Small-batch-size convolutional neural network based fault diagnosis system for nuclear energy production safety with big-data environment

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journalpeer-review

26 Scopus Citations
View graph of relations

Author(s)

  • Yuantao Yao
  • Jin Wang
  • Pengcheng Long
  • Min Xie
  • Jianye Wang

Detail(s)

Original languageEnglish
Pages (from-to)5841-5855
Journal / PublicationInternational Journal of Energy Research
Volume44
Issue number7
Online published17 Mar 2020
Publication statusPublished - 10 Jun 2020

Abstract

In nuclear energy production, with the continuous innovations and challenges in the big data and the industry 4.0 era, to guarantee the operation safety without the fault and failure will become more complex and intelligent. In this paper, a novel optimized convolutional neural network with small-batch-size processing (SCNN) was proposed and assembled in the nuclear fault diagnosis system. Eleven kinds of normal and fault conditions that include the whole 316 simulator sensor features were used to evaluate the performance of the proposed diagnosis system. The application of batch normalization with SCNN significantly optimized the model validation accuracy and loss under 100 epochs compared with normal operation and adding drop-out operation in same condition. Besides, outstanding diagnosis accuracy was highlighted by the comparison of traditional binary and multiple classification methods. This proposed diagnosis system has achieved more precise diagnosis accuracy and will provide the useful guidance to operators, assisting them to make accurate and rapid decision to ensure nuclear energy production safety.

Research Area(s)

  • convolution layer visualization, convolutional neural network, deep learning, fault diagnosis, nuclear energy production, small-batch-size processing

Citation Format(s)

Small-batch-size convolutional neural network based fault diagnosis system for nuclear energy production safety with big-data environment. / Yao, Yuantao; Wang, Jin; Long, Pengcheng et al.

In: International Journal of Energy Research, Vol. 44, No. 7, 10.06.2020, p. 5841-5855.

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journalpeer-review