You are on page 1of 8

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

net/publication/317054058

Optimal System Design for Weigh-In-Motion Measurements Using In-


Pavement Strain Sensors

Article  in  IEEE Sensors Journal · May 2017


DOI: 10.1109/JSEN.2017.2702597

CITATIONS READS
8 752

5 authors, including:

Raj Bridgelall Mu’ath A Al-Tarawneh


North Dakota State University Mu’tah University
66 PUBLICATIONS   365 CITATIONS    12 PUBLICATIONS   48 CITATIONS   

SEE PROFILE SEE PROFILE

Pan Lu
North Dakota State University
85 PUBLICATIONS   254 CITATIONS   

SEE PROFILE

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

Modeling pavement performance and preservation View project

Benefit Cost Analysis of Railroad Track Monitoring using Sensors On-board Revenue Service Trains View project

All content following this page was uploaded by Raj Bridgelall on 23 October 2017.

The user has requested enhancement of the downloaded file.


Pre-Print Manuscript of Article:
Zhiming, Z., Huang, Y., Bridgelall, R., Al-Tarawneh, M., Lu, P., “Optimal System Design for Weigh-in-Motion
Measurements Using In-Pavement Strain Sensors,” IEEE Sensors Journal, DOI:10.1109/JSEN.2017.2702597, PP(99), May 9,
2017.

Optimal System Design for Weigh-In-Motion


Measurements Using In-Pavement Strain Sensors
Zhiming Zhang, Ying Huang*, Raj Bridgelall, Mu'ath Al-Tarawneh, and Pan Lu

 require the vehicles to stop. Such static weight measurements


Abstract — Weigh-in-motion (WIM) systems are widely are time consuming and expensive. The approach poses traffic
deployed to collect real-time traffic data. They enable multiple safety hazards, and induces congestion because trucks must
applications such as truck weight enforcement, pavement design, queue along heavily traveled highways. To overcome these
freight management, traffic management, and intelligent
limitations, engineers developed various weigh-in-motion
transportation systems. Engineers currently depend on the
experiences of practitioners in the field to select parameters for (WIM) systems using strain sensors embedded inside the
the sensor and system design. This paper introduces a pavement. Such sensors incorporate piezoelectric transducers,
comprehensive design framework that optimizes the system bending plates, load cells, and fiber optic devices [4-7]. These
design of in-pavement WIM sensors to meet high-level application devices are all capable of estimating the gross weight of
requirements. The framework establishes the relationships moving vehicles by measuring their dynamic tire forces [8]. In
between the sampling rate, measurement resolution,
combination with static weight stations, the highway patrol
signal-to-noise ratio, sensor sensitivity, and the measurement
dynamic range. The simulations and case study presented uses WIM technologies to enforce truck weight limits, reduce
demonstrates an example of a specific design to accommodate traffic jams, increase the mobility efficiency [9, 10], and to
high-speed truck weight measurements. collect traffic flow information [3].
WIM operation at mainline speeds poses challenges to
Index Terms — weigh-in-motion (WIM), intelligent system designers. For speeds up to 80 mph (35.8 m/s), a short
transportation system (ITS), optimal system design, in-pavement vehicle axle-spacing produces high bandwidth WIM signals. A
strain sensors
signal bandwidth increase to accommodate high-speed vehicle
crossings, on the other hand, increases the signal noise that
I. INTRODUCTION ultimately limits the measurement accuracy and precision.
Therefore, there is a need to determine a design optimization
M EASURING the weights of passing vehicles is an important
activity for highway administrators and engineers [1].
Agencies use such measurements to help make decisions about
procedure that balances the data acquisition sampling rate with
an acceptable accuracy of weight measurement [11]. The
state-of-the-art WIM system design focuses a development of
pavement design and maintenance. According to the design algorithms to optimize multi-sensor sensor layout
guide provided by the American Association of State Highway configurations to increase the measurement accuracy [12-15].
and Transportation Officials (AASHTO), pavement design For example, Cebon derived a statistical approach to estimate
must consider traffic load accumulation by calculating the the accuracy of a WIM system with multiple, evenly spaced
equivalent single axle load (ESAL) [2]. Recently, the National force transducers, subject to heavy vehicle generated random
Cooperative Highway Research Program (NCHRP) of the and dynamic tire forces [13]. As part of a European research
Transportation Research Board (TRB) proposed the use of load project called 'WAVE', Dolcemascolo and Jacob [14]
spectra to include traffic loads into the pavement design process calibrated the measurement accuracy of a multi-sensory WIM
[3]. Either way, the design community agrees that traffic load is system from 100 runs of two pre-weighed test trucks.
a key input to pavement designs. The availability of accurate Stergioulas et al. [15] improved the accuracy of a multi-sensory
traffic load information, however, relies significantly on on-site WIM system by applying a probabilistic estimation method to
measurement solutions. two generic vehicle models including a ‘quarter-car’ and a
Traditionally, agencies collect vehicle weight information by ‘walking beam’ model. Other researchers studied sampling
directing select vehicles off the roadway to weigh stations that optimization for WIM measurements from the passage of a
single tire load [16-19]. It is evident from the literature review
This work was partially supported by the National Science Foundation ND that the field lacks a systematic approach to optimize WIM
EPSCoR under agreement IIA-1355466, and NDSU FORWARD program
FAR0026087. Zhiming Zhang, Mu'ath Al-Tarawneh, and Ying Huang* are in
system designs capable of supporting multiple axle crossings.
the department of Civil and Environmental Engineering, North Dakota State This paper introduces a comprehensive and systematic
University, USA. approach to optimize the design of a high-speed WIM system
Raj Bridgelall and Pan Lu are in Upper Great Plains Transportation Institute
that considers multiple-axle tire loads. The system design
(UGPTI), North Dakota State University, USA.
* Corresponding author: ying.huang@ndsu.edu approach determines the minimum required data sampling rate,
Page 1/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

the maximum achievable system resolution, the measurement the dependent parameters of the sensor and system. The former
dynamic range, the maximum tolerable input noise level, and includes the minimum and maximum axle spacing (Smin, Smax)
other important system design parameters. The organization of of the target vehicles, the maximum supported vehicle speed
this paper is as follows: Section 2 demonstrates the framework (vmax), and the minimum measurable axle weight (Wmin).
of the design methodology and develops models of the signal Together, these three independent parameters determine the
generated from the embedded strain transducers; Section 3 minimum strain signal (I0) that the system must measure. The
presents numerical simulations of an example system other independent parameters are the strain transducer
implementation to demonstrate the framework and Section 4 sensitivity, the maximum axle weight (Wmax) measurable, the
describes a case study for concrete pavements. Section 5 maximum allowable signal distortion (Dist), and the maximum
provides the conclusion and future work. design voltage input to the A/D converter (Vmax). In this
framework, the full-scale A/D output is a digital representation
II. METHODOLOGY of Vmax.
The dependent parameters are the maximum noise level at
A. WIM System Design Framework
the input of the strain transducer (n0), the measurement
A high-speed WIM system has three major components: the sensitivity of the WIM system (Sens), the cut-off frequency of
embedded strain transducer that produces a strain signal that is the low-pass filter (fc), the sampling rate (fs) of the A/D
proportional to the applied weight, a low-pass filter that converter, the number of A/D converter bits (N-bits), and the
includes signal amplification and an analog-to-digital (A/D) dynamic range of the weight measurement. The sampling rate
converter. The low-pass filter and the signal amplifier operate of the A/D converter and the type of A/D converter generally
on the analog strain voltage to suppress random noise from the determines the minimum signal-to-noise ratio (SNRmin) that the
output of the in-pavement transducer. Finally, the A/D A/D converter supports. The minimum SNR refers to a
subsystem digitizes the analog signal for subsequent recording, maximum tolerable noise at the input of the A/D converter (n1),
analysis, and the display of the weight measure via a separate which in turn propagates back to the maximum noise tolerable
digital sub-system. This study developed a design framework at the input of the transducer.
to optimize the in-pavement WIM sensor system. The The theoretical derivations begin with identifying the
framework guides the design optimization by linking key minimum strain signal (I0) that an embedded strain transducer
system level requirements to independent variables of the can generate.
sensor system.
The optimization requires identifying the independent and

Fig. 1. Framework to optimize the WIM system design.

as derived in [16] and shown in Equation (1).


1) Minimum identifiable axle weight
To derive the theoretical transducer output signal, the authors I (t )  v 
 PL (t   )S L ( )d
(1)
simplify the tire load of a vehicle to a constant. Hence, the
passage of a single tire across the sensor produces a strain Vehicle traversals at a velocity of v yield the temporal
signal I(t), which is a convolution of the load function PL(t) and functions from the spatial load and sensor functions. Given the
the sensitivity function SL(t) multiplied by the vehicle’s speed v tire load simplification, the strain signal, I(t), is a trapezoidal

Page 2/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

shape as shown in Fig. 2.


The output noise at an A/D converter, n1, has a power of Pn1,
dB where

Pn1, dB = 20log10RMS(n1). (8)

Thus, the minimum required output power to distinguish a


strain signal P0 after the filter is

P0= Pn1, dB + SNRmin. (9)


Fig. 2. Strain signal from a single tire load

Substituting Equations (9) into (4) yields


Hence with multiple axles crossings, the strain signal 2
T t
becomes a periodic pulse train, I (t ) . Fig. 3 shows the 1 0 0
. (10)
P0  
2(T0  t 0 ) T0  t 0
I (t ) dt
theoretical pulse train from the continuous crossings of equally
spaced tire loads. T1 is the time interval between adjacent axles
having a maximum and minimum axle spacing of Smax and Smin, Solving Equation (10) then determines the minimum input
respectively. T1 is a function of the axle spacing (S) and the strain signal, I0.
speed of the passing vehicle (v) such that
2) Optimal system design
T1=S/v. (2) A filter bandwidth well beyond the fundamental frequency
that multiple axle crossings induce will reduce the signal’s
distortion. On the other hand, increasing the filter bandwidth
The theoretical pulse train, I (t ) is: decreases the SNR because the filter integrates noise at higher
frequencies. Therefore, a tradeoff that maximizes the SNR
 while minimizing the signal distortion level (Dist) will yield an
I (t )   I (t  kT1 ), k = 0,  1,  2, … (3) optimal solution.
k 
By Nyquist’s Theorem, the A/D sampling rate (fs) must be at
least twice the optimal cutoff frequency of the filter (fc) such
where k is the number of axle crossings.
that

fs = 2 × f {Max[SNR(fc)], Min[Dist(fc)]} (11)

The sensitivity function of the selected strain transducer and


the minimum strain signal (I0) determines the minimum weight
Fig. 3. Strain signal from equally spaced multiple tire loads Wmin such that

Wmin  I 0  Sens. (12)


Subsequently, the power of the strain signal over the period
[-T0+t0, T0+t0] is
The maximum weight, Wmax, that the system can measure is
T0  t 0 2
1 (4) Wmax = Wmin 2N-bit .
2(T0  t 0 ) T0 t 0
P I (t ) dt (13)

These formulations yield the dynamic range [Wmin, Wmax] of


Given an A/D converter of N-bits, its minimum SNR the WIM system. To simplify the analysis and illustrate the
requirement is [16] framework, the authors assume a constant sensitivity factor so
that the weight is a linear transform of the strain level.
SNRmin = N-bit · 6.02 +1.76. (5) Consequently, the measurement resolution, Rweight, for the WIM
system is
The maximum allowable noise at the input of an anti-aliasing
filter, n0(t), produces the noise power in dB, Pn0, dB, of Rweight  Wmax 2 Nbits . (14)

Pn0, dB = 20log10RMS(n0). (6) Subsequently, the equivalent error of weight measurements


is
Given an input noise of n0, and an anti-aliasing filter with an
impulse response of h(t), the output noise, n1, must be Eweight = Wmax/2N-bit+1 . (15)

n1(t) = n0h(t) (7)

Page 3/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

Noise Level = -13 dB


40 Noise Level = -3 dB
III. APPLYING THE FRAMEWORK Noise Level = 17 dB

SNR, dB
This section conducts a numerical sensitivity analysis by 20
applying the proposed framework to a simulated load from a
truck axle. The load P is assumed to be 2,630.8 kg (5,800 lb). 0
Table 1 lists all the parameters used for the numerical analysis.
160 Hz
The parameters, a, b, and h are the length, width, and thickness -20
of the pavement panel. L0 and B0 are the length and width of the 0 500 1000 1500
tire contact area with the pavement. E and u are the modulus of fc, Hz
Fig. 5. SNR change with filter cutoff frequencies for S = 1.5-7.9 m (5-26 ft)
elasticity and Poisson’s ratio for the concrete. The parameter k
is the subgrade stiffness. The parameters x0, y0, and h0 are the
The authors define the distortion level of the strain signal as
location coordinates of the installed sensor in the panel. P is the
the pulse width at the mean level between the maxima and the
weight of the tire, v is the truck speed, which is 17.4 m/s (39
minima of the pulse train. Fig. 6 shows the mean pulse width as
mph), and S is the distance between the adjacent axles. In
practice, most vehicles have an axle space longer than 1.5m (5 a function of the filter cutoff frequency. The exponential decay
ft.) and shorter than 7.9 m (26 ft.). Thus, the simulations used trend establishes that there is a point of diminishing returns. For
random vehicle axle spaces, S, between 1.5-7.9 m (5-26 ft). the scenario of a 17 dB input noise level, increasing of the filter
Fig. 4 shows the simulated strain signals from the passing cutoff frequency beyond 160 Hz will diminish the SNR below
axles. the desired 3 dB level without yielding significant
improvements in distortion. Therefore, a design that sets the
TABLE I cutoff frequency at 160 Hz will yield the optimum design for
PARAMETERS FOR NUMERICAL SIMULATION this scenario. Subsequently, the corresponding sample rate
Parameter Value Unit Parameter Value Unit must be at least twice the cutoff frequency.
a 1.8 m k 29.8 × 10 6
kg/m3
Pulse Width, 1/10000 sec

400
b 1.8 m x0 1.3 m Signal without Noise
350 Noise Level = -3 dB
h 0.076 m y0 1.1 m Noise Level = -13 dB
300
L0 0.35 m h0 0.07 m
250
B0 0.23 m P 2,630.8 kg
200 160 Hz
E 15000 MPa v 17.4 m/s
150
 0.16 - S 1.5-7.9 m 0 500 1000 1500
fc, Hz
Fig. 6. Pulse width at different cutoff frequencies for S = 1.5-7.9 m (5-26 ft)
0
Using an 8-bit A/D converter and a filter cut-off frequency of
-20 160 Hz and Equation (10), Fig. 7 plots the minimum input
Strain, 

signal power as a function of input noise power. It shows that


-40 for each dB decrease of the input noise power, the minimum
detectable strain signal power increases linearly by 12.2%.
-60
Minimum Input Signal Power, dB

-80 140
0 10000 20000 Axle Distance = 5 ft
t, 1/10000 sec Axle Distance = 26 ft
Fig. 4. Strain signal from random axle spacing between 1.5 and 7.9 m (5-26 ft) 120 Axle Distance = 5-26 ft

To analyze the influence of sampling frequencies, the 100


authors applied a 7th order Bessel filter as the anti-aliasing filter
at different cutoff frequencies. Fig. 5 illustrates the expected
80
decrease in SNR as a function of the filter cutoff frequencies.
The simulations used three arbitrary input noise levels of -13
60
dB, -3 dB, and 17 dB to illustrate the linear time-invariance 0 20 40 60 80
property of the system. That is, for any given cutoff frequency, Noise Power, dB
Fig. 7. P0 change with the increase of n0 level
a higher noise results in a lower SNR. The chart also provides
guidance for setting the cutoff frequency based on an absolute
From Equation (12), Fig. 8 plots that the minimum
input noise level and an SNR requirement. For example, a 3dB
measureable weight as a function of the cutoff frequency for
SNR will require a cutoff frequency lower than 160Hz when
four different input noise levels, -10 dB, 0 dB, 10 dB, and 20
the input noise level is less than 17 dB.

Page 4/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

dB. It is evident that the minimum measurable tire weight 40, which contains a concrete pavement with panel dimensions
generally increases with the cutoff frequency at different input (a × b × h) of 1.8 m × 1.8 m × 0.076 m (6 ft × 6 ft × 3 in). The
noise levels. It is also evident that the filter has a greater impact concrete used for the pavements had a Young’s modulus, E, of
at the higher input noise levels. For a cut-off frequency of 160 15,000 MPa (2,175,566 psi) and a poison ratio, μ, of 0.16. The
Hz, the minimum measurable tire weight and resolution of the subgrade under the pavement had a stiffness, k, of 29,838,932.2
system is 45.36 kg (100 lbs), 90.72 kg (200 lbs), 249.48 kg (550 kg/m3 (1,078 pci). The strain gauges, shown in Fig. 11(a), are
lbs), and 816.47 kg (1800 lbs) for noise levels of -10 dB, 0 dB, located 1.3m (51.2 in.), x0, away from the middle of the
10 dB, and 20 dB, respectively. From Equation (13), the pavement and 1.1m (43.3 in.), y0, away from the longitudinal
maximum identifiable weights are 11,611.97 kg (25,600 lbs), pavement panel edge. Fig. 11(b) shows the installed position. In
23,223.93 kg (51,200 lbs), 63,865.81 kg (140,800 lbs), Fig. 11(c) is a photograph of the onsite data collection
209,015.36 kg (460,800 lbs) for input noise levels of -10 dB, 0 equipment used to sample the strain signal output at a rate of
dB, 10 dB, and 20 dB, respectively. 1.2 kHz. This rate was sufficiently high to examine the effects
Fig. 9 categorizes the weights of standard classifications for of sampling rate reduction by post filtering.
vehicles in the United States [20]. Hence, the weight measured
from individual axle crossings will be at least half the weights
shown. It is evident that the example WIM system with an input
noise level below 0 dB can weigh almost all the classes of
vehicles in United States.

(a) MnROAD facility

Fig. 8. Minimum identifiable weight at different cutoff frequencies for S =


1.5-7.9 m (5-26 ft)

(b) Concrete pavement of the simulation line


Fig. 10. Road sections at MnROAD for case study.

(a) Strain gauge


Fig. 9. Vehicle weights by class

IV. CASE STUDY


This section introduces a case study that applies the
framework, using the simulation parameters from Table 1. The
authors collected data from the sensor site at cell 40 of the
MnROAD Cold Weather Road Research Facility in Minnesota
[14, 15]. As shown in Fig. 10(a), MnROAD, consists of two
unique roadways: a two-lane low-volume loop that is loaded
with a 5-axle 36,287 kg (80,000 lb) semi-truck, and a section of
interstate I-94 ‘‘mainline’’ that contains two westbound lanes (b) Strain gauge configuration in the concrete pavement
with live traffic [16]. Fig. 10(b) shows the road section of Cell

Page 5/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

fractional noise equivalent bandwidth Bf = 160 Hz/600 Hz =


0.267. Therefore, the maximum tolerable strain signal with
noise before the anti-aliasing filter is -49.9 – (-11.5) = -38.4 dB.
The full scale of the strain signal input at the A/D was set to
5,000 to benefit from their linearity within that range. The
measured sensitivity of the installed WIM sensor was 0.03
/kg (0.014 /lb) [13]. Thus, Equation (11), the system was
(c) Data collection equipment capable of a measuring a maximum weight of 166,665 kg
(357,143 lb).
Fig. 11. The WIM in-pavement strain sensors, installation layout in pavement From Equation (10), the minimum measurable weight or
panel, and data collection equipment.
weight resolution of the WIM system is 651 kg (1,395 lb).
Fig. 12(a) is a diagram of the 5-axle truck used to provide the Therefore, the strain signal level from the minimum weight is
multi-axle loads. The first axle of the truck weighs 2,630.8 kg 19.53 με. Figure 5 shows that the SNR measured at the output
(5,800 lb). The smallest axle distance of the semi-truck is 1.5 m of A/D is 14 dB, then the actual noise is 38.4 – 14 = 24.4 dB,
(4.9 ft), which is within the range of the random axle spacing which is equivalent to a signal level of 16.53  Comparing
simulated. Fig. 12(b) shows a photograph of the testing scene. this value with Fig. 8, the field testing validated the simulation
Fig. 13 shows the strain signal acquired from the strain gauges with very close values. From Equation (9), the power of the
in the unit of µɛ as the truck passes over the in-pavement strain strain signal is -37 dB. From Equation (13), this system would
sensor. The five pulses coincide with the five axle traversals of have a measurement error of ± 325.5 kg (697.5 lb).
the truck. The sampling rate for the installed WIM system was 1,200
Hz. Hence, the bandwidth should be less than 600 Hz. Figs.14
shows the sampled data filtered with the 160 Hz cut-off
frequency determined from the simulations. It is evident that
the cutoff frequency provides a good tradeoff between
distortion and SNR as expected.

(a) Truck configuration

Fig. 14 Filtered signals from the case study with cut-off frequencies of 160 Hz.
(b) Test scene
Fig. 12. WIM measurement tests at MnROAD facility. V. CONCLUSIONS AND FUTURE WORK
This study developed a comprehensive framework to guide
50
the design optimization of WIM systems using in-pavement
strain transducers. The authors derived the relationships
0 between the sampling rate, resolution, SNR, sensor sensitivity,
and the dynamic range of the WIM system. The framework
Strain, 

identified the dependent and independent variables as a


-50
function of the high-level application requirements. Design
parameters include the sampling rate of the data capture
-100 equipment, the dynamic range of the weight measurement
scale, and the maximum tolerable noise levels. The theoretical
analysis, the numerical simulations, and a case study provided
-150
0 500 1000 1500 2000 evidence that the framework will be effective in guiding the
t, 1/1200 sec design of future WIM systems. In future work, the authors will
Fig. 13. Measured signals from the strain gauge at 1,200 Hz.
collect a larger amount of WIM sensor data to evaluate the
This section uses the design framework of the Section 2 to design using real highway traffic.
evaluate the system design. From Equation (5), the SNRmin
is -49.9 dB for an 8-bit A/D converter. The noise reduction of ACKNOWLEDGMENT
the anti-aliasing filter is 20*log(Bf) = -11.5 dB with the The National Science Foundation ND EPSCoR grant

Page 6/7
Optimal System Design for Weigh-In-Motion Measurements Using In-Pavement Strain Sensors

under agreement IIA-1355466, and the NDSU FORWARD Zhiming Zhang received PhD in civil
program FAR0026087 supported this study. The findings engineering at North Dakota State University
and opinions expressed in this article are those of the authors (NDSU), Fargo, USA. He completed his B.S.
only and do not necessarily reflect the views of the sponsors. degree in civil engineering from Liaoning
The authors also would like to thank Robert Strommen, Technical University, China, in 2010 and the
Leonard Palek, Benjamin Worel, and other technicians at M.S. degree in civil engineering from
MnROAD facility, MnDOT, MN, for their extensive Southwest Jiaotong University, China, in 2013.
technical help during field tests. His research interests include structural health monitoring and
intelligent transportation systems.
REFERENCES
Mu'ath Al-Tarawneh is currently a Ph.D. student in civil
[1] R. B. Malla, A. Sen, and N. W. Garrick, “A special fiber optic
engineering at NDSU, Fargo, USA. He completed his M.S.
sensor for measuring wheel loads of vehicles on highways,”
Sensors, vol. 8, no. 4, pp. 2551-2568, 2008. degree in civil engineering in civil engineering at NDSU,
[2] T. Officials, AASHTO Guide for Design of Pavement Structures, Fargo, in 2016. His research interests include structural health
1993: AASHTO, 1993. monitoring and intelligent transportation systems.
[3] J. A. Prozzi, and F. Hong, “Effect of weigh-in-motion system
measurement errors on load-pavement impact estimation,” Journal
of Transportation Engineering, vol. 133, no. 1, pp. 1-10, 2007. Dr. Ying Huang currently is an assistant
[4] L. Zhang, C. Haas, S.L. Tighe, Evaluating Weigh-In-Motion Sensing professor in the Department of Civil and
Technology for Traffic Data Collection, in: Transp. Assoc. Canada, Environmental Engineering at NDSU. She
pp. 1–17, 2007.
[5] L. Cheng, H. Zhang, and Q. Li, “Design of a capacitive flexible
received her Ph. D. degree in Missouri
weighing sensor for vehicle WIM system,” Sensors, vol. 7, no. 8, University of Science and Technology in
pp. 1530-1544, 2007. USA in 2012 before she joined NDSU. Her
[6] Hall D L and Loshbough R C 1991 Vehicle Weighing in Motion research interests include intelligent
Apparatus and Method (Google Patents)
transportation system, pipeline corrosion
[7] Z.-Q. Shi, and D. Chung, “Carbon fiber-reinforced concrete for
traffic monitoring and weighing in motion,” Cement and Concrete mitigation and assessment, smart structures and structural
Research, vol. 29, no. 3, pp. 435-439, 1999. health monitoring, multi-hazard mitigation in harsh
[8] Austroads, Weigh-in-motion Technology, Austroads, Sydney, environments, and big data for transportation. As a researcher,
Australia, 2000.
Dr. Huang has more than 100 high quality peer-reviewed
[9] K. Wang, Z. Wei, H. Zhang, X. Huang, B. Chen, and H.-L. Cui,
“Fiber-Bragg-grating-based weigh-in-motion system using publications. She serves in five international journals as
fiber-reinforced composites as the load-supporting material,” editorial board member, five distinguishing professional
Optical Engineering, vol. 45, no. 6, pp. 064401-064401-7, 2006. society as committee member, and 30 different international
[10] MDT, Montana weigh-in-motion (WIM) and automatic traffic
recorder (ATR) strategy, Montana Department of
journals and conferences as peer reviewer.
Transportation Helena, MT, 2013.
[11] R. Bushman, and A. J. Pratt, "Weigh in motion Dr. Raj Bridgelall (M’91-SM’02) received
technology-Economics and performance." the B.S. and M.S. degrees in electrical
[12] FHWA-NJ, Implementation of Weigh-in-motion (WIM) Systems
engineering from Stony Brook University in
(Piscataway: FHWA-NJ), 2009.
[13] D. Cebon, “Design of multiple-sensor weigh-in-motion systems,” 1990 and 1991, respectively. Bridgelall
Proceedings of the Institution of Mechanical Engineers, Part D: received the Ph.D. degree in transportation
Journal of Automobile Engineering, vol. 204, no. 2, pp. 133-144, and logistics at the NDSU, Fargo in 2015. He
1990.
served as the RFID Chief Technologist at
[14] V. Dolcemascolo, and B. Jacob, "Multiple sensor
Weigh-In-Motion: optimal design and experimental study." Motorola until 2004, as Vice President of Research &
[15] L. Stergioulas, D. Cebon, and M. Macleod, “Static weight Development at Alien Technology until 2006, and as Chief
estimation and system design for multiple-sensor Technical Officer at Axcess International until 2010. He is
weigh-in-motion,” Proceedings of the Institution of Mechanical
presently Program Director of the SMARTSeSM Intelligent
Engineers, Part C: Journal of Mechanical Engineering Science,
vol. 214, no. 8, pp. 1019-1035, 2000. Transportation Systems Center at North Dakota State
[16] Z. Zhang, Y. Huang, R. Bridgelall, L. Palek, and R. Strommen, University. He is the author of more than 57 peer-reviewed
“Sampling optimization for high-speed weigh-in-motion articles and holds more than 140 patents issued or pending. His
measurements using in-pavement strain-based sensors,”
Measurement Science and Technology, vol. 26, no. 6, pp. 065003,
research interests include intelligent transportation systems and
2015. smart city developments. Mr. Bridgelall was a recipient of
[17] Z. Zhang, Y. Huang, L. Palek, and R. Strommen, “Glass fiber– Sensors Magazine Best of Sensors Rising Star Award in 2014,
reinforced polymer–packaged fiber Bragg grating sensors for the IEEE Charles Hirsch Award in 2004, and the IEEE
ultra-thin unbonded concrete overlay monitoring,” Structural
Outstanding Young Engineer Award in 1999.
Health Monitoring, vol. 14, no. 1, pp. 110-123, 2015.
[18] Z. Zhang, F. Deng, Y. Huang, and R. Bridgelall, “Road roughness
evaluation using in-pavement strain sensors,” Smart Materials and Dr. Pan Lu received the Ph.D. degree in transportation and
Structures, vol. 24, no. 11, pp. 115029, 2015. logistics at NDSU, Fargo in 2009. In 2010, Pan became a
[19] Z. Zhang, F. Deng, Y. Huang, and R. Bridgelall, "Field validation of
road roughness evaluation using in-pavement strain sensors." pp.
Research Analyst at Upper Plains Transportation Institute
98031E-98031E-8. (UGPTI), at NDSU, focusing in the area of asset management,
[20] Energy, U.D.o., 2012. Vehicle Weight Classes & Categories. freight transportation, sustainable transportation, rail
transportation and GIS-T applications. From 2012, Pan has
served as Associate Research Fellow at UGPTI.
Page 7/7

View publication stats

You might also like