PHOTONIC SENSORS / Vol. 15, No.

1, 2025: 250121

Railway Slope Monitoring Based on

Dual-Parameter FBG Sensor
Hongbin XU1,2,4, Weiwei WANG3, Feng LI1,2,4*,
Yanliang DU1,2,4, Hongliang TU5, and Chuanrui GUO1,2,4
1
College of Civil and Transportation Engineering, Shenzhen University, Shenzhen 518060, China
2
National Key Laboratory of Green and Long-Life Road Engineering in Extreme Environment, Shenzhen University,
Shenzhen 518060, China
3
School of Civil Engineering, Shijiazhuang Tiedao University, Shijiazhuang 050043, China
4
Institute of Urban Smart Transportation & Safety Maintenance, Shenzhen University, Shenzhen 518060, China
5
State Key Laboratory of Geomechanics and Geotechnical Engineering, Institute of Rock and Soil Mechanics, Chinese
Academy of Sciences, Wuhan 430071, China
*
Corresponding author: Feng LI E-mail: [email protected]

Abstract: A large number of slopes appear along the line during railway construction, which will
pose a threat to railway safety operation. Slope monitoring plays an important role in ensuring the
safety of railway operation. Aiming at the difficulties of sensor multiplexing, low accuracy, and large
disturbance by trains, this paper proposes a railway slope monitoring method based on integrated
fusion detection of inclination and vibration. Instability and failure characteristics of the K3 slope in
Shuohuang Railway and dynamic characteristics under the excitation of the train load are analyzed
by the finite element method (FEM) analysis. Based on the above analysis, a slope monitoring system
is established utilizing the self-developed dual-parameter fiber Bragg grating (FBG) sensor. The
monitoring data of the past four years show that the slope is in a relatively stable state at present. The
monitoring data are consistent with the results of the FEM. The feasibility of the damage
identification method based on inclination and vibration characteristics is verified, which provides a
new method for railway slope monitoring.
Keywords: Fiber Bragg grating; railway slope; monitoring; inclination; vibration

Citation: Hongbin XU, Weiwei WANG, Feng LI, Yanliang DU, Hongliang TU, and Chuanrui GUO, “Railway Slope Monitoring
Based on Dual-Parameter FBG Sensor,” Photonic Sensors, 2025, 15(1), 250121.

weather, which has become the main hidden danger

1. Introduction
affecting safe operation of the transportation
China’s mountainous area accounts for about 2/3 infrastructure. Reinforcement and monitoring are the
of the country’s total area. With the rapid main means of preventing and controlling disasters
development and construction of China’s railways, a [1−3]. It is particularly necessary to monitor the high
large number of natural and engineering slopes have and steep slopes along the railway in real time in
appeared along the railways. Landslides, mudslides, order to ensure a safe and smooth railway and
and other geological disasters often occur in the bad improve the transportation efficiency. Much research

Received: 18 August 2023 / Revised: 4 January 2024

© The Author(s) 2024. This article is published with open access at Springerlink.com
DOI: 10.1007/s13320-024-0718-0
Article type: Regular
Page 2 of 14 Photonic Sensors

has focused on the slope slip monitoring technology. has also been applied to deep deformation
Among them, the inner deformation is the most monitoring of slopes [17, 18], especially the fiber
intuitive and effective parameter to reflect the Bragg grating (FBG) sensing technology, which was
stability of the slope. Inner deformation monitoring once considered by some scholars as an alternative
can not only provide important information for the technology for slope monitoring and landslide early
early warning of the slope instability but also help in warning [19]. At present, much research is based on
predicting the sliding position and sliding volume of the FBG [20−22], Brillouin optical time domain
the slope. analyzer (BOTDA) [23], Brillouin optical time
At present, the most commonly used technology domain reflectometer (BOTDR) [24], and optical
for slope-deep deformation monitoring is the time-domain reflectometry (OTDR) [25, 26]
borehole inclinometer technology based on an technology combining the fiber grating [27] or
electrical inclinometer, which was put forward and optical fiber [28] on the surface of the inclinometer
developed by Wilson of Harvard University in 1952 tube, which produces strain along with the bending
[4]. The key point of this technology is to solve the of the inclinometer tube, resulting in the change of
accuracy problem of the sensor. With the the reflection spectrum and scattering spectrum in
development of the microelectronics technology the optical fiber. Through the fiber grating
represented by the micro-electro-mechanical system demodulators and distributed demodulation
(MEMS) [5, 6], the slope deep deformation technology, the deep deformation of the slope can be
monitoring technology based on an electrical monitored. This method directly measures the
inclinometer is becoming more and more mature, surface strain of the inclinometer tube and obtains
and the monitoring method has also developed from the deep deformation of the slope through
the sliding monitoring technology [7] to the fixed integration, which has a large monitoring error,
monitoring technology [8], realizing the especially for high and steep slopes. The bottom of
leap-forward development from traditional manual the inclinometer tube cannot be fixed, and the
operation to automatic real-time monitoring. measurement results are easily distorted.
However, due to the limitation of the transmission In addition, the train load vibration causes
distance of electrical signals, it is difficult to realize serious interference with the conventional railway
large-scale slope monitoring. At the same time, the slope monitoring system. With an increase in the
number of sensors in each hole is limited due to the traffic density and an increase in the vehicle speed
diameter of the signal line and sensor tube, resulting and load, the interference of the train load vibration
in low spatial resolution accuracy for inner to the monitoring system becomes more and more
deformation monitoring. In addition, the electrical serious. Through the study of the slope monitoring
displacement sensors [9] and time domain reflection method based on the combination of the inclination
technology [10, 11] are also used to monitor the angle and vibration (acceleration), not only the
inner deformation of the slope, but there are also problem of train load vibration interfering with the
problems, for example, poor electromagnetic monitoring system can be solved, but also the
interference resistance, large transmission loss, and internal damage of the slope can be analyzed
short transmission distance. through the slope vibration response.
The optical fiber sensing technology has the 2. Shuohuang Railway
advantages of anti-electromagnetic interference
2.1 Slope status
[12−14], long-distance transmission [15], and easy
reusing [16]. The optical fiber sensing technology Shuohuang Railway starts from Shenchi South
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 3 of 14

Station in Shenchi County, Shanxi Province, China texture, which is easy to be eroded by running water.
in the west of China and ends at Huanghua Port There are many bulges and cracks on the surface of
Freight Yard in Cangzhou City, Hebei Province, the slope protection. The cracks are mostly
China in the east of China. It was built on November distributed longitudinally along the line, and most of
25th, 1997, and was opened to traffic on November them are located in the middle of the slope
1st, 1999. The total length of the main line of the protection and the contact part of the steps. The
whole line is about 598 km. It is designed as a surface of the unprotected slope is seriously scoured
heavy-duty subgrade, a double-track electrified by rain, and many scoured cracks and shallow
railway, and a national Class I trunk line. It is the slumps can be seen. Generally speaking, there are
second largest channel for transporting coal from the many diseases on the slope surface, and there are
west to east in China and plays an important role in some hidden dangers that affect the safety of train
the national economic development. Shuohuang operation.
Railway passes through Taihang Mountain,
Hengshan Mountain, Yunzhongshan Mountain, and
other mountains, with complex topography, high
mountains, and deep valleys, many high-filled
Fig. 1 Shuohuang Railway in K3.
embankments, and high-steep slopes, and there are
K3 of Shuohuang Railway is located at the end
potential safety hazards such as landslides or of the turn-back section of Shenchi South Station,
collapses. Therefore, it is of great significance to where there are many railway lines, among which
build a monitoring system for high-steep slopes along the main line of the ascending line is closest to the
the railway to ensure the safe operation of the railway. toe of the slope, and the outer rail is about 7.6 m
The Shenchi−Ningwu section of Shuohuang away from the concrete retaining wall at the toe of
Railway belongs to a temperate continental the slope. In this chapter, the K3+200 section with
monsoon climate, with long and cold winters, dry severe erosion of unprotected parts is selected as the
and windy springs, mild summers without intense research object, which is a secondary slope with the
heat, cool and rainy autumn, and obvious seasonal new loess soil, and the section size is shown in Fig. 2.
changes. The hillside in this area is gentle. The gully
Secondary
is developed, and there are a large number of 10 m
slope
35° 2m
artificial high-cut slopes along the line. Among them, First grade
slope

3 km of Shuohuang Railway is located in Longquan 16.5 m 0.8 m

Town, Shenchi County, Shanxi Province (about 44°

7.6 m

3 km away from Shenchi County), China, which is

2m Concrete
located in the loess hilly area. It belongs to the deep retaining wall

cutting slope. The slope is located on the side of the 1.5 m

railway uplink line, in east-west direction, with a

Fig. 2 Diagram of the slope at K3+200 of Shuohuang
height of nearly 30 m, as shown in Fig. 1. Some Railway.
areas of the slope are protected by concrete frame
2.2 Static characteristic analysis
beams, the surface of the protected part is covered
with the mortar rubble, and the surface of the The block discrete element program is used to
unprotected part has sparse vegetation. The slope is simulate and analyze the whole process of the slope
new loess with the uniform soil quality and loose landslide. The mechanical parameters of the loess
Page 4 of 14 Photonic Sensors

and structural plane are listed in Table 1. The whole slope surface is a free boundary. Under the condition
slope adopts the elastic-plastic constitutive model of a rainstorm, seepage boundary conditions are: the
and Mohr-Coulomb yield criterion [29]. Under impermeable boundary is set on the left and right
natural working conditions, the displacement boundaries, the pore water pressure is set at 200 Pa
constraints include: horizontal constraints are set on on the slope (2 cm of surface water represents heavy
the left and right boundaries, vertical and horizontal rainfall), and the pore water pressure at the bottom
constraints are set on the lower boundary, and the boundary is set at 0 Pa.
Table 1 Material properties of the slope.

Attribute Contact property

Geotechnical Tensile strength
modulus of elasticity jkn Jks Cohesion (kPa) Friction angle (°)
type Density (kg/m3) (MPa)
(MPa) (GPa/m) (GPa/m)
Loess 1 410 23 9.5 9.5 16.8 25 0.25

Under the dead weight stress, the unbalanced cohesion and friction angle of the material are
force of the model is calculated, and the maximum reduced, and the deformation of the slope when it
unbalanced force of the system is less than 1×10–5 N enters the limit equilibrium state is shown in Fig. 4
after the model runs for 8 000 steps. As shown in (the material properties of the slope are shown in
Fig. 3, the calculation converges, and the system is Table 1). As shown in the figure, it can be seen that a
considered to be in equilibrium at this time, penetrating slip plane is formed from the toe to the
indicating that the loess slope is stable as a whole. top of the slope. The reduction factor (slope safety
By using the strength reduction method, the factor) at this time is 1.32.

4040000
000
unbalanced force
Unbalanced force
Displacement (m)
3535000
000
0.000 20
3030000
000
unbalanced force (N)

0.000 18
Unbalanced force (N)

2525000
000 0.000 16
0.000 14
000
2020000 0.000 12
0.000 10
1515000
000
0.000 08
1010000
000 0.000 06
0.000 04
000
55000

00
0 20 40 60 80
3
Steptimes
step (×10 )
times(thousand)

Fig. 3 Unbalanced force of the slope. Fig. 4 Displacement of slope at the critical state of stability.

By applying the 200 Pa pore water pressure on deformed first, and the deformation starts at the
the surface to simulate heavy rainfall, in the initial foot of the slope and eventually spreads to the
stress field, that is, after the self-weight calculation whole slope surface. The initial deformation rate
converges, the seepage calculation is started, and the is slow. When the maximum deformation value
lateral displacement changes at different depths of the first-grade slope reaches 0.6 m, the surface
from the tops of the first and second-grade slopes of the slope will suddenly collapse with a
are shown in Fig. 5. It can be seen that, in the certain depth, and the collapse depth is about
case of heavy rainfall, the first-grade slope is 6 m.
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 5 of 14

0.1
0.1
0.0
0.0

0.0
0.0
−0.5
-0.5
−0.1
-0.1 25
25
00 23
23
deflection(m)

deflection(m)
Deflection (m)
Deflection (m)

−1.0
-1.0 22 21
−0.2
-0.2 21
44 19
19
66 17
17
88 −0.3
-0.3 15
−1.5
-1.5 15
10
10
13
13
12
12
−0.4
-0.4
14
14
−2.0
-2.0
−0.5
-0.5

−2.5
-2.5 −0.6
-0.6
135
135 140
140 145
145 150
150 155
155 160
160 135
135 140
140 145
145 150
150 155
155 160
160
stepStep
timestimes (×103)）
（thousand stepStep times
times (×103)）
（thousand

(a) (b)
Fig. 5 Displacement curve of the slope in different depths: (a) first grade slope platform and (b) top of the secondary slope.

2.3 Dynamic characteristic analysis of the C80 freight car), and the normal slope model
In this paper, the complete method is used for and the slope model with 0.8 m thick soft layer (slip
harmonic response analysis. A total of 250 kN surface) are constructed, respectively. The free grid
excitation load is applied to 10 nodes within the is used to divide the cells, and the cell size is
range of 7.6 m–9.4 m from the concrete retaining controlled at 0.2 m. The models after cell division
wall (calculated according to the axle load of 25 tons are shown in Fig. 6.

(a) (b)
Fig. 6 Harmonic analysis of the model: (a) normal slope model and (b) slope model with the weak layer (defect).

The LS-DYNA model is constructed by using deformation range of small strain, and the
plane162. The parameters, such as the model size amplification of the soil layer can increase the
and load, are the same as those in frequency domain vibration amplitude near the surface [30], especially
analysis. The acceleration of different positions the low-frequency component. The thicker the soil
(nodes) in two slope structural models with or layer is, the closer it is to the surface, and the more
without weak layers under 20 Hz loading is shown obvious the effect is. The vibration intensity of the
in Fig. 7. As can be seen, on the whole, the closer it slope decreases obviously at the position of the
is to the surface, the greater the acceleration weak layer and increases after passing through the
amplitude of the slope is. This is because in the case weak layer, but it is less than that before passing
of micro-vibration, the soil is in the elastic through the weak layer. The acceleration cloud map
Page 6 of 14 Photonic Sensors

is shown in Fig. 8. As shown in Fig. 8, the model with a weak layer, there is an obvious
acceleration distribution gradient is consistent boundary at the weak layer of acceleration,
with the propagation direction of the vibration which is caused by the obvious attenuation of
wave, and the closer it is to the slope surface, acceleration when it passes through the weak
the greater the acceleration peak value is. In the layer.
15
15 25
normal model
Normal model

defect model
Defect model
20 normal model
Normal model
10
10
weak layer
Weak layer defect model
Defect model

15

position(m)
weak layer
Weak layer
55
position(m)
Position (m)

(m)
10

Position
00 55

00
−5
-5
−5
-5

−10
-10 −10
-10
0.000
0.000 0.004
0.004 0.008
0.008 0.012
0.012 0.016
0.016 0.020
0.020 0.024
0.024 0.00
0.000 0.01
0.010 0.02 0.030
0.020 0.03 0.040
0.04 0.05
0.050
2
acceleration m/s2))
Acceleration ((m/s (m/s)2)
Acceleration (m/s
acceleration
2

(a) (b)
Fig. 7 Acceleration response of the slope at different positions: (a) acceleration response of the first grade slope and (b) secondary
slope acceleration response.

(a) (b)
Fig. 8 Acceleration cloud map of the slope model: (a) normal model 0.58 s and (b) defect model 0.635 s.

3. Monitoring scheme and system communication module, and monitoring center. The
construction data are collected by the FBG demodulator based on
Combining with the above simulation results, a the BaySpec module. The sensor and demodulator
monitoring system is built in the middle of the are connected by an optical cable. A wireless
unprotected area with the serious surface erosion. communication module is utilized for signal
The monitoring scheme is shown in Fig. 9, including transmission between the demodulator and
the in-situ dual-parameter FBG sensor [31], monitoring center. The on-site monitoring system is
demodulator, optical cable, solar energy, 4G powered by the solar energy.
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 7 of 14

about 8 m, the slurry in the borehole leaks. The

Base station
slurry is injected into the borehole no longer flowing
back to the ground, so it can be inferred that there
are unknown cracks in the slope. To ensure the depth
FBG Optical cable of the borehole and minimize the influence of
demodulator
sediment precipitation in the mud on the depth of the
Monitoring center
borehole, it is necessary to clean the borehole before
lifting the drill. Since the slurry leakage of the slope
Inclinometer tube Vibrating wave
cannot flow back, to avoid the influence of
over-grouting on the slope stability, the acrylonitrile
FBG sensor butadiene styrene (ABS) inclinometer tube with
a sealed bottom end should be lowered in time
Fig. 9 Schematic diagram of the monitoring system. after drilling. When the ABS inclinometer tube
The sensors are installed in a drilled hole, which encounters resistance during lowering, the
is located at 1 m away from the front edge of the orientation of the ABS inclinometer tube should be
slope. Because of the hard soil, the rotary resistance adjusted in time, so that two pairs of guide grooves
of the drilling rig is very large. Water is used to distributed at 90 degrees in the ABS inclinometer
reduce the resistance and prevent collapse during tube are vertical and parallel to the slope,
drilling. respectively. After adjusting the orientation, press
In the process of drilling, when the borehole the ABS inclinometer hard until the lowering length
is drilled to about 7.5 m, gravel is encountered. of the ABS inclinometer is close to the depth of the
As shown in Fig. 10, when the borehole is drilled to measuring hole.

(a) (b) (c)

Fig. 10 Gravel and installation of the ABS inclined pipe: (a) rhinestones carrying gravel, (b) gravel carried by the diamond, and
(c) ABS inclinometer is lowered and installed.

To avoid the hole collapse, the gap between the ABS inclinometer tube is lowered. To ensure that
ABS inclinometer tube and the hole wall of the the slope deformation and vibration signals can be
measuring hole should be backfilled in time after the accurately transmitted to the ABS inclinometer, the
Page 8 of 14 Photonic Sensors

backfill material should have the same rigidity as the

original soil. If the rigidity of the backfill material is
too large or just too small, the deformation of the
inclinometer will not match the deformation of the
original soil, which will affect the monitoring
accuracy. In the loess area, because the inclinometer
hole is deep, it cannot be pounded, and the backfill
state of the original soil is loose, so it is difficult to
Fig. 11 Backfilling of the borehole.
reach the original stiffness. The backfill mud in the
loess is difficult to consolidate in a short time, and it
is even more difficult to meet the original soil
stiffness requirements. To ensure the backfill quality
and prevent the backfill from being false, the
backfill material should have certain fluidity,
self-compaction, and high density. Therefore, gravel
with a nominal diameter of 0.5 cm–1 cm is selected
for backfilling, as shown in Fig. 11. To prevent the
gravel from impacting the borehole wall and other
factors from causing false backfilling, the Fig. 12 Installation photo of the FBG sensors.
backfilling speed should be slow, and the ABS 4. Monitoring results and analysis
inclinometer should be vibrated properly during the
backfilling process to ensure the backfilling 4.1 Deep deformation analysis of the slope
compactness. The sensor is installed by connecting The monitoring system was installed in
the rod and guide wheel, as shown in Fig. 12. The November 2016, and the monitoring data of internal
sensor spacing is 2 m, and 9 FBG dual-parameter deformation of the slope are shown in Figs. 13 and
sensors are installed. 14. It can be seen that from November 2016 to
Under the influence of the scouring effect of the August 2018, during the slope deformation stage,
drilling rig circulating mud and hole collapse, the the internal displacement of the slope gradually
diameter of the borehole is usually larger than that increased, the maximum displacement was about
of the drill bit, which leads to the actual amount of 5.7 mm, and the average deformation rate was
the backfill material far exceeding the calculated 3.25 mm/y. After August 2018, the slope basically
amount. Therefore, if the backfill amount is less than turned into a stable period, and the internal
the calculated amount, it can be determined that the displacement basically did not change. The reason
backfill is not dense. The drilling depth of the for slope deformation may be that a large amount of
borehole is about 20 m, and the pipe running depth slurry leaked during drilling and stored inside the
is 18.5 m. The theoretical backfill material is slope, which led to a decrease in the loess strength
0.14 m3, and the actual material is nearly 1 m3. This inside the slope and even the collapse of loess, thus
is not only the factor of the borehole diameter causing the internal deformation of the slope. With
scouring and hole collapse, but also the gravel time, the moisture inside the soil gradually migrated
backfilling into the internal cracks of the slope and dissipated, the stability of the soil gradually
through which the borehole passes. recovered, and the internal deformation rate of the
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 9 of 14

slope gradually decreased until it reached a stable 2020.08.15

2020.08.15

state. From Fig. 13, it can be seen that the

2020.06.01
2020.06.01
deformation size of the slope turns at 11.5 m, and
2020.03.15
2020.03.15
the deformation is larger at the position above
11.5 m. As far as the whole slope is concerned, there 2020.01.01
2020.01.01

is no obvious slip surface on the slope. 2019.10.15

2019.10.15

displacement(mm)
Displacement (mm) 2019.08.01.
2019.08.01
0
0 11 22 33 44 55 66
00 2019.05.15
2019.05.15

−2
-2 2019.03.01
2019.03.01

−4
-4 2018.12.15
2018.12.15

Time
time
−6
-6 2018.10.01
2018.10.01

−8
(m)

2018.07.15
2018.07.15
-8
depth(m)
Depth

2018.05.02
2018.05.02
−10
-10 2016.11.16
2016.11.16
2017.05.15
2017.05.15 2018.02.15
2018.02.15
−12
-12 2017.11.15
2017.11.15
2018.05.16
2018.05.16
2017.12.02
2017.12.02
2018.11.15
2018.11.15
−14
-14
2019.05.15
2019.05.15
2019.11.15 2017.09.17
2017.09.17
2019.11.15
−16
-16 2020.05.15
2020.05.15
2020.10.01
2020.10.01 2017.07.02
2017.07.02 00

−18
-18 1.5
1.5
3.5
3.5
2017.04.15
2017.04.15 5.5
5.5
7.5
7.5
Fig. 13 Time-varying curve of the displacement in the 9.5
9.5
2017.01.18
2017.01.18
primary slope. 11.5
11.5
13.5
13.5

From the above analysis, it can be seen that the 00 22 44 66

displacement(mm)
Displacement (mm)
slope is in a stable state as a whole, and the Fig. 14 Time-varying curve of the displacement in the
construction along the slope will have a certain primary slope.
impact on the slope stability. In the process of the 4.2 Analysis of slope vibration characteristics
slope construction (including sensor installation),
Under the excitation of the train, the strongest
the impact on the slope should be reduced,
excitation frequency of each measuring point on the
especially during the drilling operation before sensor
slope at different time points is shown in Fig. 15. As
installation, and the use of slurry should be shown in the figure, the strongest frequency at the
minimized. Using miniaturized sensors and reducing depth of 7.5 m decreases with time, with a variation
the diameter of drilling holes can effectively reduce range of 21.19 Hz−28.82 Hz and a frequency
the difficulty of drilling holes and the influence of fluctuation range of about 7.63 Hz. The strongest
drilling holes on slopes. frequency at the depth of 9.5 m on the slope
It should be noted that the 13.5 m depth sensor increases with time, with a variation range of
of the slope failed in January 2017, and the 17.5 m 9.42 Hz−21.08 Hz and a frequency fluctuation range
depth sensor failed in May 2020. After the sensor of about 11.66 Hz. The strongest response frequency
fails, when calculating the deep deformation of the of other measuring points under the excitation of the
slope, the deformation at the position of the sensor is train does not change obviously and is the strongest.
calculated according to the effective deformation Under the excitation of the train, the vibration
before the sensor fails. amplitude of each measuring point on the slope is
Page 10 of 14 Photonic Sensors

shown in Fig. 16. As shown in the figure, the that there is no obvious slip surface on the slope,
amplitude change is not obvious, which also shows which is mutually confirmed by the deformation data.
Frequency(Hz)
Frequency (Hz) Acceleration (m/s2))
acceleration(m/s 2

10 15 20 25 30
00 0 5 0.0 0.1 0.2 0.3 0.4 0.5
0.5
00

−2
-2 −2
-2 2017.09
2017.09
2017.09
2017.09
2018.09
2018.09
2018.09
2018.09
−4
-4 −4
-4 2019.06
2019.06
2019.06
2019.06
2020.10
2020.10
2020.10
2020.10
−6
-6 −6
-6

−8
-8 −8
-8

Depth (m)
depth(m)
Depth (m)
depth(m)

−10
-10 −10
-10

−12
-12 −12
-12

−14
-14 −14
-14

−16
-16 −16
-16

−18
-18 −18
-18

−20
-20 −20
-20

Fig. 15 Strongest frequency of each measuring point in the slope. Fig. 16 Peak acceleration of each measuring point in the slope.
Typical spectrum characteristics of soil vibration it is insensitive to the vehicle speed, vehicle type, and
are shown in Figs. 17−19. Under the excitation of vehicle characteristics on the uplink and downlink lines.
the train along the line, the vibration of the slope At different times, the maximum response
with no defect position is shown in Fig. 17. As shown frequency of the train load excitation frequency
in the figure, the main vibration frequency is single changes greatly at the depths of 7.5 m and 9.5 m on
and obvious, the signal-to-noise ratio is strong, and the slope shown in Figs. 18 and 19. As shown in the
1.2 1.6

1.4 X: 20.810
1.0 Y: 1.449 0
X: 21.670
Y: 1.019 0 1.2
0.8
1.0
Amplitude
Amplitude

0.6 0.8

0.6
0.4
0.4
0.2
0.2

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(a) (b)
3.0 2.0

1.8
X: 23.060
2.5 1.6 Y: 1.833 0
X: 22.520
Y: 2.556 0
1.4
2.0
1.2
Amplitude
Amplitude

1.5 1.0

0.8
1.0
0.6

0.4
0.5
0.2

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(c) (d)
Fig. 17 Amplitude-frequency diagrams of slope in the depth of 17.5 m: (a) 2017.09, (b) 2018.09, (c) 2019.06, and (d) 2020.10.
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 11 of 14

3.0
3.5
X: 28.820 X: 28.720

3.0 Y: 3.442 0 2.5 Y: 2.794 0

2.5 2.0

Amplitude
Amplitude
2.0
1.5
1.5 X: 23.130
Y: 0.933 1
X: 22.920 1.0
1.0 Y: 0.809 2

0.5
0.5

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(a) (b)
2.5
2.0
X: 22.980
1.8
Y: 2.302 0
X: 21.190 2.0
1.6 Y: 1.493 0

1.4
1.5

Amplitude
1.2 X: 29.530
Amplitude

Y: 1.303 0 X: 29.320
1.0 Y: 1.068 0

1.0
0.8

0.6

0.4 0.5

0.2

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(c) (d)
Fig. 18 Amplitude-frequency diagrams of slope in the depth of 7.5 m: (a) 2017.09, (b) 2018.09, (c) 2019.06, and (d) 2020.10.
2.0 4.0

1.8 X: 9.415 3.5 X: 9.806

Y: 1.850 0 Y: 3.623 0
1.6
3.0
1.4
2.5
1.2 X: 15.230
Amplitude
Amplitude

Y: 1.972 0
1.0 2.0

0.8 1.5
0.6
1.0
0.4
0.5
0.2

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(a) (b)

2.0
X: 21.080
X: 16.220 3.0 Y: 2.224 0
X: 9.561 X: 10.010
Y: 1.896 0
Y: 1.822 0 Y: 2.483 0
1.5 2.5

X: 19.380
Amplitude

2.0
Amplitude

Y: 1.369 0

1.0
1.5

1.0
0.5

0.5

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Frequency (Hz) Frequency (Hz)
(c)
(d)
Fig. 19 Amplitude-frequency diagrams of the slope in the depth of 9.5 m: (a) 2017.09, (b) 2018.09, (c) 2019.06, and (d) 2020.10.
Page 12 of 14 Photonic Sensors

figures, there are more obvious response frequencies, than that of the theoretical calculation. These results
and with time, its strongest response frequency is verify the existence of unknown cracks and defects
closer and closer to the strongest response frequency in the slope at the depth of 8 m.
in other positions of the slope. Due to slurry leakage (3) A slope monitoring method based on the
in the drilling process, it can be judged that there are inclination and vibration characteristics of the slope
unknown cracks, and a large amount of slurry is is proposed. The feasibility of this method is verified,
poured into the drilling process. After the which provides a new method for safety monitoring
inclinometer is installed and backfilled, it is of the railway slope.
surrounded by a mixture of the slurry and gravel, Acknowledgment
which leads to a change in the coupling state
between the inclinometer and soil. The water in the This work was supported in part by the National
mixture of mud and gravel diffuses with time, and Key Research and Development Program of China
the soil state gradually returns to and approaches the (Grant No. 2021YFB2600800), National Natural
original state. This also verifies the feasibility of Science Foundation of China (Grant No. 52378309),
verifying the internal damage of the slope through Shenzhen Science and Technology Program
the frequency change. (Grant Nos. JCYJ20220818095608018 and
KQTD20180412181337494), China Postdoctoral
5. Conclusions Science Foundation (Grant Nos. 2022M722188,
In this paper, a railway slope monitoring scheme 2022TQ0218, and 2022M722187), Natural Science
utilizing FBG inclination and acceleration integrated Foundation of Hebei Province (Grant No.
sensor is proposed. Based on the results of finite E2021210032), Visiting Researcher Fund Program
element method (FEM), a slope monitoring system of State Key Laboratory of Water Resources
is constructed, and the monitoring data of the past Engineering and Management (Grant No.
four years are analyzed. The main conclusions are as 2022SGG05), and Tianjin Key Laboratory of Rail
follows: Transit Navigation Positioning and Spatio-temporal
Big Data Technology (Grant No. TKL2024B10).
(1) The FEM shows that the safety factor of the
slope is 1.32. The inclination monitoring data show Declarations
that the slope deformation curve has no obvious Conflict of Interest The authors declare that they have
turning point, and the acceleration monitoring data no competing interests.
show that the acceleration amplitude at different Permissions All the included figures, tables, or text
positions of the slope has no obvious change. The passages that have already been published elsewhere have
three results not only verify each other, but also obtained the permission from the copyright owner(s) for
both the print and online format.
show that the slope is in a stable state at present,
and there is no obvious weak inter-layer and slip Open Access This article is distributed under the terms
of the Creative Commons Attribution 4.0 International
surface. License (http://creativecommons.org/licenses/by/4.0/),
(2) The acceleration monitoring data show that which permits unrestricted use, distribution, and
reproduction in any medium, provided you give
the initial maximum response frequency of the slope
appropriate credit to the original author(s) and the source,
at the positions of 7.5 m and 9.5 m is different from provide a link to the Creative Commons license, and
other positions. This difference decreases with the indicate if changes were made.
time, which is consistent with the occurrence of References
slurry leakage in the slope drilling process and the [1] B. Shi, H. H. Zhu, D. Zhang, and G. Cheng, “From
fact that the actual backfill gravel volume is greater in-situ rock-and-soil testing, exploration, monitoring
Hongbin XU et al.: Railway Slope Monitoring Based on Dual-Parameter FBG Sensor Page 13 of 14

to sensing,” Journal of Engineering Geology, 2022, [15] Y. Cheng and Z. Shi, “Permanent deformation and
30(6): 1811−1818. temperature monitoring of subgrades using fiber
[2] Q. Xu, D. L. Peng, S. Zhang, X. Zhu, C. Y. He, X. Bragg grating sensing technology,” Journal of
Qi, et al., “Successful implementations of a real-time Sensors, 2021, 2021: 8824058.
and intelligent early warning system for loess [16] D. F. Cao, H. H. Zhu, B. Wu, J. C. Wang, and
landslides on the Heifangtai terrace, China,” S. K. Shukla, “Investigating temperature and
Engineering Geology, 2020, 278: 105817. moisture profiles of seasonally frozen soil under
[3] Z. W. Wang, F. Li, and G. D. Mei, “OpenMP parallel different land covers using actively heated fiber
finite-discrete element method for modeling Bragg grating sensors,” Engineering Geology, 2021,
excavation support with rockbolt and grouting,” 290: 106197.
Rock Mechanics and Rock Engineering, 2024, 57: [17] X. L. Kuang, H. Zhang, L. P. Zhang, X. M. Yang,
3635−3657. and M. H. Zhang, “Development and application of
[4] T. D. Stark and H. Choi, “Slope inclinometers for fibre grating inclinometer for monitoring the rock
landslides,” Landslides, 2008, 5(3): 339−350. and soil slope deformation,” Highway, 2015, 60(11):
[5] Y. Q. Shi, Y. F. Li, and L. F. Che, “Design of 171−176.
inclinometer based on MEMS accelerometer array,” [18] J. Cui, D. S. Gunawardena, Z. Liu, Z. Zhao, and H. Y.
Transducer and Microsystem Technologies, 2020, Tam, “All-fiber two-dimensional inclinometer based
39(09): 66−68. on Bragg gratings inscribed in a seven-core
[6] Y. K. Cheung and H. Yu, “Electrochemical tilt multi-core fiber,” Journal of Lightwave Technology,
sensors with symmetric concentric electrode pairs,” 2020, 38(8): 2516−2522.
IEEE Sensors Journal, 2021, 21(19): 21322−21329. [19] C. Y. Hong, Y. F. Zhang, M. X. Zhang, L. M. G.
[7] G. W. Li, L. S. Hu, R. Wang, and D. Z. Chen, Leung, and L. Q. Liu, “Application of FBG sensors
“Testing of slip inclinometer and error treatment for geotechnical health monitoring, a review of
methods,” Journal of Hohai University (Natural sensor design, implementation methods and
Sciences), 2013, 41(6): 511−517. packaging techniques,” Sensors and Actuators A:
[8] H. Xu, F. Li, W. Zhao, S. Wang, Y. Du, and C. Bian, Physical, 2016, 244: 184−197.
“A high precision fiber Bragg grating inclination [20] H. C. Li, C. T. Liu, and Y. B. Feng, “Application of
sensor for slope monitoring,” Journal of Sensors, FBG bending sensors in landslide forecasting and
2019, 2019: 1354029. monitoring,” Journal of Optoelectronics Laser, 2015,
[9] B. Q. Wu, W. D. Zhang, and Y. J. Yang, “Analysis on 26(2): 309−314.
several commonly used side slope deep horizontal [21] M. S. M. Sa’ad, H. Ahmad, M. A. Alias, M. K. A.
displacement monitoring technology,” Shanxi Zaini, K. S. Lim, S. W. Harun, et al, “Surface-
Architecture, 2014, (19): 53−54. mounted tilt sensor using fiber Bragg grating
[10] H. H. Tan and H. L. Fu, “Testing study of application technology for engineered slope monitoring with
of time domain reflectometry to highway slope temperature compensation,” IEEE Sensors Journal,
monitoring,” Rock and Soil Mechanics, 2010, 31(4): 2023, 23(12), 12828−12837.
1331−1336. [22] J. Wang, W. Dong, W. Yu, C. Zhang, and H. Zhu,
“Numerical and experimental investigation of slope
[11] D. F. Li, B. W. Zhang, G. Wang, K. B. Yang, L. M.
deformation under stepped excavation equipped with
Wang, and X. J. Xu, “Time domain reflectometry
fiber optic sensors,” Photonics, 2023, 10(6): 692.
calculation model of landslides slippage,” in
[23] X. L. Yi, H. M. Tang, Y. P. Wu, Y. Ge, X. Fan, and S.
Proceedings of the Information Technology and
Zhang, “Application of the PPP-BOTDA distributed
Intelligent Transportation Systems, Xi’an, China,
optical fiber sensor technology in the monitoring of
2017, 327−336. the Baishuihe landslide,” Chinese Journal of Rock
[12] A. I. Azmi, A. S. Abdullah, M. M. Noor, M. H. Mechanics and Engineering, 2016, 35(S1):
Ibrahim, R. R. Ibrahim, T. Tan, et al., “Dynamic 3084−3091.
bending and rotation sensing based on high [24] Y. L. Liu, Y. Q. Shang, and Y. Yu, “Application of
coherence interferometry in multicore fiber,” Optics surface deformation monitoring of slope using
& Laser Technology, 2021, 135: 106716. BOTDR technology,” Journal of Jilin University
[13] J. J. Pan, L. Y. Wang, W. Hou, and H. Y. Lv, “Design (Earth Science Edition), 2011, 41(3): 777−783.
and investigation of a high-sensitivity tilt sensor [25] Z. Yu, H. Dai, Q. Zhang, M. Zhang, L. Liu, J. Zhang,
based on FBG,” Photonic Sensors, 2023, 13(2), et al., “High-resolution distributed strain sensing
230228. system for landslide monitoring,” Optik, 2018, 158:
[14] A. P. Zhang, S. Gao, G. Yan, and Y. Bai, “Advances 91−96.
in optical fiber Bragg grating sensor technologies,” [26] T. Kogure and Y. Okuda, “Monitoring the vertical
Photonic Sensors, 2012, 2(1): 1−13. distribution of rainfall-induced strain changes in a
Page 14 of 14 Photonic Sensors

landslide measured by distributed fiber optic sensing “Stability analysis for soil slopes with weak
with Rayleigh backscattering,” Geophysical interlayers using the finite element upper bound limit
Research Letters, 2018, 45(9): 4033−4040. analysis,” Journal of Railway Science and
[27] F. Li, W. T. Zhang, F. Li, and Y. L. Du, “Fiber optic Engineering, 2019, 16(2): 351−358.
inclinometer for landslide monitoring,” Applied [30] Y. T. Ho, A. B. Huang, and J. T. Lee, “Development
Mechanics & Materials, 2012, 166−169: 2623−2626. of a fibre Bragg grating sensored ground movement
[28] W. J. Pang, Q. L. Deng, J. Xiong, and H. W. Wang, monitoring system,” Measurement Science and
“Application study of optical fiber sensing Technology, 2006, 17(7): 1733−1740.
technology to slope deformation monitoring based [31] H. Xu, F. Li, Y. Gao, and W. Wang, “Simultaneous
on BOTDA,” Safety and Environmental Engineering, measurement of tilt and acceleration based on FBG
2012, 19(6): 28−33. sensor,” IEEE Sensors Journal, 2020, 20(24):
[29] X. Q. Pi, L. Li, G. P. Tang, R. Zhang, and L. H. Zhao, 14857−14864.