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

net/publication/321892730

IoT based rail track health monitoring and information system

Conference Paper · August 2017

DOI: 10.1109/ICMDCS.2017.8211548

CITATIONS READS
15 3,365

4 authors, including:

Chellaswamy Chellaiah Vanathi Am

Rajalakshmi Institute of Technology Rajalakshmi Institute of Technology
59 PUBLICATIONS   327 CITATIONS    6 PUBLICATIONS   39 CITATIONS   

SEE PROFILE SEE PROFILE

Saravanan Lakshmanan
Rajalakshmi Institute of Technology
11 PUBLICATIONS   26 CITATIONS   

SEE PROFILE

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

Respected Sir, I am working on Solar cell parameter extraction and crack identification View project

Solar cell crack indentification View project

All content following this page was uploaded by Chellaswamy Chellaiah on 13 February 2018.

The user has requested enhancement of the downloaded file.

 IoT Based Rail Track Health Monitoring and
Information System
Chellaswamy C Balaji L Vanathi A Saravanan L
Department of ECE, Rajalakshmi Department of ECE, Velammal Department of ECE, Rajalakshmi Department of ECE, Rajalakshmi
Institute of technology Institute of technology Institute of technology Institute of technology
Chennai, India Chennai, India Chennai, India Chennai, India
e-mail: chella_info@yahoo.co.in e-mail: mailbala81@yahoo.co.in vanathi.a@ritchennai.edu.in saravanan@ritchennai.edu.in

Abstract—This paper describes an Internet of things (IoT) when compared to force balance servo accelerometers. When
based track monitoring system for increasing the comfort of rail force-balance servo accelerometers are subjected to shock
track transportation. Usually, rail tracks are inspected only at inputs they are easily damaged and thus has reduced lifetime
midnight and also the inspection is carried out only once in a [3]. Jun Seok has proposed a mixed filtering approach to
month or lesser. Unfortunately, any emergency issues not detect the track irregularities. Acceleration data in lateral and
detected within the stipulated time leads to train derailment and
other safety concerns. Hence in-service vehicle has been
vertical directions are measured from the accelerometers
introduced to increase the safety of train transportation. In this installed on the axle box and the bogie of a fast moving train.
paper, track irregularities are detected using bogie and car body From these measured data displacement is determined using
acceleration measurements and these measurements is further Kalman filter, waveband classification is done using band pass
optimized using particle swarm optimization algorithm (PSO). filters and amplitude and phase compensation is performed by
The proposed system is compared with the conventional track compensation filters. This method is efficient when compared
geometry measurement system. The monitoring and control to higher order adaptive filters [4]. Axle box acceleration
information can be sent through IoT. Simulation results are (ABA) measurements can also be used in monitoring the
performed for two different speed scenarios and it shows that the insulated rail joints. ABA measurements are analyzed in time
proposed system produces better results than the traditional
track geometry measurement system and the effectiveness of the
and frequency domain to determine the condition of insulated
proposed method. rail joints [5]. The Siemens Tracksure sensor cards are
introduced to detect railway under-track voids. It can be
Keywords—Internet of things, rail track monitoring, geometry installed in the existing GSM-R cab radio. Tracksure can
measurement, particle swarm optimization detect voids by analyzing the acceleration measured in the
vehicle cab.GPS location can direct the maintenance staff to
I. INTRODUCTION the exact location. Installing Tracksure in multiple trains
reduces false alarms and increases accuracy in finding the
Rail corrugation is a frequently occurring rail wear pattern location of voids [6]. Wireless Sensor Networks (WSN) can
which results in reduced rail and wheel lifetime. Continuous be used in railway industry for monitoring rail tracks, track
measuring trolley can be used to continuously measure the rail beds, bridges, tunnels etc. A sensor node comprises one or
surface roughness. Tanaka has proposed axle-box acceleration more sensors, microcontrollers, memory, and transceiver
measurement to detect rail corrugation. The accelerometer is transmits data to the base station. The base station controls the
placed on the axle box of the track inspection car and the data sensor node and transmits data to a remote server [7].
is measured. Data can be measured either on the leading axle Tzuu-Hseng has proposed an IoT-based automated e-
or trailing axle based on the measuring direction of the track fulfillment packaging system. Three-dimensional adaptive
inspection car. However, the accuracy of the detection by the particle swarm optimization (PSO) based packing algorithm is
trailing axle is lower when compared to leading axle. The used. When the customer orders the item, PSO algorithm is
author also examined that high correlation exists between the executed. The suitable box is chosen and this algorithm checks
rail surface roughness data and the axle box acceleration data whether all items can be placed in the box. Adaptive PSO is
[1]. used to decide the position of each item in the box. If all the
Micro electro mechanical system (MEMS) technology is items cannot be packed in a single box, it can be divided into
used to detect the track irregularities. Sensors are placed on two groups and packed as a separate box. These results are
the bogie and the axle-box. Data from these sensors are then sent to the delivery robot and robot manipulator for
compared with the predefined data and if any irregularity is packing [8]. Internet of Things (IoT) can be used to monitor
present, the location of irregularity is stored automatically the reliability of the train service, fault identification;
using GPS. The exact location of irregularity can be Preventive maintenance and asset management [9].Safety of
determined with low-cost accelerometers with continuous trains depends on the running surface of the wheels and a
wavelet transform are introduced by Chellaswamy et al. [2]. collision avoidance system is described in [10]. Any defects in
MEMS accelerometers are easily available and are at low cost the wheels can be detected using an optical based system.

978-1-5386-1716-8/17/$31.00 ©2017 IEEE

Four high-resolution cameras are used to capture the images irregularities due to rail top defect and earth problems are
of wheel tread and flange. These images are processed to shown in Fig 2. In this paper, vertical and lateral irregularities
produce a vertical image of the circumference of the wheel of bogie and axle box due to squat defect are analyzed using
[11]. An IoT-based online monitoring system has been adaptive filtering. Track irregularities are detected using
proposed for continuous steel casting. Status information accelerometer which is connected with the ARM-Cortex
about the equipments collected from different sensors and processor. Location of the irregularity is determined using
transmitted to a remote server through wireless global position system (GPS) which is interfaced with the
communications [12]. IoT-based residence energy control controller. Data measured by the sensors and the GPS system
system is proposed to reduce the energy consumption of home is stored in the cloud and can be controlled through IoT. Since
appliances without using any sensors. Smart sockets are used GPS units are low quality and do not have an unobstructed
to collect the energy consumption of appliances [13, 14]. view of the sky they provide low accuracy position estimates.
But GPS units are inexpensive (<$30) and can be easily
Table I. Various accident types on different track for class-I deployed.
freight railroad between 2000-2010 [9].

Accident Type

Track Type Derailment Collision Others total

Main 37,456 989 584 39029
Siding 2,477 85 48 2610
Yard 13,364 527 513 14404
Industry 1,594 23 25 1642

A small crack in a rail track will create a significant risk

and it is not being detected suitably. The damages present in
the track leads to the derailment and causes heavy loss. The
signal which is received from the track measurement is quite Fig. 1. IoT-based track monitoring system
small for fault identification. A small crack in a rail track will
create a significant risk and it is not being detected suitably. An account is created in smartliving.io. Raspberry Pi
The damages present in the track leads to the derailment and device and other required assets (sensing input information)
causes heavy loss. Various accident types on a different track are added. Unique IDs and names are created for the device
for class-I freight railroad is shown in Table I. It motivates us and each asset in order to identify the information coming
to monitor the track and identify the fault and manages through from particular device and asset. Client id and client key are
IoT technology. The proposed IoT-based track monitoring also generated for the device. In this model, we divide the
system has different sensors which are deployed on the train tracks into five segments and these five assets are given
and they are integrated with Raspberry Pi. Any irregularities unique asset id and name. Asset ids and names are declared in
are automatically sent to the railway authorities through the python code. Each module is interconnected with
WAY2SMS. Additionally, sound alert and visual information Raspberry Pi module and sends the data to the cloud. Cloud
are displayed for the driver. The rest of this paper is organized then automatically sends SMS to the railway authorities.
as follows: Section 2 describes an overview of the proposed Hubs present at every cluster and trains can be used to
system with sensor arrangement. In section 3, describes the collect data at any time. Big data processing technologies
PSO algorithm. In Section 4, the simulation and hardware provide us the real-time situation of the tracks. Hence they are
implementation is presented. Finally, conclusions are given in widely used in the rail and other means of transportation. This
section 5. reduces the cost and improves the comfort and safety of the
passengers. Smart living cloud is used to automate the system.
II. PRINCIPLE OF THE PROPOSED SYSTEM
If there are any irregularities in the tracks it sends SMS
A. System Overview automatically to the railway authorities to stop the train.
Rail vehicle has two bogies and each has a two wheelsets. Further, necessary action can be taken thus avoiding any
These wheelsets are connected through axle box and loss of human lives.
suspension system. The suspension system is divided into The proposed system has the following novelties:
primary suspension and secondary suspension. The primary • Vertical and lateral accelerations are measured and
suspension connects the bogie and wheelsets, and the compared with the threshold value.
secondary suspension connects the vehicle body and axle box. • Micro-electro-mechanical sensor (MEMS)
The IoT-based track health monitoring system is shown in Fig accelerometer increases the system accuracy and
1. Track irregularities may cause serious safety issues. The reliability.
factors which affect the irregularities are 1) Heavy traffic load • A new automatic tracking method is introduced for
and earth problems which leads to deformation of the track. 2) calculating the location of vibration if the signal level
Squat defect due to dynamic force of rail-wheel interface. The of GPS is low.
 • If any vibration is measured information is signal dynamic behavior of this model can be investigated by
automatically sent to the central office. a work out the subsequent equation [17] as:
• The optimization algorithm is introduced to increase ‫ݔܯ‬ሷ ൅ ‫ݔܤ‬ሶ ൅ ‫ ݔܭ‬ൌ ‫ܨ‬௘௫௧ (7)
the performance of the system.

B. RLS Filter
The recursive least square (RLS) algorithm is used for
faster convergence instead of statistical approach. The
algorithm introduced time index because it is a recursive in
time. The input data which is received from the accelerometer
is pre-windowed. So that x(n) = 0, for n<0. The input signal
can be represented in vector form at time n is [15,16]:

‫ݔ‬ሺ݊ሻ
‫ۍ‬ ‫ې‬
‫ێ‬ ‫ݔ‬ሺ݊ െ ͳሻ ‫ۑ‬
ܺ௞ ሺ݊ሻ ൌ  ‫ݔ ێ‬ሺ݊ െ ʹሻ ‫ۑ‬ (1)
ǤǤ
‫ێ‬ ‫ۑ‬
‫ݔۏ‬ሺ݊ െ ݇ ൅ ͳሻ‫ے‬

The coefficient can be represent in vector form at time n is

݄ሺͲǡ ݊ሻ Fig. 2. Simplified model of capacitive MEMS domain

‫ۍ‬ ‫ې‬
‫݄ ێ‬ሺͳǡ ݊ሻ ‫ۑ‬
݄௞ ሺ݊ሻ ൌ  ‫݄ ێ‬ሺʹǡ ݊ሻ ‫ۑ‬ (2)
ǤǤ where M, B, and K represents the mass of the movable part,
‫ێ‬ ‫ۑ‬ damping factor, and the spring constant respectively. The
‫݄ۏ‬ሺ݇ െ ͳǡ ݊ሻ‫ے‬
excitation force can be represented as, Fext = F0 sin(Ȧt) and the
structure can be considered as a MEMS resonator. The
where k is length of the filter. The input signal observed from
the accelerometer has a length l=0,1,2,…..,n the desired resonance frequency ሺ߱‫ ݎ‬ሻ of the system can be expressed as
sequence d(l) and its estimate d(l-n). The RLS algorithm is [18]:
used to obtain the filter coefficient vector hk(n) which ߱௥ ؆ ට‫ܭ‬ൗ‫ܯ‬ (8)
minimizes the least squared error and is given by
‫ܧ‬௠ ൌ  σ௡௟ୀ଴ ߱௡ିଵ ȁ݁௠ ሺ݈ǡ ݊ሻȁଶ (3) The accelerometer contains 12 vertical walls and 6 fixed
Where ݁௠ ሺ݈ǡ ݊ሻ ൌ ݀ሺ݊ሻ െ ݀መ ሺ݈ െ ݊ሻ electrodes; the total capacitance C(t), produced by them are: .
ൌ ݀ሺ݊ሻ െ ݄௞௧ ሺ݊ሻܺ௞ ሺ݈ሻ ܿሺ‫ݐ‬ሻ ൌ ͳʹሺߝ௢ ݈௢ ‫ݐ‬௔ Ȁ݀ሻ. If the displacement is along Z-axis then
the capacitance is given by,ܿሺ‫ݐ‬ሻ ൌ ͳʹሺߝ௢ ݈௢ ሺ‫ݐ‬௔ െ ‫ݖ‬ሻȀ݀ሻ where
߱ is the weighting factor with the range between 0<߱<1. It is a and z are the vertical displacement and function of applied
used to give more weight to the most recent sample. So that, acceleration respectively. Now ‫ ݖ‬ൌ ݉ܽ‫ܮ‬ଷ ȀʹͶ‫ܫܧ‬. The change
the filter coefficients can be made adaptable to time varying of capacitance with respective to ‘a’ is
statistical characteristic. The Kalman gain vector is given by డ௖ డ
‫ כ‬ሺ௡ሻ
௉೘ ሺ௡ିଵሻ௑೘ ሺ௡ሻ௑೘
ൌ ൣͳ͸ሾߝ௢ ݈௢ ሺ‫ݐ‬௔ െ ݉ܽ‫ܮ‬ଷ ȀʹͶ‫ܮܧ‬ሻȀ݀ሿ൧ (9)
డ௔ డ௔
‫ܭ‬௠ ሺ݊ሻ ൌ  (4)
ఠା௑ೖ೟ ሺ௡ሻ௉ೖ ሺ௡ିଵሻ௑ೖ‫ כ‬ሺ௡ሻ
The inverse of correlation matrix is given by III. PSO ALGORITHM
ଵ
ܲ௞ ሺ݊ሻ ൌ  ሾܲ௠ ሺ݊ െ ͳሻ െ ‫ܭ‬௠ ሺ݊ሻ െ ௧ ሺ݊ሻܲ ሺ݊
‫ܭ‬௠ ሺ݊ሻܺ௠ ௠ െ ͳሻሿ (5) The particle swarm technique comprises of a swarm of
ఠ
The filter coefficient vector is particles that are triggered with a populace of arbitrary
candidate results. They travel cooperatively all the way
݄௞ ሺ݊ሻ ൌ  ݄௞ ሺ݊ െ ͳሻ െ ‫ܭ‬௞ ሺ݊ሻ݁௞ ሺ݊ሻ (6) through W-measurement space issue to seek the new results
C. MEMS Accelerometer that the fitness f could be measured by determining the
MEMS accelerometer is a wideband digital accelerometer number of exchange segments in the possible result. Every
which converts linear motion into an electrical signal. Here we particle has a location characterized by a location vector li (i is
use ADIS16220 for measuring the vibrations. The dynamic the key of the particle), and a speed characterized speed-vector
behavior of various capacitive MEMS are analyzed using a si. Every particle recollects its exact finest position until in a
simple lumped-parameter model [17] and is shown in Fig 2. vector li#, and its j-th dimensional esteem is lij#. The best
This model combines both the mechanical and electrical position-vector among the swarm so far is then put away in a
domain. In the mechanical domain, this model is a single- vector l*, and its j-th measurable data is lj*. At the point when
degree-of-freedom, mass-spring-damper system and the small the particle travels in a condition space confined to 0 and 1 on
every measurement, the transformed likelihood with time is IV. SIMULATION RESULTS AND IMPLEMENTATION
characterized as follows: Measured data is used to study the parameters of RHMDE.
‫ܮ‬௜௝ ሺ‫ݐ‬ሻ ൌ ݂௜௝ ሺ‫ ݐ‬െ ͳሻǡ ܵ௜௝ ሺ‫ ݐ‬െ ͳሻǡ ܵ௜௝‫ כ‬ሺ‫ ݐ‬െ ͳሻ (10) The simulation uses the measured voltage and the
where the likelihood function is accelerometer reading. The various parameters of RHMDE
ଵ
ܵ݅݃ ቀܵ௜௝ ሺ‫ݐ‬ሻቁ ൌ  షೄ೔ೕሺ೟ሻ (11) used in this simulation are as follows: the population size
ଵା௘ NP=50, maximum number of generation kmax=500, fitness
At each instance of time, every particle refreshes about its F=0.8, merging generation rate Pm=0.5, crossover rate Cr=1,
speed and travels to another location as indicated by Eq. (10) Length of individual Li=3, PSO-step size S=0.5. There are no
and Eq. (11) and guidelines in selecting these values; they are selected based on
݆ܵ݅ ሺ‫ݐ‬ሻ ൌ ‫ ݆݅ܵݍ‬ሺ‫ ݐ‬െ ͳሻ ൅ ‫ ݆݈݅ ͳܽ ͳݓ‬ሺ‫ ݐ‬െ ͳሻ ൅ ‫ ݆݅כ݈ ʹܽ ʹݓ‬ሺ‫ ݐ‬െ ͳሻ (12) the past studies. F should be greater than 0.4 and Cr should be
large so that high correlation exists among different
ͳ݂݅ߩ ൏ ‫݃݅ݏ‬ሺ݆ܵ݅ ሺ‫ݐ‬ሻሻ parameters of the double diode model.
݈݆݅ ሺ‫ݐ‬ሻ ൌ ቄ (13)
Ͳ‫݁ݏ݅ݓݎ݄݁ݐ݋‬ To measure the acceleration three different train speeds are
where w1 is a positive constant, referred to as weight of the considered such as 0-25 km, 26-50 km, and 51-75 km.
self-identification element, w2 is a positive constant, referred Acceleration measured using the MEMS accelerometers is
to as weight of the societal element, a1 and a2 are arbitrary filtered and optimized using PSO algorithm. It is then
numbers in the interim [0,1]. The variable q is called as the compared with the geometry measurement system.
inertial component, commonly used to change from 1 to 0 in Irregularities in both vertical and lateral directions are
the interaction process. ȡ is an arbitrary number in the closed measured using MEMS accelerometers mounted in the
interim [0.1]. From Eq. (6), a particle chooses where to travel vertical and lateral direction of the axle box and bogie.
next, considering its present express, its own understanding, Simulation is performed for three different train speeds in
which is the memory of its best past position, and the practice MATLAB 2015. Fig 9 shows the irregularities shown by
of its best particle in the swarm. vertically mounted axle box accelerometer for three different
train speeds. Irregularity is shown at a distance of 7.2 km in all
the three speeds.

Fig. 3. Comparison of vertical irregularity in axle box

mounted accelerometer: (a) 0-25 km (b) 26-50 km.
Fig. 4. Comparison of lateral irregularity in axle box mounted
The particle has a precedence level as indicated by the demand accelerometer: (a) 0-25 km (b) 26-50 km.
of peers. The succession of the peers will not change amid the
iteration process. Every particle’s location shows the possible Figure 3 and 4 shows the irregularities generated by
association state. vertically and laterally mounted axle box accelerometer under
two different speeds. Track irregularity is shown at a distance
of 7.2 km in all the two speeds but it is lesser when compared
to that of the vertically mounted accelerometer. Further, the IoT-based test bed i.e. a Raspberry Pi series platform. Table II
level of irregularity and the GPS signal level are also observed shows the basic implementation environment. Intelligent
by the controller. Figure 3 and 4 compares the proposed mobile object or local processing unit simulates the Raspberry
method and the track geometry method which clearly indicates Pi II platform. We implement SHA-3 (512 bits) as the secure
that the proposed method is more efficient than the other one. hash function, while the ECC based scalar multiplication
operations and the random number generator (96 bits) are
implemented with the Bouncy Castle Crypto APIs Open JDK
and Eclipse 3.8 is used to program this model. The setup is
shown in Fig. 8.

Fig. 5. Comparison of vertical irregularity in bogie mounted

accelerometer: (a) 0-25 km (b) 26-50 km.

Similarly, MEMS accelerometers are mounted in the

vertical and lateral direction of bogie to measure the Fig. 6. Comparison of lateral irregularity in bogie mounted
irregularity. Figure 5 a and b shows the irregularity shown by accelerometer: (a) 0-25 km (b) 26-50 km.
vertically mounted bogie accelerometer for three different
speeds. Track irregularity is shown at a distance of 7.2 km in
all the two speeds. Figure 6 a and b shows the irregularity
shown by laterally mounted bogie accelerometer for two
different speeds. Track irregularity is shown at a distance of
7.2 km which is lesser when compared to the vertically
mounted accelerometer. 26-50 km speed measurement has
more clarity than 0-25 km measurement. Figure 5 and 6
compares the proposed method with geometry measurement
and RLS filtering method. The results clearly indicates that the
proposed method is more efficient than the other technique.
The proposed method handles the critical situation and
ensures the safety of the passengers. Maintenance cost is also
largely reduced in this method. Location of vibration can be
measured by the proposed method if the GPS signals are
present or not present. If GPS signal is not present, the
controller switch on the proximity and estimate the distance Fig. 7. Comparison of LOV
and finally calculate the total exact LOV. Comparison among
proposed method and two other methods are shown in Figure For secure communication, the computation cost of 12
7 which shows clearly that LOV is measured accurately in the ECC based scalar multiplication operations and 8 SHA-3
proposed method. The proposed system is implemented on an operations is required.ECC based scalar multiplication is
implemented on the Raspberry Pi II platform. In short for 12 References
ECC based scalar multiplication and 8 SHA-3 operations
4.056 ms is needed. Hence the computation cost is lower for [1] Ir Dr Tony Lee, Ir May Tso, “A universal sensor data platform modeled
for real-time asset condition surveillance and big data analytics for
the proposed method. railway systems,” IEEE International Conference on SENSORS, pp. 1-
3, 2016.
Table II. Implementation environment [2] C. Chellaswamy, Aishwarya, Sanjna L.N Murthy, C. Vigneshwar,
“Green Controller for Rail Track Condition Monitoring and Information
Environment Description System,” IEEE International Conference on Green Computing,
Communication and Conservation of Energy (ICGCE), pp. 562-567,
Raspberry Pi Broadcom BCM2836 with ARM Cortex- 2013.
A7, 1GB RAM, Class 10 SD card [3] H. Tanaka, M. Matsumoto, Y. Harada, “Application of axle-box
Programming Language Open JDK acceleration to track condition monitoring for rail corrugation
Programming IDE Eclipse 3.8 management,” 7th IET Conference on Railway Condition Monitoring,
pp. 1-7, 2016.
Operating System Raspbian 2016/03
[4] Jun Seok Lee, Sunghoon Choi, Sang-Soo Kim, Choonsoo Park, and
Crypto API Bouncy Castle Crypto APIs Young Guk Kim, “A Mixed Filtering Approach for Track Condition
Monitoring Using Accelerometers on the Axle Box and Bogie,” IEEE
Transactions on Instrumentation and Measurement, Vol. 61, No. 3, pp.
749-758, 2012
[5] Paul Hyde, Francois Defossez, Cristian Ulianov, “Development and
testing of an automatic remote condition monitoring system for train
wheels,” IET Intelligent Transport Systems, Vol. 10, pp. 32-40, 2016.
[6] Maria Molodova, Maider Oregui, Alfredo Núñez, Zili Li, Jan Moraal,
Rolf Dollevoet, “Axle Box Acceleration for Health Monitoring of
Insulated Joints: A Case Study in The Netherlands,” IEEE 17th
International Conference on Intelligent Transportation Systems (ITSC) ,
pp. 822-827, 2014.
[7] Victoria J. Hodge, Simon O’Keefe, Michael Weeks, Anthony Moulds,
“Wireless Sensor Networks for Condition Monitoring in the Railway
Industry: A Survey,” IEEE Transactions on Intelligent Transportation
Systems, Vol. 16, pp. 1088-1106, 2015.
[8] Udit Satija, B Ramkumar, M. Sabarimalai Manikandan, “Real-Time
Signal Quality-Aware ECG Telemetry System for IoT-Based Health
Care Monitoring,” IEEE Internet of Things Journal, Vol. 4, No.3, PP,
815 - 823, 2017.
[9] Tzuu-Hseng S. Li et al., “A three-dimensional adaptive PSO-based
packing algorithm for an IoT-based automated e-fulfillment packaging
system,” IEEE Access, Vol. 5, pp, 9188-9205, 2017.
[10] C. Chellaswamy, S. Arul, L. Balaji, “Design and analysis of an
Fig. 8. Hardware implementation setup. intelligent collision avoidance system for locomotives,” IET
International Conference on Sustainable Energy and Intelligent Systems,
pp. 833 - 838, 2011.
V. CONCLUSION [11] Xiang Liu, M. Rapik Saat, Christopher P. L. Barkan, “Analysis of
An IoT-based rail track condition monitoring and the status Causes of Major Train Derailment and Their Effect on Accident Rates
Transportation Research Record,” Journal of the Transportation
transmission system are presented in this paper. The Research Board, No. 2289, Transportation Research Board of the
accelerometers are mounted in the vertical and lateral position National Academies, pp. 154-163, 2012.
of bogie and axel box to find the track irregularities. The [12] Hongming Cai, Boyi Xu, Lihong Jiang, Athanasios V. Vasilakos, “IoT-
simulation is performed for the dip length of 0.5 meters with based Big Data Storage Systems in Cloud Computing: Perspectives and
two different speed scenarios. An optimization algorithm Challenges,” IEEE Internet of Things Journal, Vol. 4, pp. 75-87, 2017.
(PSO) is used to estimate the exact vibration in the entire [13] Kun-Lin Tsai, Fang-Yie Leu and Ilsun You, “Residence Energy Control
System Based on Wireless Smart Socket and IoT, IEEE Access, Vol 4,
frequency band of the adaptive filtering. Comparison among pp. 2885-2894, 2016.
the proposed method and two other methods such as geometry [14] Feng Zhang, Min Liu,Zhuo Zhou, “Weiming Shen, An IoT Based
measurement and the system with RLS filtering have been Online Monitoring System for Continuous Steel Casting, IEEE Internet
done and it shows that the proposed method is efficient for of Things Journal, Vol. 3, pp. 1355-1363, 2016.
both the available and unavailable state of GPS signal. If any [15] Senturia SD. Microsystem design. Kluwer; 2001.
vibration is identified, the system automatically sends an alert [16] Den Hartog JP. Mechanical vibrations. MacGraw-Hill; 1968.
message to the central office and to the driver so that [17] C. Chellaswamy, V. Akila, A. Dinesh Babu, N. Kalai Arasan, “Fuzzy
logic based railway track condition monitoring system,” IEEE
necessary action will be taken. Hence the proposed system International Conference oN Emerging Trends in Computing,
ensures the safety of the passengers without any manual Communication and Nanotechnology, pp. 250-255, 2013
inspection of the tracks. [18] M. J. Bagshawe, “Investigating inertial measurements using MEMS
devices in trainborne automatic track condition monitoring
applications,” IET Conference on Control and Automation, 2013:
Uniting Problems and Solutions, pp. 1-6, 2013.

View publication stats