;329:1570–1585.
   Web of Science ®Google Scholar
• Jia F, Lei Y, Lu N, et al. Deep normalised convolutional neural network for imbalanced fault classification of machinery and its understanding via visualisation. Mech Syst Signal Process. 2018;110:349–367.
   Web of Science ®Google Scholar
• Zhang W, Li C, Peng G, et al. A deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working loads. Mech Syst Signal Process. 2018;100:439–453.
   Web of Science ®Google Scholar