Advanced search
1 file | 327.36 KB Add to list

Data preparation for training CNNs : application to vibration-based condition monitoring

Author
Organization
Abstract
Vibration data is one of the most informative data to be used for fault detection. It mostly employs in the form of frequency response function (FRF) for training deep learners. However, since normally the FRFs are measured at excessive numbers of frequencies, its usage not only enforces large computational resources for training the deep learners, but could also hinder proper feature extraction. In this paper, it is shown that given a predefined deep learning structure and its associated hyperparameters, how proper data selection and/or augmentation could improve the performance of the trained model in classifying the samples. For this purpose, the least absolute shrinkage and selection operator (LASSO) and some generative functions are utilized respectively for data selection/reduction and augmentation prior to any training. The efficacy of this procedure is illustrated by applying it to an experimental dataset created by the broadband vibrational responses of polycrystalline Nickel alloy first-stage turbine blades with different types and severities of damages. It is shown that the data selection and augmentation approach could improve the performance of the model to some extent and at the same time, drastically reduce the computational time.
Keywords
Fault detection, vibrations, deep learning, data selection, LASSO

Downloads

  • 103 CameraReady Yaghoubi DCAI CR.pdf
    • full text (Published version)
    • |
    • open access
    • |
    • PDF
    • |
    • 327.36 KB

Citation

Please use this url to cite or link to this publication:

MLA
Yaghoubi Nasrabadi, Vahid, et al. “Data Preparation for Training CNNs : Application to Vibration-Based Condition Monitoring.” 1st NeurIPS Data-Centric AI Workshop (DCAI 2021), Proceedings, 2021.
APA
Yaghoubi Nasrabadi, V., Cheng, L., Van Paepegem, W., & Kersemans, M. (2021). Data preparation for training CNNs : application to vibration-based condition monitoring. 1st NeurIPS Data-Centric AI Workshop (DCAI 2021), Proceedings. Presented at the 1st NeurIPS Data-Centric AI workshop (DCAI 2021), Online.
Chicago author-date
Yaghoubi Nasrabadi, Vahid, Liangliang Cheng, Wim Van Paepegem, and Mathias Kersemans. 2021. “Data Preparation for Training CNNs : Application to Vibration-Based Condition Monitoring.” In 1st NeurIPS Data-Centric AI Workshop (DCAI 2021), Proceedings.
Chicago author-date (all authors)
Yaghoubi Nasrabadi, Vahid, Liangliang Cheng, Wim Van Paepegem, and Mathias Kersemans. 2021. “Data Preparation for Training CNNs : Application to Vibration-Based Condition Monitoring.” In 1st NeurIPS Data-Centric AI Workshop (DCAI 2021), Proceedings.
Vancouver
1.
Yaghoubi Nasrabadi V, Cheng L, Van Paepegem W, Kersemans M. Data preparation for training CNNs : application to vibration-based condition monitoring. In: 1st NeurIPS Data-Centric AI workshop (DCAI 2021), Proceedings. 2021.
IEEE
[1]
V. Yaghoubi Nasrabadi, L. Cheng, W. Van Paepegem, and M. Kersemans, “Data preparation for training CNNs : application to vibration-based condition monitoring,” in 1st NeurIPS Data-Centric AI workshop (DCAI 2021), Proceedings, Online, 2021.
@inproceedings{8732647,
  abstract     = {{Vibration data is one of the most informative data to be used for fault detection. It mostly employs in the form of frequency response function (FRF) for training deep learners. However, since normally the FRFs are measured at excessive numbers of frequencies, its usage not only enforces large computational resources for training the deep learners, but could also hinder proper feature extraction. In this paper, it is shown that given a predefined deep learning structure and its associated hyperparameters, how proper data selection and/or augmentation could improve the
performance of the trained model in classifying the samples. For this purpose, the least absolute shrinkage and selection operator (LASSO) and some generative functions are utilized respectively for data selection/reduction and augmentation prior to any training. The efficacy of this procedure is illustrated by applying it to an experimental dataset created by the broadband vibrational responses of polycrystalline Nickel alloy first-stage turbine blades with different types and severities of damages. It is shown that the data selection and augmentation approach could improve the performance of the model to some extent and at the same time, drastically reduce the computational time.}},
  author       = {{Yaghoubi Nasrabadi, Vahid and Cheng, Liangliang and Van Paepegem, Wim and Kersemans, Mathias}},
  booktitle    = {{1st NeurIPS Data-Centric AI workshop (DCAI 2021), Proceedings}},
  keywords     = {{Fault detection,vibrations,deep learning,data selection,LASSO}},
  language     = {{eng}},
  location     = {{Online}},
  pages        = {{5}},
  title        = {{Data preparation for training CNNs : application to vibration-based condition monitoring}},
  url          = {{https://datacentricai.org/neurips21/papers/103_CameraReady_Yaghoubi_DCAI_CR.pdf}},
  year         = {{2021}},
}