IOP Conference Series: Earth and Environmental Science

PAPER • OPEN ACCESS You may also like

- An educational review on distributed optic
Application of wavelet analysis in distributed fiber sensing based on Rayleigh
backscattering for damage tracking and
optical fiber Brillouin temperature strain monitoring structural health monitoring
L Chamoin, S Farahbakhsh and M
Poncelet
system
- Detection of the abnormal events along
the oil and gas pipeline and multi-scale
To cite this article: Zhou Feng et al 2018 IOP Conf. Ser.: Earth Environ. Sci. 189 032026 chaotic character analysis of the detected
signals
Zhigang Qu, Yan Zhou, Zhoumo Zeng et
al.

- Axial strain monitoring method of cast-in-

View the article online for updates and enhancements. place piles based on ultra-weak fiber
Bragg grating
Zhihui Luo, Zekun Lin, Shuai Ding et al.

This content was downloaded from IP address 1.65.223.39 on 03/11/2023 at 12:29

2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

Application of wavelet analysis in distributed optical fiber

Brillouin temperature strain monitoring system

Zhou Feng1,2 GanJianjun3* LvHuan2 CuiLigang2

1
Water Conservancy and Civil Engineering Infrastructure Safety Key Laboratory,
Nanchang Institute of Technology, Nanchang, China, 330099
2
Civil and Architectural Engineering School, Nanchang Institute of Technology,
330099 Nanchang, China
3
School of Water Conservancy and Ecological Engineering, Nanchang Institute of
Technology, 330099
E-mail：[email protected]

Abstract: In the distributed optical fiber monitoring system, the Brillouin system is a system
with the most difficult integration, the weakest signal and the most widely used range. It can
monitor the strain and temperature of single fiber cable at the same time, and the monitoring
range can exceed 100km. It is widely used in super engineering such as submarine optical
cable, submarine tunnel, oil pipeline and so on. It is of great research value since this system
can monitor the physical quantity of micro deformation, crack and temperature in real time.
After the Brillouin system is built independently, the high precision extraction of Brillouin
spectrum is realized by using the high efficiency denoising ability of wavelet. By studying the
algorithm, the system can reach the temperature identification accuracy of ± 1 degrees
centigrade, and the accuracy of strain reaches ± 20 με 。

1. Foreword
The distributed Brillouin optical fiber system can monitor the temperature and strain at the same time
by using the Brillouin scattering principle and the BOTDA technology. Such system is distributive,
real-time and accurate. Due to the spacial layout of optical fiber, undetection can be avoided. High
precision technology can effectively analyze and capture distributed distance information. It is easy to
realize automation efficiently and accurately, and the distance of monitoring fiber can even reach
hundreds of kilometers.In 2009, Zhao Lijuan[1] analyzed the key factors affecting the performance of
Brillouin system, and proposed a new method to improve the performance of Brillouin system. In
2010, Liu Jianfu[2] developed the Brillouin monitoring system, which has been applied to practice and
achieved good results. In 2011, Huang Min Shuang[3] et al. Proved that Brillouin system can detect the
physical information of the target through the frequency shift of the detected signal. In the same year,
Cao Jianmei[4] designed a Brillouin grid structure system. In 2012, Liu Yangjun[5] and others analyzed
the characteristics of Brillouin signals and proposed the gain of Brillouin signals.With the rapid
development of optical fiber sensor system, the monitoring technology of fiber sensor system for civil
engineering structures is getting closer to maturity. In 2013, the application of distributed optical fiber
sensing technique to the deformation monitoring of No. I landslide in Majiagou,The Three Gorges,
was applied by Sun Yijie[6] and so on. The results showed that the optical fiber sensor system had a

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

good ability to monitor the abnormal deformation value but lacks an early warning system.In 2014, Lu
Yi[7] used the distributed optical fiber monitoring system to monitor the ground cracks, which made
the monitoring space precision to 0.1mm, but the way of storage was fixed-point placement and failed
to be stereoscopic.
Wavelet de-noising is a popular denoising method in recent years. It can make up for the lack of
signal recognition in the low frequency segment of the traditional denoising method. It can greatly
save the response time of the system and play a good de-noising effect. Moreover, wavelet denoising
can achieve high resolution requirements. It can control multi-resolution and obtain detailed
information in time domain and frequency domain. Therefore, the time windows can be adjusted
accordingly for different signals and analysis states.In 1981, Stromberg improved Haar wavelet and
established wavelet and wavelet basis functions similar toHaar wavelets. In 1990, Cui Jintai and Wang
Jianzhong[8] used this method to construct a spline wavelet. Because the expression of the wavelet is
very clear, it can be used in the process of programming. In 2000, Yuan Changmao, Wen Hongyan[9]
and others used the wavelet denoising method to denoise the deformation monitoring data of buildings
and obtained the trend information of buildings. Due to this, the application of wavelet denoising in
buildings has become more and more frequent. Based on this, various papers and techniques emerge in
endlessly. The two difficult directions of wavelet analysis are the wavelet denoising[10]and image
compression technology[11]. Combined with the characteristics of wavelet denoising method, this paper
adopts wavelet denoising method to reduce Brillouin scattering signal denoising.

2. Experimental Parts

2.1 System Components and Devices

Brillouin system includes system components and test calibration devices. The calibration device
includes thermostat and high-precision strain testing device. The temperature range of the thermostat
is between room temperature ~ 300 degrees centigrade,the resolution and accuracy are 1 degrees
centigrade. The strain gauge accuracy of strain testing device is 20 micro strain. As shown in Figures 1,
2, and 3.

(a)System Component (b)Physical Optical

Fiber
Figure 1 BOTDA System Physical Diagram

Figure 2 Thermostat

2
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

(a) (b)
Figure 3 Strain Testing Device

2.2 Testing Methods

The distributed Brillouin optical fiber system is shown in Figure 4. The length of the distributed
Brillouin optical fiber system is 15km, the frequency range is from 10550MHz to 10850MHz, the
collected data is added 1000 times, the room temperature is 20 degrees centigrade, and at the
beginning the fiber is relaxed. A 30m range of optical fiber 10800m is being placed in a constant
temperature water bath, and it is heated to 40, 50, 60, 70, 75 and 80 degrees centigrade respectively.
Then around 10m range ofoptical fiber10830m is being woundround the strain device (Fig. 3), and
500 με , 1500 με , 2500 με , 3500 με and 4500 με of strain are applied respectively at room
temperature to find out the relationship coefficient of frequency shift with temperature, frequency shift
with strain, power with temperature, and power with strain. With the formula of relationship
coefficient between temperature and strain with frequency and power obtained through wavelet
denoising and traditional fitting method, the temperature and strain of the corresponding position point
can be analyzed.

Figure4 BOTDA System Structure Diagram

3. Experiment and Analysis

The data obtained from the Brillouin system at 70 degrees centigrade is shown in Figure 5:

Figure 5 70℃3D Magnification

Figure 6 extracts the scatter plot after the calibration of 1000m position in frequency domain direction,

3
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

and combines Lorenz's fitting with scatter plot and Lorenz fitting formula
h(ω / 2 )
2
f ( x, θ ) = (1)
(x − ν )2 + (ω / 2)2
as shown in Figure 6：

Figure6 Fitting Effect Diagram

The experiment carried out a data processing on threshold denoising in frequency

domainsrespectively under 40, 50, 60, 70 and 80 degrees centigrade, and the data of a total 30m in the
range of about 10800 wasaveraged.Frequency comparison fitted before and after noise reduction is as
shown in Table 1:

Table1Comparison of Frequency VariationCorresponding to Temperature Change Before and After

Filtering
Temperature Temperature Original Data Original Data Frequency Frequency
Value Variation（℃） Frequency Frequency Value After Variation Value
（℃） Value（MHz） Variation Denoising After
Value（MHz） （MHz） Denoising
（MHz）
20 0 10608 0 10608 0
40 20 10630 22.37 10630 22.32
50 30 10640 31.49 10639 31.39
60 40 10649 40.76 10649 40.60
70 50 10658 50.39 10658 50.25
80 60 10662 54.39 10662 54.16
Power comparison fitted before and after noise reduction is as shown in Table 2:

Table 2 Comparison of Power Variation Corresponding to Temperature Change Before and After
Filtering.

Temper Temper Origina Origina Power Power

ature ature l Data l Data Value Variatio
Value Variatio Power Power After n Value
（℃） n（℃） Value Variatio Denoisi After
（db） n Value ng（db） Denoisi
（db） ng（db）
20 0 1.0688 0 1.0687 0

40 20 1.13 0.0612 1.1295 0.0608

50 30 1.1297 0.0609 1.1299 0.0612

4
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

60 40 1.13 0.0612 1.1297 0.0610

70 50 1.1309 0.0621 1.1307 0.0620

80 60 1.1314 0.0626 1.1315 0.0628

By conducting linear fitting to the frequency and power after denoising, the relationship
coefficient of frequency shift, power and temperature can be obtained. The temperature-frequency
shift relationship coefficient were 0.8293 and 0.8236before and after denoising.By comparing with
theoretical value 1.2, we can see that it meets the basic requirement. The temperature-power
relationship coefficient is 0.000032 and that after denoising is 0.000048, differ from the theory value
0.000036. This is mainly due to the power difference in fiber optical and regulating light of different
manufacturers. Then puttherelationship coefficient of the frequency shift, power and temperature into
formula 2:
Δvb = c1 Δε + c 2 ΔΤ
100ΔPB (2)
= c3 Δ ε + c 4 Δ T
PB
The data obtained under 40℃, 50℃, 60℃, 70℃, 75℃ is analyzed, the calculated temperature is
filled in Table 3, and the temperature data after denoising is drawn in Figure 7. From Table 3, we could
see that the resolution of the temperature after filtering is higher than that of before.

Table 3 Analysis of Temperature Values Before and After Denoising

Theoretical Temperature Value Analyzed Temperature Value Analyzed Temperature Value

（℃） Before Denoising（℃） After Denoising（℃）
40 44 41.5
50 54 51
60 62.5 61
70 69 70.5
75 76 74.5

Figure7Temperature Curve

By comparison, it can be found that the error range of the analyzed temperature value after
denoising is mostly within ± 1 ℃, and that of before is far beyond ± 1 ℃. Similarly, the data
collected under500 με , 1500 με , 2500 με , and 4500 με is processed in frequency denoising domain,
and the data of a total 10m in about 10830 range isaveraged.Frequency comparison fitted before and
after denoising is as shown in Table 4:

5
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

Table 4 Comparison of Frequency Variation Corresponding to StrainChange Before and After

Filtering.

Strain Variation Original Data Original Data Frequency Value Frequency

（ με ） Frequency Value Frequency After Denoising Variation Value
（MHz） Variation Value （MHz） After Denoising
（MHz） （MHz）
0 10613.07 0 10613.17 0

500 10715.94 102.87 10716.19 103.02

1500 10747.09 134.02 10746.78 133.61

2500 10785.47 172.40 10785.66 172.49

4500 10882.29 269.22 10882.80 269.63

By conducting linear fitting to frequency and power variation after denoising, the relationship
coefficient of frequency shift, power and strain can be obtained. The strain-frequency shift relationship
coefficient were 0.04209 and 0.04221 before and after denoising. Differ from the theory value 0.048.
The main reason is that different manufacturers produce different optical fiber materials, resulting in
different degrees of optical loss. The strain-power relationship coefficient is 9.267e-005, differ greatly
from the theory value 0.000036 and this is due to the uneven strain applied by the device thus causing
stress concentration at both ends. Therefore, it is necessary to improve the device in later stage. Put the
desired relationship coefficient of frequency shift, power and strain into formula 2, then analyze the
data collected under 500 με 、1500 με 、2500 με 、3500 με 、4500 με . The calculated strain values
are then being put in Table 5, and the strain data after the denoising is drawn in Figure 8. Through
Table 4 we can see that the resolution of strain after filtering is higher than that of before.

Table5 Analysis of Strain Values Before and After Denoising

Theoretical Strain Value Analyzed Strain Value Before Analyzed Strain Value After
（ με ） Denoising Denoising
（ με ） （ με ）
500 556 518
1500 1525 1511
2500 2400 2495
3500 2600 3490
4500 4525 4520

Figure8 Strain Curve

By comparison, it can be found that most of the strain values obtained after denoising are in an error

6
2018 International Conference on Civil and Hydraulic Engineering (IConCHE 2018) IOP Publishing
IOP Conf. Series: Earth and Environmental Science 189 (2018)
1234567890 ‘’“” 032026 doi:10.1088/1755-1315/189/3/032026

range of ± 20 με , and that of before are ± 100 με . So, the threshold denoising proved to be effective
for the measurement of strain.

4. Conclusion
By combining the wavelet threshold denoising algorithm with the traditional curve fitting algorithm,
collecting and analyzing the distributed Brillouin optical fiber system based on the BOTDA
technology, the following conclusions are drawn:
1) After denoising, the precision of temperature recognition can reach ± 1 ℃, and the precision
of strain is ± 20 με . The accuracy of temperature and strain are improved evidently after denoising.
2) After studying the principle of temperature and strain measurement in the distributed Brillouin
optical fiber monitoring system and comparing the denoising effect of wavelet group such as sym and
DB, it is concluded that the db5 wavelet function is more suitable for the wavelet threshold denoising
of this system.
3) The distributed Brillouin optical fiber monitoring software is programmed with the
combination of C# language, MATLAB and SQL database. The software realizes data acquisition and
analysis to temperature and strain identification.

References
[1] Zhao Lijuan, Li Yongqian, Zhang Shue, He Yujun. Optical fiber Brillouin scattering spectrum
teaching experimental system design [J]. Experimental Technology and Management, 2011, 28
(08): 64-66.
[2] Li Rongwei, Liu Jianfu, Yang Zhi, Li Yongqian. Research progress of distributed optical fiber
temperature and strain sensing technology [J]. Study on Optical Communications, 2010 (03):
44-47.
[3] Huang Minshuang, Huang Junfen. Optical fiber frequency shift distributed Brillouin optical fiber
sensing technology [J]. Acta PhotonicaSinica, 2011, 40 (9): 1428-1432.
[4] Cao Jianmei, Ren Chunnian. Research on deformation monitoring system of optical fiber grid
tunnel based on BOTDR [J]. Computer Measurement & Control, 2011, 19 (11):
2616-2618+2628.
[5] Liu Yangjun, He Julong, Chu Duanduan, XieHongwei. Strain sensitization based on Brillouin
sensing technology [J]. Journal of Qinghai University (Natural Science), 2012,30 (04): 26-29.
[6] Sun Yijie, Zhang Dan, Tong Hengjin and so on. Application of distributed optical fiber monitoring
technology in the Majiagou landslide in the Three Gorges Reservoir Area [J]. The Chinese
Journal of Geological Hazard and Control, 2013, 24 (4): 97-102.
[7] Lu Yi, Shi Bin, Xi Jun et al. Research on distributed optical fiber monitoring technology for
ground fissure based on BOTDR [J]. Journal of Engineering Geology, 2014, 22 (1): 8-13.
[8] Huang Daren, Wang Jianzhong. The Budan-Fourier theorem of spline and its application [J].
Chinese Science Bulletin, 1982, 27 (17): 1028-1028.
[9] Yuan Changmao, Wen Hongyan, YUANChangmao, et al. Wavelet denoising method for
deformation monitoring data processing [J]. Geospatial Information, 2009, 7 (4): 136-138.
[10] Zhang Heng. Simulation and implementation of wavelet denoising technology based on Matlab
[J]. Science and Technology of West China, 2010, 09 (29): 4-5.
[11] Li Zhenwei, He Jishan, Liu Bingquan and so on. Progress in wavelet image compression [J].
Computer Technology and Development, 2004, 14 (6): 29