Acta Photonica Sinica, Volume. 50, Issue 3, 50(2021)
Phase-sensitive Optical Time-domain Reflectometric System Pattern Recognition Method Based on Wavenet
Figures & Tables(13)
Fig. 3. Waveforms of three vibration signals act separately and simultaneously at three positions
Fig. 5. Structrue of LSTM, one-dimensional convolution, causal convolution
Table 1. Test sets' recognition accuracy rate of three structures
View tableView in ArticleTable 1. Test sets' recognition accuracy rate of three structures
Network model True label Prediction label Accuracy rate/% Foot stepping/% Hand tapping/% Stick striking/% One-dimensional CNN Foot stepping 93.35 3.47 3.18 92.75 Hand tapping 1.06 95.47 3.47 Stick striking 5.44 5.14 89.42 LSTM Foot stepping 97.89 0.60 1.51 99.24 Hand tapping 0.00 100.00 0.00 Stick striking 0.00 0.15 99.85 Wavenet Foot stepping 100 0.00 0.00 99.85 Hand tapping 0.30 99.70 0.00 Stick striking 0.00 0.15 99.85
Table 2. Three structures' training and testing efficiency
View tableView in ArticleTable 2. Three structures' training and testing efficiency
Network model Iteration speed/（ms/step） One Epoch's time/s Numebers of training epoch Training time/s Testing time/ms One-dimensional CNN 12 2 60 120 12 LSTM 270 42 40 1 680 61 Wavenet 107 16 6 96 30
Tools
Get Citation
Copy Citation Text
Yaolu ZHANG, Miao YU, Tianying CHANG, Shufan LI, Zhifeng ZHENG, Yue YANG, Zhongmin WANG, Hongliang CUI. Phase-sensitive Optical Time-domain Reflectometric System Pattern Recognition Method Based on Wavenet[J]. Acta Photonica Sinica, 2021, 50(3): 50
Paper Information
Category: Fiber Optics and Optical Communications
Received: --
Accepted: --
Published Online: Jul. 13, 2021
The Author Email: CHANG Tianying (tchang@jlu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20