Optical Fiber Vibration-Sensing Event Recognition Based on CLDNN

Zichun Zhou; Kun Liu; Junfeng Jing; Tianhua Xu; Shuang Wang; Zhenshi Sun; Hairuo Guo; Tiegen Liu

doi:10.3788/AOS202141.1306019

Acta Optica Sinica, Volume. 41, Issue 13, 1306019(2021)

Optical Fiber Vibration-Sensing Event Recognition Based on CLDNN

Zichun Zhou^1,2,3, Kun Liu^1,2,3、*, Junfeng Jing^1,2,3, Tianhua Xu^1,2,3, Shuang Wang^1,2,3, Zhenshi Sun^1,2,3, Hairuo Guo^1,2,3, and Tiegen Liu^1,2,3

Author Affiliations

¹School of Precision Instruments and Optoelectronics Engineering, Tianjin University, Tianjin 300072, China

²Key Laboratory of Optoelectronic Information Technology of Ministry of Education, Tianjin University, Tianjin 300072, China

³Institute of Optical Fiber Sensing, Tianjin University, Tianjin 300072, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Schematic of DMZI distributed optical fiber vibration sensing system

Download full size

Fig. 2. Equivalent optical path of DMZI distributed optical fiber vibration sensor

Download full size

Fig. 3. Network structure of CLDNN

Download full size

Fig. 4. DMZI distributed optical fiber vibration sensing system. (a) Sensing part; (b) demodulation part

Download full size

Fig. 5. Preprocessing of knocking. (a) Schematic of superposition process of 5 frames of signals; (b) schematic of signal misalignment clipping after superposition

Download full size

Fig. 6. Characteristic maps of 5 types of intrusion event signals. (a) No intrusion; (b) knocking; (c) crashing; (d) waggling; (e) kicking

Download full size

Fig. 7. Variation of error function values of CNN in CLDNN under three convolution levels

Download full size

Table 1. Structural design parameters of CLDNN

View table

Table 1. Structural design parameters of CLDNN

Layer	Kernelsize	Stride	Outputdimension	Function
Conv	5×5	1	(100,40,40,64)	Relu
Pooling	2×2	2	(100,20,20,64)	Max Pooling
Conv	5×5	1	(100,20,20,64)	Relu
Pooling	2×2	2	(100,10,10,64)	Max Pooling
Conv	5×5	1	(100,10,10,64)	Relu
Pooling	2×2	2	(100,5,5,64)	Max Pooling
Conv	5×5	1	(100,5,5,64)	Relu
Line	-	-	(100,256)	Line
LSTM	128	-	-	Tanh
FC	128		(100,5)	-
Output	-	-	(100,5)	Softmax

Table 2. Average recognition rate and average training time under different clipping conditions in three tests
View table
Table 2. Average recognition rate and average training time under different clipping conditions in three tests
Parameter N=20 N=30 N=40 N=50
Accuracy 1 /% 78.3 88.0 96.6 95.9
Accuracy 2 /% 75.0 87.4 96.2 94.7
Accuracy 3 /% 72.9 89.3 96.1 95.1
Time /s 2034 3004 5420 8982

Table 3. Comparison of recognition results of three algorithms
View table
Table 3. Comparison of recognition results of three algorithms
Algorithm Average accuracy /% Number of events Preprocessing time /s Identification time /s
CLDNN 96.24 5 5×10^-5 0.006
RBF-EMD 85.75 4 1.14 0.510
SVM 93.82 4 0.30 0.300

Table 4. [in Chinese]
View table
Table 4. [in Chinese]
Algorithm No intrusion Knocking Crashing Waggling Kicking
CLDNN 100.0 99.8 95.2 87.5 98.7

Tools

Get Citation

Copy Citation Text

Zichun Zhou, Kun Liu, Junfeng Jing, Tianhua Xu, Shuang Wang, Zhenshi Sun, Hairuo Guo, Tiegen Liu. Optical Fiber Vibration-Sensing Event Recognition Based on CLDNN[J]. Acta Optica Sinica, 2021, 41(13): 1306019

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites