Acta Photonica Sinica, Volume. 50, Issue 9, 0906003(2021)

Optical Fiber Perimeter Intrusion Pattern Recognition Based on 1D-CNN

Houdan YU, Qiushi MI, Dong ZHAO, and Qian XIAO*
Author Affiliations
  • Optical Fiber Research Center, Department of Material Science, Fudan University, Shanghai200433, China
  • show less
    Figures & Tables(13)
    Schematic diagram of line-based Sagnac interference structure
    The intervention signal
    Structure of convolution neural network
    Schematic diagram of the experiment system
    Six types of demodulated signal and their power spectrum
    Six kernels after training
    Mel filters
    Features extracted by MFCC method
    Cutting signal after convolution
    Branch tapping signal after convolution
    Cutting features after batch normalized
    Branch tapping features after batch normalized
    • Table 1. Recognition result in the test set by 1D-CNN

      View table
      View in Article

      Table 1. Recognition result in the test set by 1D-CNN

      Type of signalCorrect classificationNumber in test setRecognition rate of 1D-CNNRecognition rate of MFCC
      Walking45346098.48%97.39%
      Branch tapping54958094.66%81.03%
      Knocking62866095.15%91.21%
      Cutting55356098.75%96.79%
      Waggling50252096.54%89.04%
      Wind blowing36338095.52%97.37%
      Total3 0483 16096.46%91.61%
    Tools

    Get Citation

    Copy Citation Text

    Houdan YU, Qiushi MI, Dong ZHAO, Qian XIAO. Optical Fiber Perimeter Intrusion Pattern Recognition Based on 1D-CNN[J]. Acta Photonica Sinica, 2021, 50(9): 0906003

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Fiber Optics and Optical Communications

    Received: Jan. 22, 2021

    Accepted: Mar. 25, 2021

    Published Online: Oct. 22, 2021

    The Author Email: XIAO Qian (ychunww@163.com)

    DOI:10.3788/gzxb20215009.0906003

    Topics