Journal of Mining and Earth Sciences 1

Application of correlation and regression analysis between

GPS-RTK and environmental data in processing monitoring
data of cable-stayed bridge
Tinh Duc Le 1,*, Hien Van Le2, LinhThuy Nguyen3, Thanh Kim Thi Nguyen4, Duy
Tien Le5
1
Faculty of mapping-surveying and land management, university of mining and geology, Vietnam, Email:
[email protected]
2
University of Transport and Communications, Hanoi, Vietnam, Email: [email protected]
3
University of Transport and Communications, Hanoi, Vietnam, Email: [email protected]
4
Faculty of mapping-surveying and land management, university of mining and geology, Vietnam, Email:
[email protected]
5 The branch of Hanoi University of Natural Resources and Enviromment in Thanh Hoa Province, Vietnam,
Email: [email protected]

ARTICLE INFO ABSTRACT

Quá trình:
Structural Health Monitoring system – SHMs has been playing a vital role
Article history: in monitoring large-scale structures during their performance in a lifetime,
Received especially with the long-span bridge such as a suspended bridge or cable-
Accepted stayed bridge. In a SHM system, many kinds of sensors are used to set up at
Available online the specific locations in order to monitor and detect any changes of
Keywords: structures in real-time based on the changes of monitoring data as well as
the changes of correlation among monitoring data types. This paper
Structural Health
proposes a method of applying the correlation and regression analysis for
Monitoring; processing displacement monitoring data acquired by GPS-RTK
cable-stayed bridge; , considering the effects of environmental factors such as temperature and
Correlation Analysis; wind-speed. The results show that the air-temperature has high correlation
Regression Analysis; with the displacements of a cable-stayed bridge acquired by GPS-RTK
GPS-RTK measurement along to specific directions while the wind-speed has low
correlation. Then the general displacement of the target bridge could be
recognized and regression equation is also built to predict the bridge
displacement under effects of the air-temperature.

1. Introduction system, there are many kinds of sensors setting

on target structures for observing different
Structural Health Monitoring (SHM) has been
objective, such as capturing dynamic or static
using successfully to monitor the super-
structural responses by using strain sensor,
structures during their operations, such as high-
stress sensor or accelerometer etc; monitoring
rise building, long-span structures. In a SHM
 2
environmental factors by using temperature and operational factors to the outcome displacement
windspeed sensors (Kaloop and Li, 2009). For data which high correlated coefficients of any
the large-scale structures, monitoring the factors show high influence to the outcome
deformation of structures is one of very displacement data (Cornwell and et al., 1999;
important task which can assess the structural Farrar and et al., 2000; Omenzetter and
health then detect any damages of structures. Brownjohn, 2005; Omenzetter and Brownjohn,
Long-span bridges such as cable-stayed bridges 2006; Sohn and et al., 2000; Hien And Mayuko,
or suspended-supported bridges have two kinds 2015; Hien and et al., 2015). Besides, regression
of deformation: long-term and short-time algorithm is effective method to use in analyzing
deformation which long-term deformation is time-series data in order to detect outlier as well
often caused by environmental factors while as using for further prediction. To define a
short-time deformation is mainly caused by regression model is a fitted model of a given
dynamic inputs such as wind, earthquake and time-series data by assessing the determination
traffic etc (Kaloop and Li, 2009; 2011; Celebi, coefficient and testing the fitted redundant
2000 ). between model and data (Sanford Weisberg,
To monitor the displacements of a structure, 2005; Shumway and Stoffer, 2010; Peter and
the interferometer or some electronic distance Annick , 1987).
measuring instruments are used. Although these
methods can provide high accruracy results, they
still have some shortcomings in application such
as they cannot measure the large displacements
of structures, especially with long-span bridge;
cannot measure in real-time or they depend on
weather condition etc (Cheng and Zheng, 2002).
Recently, Global Navigation Satellite System
(GNSS) has been using to monitor the
displacement of a super-structure in a SHM
system, especially in SHM of a long span bridge,
such as Stonecutters bridge in Hong Kong, Akashi
Kaikyo bridge in Japan, Ting Kau cable-stayed
bridge in Hong Kong etc. In Vietnam, GPS
technology have been using in some cable-stayed
bridges, such as Can Tho bridge, Tran Thi Ly
bridge, Nhat Tan bridge and Bach Dang bridge
etc. Even GPS is considered as a high cost method
in SHM system, however it has many advantages,
such as it is less affected by weather condition; it
can measure the displacements of a specific point
in 3D dimension at millimeter level of accuracy
(Kaloop and Li, 2009; Cheng and Zheng, 2002).
Considering the long-term monitoring of
structures, data processing is a vital work for
recognizing the structural changes during their
operation. Some studies showed that
environmental factors have greatly effects to the
long-term monitoring data (Sohn and et al., 1999
; Cornwell and et al., 1999 ; Farrar and et al.,
2000). To analyze long-term monitoring data,
correlation analysis method is often used to
recognize how the effects of environmental or
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13
This study aims to analyze the long-term
∑ Xi ∑ Yi ∑ XiY i

{
monitoring displacements of a real cable-stayed i i i
X́= ; Ý = ´ =
; XY
bridge acquired by GPS-RTK measurement n n n
considering the effects of environmental factors 2 2
(2)
such as air-temperature and windspeed. Firstly, a ∑X i ∑Y i
X́ 2= ; Y´ 2 =
i i
time-series monitoring data of the target cable-
n n
stayed bridge were acquired for analyzing
including GPS displacements, air-temperature Correlation coefficient calculating by equation
and windspeed. Correlation analysis was then (1) shows the relationship between two variable
adopted to figure out how the air-temperature X and Y which has value domain from -1 to +1. If
and windspeed effect to GPS displacement data, coefficient rXY is closed to +1 or -1, it means that
from that the global deformation of the target variable X and Y have a very high correlation, in
bridge could be recognized in some significant contrary, if rXY is closed to 0, it means variable X
directions. Furthermore, regression model in and Y have a very low correlation.
both monovariant and multivariant variables - Step 2: Assessing the stability of correlation
were used to describe displacement modeling of coefficient that depends on the time interval of
the target bridge. Results of regression models monitoring:
were then used to assess which environmental 1-With a large number times of monitoring
factor and which significant direction of the (n≥ 50):
target bridge are useful for analyzing the
structural changes. 1−r 2
σ r≈ (3)
√n
2. Correlation analysis Then, the correlation between X and Y
satisfies the condition follows:
The correlation among variables can be
analyzed by using two kinds of formula: single |r|≥ 3 σ r (4)
correlation and multiple correlation. 2- If n < 50 the Fisher equation is used:
2.1. Single correlation analysis 1 1+ r
Z= ln (5)
Assume {(Xi, Yi} (i=1÷n) are two random 2 1−r
variables, the correlation coefficient rXY between Variance of Z can be calculated by:
variable X and variable Y can be calculated by
1
following steps [15,17]: σ r≈ ( 6)
- Step 1: Calculating the correlation coefficient √ n−3
between X and Y: and checking the correlation condition by
r XY =∑ ( X i− X́)¿ ¿ ¿ ¿
i |Z|≥ 3 σ Z (7)
Where: Figure 1 describes the correlation between
two variables X and Y:

(a) r = 1 (b) r = -1 (c) r = 0

 2

(d) r = 0.86 (e) r = -0.88 (f) r = 0

Figure 1. Examples of correlation between two variables
Correlation coefficient can be considered as function of r that is an unbiased method. By this
a “effect coefficient”, when the correlation calculation, substitution variable z can be defined
coefficient is approximately +1 or -1, that by:
means the effect between two variables are 1 1+r
z= log ⁡ (9)
very high. 2 1−r
- 95% confidence interval of correlation Then, the standard deviation of z is
coefficient: the correlation coefficient is calculated by:
affected by the oscillation of variables. Thus, it 1
is necessary to calculate 95% confidence s z= (10)
interval of correlation coefficient. To
√ n−3
Hence, the 95% confidence interval of z can
calculating the 95% confidence interval of
be substituted to the correlation coefficient by:
correlation coefficient, we have to use the
standard deviation of correlation coefficient
calculated by:
e 2 z −1
r= ( 11)
1−r 2
sr = √ (8) e2 z+ 1
√ n−2 2.2. Multiple correlation analysis
Equation 8 shows that sr and r are dependent, Considering p randome quantities x1, x2,...,
thus, using an unbiased method is necessary. xp, which are measured independently in n times
Ronald A. Fisher showed that calculating sr of a described in Table 1.
Table 1. An sample of monitoring data
Period Random quantities
X1 X2 ... Xk ... Xp
1 x11 x21 ... xk1 ... xp1
2 x12 x22 ... xk2 ... xp2
... ... ... ... ... ... ...
i x1i x2i ... xki ... xpi
... ... ... ... ... ... ...
n x1n x2n ... xkn ... xpn

A random variable is specified by expectation Estimation of expectation, variance and

M[xi], variance and correlating moment Kij: correlating moment can be defined by (11):
n
i=1 , p 1
K= { K ij } (12) M [ xk ] = ∑ x ,(k =1,2 ,… , p)( 13 )
j=1 , p n i=1 ki
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13
n
1 [ X 2 ] a+ [ X ] b− [ XY ]=0 ( 20 )
Dx =k
∑
n−1 i=1
n
(
2
x ki−M [ x k ] ) ( 14 )
{ [ X ] a+ nb− [ Y ] =0
1
K ki = ∑ ( x −M [ x k ])(x ii−M [ x i ] )( 15 ) Solving the set equations combining with
n−1 i =1 ki equation (1), parameters a and b are then
Dividing the correlating matrix (15) to the defined:
corresponding variance σ k =√ D andσ i =√ D ,
the correlating matrix can be defined as: √
a=r XY X́ 2−¿¿ ¿
r 11 r 12 … . r 1k b=Ý −a X́

…
(
r = r 21

r k1
r 22 … . r 2 k (16)
… .. … .. … .
r k 2 … . r kk
Analysing correlation between p random
) 3.2. Establishment of multivariant regression
Regression function f(X2, X3,…Xp ) describes
linear dependent between variable Y and p
quantities (Xi, Xj, Xk), the dependence between 2 variables (X1, X2…Xp) following the least square
quantities can be determined by partial principle, as shown:
correlation coefficients (Khanh Tran and E¿
QuangPhuc Nguyen, 2010; Khanh Tran and
DuTinh Le, 2010; Duc Tinh Le, 2012) calculating When p > 2 the multivariant equation is:
by equation below: Y =f ( X 1 , X 2 , … X P ) =a 0+ a1 X 1+ … a p X p (23)
r 12,34 … ( p−1)−r 1 p ,34 … ( p−1 ) r 2 p ,34 … ( p−1)
r 12,34 … p= (17) Notation:
√ ( 1−r 21 p , 34 …( p −1 ))( 1−r 22 p , 34 …( p −1 ) )
Statistical assessment of correlation 1 x 11 … . x p 1

{ () ()
( )
coefficients is done following Fisher criterion A= 1 x 12 … . x p 2
(assume: analyse 4 random quantities): … … .. … .. ….
R 21,234 (n−m) 1 x 1 n … . x pn
F Φ= ≥ F q (18) (24)
(1−R2i )(m−1) Y1 a0
where n is the number of quantities, m is
L= Y 2 ; Z= a1
number of parameters. If condition (18) is
correct, then correlation coefficient Ri is … …
accepted. Yn ap

3. Regression Establishment According to equation (22) and condition (23)

the matrix form of set equation can be
3.1. Establishment of monovariant regression established as (Khanh Tran and DucTinh Le,
2010):
The monovariant regression is used to
describe the correlation between two variables X AT AZ− AT L=0(25)
and Y as shown in equation below: Then, the result of parameters can be solved
by:
Y =aX +b ( 19 ) Z=¿ ¿
Parameters a, b are determined by least Monovariant and multivariant regressions are
square principle applying for n measurement applied for analyzing monitoring data of a cable-
couple (Y, X) which are: stayed bridge, including displacement data
acquired by GPS-RTK, and air-temperature data
{( Y i , X i ) }= {( Y 1 , X 1) , ( Y 2 , X 2 ) , … , ( Y n , X n ) }, then and windspeed data. Variable y in each kind of
set of equations can be written as below (Khanh regression model is chosen as the coordinate of
Tran, DucTinh Le, 2010): considering point along to separate direction,
variable x depends on each case of regression:
 2
(1) in case of monovariant regression, x variable This study applies regression model for
is air-temperature or windspeed; (2) in case of monitoring data to define parameters a1, a2, a0, b,
multivariant regression, two variables x1 and x2 with displacement y is affected by air-
are denoted to air-temperate and winspeed temperature t and windspeed v. Moreover, the
respectively. In both cases, parameter a 0 is the determination coefficient R2 is used to assess the
redundant between regression equation and the consistence of regression model for analyzing,
applied data. a0 has to satisfy the condition that it assessing and predicting the displacement of the
is white noise with normal distribution and ∑a0 = specific points.
0, its variance is constant. According to statistic, 4. Experiment
the determination parameter R2 is used to define 4.1. Introduction of the SHM system of Can
the appropriate regression model, it means that Tho cable-stayed bridge
R2 is closed to 1, the defined regression model is Can Tho cable-stayed bridge was started to
the most appropriate, the defined model explains build in 2004 crossing the Hau river to connect
effectively the effects of variables (Sanford Can Tho province to Vinh Long. Figure 2 and 3
Weisberg, 2005; Shumway and Stoffer, 2010; show the target bridge and its location.
Peter and Annick , 1987). Parameter R2 is defined
by:
RSS
R2=1− (27)
SYY
where: RSS is square summation of the
redundant between model and data; SYY is
square summation of the deviation between the
displacement i and the mean value.

Figure 2. Can Tho bridge location

Can Tho bridge was completed to use in 2010
with the longest main span in the South East Asia
(550m), total length of the main bridge is
2,750m, the height of tower is 171m; its concrete
girder is 26m wide, but 210 middle length of the
main span is made by steel structure.
Structural Health Monitoring System – SHMs
was established in 2010, including many kinds of
sensors. It is considered as a modern monitoring
system in Vietnam. Figure 4 shows the sensor
locations of the SHMs of Can Tho bridge.
Figure 3. Can Tho bridge
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13

Figure 4. Diagram of sensor locations on Can Tho bridge (Farrar and et al., 2000)

Global Positioning System – GPS is applied in but the acquired GPS data are calculated to save
the SHMs of Can Tho bridge which includes 09 the average value in 1 minute, 10 minute, 1 hour
rover sensors located on some specific location and 1 day.
such as on the top of towers, on the main girder GPS technology shows various advantages in
and other piers. Two base stations are monitoring the displacements of a large-scale
established at near the management office and structure, especially in monitoring of a long-span
near the North Pylon (Fig. 5, 6 and 7). GPS bridge, such as it can measures in real-time;
equipment is Leica brand with GMX 902 version overcome all kinds of weather conditions; can
that has specific errors provided by the reaches to millimeter level in accuracy. However,
manufacture with ±10mm ± 1ppm in horizontal GPS technology has a bit high cost in its
plane and ±20mm ±1ppm in vertical direction. application and GPS data processing is still a
The principle of GPS measurement is used in challenge to assess the structural health.
SHMs is Real Time Kinematic – RTK technique.
The frequency of GPS sensor can reach to 20Hz,
 2
Figure 5. GPS sensors locations on Can Tho bridge[Farrar, and et all., 2000]

Figure 6. GPS base station location

Figure 7. GPS rover on girder location

4.2. Experimental data days (from January 2nd to 5th in 2017) are used
In the monitoring system of the target bridge, to analyse that include GPS displacement data,
all sensors acquire data in real-time at specific air-temperature data and windspeed data. Figure
frequency, then the acquired data are saved in 8 shows the experimental data in 4 specific
short-time. Furthermore, the short-time data are monitoring points that are two points on the top
then averaged value in 1 minute, 10 minute, 1 of towers and two other points on the girder: at
hour or 1 day for saving long-time. However, it is the middle of main span and at the quarter main
a challenge to store and analyze the long-time span. Figure 9 shows the environmental data
monitoring data because it is huge volume. GPS including air-temperature and windspeed.
system acquires the displacement monitoring
data of specific points along to 3 directions:
longitudinal direction (x – direction); lateral
direction (y-direction) and vertical direction (z-
direction).
In this experimental study, the 10 minute
average data of the target bridge extracted in 3
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13

(a) North tower (b) South tower

(c) Middle span (d) Quarter span

Figure 8. GPS experimental data

Figure 9. Environmental data

4.3. Experimental Results
4.3.1. Correlation analysis the 95% confident interval of each correlation
In this experimental study, the correlation coefficient was also calculated. The results of
between GPS displacement and the correlation coefficients and the 95% confident
environmental parameters was analyzed. Then
 2
interval of some specific direction were shown in Table 2.

Table 2. Results of correlation coefficients and 95% confident interval

Points Direction Correlation Coefficient 95% confident interval of
Air-temperature Windspee high coefficients
d
North Longitudinal - x 0.63 -0.29 0.53 ÷ 0.68
tower Lateral - y -0.35 -0.48
points Vertical - z 0.36 0.39
Middle Longitudinal - x 0.37 0.15
span Lateral - y 0.30 -0.38
Vertical - z -0.90 0.14 -0.93 ÷ -0.88
Quarter Longitudinal - x 0.67 0.25 0.51 ÷ 0.70
span Lateral - y 0.24 -0.24
Vertical - z -0.88 0.14 -0.91÷ -0.86
South Longitudinal - x -0.68 0.33 -0.70 ÷ -0.58
tower Lateral - y -0.40 -0.49
Vertical - z 0.46 0.37

The results of correlation coefficients between air-temperature and the girder points
between GPS data and environmental data show along to vertical direction that are very high
some discussion follows: contravariant correlation, the coefficients are
- Correlation between windspeed and GPS -0.90 and -0.88 for the middle and quarter span
data is very weak which the correlation respectively. While, the lateral direction of the
coefficients of all points along to all directions girder points shows the low correlation with the
that are less than 0.5. It can be understood that air-temperature; and the longitudinal direction
the windspeed has less effect to the of the quarter span point shows high correlation
displacements of the target bridge. Moreover, the (0.67). It can be explained that the quarter span
windspeed has correlated with GPS data along to point is non-symmetric point, thus the
the lateral direction (y – direction) that is higher movement of this point along to x-direction is
than other directions (x- and z- directions). The much larger than the middle span point.
tower points have higher correlation with the According to the discussion of correlation
GPS data than the girder points. This kind of results, it can be concluded that the air-
result is appropriate with the characteristic of a temperature has affected on the GPS data of the
cable-stayed bridge. target bridge. Then, the significant directions of
- Correlation between the air-temperature the bridge movement can be recognized that are
and GPS data is very high in some specific the longitudinal direction (x- direction) of the
directions such as vertical direction of the girder tower points; and the vertical direction (z-
points and longitudinal direction of the tower direction) of the girder points. These significant
points. Statistically, the correlation coefficients directions are then used to analyze regression
along to the longitudinal direction (x-) of the model. Moreover, the global displacement model
tower points that are larger than 0.5, and they of the target bridge could be recognized through
show reverse correlation between tow point the GPS monitoring data as shown in Figure 10.
(0.63 and -0.68). While, the lateral and vertical
directions of tower points show small correlation
coefficients with the air-temperature (less than
0.5). This results show the coincidence with the
movement of the bridge pylon, that they just
show significant movement along to the
longitudinal direction. Besides, the correlation
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13
*) Establishment of monovariant regression
model – Model 1
Monovariant regression model was applied
for the specific points on the target bridge along
to the significant directions that were recognized
such as the longitudinal direction of the tower
points (namely #Pt1 and #Pt4) and the vertical
direction of the girder points (namely #Pt2 and
#Pt3). In this application, GPS displacement data
is considered as a function of air-temperature
variable windspeed seperately, as described in
equation (28).
Figure 10. Global model of GPS displacement f =a 0+ a1 t (28)
of the target bridge The least square principle was used to define
4.3.2. Regression analysis the regression functions along to each direction.
The results of monovariant regression along to
significant directions are shown in Table 3.
Table 3. Monovariant regression results
Function #Pt1 #Pt2 #Pt3 #Pt4
parameters (x-direction) (z-direction) (z-direction) (x-direction)
Temp. Windspee Temp. Windspeed Temp. Windspeed Temp. Windspeed
d
a0 -0.040 0.045 43.150 42.627 40.238 39.954 550.467 550.393
a1 0.0033 0.0012 -0.0203 -0.0106 -0.0110 -0.0058 -0.0028 -0.0012
Figure 11 shows the monovariant effects of temperature, this is coincide with
regression line and the 95% confident interval the characteristic of the target bridge.
of the significant directions as well as the Figure 12 shows the monovariant regression
determination coefficient R2 of each case is results between GPS data and winspeed data as
also shown. It can be seen that the well as the 95% confident interval. The R2
monovariant regression model is much coefficients are also shown in the significant
coincide with the vertical direction of the directions separately. It can be seen that the
girder points which the R2 coefficients are winspeed has more effects to the vertical
direction of the girder points than the
0.80 and 0.77 respectively. While, R2
longitudinal direction of the tower points,
coefficients of the tower points (#Pt1 and however, the very low R2 coefficients show that
#Pt4) are low (0.39 and 0.47). It means that the monovariant regression with winspeed
displacements of girder point (middle span variable is not appropriate for GPS monitoring
and quarter span) are mainly caused by data.
 2

(a) #Pt 1 – x direction; R2 = 0.39 (b) #Pt 4 – x direction; R2 = 0.47

(c) #Pt 2 – z direction; R2 = 0.80 (d) #Pt 3 – z direction; R2 = 0.77

Figure 11. Monovariant regression with air-temperature variable of the significant directions

(a) #Pt 1 – x direction; R2 = 0.04 (b)

#Pt 4 – x direction; R2 = 0.05
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13

(c) #Pt 2 – z direction; R2 = 0.15 (d) #Pt 3 – z direction; R2 = 0.14

Figure 12. Monovariant regression with winspeed variable of the significant directions
*) Establishment of multivariant regression where: t is the air-temperature variable; v is
model – Model 2 the windspeed variable.
In this study, the GPS displacement along to
The least square principle was used to define
each significant direction is considered as a
the regression functions. The results of
function of air-temperature and winspeed
multivariant regression along to significant
variables, as described in equation (29):
directions are shown in Table 4.
f =a 0+ a1 t+a 2 v( 29)
Table 4. Multivariant regression results
Function #Pt1 #Pt2 #Pt3 #Pt4
parameter (x-direction) (z-direction) (z-direction) (x-direction)
s
a0 -0.052 43.186 40.257 550.476
a1 0.0039 -0.0220 -0.0120 -0.0033
a2 -0.0013 0.0039 0.0021 0.0010
Figure 13 shows the multivariant regression showed in the parameters a1 and a2 of the
plane and the 95% confident interval of the egression functions (Table 4). Moreover, the
significant directions. The determination air-temperature affects to GPS displacement
2
coefficient R of each case is also shown in each along to vertical direction of girder points that
figure. It can be seen that the windspeed has is approximately six times of the windspeed’s
effects to GPS monitoring data along to the effects. Similar to the monivariant regression
significant directions of the target bridge that case, displacements of the girder points are
is less than the air-temperature’s effects which caused mainly by the effects of temperature.
 2
(a) #Pt 1 – x direction; R2 = 0.42 (b) #Pt 4 – x direction; R2 = 0.49

(c) #Pt 2 – z direction; R2 = 0.82 (d) #Pt 3 – z direction; R2 = 0.79

Figure 13. Multivariant regression of the significant directions
4.3.3. Statistical analysis monitoring data along to longitudinal direction
of the tower points and vertical direction of the
A regression model is considered as the
girder points. This studied conclusion is fitted
fitting model if it satisfies the condition that
to the characteristic of the target bridge, and
the redundant between the regression model
the global deformation of a cable-stayed bridge
and the real data are white noise. It means that
could be recognized based on the correlation
the redundant has to have the normal
with the air-temperature. Otherwise, the
distribution and its p-value has to be less than
windspeed has a low or no correlation with the
0.05, then the GPS monitoring data have
GPS monitoring data.
statistical significance, and the regression
(3) Regression analysis showed that the
model is appropriately to describe the
monovariant regression with temperature
displacement of a structure. Therefore, the
variable or the multivariant regression with
redundant of both monovariant and
temperature and winspeed variables are
multivariant regression cases of experimental
appropriate to describe the displacement
study were then tested the white noise
model of the main span points along to vertical
condition. The results showed that the
direction.
regression models with the air-temperature
(4) Correlation analysis can figure out what
variable defined in both cases that are fitted
are the features that effect to the
model for GPS monitoring data, while the
displacements of a long-span bridge, from that
monovariant regression with the winspeed
which one is the main feature causing the
variable is not appropriate for GPS
displacements could be recognized by the high
displacement data of the target bridge.
correlation coefficients. This study shows that
the correlation analysis is an effective method
5. Conclusions
to analyze the GPS displacement data of the
The results of this study can figure out some target bridge.
conclusions below:
(1) GPS technology with the Real Time References
Kinematic technique that can monitor the
Kaloop, M. R., & Li, H, 2009. Monitoring of
displacements of large-scale structures such as
bridge deformation using GPS technique.
a long-span bridge, and GPS-RTK monitoring
KSCE Journal of Civil Engineering, 13(6),
data can be used to assess the structural health
423-431.
during its operation
(2) In the target bridge of this study, GPS Kaloop, M. R., & Li, H, 2011. Sensitivity and
monitoring data have high or very high analysis GPS signals based bridge damage
correlation with the air-temperature
 Phan Quang Văn và nnk/Tạp chí Khoa học Kỹ thuật Mỏ - Địa chất 58 (2), 11 - 17 13
using GPS observations and wavelet statistical process control. Journal of
transform. Measurement, 44(5), 927-937 . Structural Engineering, 126(11), 1356-1363.
Celebi, M, 2000. GPS in dynamic monitoring of Hien V. Le, Mayuko Nishio, 2015. Time-series
long-period structures. Soil Dynamics and analysis of GPS monitoring data from a
Earthquake Engineering, 20(5), 477-483. long-span bridge considering the global
deformation due to air temperature
Cheng, P., John, W., & Zheng, W, 2002. Large
changes. Journal of Civil Structural Health
structure health dynamic monitoring using
Monitoring, Springer 5, 415–425.
GPS technology. In FIG XXII International
Congress, Washington, DC USA. H. V. Le, M. Nishio, H. Yamada, H. Katsuchi,
2015. Statistical condition assessment of a
Fujino, Y., Murata, M., Okano, S., & Takeguchi,
cable stayed bridge using GPS structural
M, 2000. Monitoring system of the Akashi
health monitoring data. The 7th
Kaikyo Bridge and displacement
International Conference of Structural
measurement using GPS. In SPIE's 5th
Health Monitoring of Intelligent
Annual International Symposium on
Infrastructure SHMII, Italy.
Nondestructive Evaluation and Health
Monitoring of Aging Infrastructure (pp. 229- Sanford Weisberg, 2005. Applied Linear
236). International Society for Optics and Regression. Book, third edition. Wiley-
Photonics. Interscience.
Sohn, H., Dzwonczyk, M., Straser, E. G., Shumway, R. H., & Stoffer, D. S, 2010. Time
Kiremidjian, A. S., Law, K. H., & Meng, T, series analysis and its applications: with R
1999. An experimental study of examples. Springer.
temperature effect on modal parameters of
Peter J. Rousseeuw, Annick M. Leroy, 1987.
the Alamosa Canyon Bridge. Earthquake
Robust Regression and Outlier Detection.
engineering & structural dynamics, 28(8),
Book, Wiley-Interscience.
879-897.
Cao Van Nguyen et al, 2002. Theory of
Cornwell, P., Farrar, C. R., Doebling, S. W., &
probability and mathematical statistics.
Sohn, H, 1999. Environmental variability of
Publisher of education, Hanoi.
modal properties. Experimental Techniques,
23(6), 45-48. Khanh Tran, QuangPhuc Nguyen. 2010.
Structural Deformation and Displacment
Farrar C. R., Cornwell P., Doebling S. W., &
Monitoring. Publisher of Transportation.
Prime M. B, 2000. Structural Health
Monitoring Studies of the Alamosa Canyon Khanh Tran, DucTinh Le, 2010. Application of
and I-40 Bridges. Los Alamos National correlation analysis method in assessing the
Laboratory, LA-13635-MS. structural displacement. Journal of science
and technique of mining and geology (31).
Omenzetter P., & Brownjohn J. M. W, 2005. A
seasonal ARIMAX time series model for Duc Tinh Le, 2012. Application of statistical
strain-temperature relationship in an methods in analyzing the deformation of
instrumented bridge. Proceedings of the 5th hydropower projects in Vietnam. Final report
International Workshop on Structural Health of topic for supporting PhD student, code
Monitoring, 299-306. number N2010-31.
Omenzetter, P., & Brownjohn, J. M. W, 2006.
Application of time series analysis for
bridge monitoring. Smart Materials and
Structures, 15(1), 129-138.
Sohn, H., Czarnecki, J. A., & Farrar, C. R, 2000.
Structural health monitoring using