• Acta Optica Sinica
  • Vol. 44, Issue 1, 0106026 (2024)
Yin Zhang1, Ting Hu1, Youxing Li2, Jian Wang1, and Libo Yuan1、*
Author Affiliations
  • 1School of Optoelectronic Engineering, Guilin University of Electronic Technology, Guilin 541004, Guangxi, China
  • 2College of Physics and Optoelectronic Engineering, Harbin Engineering University, Harbin 150006, Heilongjiang, China
  • show less
    DOI: 10.3788/AOS231392 Cite this Article Set citation alerts
    Yin Zhang, Ting Hu, Youxing Li, Jian Wang, Libo Yuan. Pattern Recognition of Phase-Sensitive Optical Time-Domain Reflectometer Based on Conditional Generative Adversarial Network Data Augmentation[J]. Acta Optica Sinica, 2024, 44(1): 0106026 Copy Citation Text show less
    Signal collection in Φ-OTDR system
    Fig. 1. Signal collection in Φ-OTDR system
    Model architecture of generative adversarial network (GAN)
    Fig. 2. Model architecture of generative adversarial network (GAN)
    Model architecture of conditional generative adversarial network (CGAN)
    Fig. 3. Model architecture of conditional generative adversarial network (CGAN)
    Realistic view of experimental plant site
    Fig. 4. Realistic view of experimental plant site
    Network architecture of generator
    Fig. 5. Network architecture of generator
    Network architecture of discriminator
    Fig. 6. Network architecture of discriminator
    Flow chart of CGAN training
    Fig. 7. Flow chart of CGAN training
    Loss diagrams of generators and discriminators
    Fig. 8. Loss diagrams of generators and discriminators
    Display of generating a signal (200 rounds at a time)
    Fig. 9. Display of generating a signal (200 rounds at a time)
    Comparison of generated and real signals. (a) Vehicles passing by; (b) ambient noise; (c) pedestrians passing by; (d) artificial destruction
    Fig. 10. Comparison of generated and real signals. (a) Vehicles passing by; (b) ambient noise; (c) pedestrians passing by; (d) artificial destruction
    Comparison of histograms of raw and generated data. (a) Raw data; (b) generated data
    Fig. 11. Comparison of histograms of raw and generated data. (a) Raw data; (b) generated data
    Comparison of detection results between raw and generated data
    Fig. 12. Comparison of detection results between raw and generated data
    Comparison of confusion matrix plots between raw and generated data
    Fig. 13. Comparison of confusion matrix plots between raw and generated data
    Test results of real data
    Fig. 14. Test results of real data
    MethodAccuracy /%
    Ref.[892.43
    Refs.[232494.43
    Ref.[25]96.10
    Proposed method98.22
    Table 1. Comparison of detection results of different methods
    Yin Zhang, Ting Hu, Youxing Li, Jian Wang, Libo Yuan. Pattern Recognition of Phase-Sensitive Optical Time-Domain Reflectometer Based on Conditional Generative Adversarial Network Data Augmentation[J]. Acta Optica Sinica, 2024, 44(1): 0106026
    Download Citation