You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/318855806

Structural Health Monitoring of pavement assets through acoustic signature

Conference Paper · June 2017


DOI: 10.1201/9781315100333-127

CITATIONS READS

10 559

4 authors:

Rosario Fedele Filippo Praticò


Mediterranean University of Reggio Calabria Mediterranean University of Reggio Calabria
38 PUBLICATIONS   174 CITATIONS    325 PUBLICATIONS   1,561 CITATIONS   

SEE PROFILE SEE PROFILE

Riccardo Carotenuto Francesco Giuseppe Della Corte


Mediterranean University of Reggio Calabria Mediterranean University of Reggio Calabria
134 PUBLICATIONS   935 CITATIONS    266 PUBLICATIONS   2,486 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Highly linear and sensitive temperature sensors based on SiC diodes for a wide range o temperature View project

Unconventional Neural Networks View project

All content following this page was uploaded by Rosario Fedele on 27 September 2017.

The user has requested enhancement of the downloaded file.


Bearing Capacity of Roads, Railways and Airfields – Loizos et al. (Eds)
© 2017 Taylor & Francis Group, London, ISBN 978-1-138-29595-7

Structural Health Monitoring of pavement assets


through acoustic signature

R. Fedele, F.G. Praticò, R. Carotenuto & F.G. Della Corte


University Mediterranea of Reggio Calabria, Italy

ABSTRACT:  Traffic causes damages, vibrations and acoustic waves. These affect maintenance, rehabili-
tation, asset management, health issues in bystanders, and damages to historical buildings. Consequently,
structural health monitoring and damage detection are vital. The main scope of the study described in this
paper was the detection and analysis of the vibro-acoustic signature of a pavement. Traffic-induced vibra-
tions and acoustic waves were analyzed. Accelerated loading tests were carried out. A set of asphalt con-
crete slabs was built and tested through a linear kneading compactor (EN 12697-22:2007). A microphone
was placed in a hole drilled in the slab and signals were sampled and processed, before and after cracks
appearance. Comparative analyses before-after were carried out. Results are encouraging about the pos-
sibility to associate to a given pavement condition a specific acoustic signature, based on course properties
and state of health (structural health monitoring). Results can benefit both practitioners and researchers.

1  INTRODUCTION tion (Praticò et  al. 2017), and cracks and defects
of the wearing course through visual inspection or
Noise and vibrations from road traffic cause image analysis (Wang et al. 2016, Koch et al. 2015,
“health” issues for both people and infrastructures Schnebele et  al. 2015). Another approach refers
and originate in the wheel-pavement interface to the estimation of the status of the “bulk prop-
(Praticò & Vaiana 2015, Praticò et  al. 2015, De erties” of pavement layers. In this case, various
Maio & Vitetta, 2015). Hence, Structural Health devices and techniques can be used, like the ground
Monitoring (SHM) and Structural Health Assess- penetrating radar or GPR (Pitonak & Filipovsky
ment (SHA) are needed. 2016, Rodés et  al. 2015, Porubiaková et  al 2014,
Different methods for SHM/A of civil engineer- Krysiński & Sudyka 2013), the thermal imager
ing structures and infrastructures can be used: i) (Solla et al. 2014), the Light Weight Deflectometer
Acoustic emission (Rehman et  al. 2016, Behnia or LWD (Gosk 2016). A further way refers to the
et  al. 2014); ii) Impact echo (Hola & Schabowicz analysis of the response of the pavement material
2010); iii) Sonic tomography (Rehman et al. 2016, (asphalt concrete mixtures) at different stress lev-
Choi et al. 2016); iv) Ultrasonic tomography (Hola els (Gao et al. 2015, Liu et al. 2014, Mounier et al.
& Schabowicz 2010, Hoegh et al. 2011); v) Impulse 2012). Recently, sensors-based SHM systems were
response (Hola & Schabowicz 2010); vi) Acoustic developed (Miraliakbari et  al. 2014, Lajnef et  al.
emission techniques (Behnia et al. 2014); vii) Ther- 2013, Yu et al. 2013, Lu et al. 2002).
mography and ultrasound (Rayleigh waves) (Agge- The acoustic methods are gradually proving
lis et  al. 2010); viii) Pulse velocity test (IAEAV their effectiveness for the purposes cited above
2002); ix) Ultrasound Pulse Echo testing (IAEAV (Alavi et al. 2016, Praticò et al. 2016).
2002); x) Relative amplitude method (IAEAV The study presented in this paper focuses on
2002); xi) Velocity versus rebound number curves SHM for transportation infrastructures through
(IAEAV 2002); xii) Ultrasonic guided waves (Pahl- acoustic/vibrational methods. In more detail, the
avan et al. 2016, Pahlavan & Blacquière 2016). acoustic signals of road pavements are analyzed
Most of the methods above were designed for in the aim of detecting and studying their specific
SHM/A of buildings and bridges and they are properties (signature of the signal) which describe
not always applicable for the detection of inho- the actual pavement condition (fault detection).
mogeneities and cracks in road pavements. Road
infrastructures can be monitored through different
approaches. 2  OBJECTIVES
One of these is the analysis of the surface char-
acteristics in terms of texture (Praticò & Vaiana The main objectives and scopes of this study are to
2015), roughness (Mubaraki 2016), sound absorp- set up a new method for the analysis of the acoustic

869
signals generated by a pavement in order to identify Table  2.  Factorial plan of experiments (Task  5 and
its acoustic signature and to allow monitoring its Task 6).
Structural Health Conditions (SHM) and to opti-
mizing the pavement management process. This Groove Type
dimensions of load
paper illustrates the results of the preliminary stage Type of
of the research program. Test ID mm Slab ID groove kN

1 0 × 0 × 0 UC* TR** 0.7


3  DESIGN OF EXPERIMENTS 2 260 × 10 × 3 LC* TR** 0.7
3 260 × 10 × 3 HC* TR** 0.7
In pursuing the above objectives, in order to gather
the signals, the design of the experiments included * UC, LC, HC stand for Un-Cracked, Lightly, and
a first set of activities on Hot Mix Asphalt (HMA) Highly Cracked slabs respectively.
samples produced and tested in the laboratory. ** TR: Transversal to the wheel path. The grooves have
The main tasks of the research program are a length of 260 mm, a width of about 3 mm and two dif-
ferent depths, i.e. 10 mm for the LC slab and 20 mm for
shown below: the HC slab.
Task 1. Mechanistic analysis of the pavement.
Task 2. Set up of the hardware (slab, embedded/
in-hole microphone, cable, and laptop).
Task 3. Matlab® coding for acoustic signals
recording and analysis.
Task 4. Preliminary tests.
Task 5. Design of the tests.
Task 6. Tests (see Table 2).
Task 7. Results analysis.
The study presented in this paper refers to the
preliminary tests.
In Task  1, the KENPAVETM (Huang 2004)
method was used in order to study the critical lay-
ers for a given class of pavements. Based on the
analysis carried out, it was possible to derive that
for the types of pavements under investigation,
fatigue failures usually start in the interface between
the HMA bottom and the underneath un-bound
base course. These inferences positively matched
with the international literature (AASHTO 1993,
Asphalt-Institute 1991, Beskou et al. 2016, Pokor-
skia et al. 2016).
Preliminary tests helped define procedures and
methods (see Table 1).
Table 2 illustrates the dimensions of the cracks
(“groove”) created in the bottom surface of the
HMA slabs in order to associate distress type
(groove) and acoustic signature. To this end three
identical HMA slabs were produced (see Figure 2).
Accelerated loading tests (Task  5 and 6) on
asphalt concrete slabs were carried out using the
wheel tracking machine (see the following figure,
Figure 1.  Test setup.

Table 1.  HMA slabs.

Bitumen Filler EN 12697-22:2007). A microphone was placed in


Dimensions percentage Gmb AV percentage a hole drilled in the slab and it was secured to the
Mix slab surface through plasticine (modeling clay).
mm3 % g/g % % Type Afterwards, the test wheel (5) was put down
500 × 260 × 50 5 2.3 8 4 DGFC against the slab and a constant load of 0.7  kN
was imposed between slab and wheel (as per the

870
Figure 3.  Flow chart of the Matlab codes (Task 3).

For spectral analysis, it may be observed that


the power spectral density (or autospectral density
function, Gxx(  f  )) was used. As is well known, it
can be derived based on Finite Fourier Transforms
Figure  2.  The three slab types under investigation. (Bendat & Piersol 2011):
UC = Un-Cracked; LC = Lightly Cracked; HC = Highly
Cracked. 1 
Gxx ( f ) = 2 lim E X ( f ,T ) 
2
(2)
T →∞ T  

s­ tandard mentioned above). This corresponds to a or simply


pressure of about 280 kPa.

Load =
0.7 kN
=
0.7 kN
Gxx ( f ) =
2
T
E { Xf ,T } 2
(3)
0.05 ⋅ 0.05m 2 0.0025m 2 (1)
kg where Gxx( f ) is the one-sided, autospectral density
= 280 kPa ≅ 0.3 2
cm function (which is twice the two-sided autospectral
density function) and it is defined for 0  <  f  <  ∞,
It is noted that the contact pressure of light vehi- f is the frequency, T is the record length, E[ ] is
cles is about 200 kPa, while the contact pressure of the ensemble average, X( f, T ) is the finite Fourier
heavy vehicles may be as high as 500 kPa. Transform of the signal x(t) of length T, t stands
Figure 1 illustrates the test setup. In more detail, for time, the absolute value of the FFT is squared.
once the HMA slab (1) was produced in the labo- In terms of Power Spectral Density (PSD), it
ratory (EN 12697–22:2007), the same slab was results:
drilled in order of allocating the thermocouples
(2), and the microphone (3). Finally, the slab was PSD = Gxx ( f ) = 0, for f < 0 (4)
fixed in its metallic box (4), the doors of the ther-
mal camera were closed and the conditioning of The finite Fourier Transform X does not have
the material was carried out (30°C). the same units of measure (and dimensions) of the
The following figure (Figure 2) shows the slabs signal x(t), being:
under investigation, i.e. the Un-Cracked (UC) slab,
the Lightly Cracked (LC) slab, and the Highly T

Cracked (HC) slab. X ( f ,T ) = ∫ x (t ) e − j 2π ft dt (5)


0

4  MODELLING, RESULTS AND Furthermore, the two-sided autospectral den-


DISCUSSION sity function Sxx( f ) can be defined for −∞ < f < ∞,
by setting (Bendat & Piersol 2011):
Figure 3 summarizes the main phases of modeling
1
and compiling process. Power spectra were ana- Sxx ( f ) = ⋅ Gxx ( f ) , when f > 0 (6)
lyzed and centroids were derived. 2

871
and

Sxx ( − f ) = Sxx ( f ) (7)

If the Fast Fourier Transforms (FFT) are used,


then the sampling frequency must be used and the
above formula becomes:
2
FFT
PSD = 2 ⋅ (8)
N ⋅ Fs

where Fs is the sampling frequency (Hz) and N is


the length of the recorded signals.
Based on the literature and based on the above,
PSD expresses the power (Watt) per unit of fre-
quency. The logarithm scale is usually used for the
y-axis.
The analysis of frequencies was carried out in
terms of spectral centroids. The following algo-
rithm was used (Schubert & Wolfe 2006):
N −1

∑p n ⋅ fn
fc = n=0
(9) Figure  4.  Example of test results for the Un-Cracked
N −1
(UC) slabs with the y-axis in: a) logarithmic scale; b) lin-
∑ pn
n=0
ear scale.

where fc is the spectral centroid, i.e. the “center of


mass” of the spectrum; N is the sample length of
the acoustic signal; pn represent the weights, i.e. the
values on the y-axis; fn are the frequencies (x-axis
of the spectrum).
Figures  4 to 6 illustrate the results obtained.
Periodograms (PSD versus Frequency) refer to the
tests carried out on the three classes of slabs under
investigation.
The high sampling frequency used (i.e., 192 kHz)
allowed obtaining periodograms in the frequency
range 0–96 kHz. In the pursuit of better analyzing
the plots, when using the logarithmic scale for the Figure 5.  Example of test results on the Lightly Cracked
y-axis (Figures 4a, 5, 6), a value of −60 dBW/Hz (LC) slabs.
was considered as the lower bound.
In Figure 4a, six main peaks are marked. Note
that the terms peak is herein used in the sense of
area of the plot between relevant minima (valleys).
Figure 4b refers to the same signal, but in this
case the y-axis is the PSD and not its logarithm.
It is important to note that, in this study, all the
passages of the wheel over the UC slab (about 200
passages) were analyzed. In Figures  4a, 5, 6, the
x-axis reports the frequencies (Hz), while the y-axis
refers to the PSD in a logarithmic scale.
Figure  5 refers to the analysis of the recorded
signals and power spectra once slabs were lightly
cracked (Table 2).
These cracks had the following dimensions: Figure 6.  Example of test results on the Highly Cracked
depth 10 mm, length 260 mm, width 3 mm. (HC) slabs.

872
The cracks crossed the slabs in a direction and x-axis (834–892  Hz), becoming almost
orthogonal to the wheel path and were set in the negligible.
bottom part of the slab (Test 2 in Table 2). 6. For frequencies higher than 1 kHz, minor peaks
This configuration of the slab was called Lightly appear for Lightly Cracked slabs.
Cracked (LC).
Figure 6 illustrates an example of output for the
Highly Cracked (HC) slabs. 5  DISCUSSION
The crack crossed the slab in a direction orthog-
onal to the wheel path and was set in the bottom Based on the in-lab tests, procedures and setup have
part of the slab. been greatly improved but several circumstances
This crack had the following dimensions: depth call for a sound and more comprehensive study:
20 mm, length 260 mm, width 3 mm.   i. Spectrum dependency on a number of factors
Table  3  summarizes the preliminary analy- (slab small scale, mixture typology, intrinsic
ses carried out in terms of spectral centroids properties of the mixes, see Praticò & Moro
(Equation 9). 2011, and experimental set up) implies a set
Note that due to the dependence of the centro- of boundary conditions which call for further
ids on p (y-axis), the analyses were carried by refer- research and on–site testing;
ring to f-PSD plots and data, i.e., without using ii. An in-depth study about the best way to cou-
the logarithm. ple the microphone(s) to the pavement has to be
Based on the analyses of the acoustic signatures carried out because of the issues which might
of the slabs under investigation (Figures 4–6), i.e., emerge when implementing the method in full
in terms of PSD variation when cracks are present scale;
(see Table 3), it can be noted that: iii. Frequency range affects the reliability of results
1. The peak N. 1, which belongs to the frequency shown above. It should be noted that the analy-
range 20–200 Hz, does not undergo appreciable ses were carried out in the range 20–1.5  kHz
variations. The maximum is almost constant and not in the full frequency range.
(−25 dBW/Hz), as well as its frequency position
(47–54 Hz).
6  CONCLUSIONS AND
2. The peak N.  2, located in the frequency range
RECOMMENDATIONS
200–380  Hz, increases of about 5  dBW/Hz,
while its spectral position moves from 305 Hz to
Based on the results obtained at this stage, it is pos-
356 Hz.
sible to state that cracks seem to imply a reduction
3. The peak N. 3, which refers to the range 380–
of the peak frequency and an increase of the peak
500  Hz, increases of about 10  dBW/Hz, while
amplitude of part of the vibration phenomenon.
its frequency position stays almost constant
This latter is generated in the wheel-pavement inter-
(427–436 Hz).
face, propagates through the slab, and it is recorded
4. The peak N. 4 (500–750 Hz) undergoes signifi-
by the microphone. Furthermore, higher loads
cant deviations in terms of both level and fre-
seem to correspond to higher peak amplitudes. In
quency. Its maximum decreases of 10 dBW/Hz,
contrast, peak frequency does not undergo appre-
while its frequency decreases (620 to 539 Hz).
ciable variations based on load. Finally, the PSD
5. The peak N.  5 (750  Hz to 1000  Hz) has vari-
peaks level, power spectrum shape, and peak posi-
ations in terms of both y-axis (−10  dBW/Hz)
tion seem to undergo appreciable variations when
considering Un-Cracked (UC), Lightly Cracked
(LC), or Highly Cracked (HC) slabs.
Results are encouraging about the possibility to
Table 3.  PSD peaks and spectral centroids ( fc ).
associate to a given pavement its acoustic signa-
PSD-peaks ID ture, based on course properties and state of health
(structural health monitoring). Results can benefit
1 2 3 4 5 6 both practitioners and researchers.
Further research is needed. Based on the above,
20–200 200–380 380–500 500–750 0.75–1 1–1.5 other in-lab and on-site tests will be carried out in
the future for three main purposes, i.e., 1. To verify
fc Hz Hz Hz Hz Hz kHz the effectiveness of the new method; 2. To collect
UC 47 305 427 620 834 1089
acoustic signals in order to create a sound data-
LC 54 333 436 551 892 1062
base of acoustic signatures of different types of
HC 53 356 430 539 880 1274 road pavements; 3. To jointly consider mixes prop-
erties, including thickness and remaining factors.

873
REFERENCES Krysiński, L., and Sudyka, J. 2013. GPR abilities in
investigation of the pavement transversal cracks. Jour-
AASHTO 1993. AASHTO Guide for Design of Pave- nal of Applied Geophysics 97: 27–36.
ments Structures. American Association of State High- Lajnef, N., Chatti, K., Chakrabartty, S., Rhimi, M., and
way and Transportation Officials, Washington, DC. Liu, T., Zhang, X-N., Li, Z., and Chen, Z-Q. 2014.
Aggelis, D.G., Kordatos, E.Z., Soulioti, D.V., Matikas, Research on the homogeneity of asphalt pavement
T.E. 2010. Combined use of thermography and ultra- quality using X-ray computed tomography (CT) and
sound for the characterization of subsurface cracks fractal theory. Construction and Building Materials 68:
in concrete. Construction and Building Materials 24: 587–598.
1888–1897. Lu, J.J., Mei, X., & Gunaratne, M. 2002. Develop-
Alavi, A.H., Hasni, H., Lajnef, N., and Chatti, K. 2016. ment of an automatic detection system for meas-
Continuous health monitoring of pavement systems uring pavement crack depth on Florida roadways.
using smart sensing technology. Construction and Project sponsored by the Florida Department of
Building Materials 114: 719–736. Transportation.
Asphalt-Institute. 1991. Thickness Design­—Asphalt Miraliakbari, A., Hahn, M., and Maas, H-G. 2014.
Pavements for Highways and Streets. Manual Series 1 Development of a multi-sensor system for road condi-
(MS-1), Asphalt Institute, Lexington, KY. tion mapping. The International Archives of the Pho-
Behnia, A., Chai, H.K., Shiotani, T. 2014. Advanced togrammetry, Remote Sensing and Spatial Information
structural health monitoring of concrete structures Sciences, vol. XL-1.
with the aid of acoustic emission. Construction and Mounier, D., Di Benedetto, H., and Sauzéat, C. 2012.
Building Materials 65: 282–302. Determination of bituminous mixtures linear proper-
Bendat, J.S. & Piersol, A.G. 2011. Random Data Analy- ties using ultrasonic wave propagation. Construction
sis and Measurement Procedures. Fourth Edition. and Building Materials 36: 638–647.
John Wiley & sons, inc., publication. Mubaraki, M. 2016. Highway subsurface assessment
Beskou, N.D., Tsinopoulos, S.V., Hatzigeorgiou, G.D. using pavement surface distress and roughness data.
2016. Fatigue cracking failure criterion for flexible International Journal of Pavement Research and Tech-
pavements under moving vehicles. Soil Dynamics and nology 9: 393–402.
Earthquake Engineering 90: 476–479. Pahlavan, L., and Blacquière, G. 2016. Fatigue crack siz-
Choi, P., Kim, D.H., Lee, B.H., and Won, M.C. 2016. ing in steelbridge decks using ultrasonic guided waves.
Application of ultrasonic shear-wave tomography NDT&E International 77: 49–62.
to identify horizontal crack or delamination in con- Pahlavan, L., Mota, M.M., and Blacquière, G. 2016.
crete pavement and bridge. Construction and Building Influence of asphalt on fatigue crack monitoring in
Materials 121: 81–91. steel bridge decks using guided waves. Construction
De Maio, M.L., Vitetta, A. 2015. Route choice on road and Building Materials 120: 593–604.
transport system: A fuzzy approach. Journal of Intel- Pitonak, M., and Filipovsky, J. 2016. GPR application
ligent and Fuzzy Systems 28: 2015–2027. – non-destructive technology for verification of thick-
Gao, H., Yang, X., Zhang, C. 2015. Experimental and nesses of newly paved roads in Slovakia. Procedia
numerical analysis of three-point bending fracture of Engineering 153: 537–549.
pre-notched asphalt mixture beam. Construction and Pokorskia, P., Radziszewski, P., Sarnowski, M. 2016.
Building Materials 90: 1–10. Fatigue life of asphalt pavements on bridge decks.
Gosk, W. 2016. Stiffness Estimation of the Soil Built-in Procedia Engineering 153: 556–562.
Road Embankment on the Basis of Light Falling Porubiaková, A., Grin, M., and Slabeja, M. 2014. Non-
Weight Deflectometer Test. Procedia Engineering 143: invasive and Effective Method of the Pavement
395–402. Investigation: Ground Penetrating Radar. Procedia
Hoegh, K., Khazanovich, L., Yu, H.T. 2011. Ultrasonic Engineering 91: 464–468.
Tomography Technique for Evaluation of Concrete Praticò, F. G., and Moro, A. 2011. In-lab and on-site
Pavements. Transportation Research Record: Journal measurements of hot mix asphalt density: Conver-
of the Transportation Research Board 2232: 85–94. gence and divergence hypotheses. Construction and
Hola, J. & Schabowicz, K. 2010. State of the art non- Building Materials 25(2): 1065–1071.
destructive methods for diagnostic testing of building Praticò, F.G., Della Corte, F., Merenda, M. 2016. Self-
structures–anticipated development trends. Archives Powered Sensors For Road Pavements, 4th Chinese
of civil and mechanical engineering 10(3). European Workshop (CEW) on Functional Pavement
Huang, Y.H. 2003. Pavement Analysis and Design. Pear- Design. Delft, the Netherlands. Edited by Sandra
son Prentice Hall, Upper Saddle River, NJ, USA. Erkens et al., Taylor & Francis Ltd, London, ISBN13
Huang, Y.H. 2004. Pavement Analysis and Design. 2nd 9781138029248.
Edition, New Jersey, Prentice Hall, USA. Praticò, F.G., Fedele, R., Vizzari, D. 2017. A Com-
International Atomic Energy Agency Vienna (IAEAV). prehensive study on the precision and accuracy of
2002. Guidebook on non-destructive testing of con- in-lab and on-site measurement of pavement acous-
crete structures. Training Course Series (17). tic absorption. Transportation Research Board 96th
Koch, C., Georgieva, K., Kasireddy, V., Akinci, B., and Annual Meeting 2017.
Fieguth, P. 2015. A review on computer vision based Praticò, F.G., Vaiana, R. 2015. A study on the relation-
defect detection and condition assessment of concrete ship between mean texture depth and mean profile
and asphalt civil infrastructure. Advanced Engineering depth of asphalt pavements. Construction and Build-
Informatics 29: 196–210. ing Materials 101: 72–79.

874
Praticò, F.G., Vaiana, R., Iuele, T. 2015. Macrotexture Schubert, E. & Wolfe, J. 2006. Timbral brightness and
model-ing and experimental validation for pavement spectral centroid. Acta Acustica United with Acustica
surface treatments. Construction and Building Materi- 92: 820–825.
als 95: 658–666. Solla, M., Lagüela, S., González-Jorge, H., and Arias, P.
Rehman, S.K.U., Ibrahim, Z., Memonb, S.A., Jameel, 2014. Approach to identify cracking in asphalt pave-
M. 2016. Nondestructive test methods for concrete ment using GPR and infrared thermographic meth-
bridges: A review. Construction and Building Materi- ods: Preliminary findings. NDT&E International 62:
als 107: 58–86. 55–65.
Rodés, J.P., Pérez-Gracia, V., and Martínez-Reguero, A. Wang, T., Gopalakrishnan, K., Somani, A., Smadi, O.,
2015. Evaluation of the GPR frequency spectra in and Ceylan, H. 2016. Machine-Vision-Based Road-
asphalt pavement assessment. Construction and Build- way Health Monitoring and Assessment: Develop-
ing Materials 96: 181–188. ment of a Shape-Based Pavement-Crack-Detection
Sarkar, P. 2013. Smart Pavement Monitoring System. Approach. Final Report 2016. Institute for Transpor-
FHWA-HRT-12-072. tation, Iowa State University.
Schnebele, E., Tanyu, B. F., Cervone, G., and Waters, N. Yu, Y., Zhao, X., Shi, Y., and Ou, J. 2013. Design of a
2015. Review of remote sensing methodologies for real-time overload monitoring system for bridges and
pavement management and assessment. Eur. Transp. roads based on structural response. Measurement 46:
Res. Rev. 7: 7 DOI 10.1007/s12544-015-0156-6. 345–352.

875

View publication stats

You might also like