Fault Detection and Diagnostic Methods for Railway Systems – A Literature Survey

Full text

361
INTRODUCTION
Railway systems are essential components of
modern transportation networks, providing e-
cient and reliable means of moving people and
goods over long distances. However, these sys-
tems are not without their challenges, and various
components within railway networks are suscep-
tible to failure. Locomotive and railway wagon
failures challenge both the safety and eciency
of railway transportation systems. Identifying
and addressing the root causes of failures, im-
plementing preventive measures, and adhering
to safety standards are essential steps in ensuring
the reliability and safety of railway operations.
Through ongoing advancements in technology
and a commitment to safety, the railway industry
will continue to provide a vital and sustainable
mode of transportation for the future. Track fail-
ures, signaling and communication issues, rolling
stock malfunctions, infrastructure problems, and
various other challenges can disrupt operations,
pose safety risks, and cause substantial economic
losses, and addressing these common failures re-
quires a concerted eort from railway operators,
government agencies and researchers. By investing
in modernization, rigorous maintenance programs,
enhanced safety protocols, and advanced diagnos-
tic systems, failure risks can be mitigated, ensuring
that railway systems continue to serve as safe and
reliable modes of transportation. The most com-
mon failures of railway systems, as well as their
impact on operations, safety, and the economy,
are described in further detail within this paper.
Fault Detection and Diagnostic Methods for Railway
Systems – A Literature Survey
Jakub Wróbel1*, Paweł Bury1, Mateusz Zając1, Artur Kierzkowski1,
Sławomir Tubek2, Jędrzej Blaut3
1 Department of Technical Systems Operation and Maintenance, Faculty of Mechanical Engineering, Wroclaw
University of Science and Technology, ul. Łukasiewicza 7/9, Wroclaw, Poland
2 Tankwagon Sp. z o.o., Św. Ducha 5A/15 Str., Szczecin, Poland
3 Department of Mechanical Engineering and Robotics, AGH University of Krakow, al. A. Mickiewicza 30,
Krakow, Poland
* Corresponding author’s e-mail: jakub.wrobel@pwr.edu.pl
ABSTRACT
This paper presents a systematic literature survey on diagnostic methods used for railway materials and systems.
The authors analyze various railway accident reports, focusing on the types of failures described and their causes.
Previous review papers have addressed various aspects of railway systems diagnostics; however, most of the exist-
ing research focuses on specic parts of the rail vehicle or subsystems. In contrast, this survey focuses on railway
diagnostic systems rather than general diagnostic methods used in mechanical and electrical engineering. The authors
classify the types of failures and diagnostic methods that are used in rail transport into two categories: infrastructure
and rolling stock. The purpose of this paper is to systematize the types of failure that occur in railway transport sys-
tems; identify the state-of-the-art means and methods of diagnostics in railway materials and systems, with particular
focus on new research ndings; and identify trends and possible research gaps in need of further development.
Keywords: railway, diagnostics, failure detection, rolling stock, railway track, railway wheel, railway bogie, rail-
way failure.
Received: 2024.05.15
Accepted: 2024.09.15
Published: 2024.09.08
Advances in Science and Technology Research Journal, 18(6), 361–391
hps://doi.org/10.12913/22998624/191762
ISSN 2299-8624, License CC-BY 4.0
Advances in Science and Technology
Research Journal

362
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
This paper provides a detailed survey of the
current state-of-the-art in railway track condition
monitoring systems. The focus is on exploring
a range of technologies and methodologies em-
ployed for track health assessment. Ecient and
reliable railway systems are crucial for transport-
ing people and goods over long distances, un-
derscoring their importance. However, they face
signicant challenges due to their susceptibility
to failures in various components, which impact
safety, eciency, and the economy.
Identifying and addressing the root causes of
these failures, implementing preventive measures,
and adhering to stringent safety standards are es-
sential to maintaining the integrity of railway sys-
tems. The future of the railway industry lies in con-
tinued advancements in technology and a steadfast
commitment to safety, ensuring a sustainable and
ecient mode of transportation for years to come.
Figures 1 and 2 present an assessment of the
ten-year accident report of the U.S. Department
of Transportation, Federal Railroad Administra-
tion Oce of Safety Analysis. The main cause of
train accidents over the ten-year reporting period
is described as the “Human Factor”. The second
most common type of train accident is “Track
caused”, followed by “Other types”, “Miscella-
neous caused” and “ Motive power/equipment
caused” with “Signal caused” accidents closing
the ranking. The number of railroads included
into the report for each year varies from 811 to
847 railroads; approximately ± 2% distribution
comprises the average number of railroads taken
into account. Only partial yearly data are includ-
ed for 2023. The number of annual accidents re-
mains similar, irrespective of the cause.
Derailment was the most common type of ac-
cident for every recorded year. The cause of said
derailment was not specied in the report. Based
on the data displayed in Figure 2, it can be assumed
that most derailments are caused by human factors,
followed by track-related issues and other causes.
The number of annual railway accidents that
have occurred in the US in the last decade did
not decrease signicantly. Implementing railway
operator assistance and monitoring technologies
could help reduce the prevalence of human-re-
lated railway accidents. Acts of vandalisms or
Figure 1. Causes of train accidents from 2014 to 2023 [1]
Figure 2. Types of railway accidents from 2014 to 2023 [1]

363
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
random events are more dicult to detect and
monitor. Railway infrastructure and rolling stock
conditions, on the other hand, present many pos-
sibilities for improvement which could not only
reduce the number of railway accidents, but also
increase passenger and freight transport reliabili-
ty and make the railways more cost-eective.
Multiple review papers have discussed fac-
tors such as condition monitoring of rail vehicle
dynamics [2], wireless sensor networks for rail
transport [3], track condition monitoring [4, 5],
[6] and crossing and switch systems [7-9]. Most
of the mentioned studies focused on a specic
aspect of the railway transportation system. The
purpose of this literature review is to systematize
the types of damage that occur in rail transport
systems; identify the state-of-the-art means and
methods of diagnostics in railway systems, espe-
cially new ndings and research; and last, but not
least, identify possible research gaps for poten-
tial development. This paper presents common
railway failures, which are divided into two main
groups: infrastructure and rolling stock. The same
approach was adopted for diagnostic systems pre-
sented in later sections of the paper. Rewired work
is presented chronologically, focusing strictly on
railway diagnostic systems rather than general di-
agnostic methods found in mechanical and elec-
trical engineering.
The annual number of publications registered
in Web of Science and Scopus databases in last
two decades was reviewed in the context of rail-
way diagnostic systems. Two phrases – “railway
fault detection” and “railway diagnostic” – were
searched for in the titles, abstracts and key words
of published papers. Figure 3 shows the annual
number of publications containing the phrase
“railway fault detection” and Figure 4 shows the
annual number of papers containing the phrase
“railway diagnostic” recorded in the WoS and
Scopus databases from 2000 to 20 October 2023.
The number of publications relating to rail-
way diagnostics has increased annually since
2011, indicating an increase in research interest.
Figure 3. The annual number of publications containing the phrase “railway fault detection” recorded in the
WoS and Scopus databases from 2000 to 20 October 2023
Figure 4. The annual number of publications containing the phrase “railway diagnostic” recorded in the WoS
and Scopus databases from 2000 to 20 October 2023

364
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
However, despite increased research interest,
and the various new diagnostic methods and
techniques that have been proposed, the railway
accident rate has not decreased signicantly. De-
veloping new measurement techniques, digital
signal processing methods and big data analysis
takes time, especially when it comes to realizing
the real-life application of laboratory setups, as
strict railway safety regulations, compatibility
with other systems and data transfer capabilities
must also be considered.
FAILURES
The consequences of locomotive and wagon
failure can have far-reaching impacts on railway
operations, safety, and the economy. Delays and
disruptions can aect schedules and logistics,
with a cascading eect on the entire rail network,
causing congestion and reduced eciency. Brake
failures, wheel issues, or track derailments pose
signicant safety hazards, leading to accidents,
injuries and even loss of life. Repairing or replac-
ing locomotives, addressing damage caused by
derailments, and compensating for service dis-
ruptions lead to substantial nancial burdens.
INFRASTRUCTURE FAILURES
Tracks
One of the most prevalent areas of failure in
railway systems is the track itself. The railway
track, comprised of rails, sleepers (ties), ballast, and
subgrade, provides the guided pathway for trains.
Track failures can take various forms, including
buckling (Fig. 5) due to temperature uctuations;
rail defects such as cracks (Fig. 6) and wear; ballast
degradation, leading to uneven track support; and
irregular track geometry. All of the above can cause
derailments. These failures can disrupt railway op-
erations, result in costly repairs and pose signicant
safety risks to passengers and personnel.
Signaling and communication
Railway systems rely on signaling and com-
munication systems to ensure safe and ecient
train movements. These systems use visual or
electronic signals to communicate instructions
to train drivers regarding speed, route, and up-
coming track conditions. Signal failures, such as
broken signals or incorrect indications, can lead
to accidents and delays. Communication system
failures can disrupt train control and dispatch, po-
tentially causing collisions or operational chaos.
Infrastructure components
Infrastructure components such as switches
and crossings, bridges, tunnels, and level crossings
are essential elements of railway networks that
must be maintained in optimal condition to ensure
safe and uninterrupted train operations. Failures in
these areas, whether they are the result of struc-
tural issues or wear and tear, can result in service
interruptions, repair costs, and, in the worst cases,
accidents that endanger passengers and crew.
Power supply
In electried rail systems, the power supply
consists of substations, transformers, and overhead
Figure 5. Buckled rails damaged by heatwaves [10]
Figure 6. A broken rail caused by a tiny defect in the
foot of the rail and repeated bending under traffic [11]

365
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
catenary systems that deliver uninterrupted electri-
cal energy to trains. Any disruptions in this supply
can halt train operations, causing signicant delays
and inconvenience for passengers. Therefore, en-
suring a reliable power supply is crucial for main-
taining the smooth functioning of railways.
Wear and tear
Railway infrastructure consists of various
components such as tracks, switches, bridges,
tunnels, and level crossings. Due to heavy use,
particularly in the case of heavy freight transpor-
tation, railway infrastructure is subject to con-
tinuous wear and tear. This necessitates regular
maintenance and replacement of components to
ensure safe and reliable operations.
Rolling stock failures
Rolling Stock refers to the collection of locomo-
tives, railcars, and other vehicles that move along the
railway tracks. These vehicles consist of mechanical
and electrical systems essential for propulsion, brak-
ing, and passenger or freight transport. The mechan-
ical and electrical components of locomotives and
railcars are prone to failure. Locomotive failures can
occur for various reasons, disrupting the operation
of trains and posing safety hazards.
Drive failures
Locomotives are rail vehicles powered by
engines, typically diesel or electric, that provide
the necessary traction to haul trains along railway
tracks. They are equipped with diverse mechani-
cal and electrical systems for propulsion, braking,
and control, crucial for ensuring safe and ecient
railway operations. Diesel-powered locomotives
can encounter issues such as engine overheating,
fuel system problems, and injector failures, with
critical reliance on fuel pumps; failures in these
pumps can cause engine shutdowns. Addition-
ally, overheating due to coolant system problems
can lead to engine damage. Electric locomotives,
while not immune to failure, often experience
motor issues such as electrical shorts, overheat-
ing, bearing failures, or worn-out components.
Mechanical failures
Engineering practice shows that locomotives
transmission systems can exhibit various prob-
lems, including gear or clutch failures. Wheel
fractures or at spots can lead to rough rides or
even derailments, while fractured or bent axles can
compromise the structural integrity of the wagon
and pose a serious safety hazard. Meanwhile, bo-
gie axle bearing failures can occur due to poor lu-
brication, contamination, manufacturing defects
or simple wear. When bearings fail, they tend to
overheat, increase friction and potentially lead
to derailments, posing a signicant safety risk.
The suspension system in bogies plays a crucial
role in absorbing shocks and vibrations, pro-
viding a safe and comfortable ride. Suspension
system failures can occur due to worn-out com-
ponents, such as springs or dampers. Scenarios
involving broken or shortened springs, as well
as softening, which includes individual spring
losses from a nest or crosssection degradation
due to corrosion, can lead to an uncomfortable
ride and increase wear on the track, requiring
more frequent maintenance, while malfunction-
ing couplers can result in decoupling between
wagons, disrupting the train’s stability.
Locomotive control systems
Issues with the locomotive’s control systems, in-
cluding the computerized control unit, sensors, and
communication systems, can cause failures. Faulty
wiring, loose connections, or electrical shorts can
disrupt the locomotive’s electrical systems, while
problems with the air brake system, such as leaks
or malfunctioning valves, can compromise the abil-
ity to stop the locomotive and wagons safely. Brake
failures can lead to dangerous situations, such as
loss of control or runaway trains. Brake shoes, air
compressors, or brake lines can also fail. Air com-
pressors are crucial for maintaining air pressure in
the locomotives and wagons systems.
Human factor and consequences failures
Operator errors such as improper operation or
failure to follow safety procedures, as well as inad-
equate maintenance or repairs, can result in loco-
motive or wagon breakdowns. Deliberate acts of
vandalism and sabotage can target railway systems,
causing costly damage and disruptions. Protection
of railway assets from such threats is essential for
maintaining safety and operational continuity. Cy-
bersecurity threats are another emerging concern
for railway systems, as targeting control systems,
signaling, or communication networks can com-
promise safety and service integrity.

366
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
The consequences of locomotive and wagon
failures can have far-reaching impacts on rail-
way operations, safety, and the economy. De-
lays and disruptions can aect schedules and
logistics, with a cascading eect on the entire
rail network, causing congestion and reduced ef-
ciency. Brake failures, wheel issues and track
derailments pose signicant safety hazards that
can lead to accidents, injuries, and even loss of
life. Meanwhile, repairing or replacing locomo-
tives, addressing damage caused by derailments,
and compensating for service disruptions come
with substantial nancial burdens.
RAILWAY TRACKS
In the realm of railway diagnostics, the de-
velopment and application of various methods
have evolved signicantly over time. Initially,
railways relied heavily on manual visual inspec-
tions, where personnel would physically examine
tracks and rolling stock to identify any potential
issues. While these inspections were crucial in
ensuring safety, they were labor-intensive and
often limited in scope. As technology advanced,
periodic inspections became the norm, allowing
for more detailed and systematic assessments.
This phase saw the introduction of specialized
equipment to measure track geometry and other
critical parameters. The emergence of in-service
vehicle monitoring systems marked a pivotal
shift, utilizing existing rolling stock to continu-
ously gather real-time data on track conditions.
Further innovations brought ber optic sensors,
which provided high sensitivity and long-range
monitoring capabilities. Onboard monitoring
systems for rolling stock, along with vibration
analysis techniques, oered non-intrusive ways
to detect component health and potential failures.
These advancements paved the way for integrated
systems that combine multiple sensor technolo-
gies, enabling comprehensive diagnostics and
predictive maintenance. The Figure 7 visually
maps the progression from early methods such as
visual inspections to the future direction of inte-
grated systems utilizing multiple sensor technolo-
gies for comprehensive diagnostic capabilities.
In 2003, Bogue [12] delineated the develop-
ment and utilization of a wireless-based stress
monitoring system. Its scope extends to various
applications, including the monitoring of railway
tracks, rolling stock and substantial structures
such as oil rigs and bridges.
In 2005 and 2006, Kojima et al. [13, 14] dis-
cussed the conventional methods of measuring
the conditions of railway tracks using inspection
vehicles. They suggested that it might be more
Figure 7. Flowchart illustrating the evolution of diagnostic methods in railway systems over time

367
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
ecient to use simple sensors, such as accelerom-
eters, attached to commercial vehicles to evaluate
track conditions. Specically, the authors focused
on rail corrugation detection through the analy-
sis of vertical acceleration of a vehicle body over
time using time–frequency analysis. The research
involved conducting actual vehicle tests on a com-
mercial rail line, which involved measuring the
vertical acceleration of axle-boxes and the vehicle
body. The paper demonstrated that rail corrugation
could be detected based on the vertical accelera-
tion of a vehicle body using multi-resolution anal-
ysis (MRA) with the wavelet transform technique.
In their 2006 study, Weston et al. [15] focused
on the miniaturization of track recording vehicle
equipment, enabling its installation on in-service
vehicles for the purpose of monitoring track geom-
etry. An alternative approach, as presented in the
paper, involved attaching robust sensors, such as
accelerometers and rate gyroscopes, to the bogie
and axle-boxes of in-service vehicles. The sensors’
data, however, had limitations, particularly in terms
of gauge measurement, and the lateral movement of
the wheelsets with respect to the track introduced
inaccuracies. Despite these constraints, the paper
emphasized that valuable information could still
be derived, including estimates of mean vertical
and lateral alignment standard deviations and the
detection of certain track geometry faults. This in-
formation could help to inform track maintenance
practices. Furthermore, the paper highlighted the
potential for the motion of the bogie and wheelsets
in relation to the track to provide insights into the
interaction between specic vehicles and the track
itself. While this information might not explic-
itly reconstruct track geometry, it could be used to
monitor vehicle/track interaction. The paper pro-
vided illustrative examples of features observed in
the bogie and axle-box data obtained from sensors
installed on a Tyne and Wear Metro vehicle. These
features oered insights into the vehicle’s interac-
tion with the track, although they did not directly
reconstruct the track geometry.
Naderi and Mirabadi [16] discussed the ap-
plication of ber optic sensors in the railway in-
dustry. Their research evaluated the limitations
and capabilities of dierent sensor types, includ-
ing intensity-, phase-, and wavelength-modulated
sensors, along with their signicant parameters
for railway applications. The paper also presented
simulation results for Fiber Bragg Grating (FBG)
and Fabry–Perot Interferometer (FPI) sensors de-
signed for railway use, with a focus on modeling
the railwheel interaction using ANSYS software.
Furthermore, the authors detailed the modeling
and performance evaluation of a sensor created
to measure train weight under various conditions.
Hayashi et al. [17] described various meth-
ods for detecting faults in railway vehicles and
tracks. They employed a model-free approach us-
ing multi-resolution analysis for fault detection
in tracks from on-board measurement data. For
vehicle fault detection, they utilized the interact-
ing multiple-model (IMM) algorithm as a model-
based approach. Their research included simula-
tion studies and eld tests with actual vehicles,
which indicated that the proposed methods could
eectively detect both track and vehicle faults.
In 2007, Weston et al. [18] focused on main-
taining the alignment of railway tracks to ensure
safe and ecient travel. The paper highlighted
that poor track alignment could lead to issues
such as ride quality problems, ange contact, and
even ange climb, all of which could pose safety
risks. The paper proposed that mean track align-
ment be estimated using sensors installed on the
bogie of an in-service railway vehicle, without
the need for optical or contact sensors. It outlined
how either bogie lateral acceleration or yaw rate
could be processed to estimate mean lateral track
irregularity, with a yaw rate gyro being particu-
larly eective, especially at lower vehicle speeds,
and it did not require compensation for bogie roll
eects. The paper also described an improved es-
timation technique achieved by inversely relating
mean track alignment to bogie yaw motion. The
eectiveness of this method was demonstrated
using results obtained from a Class 175 vehicle.
The research emphasized that continuously moni-
toring the lateral response of a bogie on in-service
vehicles, relying solely on a yaw rate gyro, could
provide valuable data regarding which areas of
maintenance to prioritize, contributing to the
safety and eciency of railway operations.
Attivissimo et al. [19] developed a railway
measurement system to measure wheel–rail in-
teraction quality in real time. They specically
focused on equivalent conicity, as dened by the
UIC 518 Standard. Their system used contactless
optical data processing and met the accuracy re-
quirements of the UIC 519 Standard.
In 2008, Ho [20] introduced the principles
of photonic distributed sensors and emphasized
the advantages of Fiber Bragg grating sensor ar-
rays in railway applications. Their work presents
initial results from eld measurements conducted

368
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
with local railway partners, demonstrating the
eectiveness of these sensors for smart railway
condition monitoring systems. Two commercial
systems have been developed and are in use lo-
cally—one on rail tracks to detect wheel/rail in-
terface responses and another onboard trains with
thermal monitoring. There are ongoing eorts to
promote these systems to railway operators and
consultants worldwide.
Mizuno et al. [21] developed a mobile sensing
unit and created a prototype for monitoring railway
tracks. This unit consisted of a compact PC, a GPS
receiver, an accelerometer, and an Analog/Digital
Converter (ADC). It was designed to track routes
and capture acceleration data from passenger vehi-
cles, oering more frequent and higher-quality data
compared to traditional railway track inspection
equipment. The unit’s accurate location determina-
tion, which incorporated GPS data, existing land-
marks, and vehicle acceleration responses, was a
signicant advantage. The researchers believed
that their unit held promise for ecient railway
property management. The prototype’s ndings
suggested a correlation between car acceleration
responses and low-frequency track displacements,
indicating that placing sensors on the vehicle oor,
rather than on axles or bogies, was eective for
capturing vertical track displacements.
In 2010, Mori et al. [22] designed a portable
system for monitoring track conditions which was
specically tailored for easy installation on in-ser-
vice vehicles. The system relied on the car body’s
vertical and lateral acceleration data to estimate
rail irregularities. It employed a rate gyroscope to
calculate the car body’s roll angle, allowing for dif-
ferentiation between line and level irregularities.
Rail corrugation was detected by analyzing cabin
noise and identifying spectral peaks. To precisely
locate track faults, the system utilized GPS tech-
nology and a map-matching algorithm. Field tests,
conducted with in-service vehicles, veried the
system’s ability to eectively estimate rail irregu-
larity and rail corrugation conditions.
In 2011, Ward et al. [23] explored the chal-
lenges stemming from the global increase in rail-
way passengers, which has put pressure on im-
proving the entire system’s capacity, punctuality,
and cost eciency. They underscored the role of
condition monitoring in meeting these demands.
The article specically investigated the use of
sensors mounted on rolling stock to monitor both
infrastructure and the rolling stock itself. This ap-
proach was considered in light of contemporary
rolling stock equipped with advanced communi-
cation systems and multiple sensors, oering the
potential for sophisticated data analysis. The ar-
ticle consolidated related research that employed
a common set of rolling stock sensors, covering
topics such as their general application and ad-
vantages, track defect detection methods, moni-
toring running gear conditions and detecting ab-
solute train speeds.
Lee et al. [24] conducted a study that compared
the use of axle-box and bogie-mounted accelerom-
eters for monitoring track conditions with in-ser-
vice high-speed trains. They introduced a method
that relied on Kalman lters, band-pass lters and
compensation lters to estimate lateral and vertical
track irregularities based on data from either axle-
box or bogie-mounted accelerometers. They also
used rail vehicle dynamics software to analyze the
estimated results and validate their methodology.
In another study in 2012, Lee et al. [25] pre-
sented a method of estimating railway track irreg-
ularities using acceleration data from high-speed
trains. Track irregularities can lead to train vibra-
tions, making their monitoring essential for ride
quality. Their method involved applying lters
for stable displacement estimation and waveband
classication of acceleration data. Accelerom-
eters placed on the axle box and bogie of high-
speed trains captured lateral and vertical accel-
eration. The study compared their approach with
commercial track geometry measurement results
and discussed the accelerometer placement’s im-
pact on estimated track irregularities.
Tesfa et al. [26] aimed to address the issue of
bolted joint failures in industrial structures related
to railway infrastructure. These bolted connections
are crucial to the safe and reliable operation of tracks
and trackside equipment. Current manual mainte-
nance procedures, often involving tens of thousands
of bolted joints, are costly, disruptive and suscep-
tible to human error. The objective of Tesfa et al.’s
study was to develop a sensor-equipped washer that
could automatically measure the clamping force of
each individual bolted joint. This technology aimed
to provide a more ecient and reliable solution for
maintaining the integrity and safety of these struc-
tures, reducing maintenance costs and minimizing
disruptions. The paper outlines the development of
the sensor technology to be incorporated into the
washer, with a focus on meeting specic criteria for
accurate clamping force measurement.
Ngigi et al. [27] focused on modern railway
systems and their advanced monitoring techniques

369
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
for maintenance. They explored condition moni-
toring methods that use sophisticated approaches
like ltering and signal analysis for fault detec-
tion. These methods are adept at handling system
complexities and variations without the need for
intricate mathematical models. In practice, sen-
sors are deployed either on the track or rolling
stock, depending on the specic monitoring re-
quirements. For instance, track-mounted sensors
can monitor wheelset dynamics, while vehicle-
based sensors are used to oversee the train infra-
structure. The paper aimed to compile and assess
contemporary techniques for monitoring railway
vehicle dynamics, providing a critical evaluation
of their advantages and limitations.
Chellaswamy et al. [28] aimed to enhance
railway passenger comfort by reducing noise
and vibration during travel. They introduced the
Fuzzy Track Monitoring System (FTMS), which
is to estimate track irregularities in real time. Vi-
bration sensors on the train’s axle box and bogie
measured acceleration in vertical and lateral di-
rections, allowing vibration data to be tracked and
relayed to a central oce. This showcased this
method’s potential to obtain real-time measure-
ments and thus improve ride quality.
Bagshawe [29] explored the feasibility of us-
ing readily available MEMS (Micro-Electro-Me-
chanical Systems) sensors in existing train-borne
track condition monitoring systems. These sys-
tems often rely on costly and relatively large in-
ertial measurement units (IMUs). The goal was to
reduce support costs and improve spare parts avail-
ability by considering MEMS sensor substitutes.
The study compared a candidate MEMS acceler-
ometer with a standard accelerometer, particularly
in relation to inertial measurements in the vertical
prole. The research determined the minimum per-
formance requirements for the replacement accel-
erometer based on measurement system specica-
tions. A suitable MEMS device was integrated into
the system. Trial data collected at various speeds
were analyzed, and the results indicated that the
MEMS accelerometer was theoretically suitable
for use at speeds as low as 25 mph. Despite some
practical limitations related to its installation, the
trial showed that the MEMS accelerometer closely
matched the performance of the standard acceler-
ometer, especially at speeds greater than 50 mph.
Further trials with improved installation positions
were suggested for future investigation.
In 2013, Chellaswamy et al. [30] aimed to
address the primary issue aecting passenger
comfort in railway transportation: vibrations,
predominantly stemming from track variations.
Their work focused on reducing these vibrations
by eliminating track irregularities through an in-
novative system known as the Intelligent Track
Monitoring System (ITMS), which is designed to
identify and rectify track aberrations. The system
uses MEMS technology, which is known for its
reliability, employing sensors placed on both the
axle box and bogie in both vertical and lateral di-
rections. A controller continuously monitors GSM
signals and processes data accordingly, with GPS
used to determine location when the GSM signal
strength is low. ITMS autonomously detects ir-
regularities and communicates their locations to
a central oce. The system utilizes Continuous
Wavelet Transform (CWT) to estimate output,
and the study’s results conrmed that the pro-
posed monitoring system signicantly improves
passenger ride quality.
Qin et al. [31] created an onboard device
for diagnosing track faults in railway condition
monitoring. The device, which is designed to be
installed on in-service vehicles, identies track
faults by analyzing the unique vertical and lat-
eral acceleration patterns of the axle-box, bogie,
and car-body. To assess the track’s condition, the
vibration signal is initially transformed from the
time domain to the frequency domain. Principal
component analysis (PCA) and support vector
machines (SVM) are then used to calculate prob-
abilities of track faults. The Dempster–Shafer
(D-S) evidence theory is employed to combine
information from various sources for track fault
diagnosis. When a fault is detected, the system
immediately sends an alert to the monitoring cen-
ter, ensuring the safety of subsequent vehicles.
Experiments using three accelerometer signals
from dierent positions validated the algorithm’s
eectiveness in estimating rail irregularities and
diagnosing track faults.
In 2014, Tsunashima et al. [5] developed a
track condition monitoring system for conven-
tional railways. In their work, they estimate track
irregularities by analyzing vertical and lateral
acceleration, as well as the roll rate of the train
car. They also detect rail corrugation by analyz-
ing cabin noise through spectral peak calculation.
To pinpoint the location of track faults, they use
a combination of GPS and a map-matching algo-
rithm. The authors have also introduced a new
and improved device designed for practical use,
oering higher performance. Data collected by

370
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
the system can be stored in an onboard memory
or transmitted to a data server via a cellular con-
nection. This compact on-board device ensures
regular monitoring of railway tracks to maintain
and secure the safety of the railway system.
In another study, Tsunashima et al. [32] focused
on enhancing railway safety and comfort through
track maintenance based on track geometry data.
They developed a track condition monitoring sys-
tem that was primarily used to identify areas requir-
ing track tamping for smoother rides. Their inno-
vation involved using simple car-body acceleration
measurement devices, which simplify maintenance
procedures. While car-body acceleration data dier
signicantly from track geometry data, the paper
demonstrates the feasibility of estimating Shinkan-
sen track geometry solely from car-body motion.
They employed a Kalman lter (KF) in an inverse
problem to estimate track irregularities from car-
body motions, achieving accurate vertical track ir-
regularity estimation for practical use.
Reiterer et al. [33] focused on the impor-
tance of regular condition monitoring for railway
maintenance and safety. They recommended us-
ing running inspection trains or stationary check-
points equipped with advanced measurement
systems. These systems must operate swiftly and
accurately under challenging conditions. Laser
scanning was identied as an ecient method for
precise measurement of railway infrastructure,
oering high point densities and millimeter level
accuracy. The paper provides an overview of la-
ser scanning methods, their advantages and their
disadvantages, and describes specic railway
measurement systems developed by the authors.
Yeo et al. [34] utilized an inertial measure-
ment unit (IMU) with three high-quality acceler-
ometers and three gyroscopes attached to the bo-
gie of a working railway vehicle. They processed
the IMU data along with a tachometer signal and
GPS information to monitor the location, orienta-
tion and trajectory of the bogie, which allowed
them to assess the condition of the track geometry.
They continuously monitored these data as the
vehicle traveled along its normal routes, enabling
the detection of changes in track geometry with
ne temporal granularity. The research described
the applications of such precise track geometry
data and the development of automated process-
ing methods to extract the desired information.
The primary goal was to observe track geom-
etry through various stages, including renewals,
degradation, and maintenance, to understand the
development of faults and the eectiveness of
maintenance eorts.
In their eld trial, Roveri et al. [35] utilized a
Fiber Bragg Grating (FBG) sensor array system to
conduct real-time monitoring of railway trac and
assess the structural health of both the railway track
and train wheels. They executed these tests on Mi-
lan’s second line of the metropolitan underground,
deploying over 50 FBG sensors along a 1.5 km rail
track. These tests were conducted during daily pas-
senger rail transport, with trains reaching speeds of
approximately 90 km/h. Measurements were con-
tinuously taken over a six-month period at a sam-
pling rate of approximately 400 Hz. The abundance
of data and sensors enabled precise statistical anal-
ysis of measurement data. Dedicated algorithms al-
lowed for the estimation of rail and wheel wear, as
well as key trac parameters such as axle count,
train speed, load, and the potential identication of
localized imperfections in the near future.
Hodge et al. [36] explored the recent advance-
ments in sensing technologies and the decreasing
cost of sensor devices, which have driven the
widespread adoption of condition monitoring in
various domains, including systems, structures,
vehicles and machinery. They highlighted the sig-
nicance of wireless communication and mobile
ad hoc networking technologies, along with the
ability to integrate devices, leading to the use of
wireless sensor networks (WSNs) for monitoring
railway infrastructure and vehicle health. Condi-
tion monitoring through WSNs reduces the need
for human inspections, enables earlier fault detec-
tion, preventing escalation, and enhances safety
and reliability in the railway industry. The paper
provided a comprehensive survey of WSN tech-
nology for railway monitoring, focusing on prac-
tical engineering solutions, the types of sensor
devices employed, their specic applications, and
the congurations and network topologies used.
It also compared and contrasted the motivations
and advantages and disadvantages of these sensor
setups in the context of railway monitoring.
In 2015, Weston et al. [37] discussed the cur-
rent status of monitoring track geometry condi-
tion via in-service vehicles. They described the
technology used, the challenges associated with
processing, location determination and how this
eld has evolved over the past decade. The pa-
per emphasized the need for ongoing research.
While in-service vehicle-based track geometry
monitoring has already become a reality, there is
room for improvement regarding this technique.

371
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
The paper highlighted the value of repeated track
observations for detecting geometry degradation,
informing maintenance decisions and evaluat-
ing maintenance eectiveness. It mentioned un-
attended track geometry measurement systems
used worldwide, with underutilized data primar-
ily generating reports. The paper also described
dierent monitoring approaches, including sim-
pler systems, experimental methods and systems
indirectly assessing poor track geometry.
In 2015 and 2016, Tsunashima et al. [38],
[39] developed a track condition monitoring sys-
tem for enhancing railway safety. They achieved
this by equipping in-service vehicles with sen-
sors and GPS systems to capture real-time ve-
hicle vibrations. Their innovation lies in the
use of a compact on-board sensing device and
specialized diagnosis software. The software
detects track faults by analyzing the root mean
square (RMS) of the car-body’s acceleration and
conducts time–frequency domain analysis using
wavelet transforms. The researchers also applied
a Kalman lter to solve an inverse problem, es-
timating track irregularities based on car-body
acceleration data. This estimation method was
employed to assess track irregularities in terms
of track geometry and 10 m-chord versine in the
longitudinal direction. The results demonstrated
that this estimation technique eectively sup-
ports track condition monitoring with acceptable
accuracy for conventional railways. Field tests
conducted on local railway lines conrmed the
system’s practical eectiveness.
In 2016, Ilie and Stancalie [40] addressed the
safety concerns associated with railway defects,
particularly broken rails, which jeopardize railway
operations. Their goal was to establish a reliable
and accurate railway monitoring system with po-
tential for remote use. They conducted a series of
extensive experiments to explore the feasibility of
using optical ber sensors for railway monitoring,
with particular focus on how optical ber sensors
respond to temperature and strain variations to as-
sess railway expansion, aiming to create a proof of
concept for a dependable monitoring system.
Maddison and Smith [41] developed wireless
sensor networks for remote geotechnical moni-
toring, incorporating low-cost, self-contained,
and self-conguring wireless sensors. These
sensors are easily installed without the need for
extensive wiring. They have made signicant
advancements in areas such as extending sensor
node battery life, improving network robustness,
data throughput, and sensor precision. This tech-
nology has performed successful monitoring in
various challenging environments, such as rail
tunnels and trackbeds, fullling the needs of as-
set holders such as London Underground and
Network Rail. The system demonstrated precise
measurements of tunnel deformation during en-
gineering projects, and it can also eectively
monitor railway tracks, measuring changes in
track cant, twist and longitudinal rate of change
with high precision and stability. The wireless
solution was preferred over optical options due
to its stability, repeatability and resistance to en-
vironmental challenges. A growing trend toward
wireless monitoring of rail assets is anticipated
due to the advantages it oers.
Lienhart et al. [42] have introduced innova-
tive methods for the continuous monitoring of
railway tracks and vehicles across extensive dis-
tances in the European railway network, utiliz-
ing distributed ber optic sensing (DFOS) tech-
niques. They began by axing ber optic strain
sensing cables to railway tracks to detect strain
changes caused by rail deformations. This was
monitored through distributed Brillouin measure-
ments (BOTDA, BOFDA). This system enables
the early identication of potential damage from
natural events, such as mudows, avalanches,
oods and landslides, allowing for swift imple-
mentation of countermeasures. In a second ap-
proach, optical communication cables already in
place alongside modern rail infrastructure were
used to detect at spots on railway wheels. These
at spots, if left unaddressed, can damage tracks
and may lead to derailment. The researchers used
distributed acoustic sensing (DAS) for continu-
ous monitoring of trains and extraction of indi-
vidual proles indicating at spots. Importantly,
this system requires no additional infrastructure,
apart from the optical time-domain reectometry
(OTDR) instrument. Through eld installations
in the Austrian railway network, the eectiveness
of both BOFDA and OTDR systems is demon-
strated, allowing for the ongoing monitoring of
incidents and deformations. This approach en-
sures the continuous assessment of railway tracks
and vehicles over long distances.
Xu et al. [43] developed a method for iden-
tifying track irregularity faults that can cause ab-
normal train vibrations, leading to poor ride qual-
ity and derailment. Their approach, based on Evi-
dence Reasoning (ER) rules, uses data from accel-
erometers mounted on the axle-box and car-body

372
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
of in-service trains. Their method uses statistical
analysis of sample data to generate diagnostic evi-
dence, combines this evidence using ER rules and
subsequently estimates irregularity displacement.
It then identies the dynamic levels of track ir-
regularity based on the estimated displacement and
relevant management standards. An experiment
conducted on a Chinese railway line demonstrated
the superior accuracy of this method compared to
classical neural net-work-based approaches.
Balouchi et al. [44] collaborated with the
Rail Safety and Standards Board and the Insti-
tute of Railway Research to develop the Siemens
Tracksure track monitoring system. This system
leverages the existing GSM-R cab radio in all
UK trains, equipped with a sensor card to detect
track condition via three-axis train vibrations.
Tracksure’s onboard signal processing reduces
data transfer requirements. For voided switches
and crossings (S&C), the system uses GPS loca-
tion to pinpoint asset numbers, increasing accu-
racy and streamlining maintenance eorts. The
Ground System combines data from multiple
trains to achieve more precise void detection and
fewer false alarms. It also oers automated void
reporting, facilitating ecient maintenance plan-
ning. Nexus Tracksure can be easily activated on
all UK trains, forming a comprehensive network-
wide track monitoring system. Recent trials dem-
onstrated its eectiveness in detecting voided
sleepers under various track types, with excellent
repeatability. Upcoming Network Rail upgrades
will enhance the system with GPS connectivity,
improving preventative maintenance and safety.
In 2017, Chellaswamy et al. [45] proposed a
new method for monitoring rail track irregulari-
ties, enhancing transportation safety. Traditional
track inspections are infrequent and often occur at
night when service is minimal, meaning that criti-
cal issues may be overlooked. Thus, the focus has
shifted to in-service vehicle monitoring. These
researchers’ approach involves using acceleration
data from both the bogie and car body to detect
track irregularities. These data inform track align-
ment analysis through mathematical modelling
and frequency response evaluation. Both simula-
tions and real-world tests conrmed the system’s
eectiveness. They dierential evolution (DE) al-
gorithm was used to optimize irregularity values
from accelerometers in the axle box and bogie.
Simulation results at various train speeds demon-
strate the ecacy of their approach compared to
traditional track geometry measurements.
Seraj et al. [46] have developed RoVi, a
smartphone-based framework for the continu-
ous monitoring of ground transport infrastruc-
tures. RoVi tracks various indicators such as rail-
road track geometry features and road/bike path
roughness, providing real-time, ne-grained data
through smartphones’ inertial sensors. It lls the
gaps in monitoring caused by infrequent inspec-
tions using a crowd-sensing approach. RoVi is
a valuable tool for engineers and maintenance
planners, oering features and indicators for as-
set management and maintenance planning. It
extracts this information from smartphone data,
using adaptive signal processing and geolocation
visualization. The system’s performance has been
rigorously evaluated and has proven eective in
continuous infrastructure monitoring.
Muthukumar and Nallathambi [4] have in-
vestigated the growing use of sensing technolo-
gies in recent years, as these technologies have
become more cost-eective. This expansion has
led to increased condition monitoring of various
systems, structures, vehicles, and equipment.
They highlight the role of advancements in net-
working technologies, such as wireless commu-
nication and mobile ad hoc networking, in con-
junction with the integration of sensor devices.
Wireless sensor networks (WSNs) have found
applications in the railway industry for monitor-
ing infrastructure like bridges, rail tracks, track
beds, and track equipment, as well as vehicle
safety monitoring. This technology reduces the
need for manual inspections, identies issues be-
fore they escalate, and enhances safety and reli-
ability, which are critical to the development and
expansion of railway systems. Their project ex-
plores the use of WSN technology in the railway
sector, focusing on systems, structures, vehicles,
and machinery. It also examines practical engi-
neering solutions, including the types of sensor
devices used, their applications, and various sen-
sor congurations and network topologies. The
study aims to provide a comparative review of
their advantages and disadvantages.
Groos et al. [47] have addressed the signi-
cant maintenance costs associated with railway
tracks, which often involve reactive measures.
In their paper, they propose a solution utilizing
embedded sensors on in-service vehicles for daily
track condition monitoring. This shift from reac-
tive to proactive maintenance aims to reduce costs
signicantly. Their work outlines a framework
and the initial results of a prototype system for

373
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
quasi-continuous monitoring of short-wavelength
defects in railway tracks, like rail corrugation.
The prototype was tested on a shunter locomotive
in Germany, collecting data over a four-month
period for algorithm development and evalua-
tion. Data from acceleration sensors, combined
with infrastructure information, pass through a
data management system. This process involves
georeferencing, feature extraction, and intelligent
data analysis, with results visualized for infra-
structure operators through a web interface.
Chudzikiewicz et al. [48] focused on demon-
strating the feasibility of estimating track con-
dition using acceleration signals derived from
axle-boxes and car-body motions. They present
the results of a preliminary investigation on a
test track and supervised runs on Polish Railway
Lines, involving an Electric Multiple Unit (EMU-
ED74) with a prototype track quality monitoring
system installed onboard. Their prototype uses a
track quality indicator (TQI) algorithm based on
a modied Karhunen–Loève transformation to
process acceleration signals and extract princi-
pal dynamics from the measurement data. They
compare these results with other methods used to
evaluate track quality, including the synthetic co-
ecient Jsynth and ve defectiveness parameters
(W5). Their ndings indicate that estimating track
condition with acceptable accuracy is achievable
for in-service applications, facilitating the devel-
opment of cost-eective maintenance strategies.
Zhang et al. [49] have implemented Fib-
er Bragg Grating (FBG) sensing technology for
real-time monitoring and early warning of high-
speed railway track conditions in China. They es-
tablished a sensor network with FBG sensors on
the tracks to continuously monitor variables such
as track temperature, displacement, and strain.
The collected data are processed and analyzed
using FBG demodulators. They also developed a
temperature prediction model based on the rele-
vance vector regression algorithm, enhancing the
prediction accuracy. This system is currently in
use on the Guangzhou–Shenzhen–Hong Kong
high-speed railway, successfully providing early
warnings regarding track conditions.
In 2018, Ngamkhanong and Kaewunruen [50]
focused on the role of railway systems in modern
transportation, highlighting the increased demand
for both passenger and cargo transport. They em-
phasized the need for condition monitoring to as-
sess railway track health, particularly in the face
of harsh conditions and heavy loads. Their work
reviews the various sensors used for monitoring
railway track infrastructure, oering insights into
their applications during extreme events. The aim
of their research was to improve track inspection,
damage detection, and predictive maintenance
strategies, ultimately supporting cost-eective
management within the railway industry.
Känsälä et al. [51] explored the application of
cost-eective real-time sensors and the Industrial
Internet of Things (IIoT) for proactive asset man-
agement in the rail trac industry. Recognizing
that railways are long-lasting infrastructure assets,
even small improvements in eciency and cost
can have a substantial impact on overall life-cycle
costs. Their study demonstrates the use of wireless
three-dimensional acceleration sensors to monitor
track conditions. Data collection was conducted in
October 2016 on a Finnish Railways-operated rail-
way line. A sensor was attached to a train unit, and
train acceleration on a track segment was repeatedly
measured at varying speeds. To enhance the col-
lected data, map-matching and Bayesian ltering
techniques were applied to improve the accuracy of
Global Positioning System (GPS) location data. The
ltered acceleration signals were analyzed, and any
anomalies detected were compared to known pa-
rameters such as bridges and switches. The results
of their testing conrm the feasibility of this con-
cept. They also discuss the potential implications
of this approach for proactive asset management of
track networks and the application of statistical pro-
cess control-based monitoring for track condition.
Milne et al. [52] addressed the use of low-fre-
quency vibrations for assessing railway track
condition and performance. Typically, the dis-
placements that describe track movement under
train loads are derived from velocity or accelera-
tion signals. However, signal processing artifacts
and wheel-to-wheel variability make interpreting
these measurements challenging. Consequently,
track deections are often inspected rather than
analyzed through systematic methods, limiting
the practical utilization of track vibration data for
condition or performance monitoring. Their study
introduces a novel approach that leverages the cu-
mulative distribution function of track deection
to identify the at-rest position and interpret typi-
cal track movement based on displacement data.
This technique can correct at-rest position shifts
in velocity or acceleration data, determine the
ratio of upward and downward movement, and
align data from multiple sensors to a common
reference point to visualize deection relative to

374
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
distance along the track. By automating the char-
acterization of track displacement, this method
enables the utilization of large volumes of track
vibration data for condition monitoring.
Tešić et al. [53] introduced the Vehicle/Track
Interaction Monitor (V/TI Monitor), a modern
system that employs accelerometers to assess the
railway infrastructure’s quality. This technology
captures and characterizes the dynamic behavior
of vehicles interacting with the track. The VTI-
TQI software was created to identify and analyze
numerous common scenarios in rail transporta-
tion in which various irregularities coincide. This
system can also predict the future progression and
expansion of these irregularities by generating
appropriate deterioration trends.
Roth et al. [54] illustrated how positioning
concepts can facilitate the real-time condition
monitoring of railway tracks. In their study, they
emphasize that precise georeferencing of monitor-
ing data is achievable through the fusion of GNSS
and IMU measurements with a railway network
map. Their approach addresses this oine posi-
tioning challenge in two stages. They begin by
estimating path hypotheses based on GNSS data
and the railway map. Next, they employ a non-
linear Rauch–Tung–Striebel smoother to provide
on-track positions and speeds in path coordinates.
These methods are integral to a track condition
monitoring system developed at DLR and have
been validated using actual data from the harbor
railway network in Braunschweig, Germany.
Tam et al. [55] introduced a railway health
condition monitoring system founded on an optical
ber sensing network. This system has the poten-
tial to enable predictive maintenance in the railway
industry. The researchers harnessed machine learn-
ing to create models capable of detecting and clas-
sifying various track defects, including rail corru-
gations, dipped weld joints, and rail crossings.
In 2019, Chia et al. [6] conducted a compre-
hensive review of the application of inertial sen-
sors, including accelerometers and gyroscopes,
for monitoring transportation asset conditions.
The study explored aspects such as sensor spec-
ications, sensor placement, and signal process-
ing techniques to provide valuable insights into
this eld, with particular focus on evaluating
the associated challenges and opportunities for
improving railroad track condition monitoring.
Lu et al. [56] developed a wireless rechargeable
sensor node system for monitoring urban rail
corrugation. Their system includes an energy
generator that uses electromagnetic induction, a
DC-DC booster converter, wireless sensor nodes
and an analysis interface using Littlewood–Paley
wavelet transform methods. They established a
vehicle–track interaction model to predict rail-
way track responses with rail corrugation. Field
testing was conducted to validate their theoret-
ical predictions, and the power consumption
of sensor nodes was assessed. Their case study
demonstrated the identication of rail corruga-
tion defects by analyzing rail acceleration signals
using Little-wood–Paley wavelet analysis.
Chandran et al. [57] developed a train-based
dierential eddy current (EC) sensor system to de-
tect rail fasteners. This system uses electromagnet-
ic induction, with an alternating current-carrying
coil to create an EC on the rail and nearby conduc-
tive materials, and a pick-up coil to measure the
resulting eld. Their paper describes the theoreti-
cal background and application of this EC sensor
system for rail fastener condition monitoring, with
experimental results from both laboratory and eld
measurements. Field tests were conducted along a
heavy-haul railway line in Sweden, and the results
showed that the method could detect individual
fastening systems from a height of 65 mm above
the rail. The system also identied missing clamps
within fastening systems by analyzing a time-do-
main feature of the measurement signal.
Hamadache et al. [7] conducted a compre-
hensive review of fault detection and diagnosis
(FDD) techniques for railway switch and crossing
(S&C) systems. These complex systems involve
various components and technologies, making
them susceptible to failures and malfunctions that
may disrupt railway operations and safety. Their
paper serves as an overview of the current state
of FDD methods for railway S&C systems, pro-
viding valuable insights for researchers, railway
operators, and experts. It aims to facilitate the de-
velopment and adoption of eective FDD tech-
niques, contributing to the advancement of condi-
tion-based maintenance and the safe operation of
high-speed trains in the railway industry.
Farkas et al. [58] conducted a study driven
by the increasing frequency of passenger and
freight trains globally, which emphasizes the need
for more extensive track monitoring and rail in-
spection. They focused on the wheel–rail contact
forces, which result from both static axle loads
and dynamic eects stemming from track super-
structure vibrations, as signicant contributors to
railway track degradation. Their research delved

375
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
into measurements of track irregularities, which
are traditionally used to assess the current condi-
tion and quality of railway lines, with a particular
focus on compact inertial measurement systems
(IMUs). This paper explored the components,
installation, and fundamental measures of track
quality using motion sensors such as accelerom-
eters and gyroscopes, which were strategically
placed on the vehicle. It also briey touched upon
the basics of inertial navigation and the kinemat-
ics of translational and rotational train motions for
obtaining orientation, velocity and position data.
In 2020, Bhardwaj et al. [59] developed a con-
tinuous condition monitoring system to detect and
locate irregularities in railroad tracks using inertial
sensors on revenue service trains. They addressed
the challenges of inaccurate geospatial position es-
timates from GPS receivers and non-uniform sensor
sampling, which introduced noise and reduced sig-
nal strength. To improve the signal-to-noise ratio,
they introduced a method suitable for various signal
ltering approaches. By analyzing the frequency
window of the energy and variance of ensemble av-
eraged Fast Fourier Transforms (FFTs), the method
determined the best cut-o frequency for ltering.
Their results showed that applying a low-pass nite
impulse response lter with the selected cuto fre-
quency improved the signal-to-noise ratio, demon-
strating the method’s eectiveness and practicality.
Aung et al. [60] aimed to address the trac
congestion issues in Yangon, which have inten-
sied due to a rapid increase in the city’s pop-
ulation and car numbers. They focused on the
importance of maintaining railway tracks for e-
cient rail transportation as a solution. To monitor
rail track conditions for early damage detection,
they employed onboard sensor measurements us-
ing a smartphone’s accelerometer, which could
sense irregularities during train travel. They also
harnessed satellite image analysis, particularly
phased-array-type L-band synthetic aperture radar
images, to detect rail track irregularities using the
interferometric technique. This approach allowed
for eective estimation of rail track conditions.
Liu and Markine [61] investigated the reasons
behind the rapid deterioration of a railway cross-
ing. They utilized sensor-based instrumentation
to assess the crossing’s dynamic performance and
validated their ndings using a multi-body sys-
tem (MBS) model that simulates the interaction
between vehicles and the crossing. Through eld
inspections, measurements, and simulations, they
identied that the rapid degradation of the crossing
was primarily attributed to high wheel–rail impact
forces caused by the hunting motion of passing
trains. Furthermore, they revealed that track geom-
etry misalignment ahead of the crossing triggered
this hunting motion. These ndings emphasized
that crossing degradation might not solely originate
from issues within the crossing itself but could also
stem from problems in nearby track structures. The
study’s outcomes were integrated into a railway
crossing condition monitoring system, allowing
for timely and targeted maintenance actions.
Wilk et al. [62] addressed the challenge of dig-
ital ltering in the context of mobile global navi-
gation satellite system (GNSS) measurements to
accurately determine railway track coordinates.
They introduced a measurement technique em-
ploying a platform equipped with multiple stra-
tegically placed GNSS receivers, two of which
determined the directional base vector of the
platform. These receivers operated in real-time
kinematic (RTK) mode with a high measure-
ment frequency and allowed result correction in
post-processing. The article also delved into the
assessment of measurement quality from GNSS
receivers and their preparation for further process-
ing, which involved geometrically constrained pa-
rameters of the base vector and specialized digital
ltering. Their results conrmed the eectiveness
of this GNSS signal processing approach.
In 2021, Shah et al. [63] conducted a study
focused on enhancing fault diagnosis in railway
condition monitoring. They aimed to improve
the traditional methods used in underdeveloped
countries, which rely on manual extraction of
track data using push trolley/train-based track
recording vehicles (TRV). This manual process
is labor-intensive and subjective, impacting the
accuracy of fault diagnosis. To address this is-
sue, the researchers introduced a prototype called
“Muhaz,” an automated and portable TRV with
a novel design based on axle-based acceleration
methodology. Through site-specic experimenta-
tion, they demonstrated that Muhaz is 87% more
ecient than the traditional push trolley-based
TRV system, signicantly improving railway
condition monitoring and safety.
Kukulski et al. [64] aimed to diagnose the con-
dition of continuous welded tracks, with particular
focus on measuring rail displacements during oper-
ation. Their work addressed the inuence of thermal
stresses on rails, which are aected by temperature
changes and climatic conditions. They introduced
an original and eective analytical method based

376
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
on experimental research to diagnose the track’s
condition and provide recommendations for repair
or maintenance. Two scenarios were considered:
tracks under load and tracks without load. The au-
thors derived empirical formulas for calculating rail
temperature and longitudinal force based on ambi-
ent temperature, particularly for straight tracks with
60E1 rails, no engineering structures, conventional
surfaces, wooden sleepers, and high train trac.
The results demonstrated very high accuracy, with
a correlation coecient (R2) of ≥ 0.995, which val-
idated the precision of their proposed method.
Zvolenský et al. [65] conducted research with
a focus on railway vehicle maintenance for safe-
ty, passenger comfort and cost-eciency. Their
study explored the application of acoustics to di-
agnose the technical condition of railway vehicles
while they are in operation. They monitored the
condition of individual carriages and compared
their noise levels. The paper also discussed poten-
tial practical applications for eectively reducing
noise in railway carriage operations.
Balcı et al. [66] explored the integration of ad-
vanced technologies such as articial intelligence
(AI), the internet of things (IoT) and big data in
the context of railway systems. They emphasized
the importance of eectively detecting track faults
and conducting maintenance to ensure the safety
of railway operations. The researchers highlighted
the current use of image processing and machine
learning for automated track inspections but iden-
tied shortcomings regarding the integration of
these technologies into railway tracks. Their work
compared traditional and smart approaches to track
inspection and maintenance, pinpointing areas in
need of improvement. The researchers also dis-
cussed the potential impacts of using smart sys-
tems on the overall life cycle of railway structures.
In 2022, Daniyan et al. [67] developed an in-
spection and diagnostic robot for enhancing rail
infrastructure integrity. This robot was designed
to detect various issues such as cracks, corrosion,
missing clips and wear on rail tracks. It incorpo-
rates infrared and ultrasonic sensors for obstacle
avoidance and crack detection, 3D prolometers
for wear detection, high-resolution cameras for re-
al-time imaging and colour sensors for corrosion
detection. The robot’s image processing capabili-
ties allow in-depth analysis of detected cracks and
corrosion. The study involved computer-aided
design and modeling of the robot, and simulation
in MATLAB 2020b. The results presented frame-
works for wear, corrosion, missing clips, and crack
detection, along with design data for the integrated
robotic system. The simulation results demonstrat-
ed the system’s signicant sensitivity and accura-
cy in fault detection. The work introduces a novel
autonomous system for proactive rail track mon-
itoring and defect detection, with the potential to
increase rail network capacity and availability.
Bolshakova et al. [68] conducted an analysis
of small-sized inertial measuring systems that
utilize sensors placed near the wheel/rail contact
area. This design enables the simultaneous mon-
itoring of mechanical and acoustic eects gener-
ated by the passage of faulty track elements using
a single device. The study examined the mechan-
ical and acoustic characteristics of the car/rail
track system and explored methods for kinematic
analysis and vibroacoustics, which can be applied
to assess the condition of structural components
in both the railcar and rail track.
In 2023, Tsunashima et al. [69] developed a
system for monitoring railway track conditions,
which is crucial for ensuring safety. Tradition-
al track maintenance on regional railways faces
nancial challenges and a lack of manpower. To
address this, the researchers created a diagnostic
system involving onboard sensors on in-service
vehicles that measure car body vibration accelera-
tion to assess the condition of the track. Long-term
measurements were conducted, and changes in
track conditions were evaluated using the vibration
data. The study successfully identied degraded
track sections and demonstrated the system’s eec-
tiveness in conrming track maintenance results.
They also proposed a method for improving train
position determination using yaw angular veloci-
ty and introduced a technique for more precisely
assessing track condition through time–frequency
analysis of car-body vertical acceleration. Zanelli
et al. [70] addressed the critical need for railway
infrastructure monitoring to ensure transportation
reliability and safety. While diagnostic trains are
typically used for this purpose, they are infrequent
on main lines. The researchers developed a wire-
less system capable of monitoring vehicle dy-
namics and detecting potential track issues in real
time. They achieved this by analyzing acceleration
RMS values within specic frequency ranges, us-
ing data collected during extended experimental
campaigns. This method enables continuous mon-
itoring and oers an aordable and easily installed
solution for freight wagons. Ultimately, the sys-
tem can enhance maintenance strategies for con-
ventional lines, along which diagnostic train runs

377
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
occur at long intervals. In 2024 Pal and Datta [71]
study tackles the challenge of timely fault detection
in railway tracks, crucial for safety and eciency.
Traditional methods are complex, and recent AI
approaches lack a systematic weighting system for
optimal performance.
This research bridges the gap by assigning
weights to key AE signal parameters (ampli-
tude, frequency, etc.) based on their importance.
This guides an articial neural network (ANN)
model to focus on critical data points for im-
proved fault detection accuracy. Extensive testing
Table 1. Types of measurements/method used for track diagnostics
Main classication Type of measurement/method References
Onboard
Accelerometer
Kojima et al. [13], [14]
Mizuno et al. [21]
Lee et al. [24]
Lee et al. [25]
Chellaswamy et al. [28]
Bagshawe [29]
Chellaswamy et al. [30]
Qin et al. [31]
Tsunashima et al. [32]
Tsunashima et al. [38], [39]
Xu et al. [43]
Balouchi et al. [44]
Chellaswamy et al. [45]
Groos et al. [47]
Chudzikiewicz et al. [48]
Känsälä et al. [51]
Teši et al. [53]
Roth et al. [54]
Shah et al. [63]
Zanelli et al. [70]
+ gyroscope
Weston et al. [15], [18]
Ward et al. [23]
Yeo et al. [34]
Seraj et al. [46]
Chia et al. [6]
Farkas et al. [58]
Bhardwaj et al. [59]
Aung et al. [60]
Bolshakova et al. [68]
Tsunashima et al. [69]
+ microphone Hayashi et al. [17]
Mori et al. [22]
Optical/visual Attivissimo et al. [19]
Reiterer et al. [33]
Others Chandran et al. [57]
Wilk et al. [62]
Track side
Strain gauge Bogue [12]
Fiber optic
Naderi and Mirabadi [16]
Ho [20]
Roveri et al. [35]
Ilie and Stancalie [40]
Lienhart et al. [42]
Zhang et al. [49]
Tam et al. [55]
Force Tesfa et al. [26]
Kukulski et al. [64]
Vibration
Milne et al. [52]
Lu et al. [56]
Liu and Markine [61]
Others Maddison and Smith [41]
Muthukumar and Nallathambi [4]
Others works
Nigi et al. [27]
Hodge et al. [36]
Ngamkhanong and Kaewunruen [50]
Hamadache et al. [7]
Balcı et al. [66]
Daniyan et al. [67]

378
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
demonstrates the eectiveness of this approach
in both lab simulations and real-world scenarios.
Table 1 categorizes track diagnostic methods into
onboard (mounted on trains) and trackside (xed
to the track) systems. It details specic tech-
niques like accelerometers, ber optics, and vi-
bration analysis, along with references for further
exploration.
Rolling stock diagnostic systems can be di-
vided into three main categories: wheels, bo-
gies, and other. Wheel surface damage and wear
detection systems are among the most well-
researched topics. Recently, there has been in-
creasing focus on suspensions systems, as well
as diagnostics of freight wagons and other loco-
motive or wagon sub systems.
Table 2 provides a comprehensive compari-
son of various diagnostic methods, highlighting
their respective strengths, weaknesses, and ideal
applications in the context of railway system
maintenance and monitoring.
Wheels
In the early 1980s, Battelle-Columbus Labo-
ratories in the United States designed a dynamic
vertical wheel loads measurement system. AM-
TRAK installed the initial version of this sys-
tem in mid-1983 along the North East Corridor
to identify the vertical forces generated by train
wheels. This equipment eectively pinpointed
and agged dynamic wheel loads exceeding a
predened threshold, which was caused by ir-
regularities in the train wheels as they passed by
[72]. The measurement systems were microcom-
puter-based and were designed for monitoring,
analyzing and reporting dynamic forces between
train wheels and rails. When the forces generated
by passing train wheels exceeded user-dened
thresholds, the system identied the axle posi-
tion within the train and transmitted information
via a modem link to a remote terminal, simulta-
neously triggering an audible alarm. The system
compiled and stored statistical tables, includ-
ing peak, mean, and dynamic forces, as well as
peak-to-mean ratios for all passing wheels. Ver-
tical wheel forces were measured using resis-
tance strain gauges xed to the web of the rail.
Forces were measured at eight sequential crib
locations. Each measurement circuit comprised
four shear gauge pairs linked to a front-end pro-
cessor responsible for recognizing and transmit-
ting the highest force values from each succes-
sive wheel to a central computer. The central
computer, in turn, identied and communicated
any forces that surpassed pre-established limits
while also keeping statistical records up to date.
There were ve available threshold levels for
peak force, dynamic force, and peak-to-mean ra-
tio. An audible alarm at the reporting destination
Table 2. Summarizing the key characteristics, advantages, and disadvantages of each diagnostic method
Diagnostic method Advantages Disadvantages Suitability
Track geometry
measurement systems
High accuracy, detailed
track prole data, suitable
for preventive maintenance
planning.
Expensive, requires dedicated
vehicles or trackside
equipment, may disrupt
operations.
Regular inspections, post-
maintenance evaluation.
In-service vehicle monitoring
systems
Cost-e󰀨ective, utilizes existing
rolling stock, provides real-time
data.
Lower accuracy compared
to dedicated systems, data
interpretation requires
advanced algorithms.
Continuous monitoring,
early detection of track
irregularities.
Fiber optic sensors
High sensitivity, long-range
monitoring, immunity to
electromagnetic interference.
Relatively new technology,
requires infrastructure
installation, data analysis
complexity.
Monitoring critical
structures (bridges,
tunnels), detection of
internal cracks or stress.
Onboard monitoring systems
Continuous monitoring, real-time
data on component health, fault
detection and isolation.
Requires installation of sensors
on various subsystems, data
interpretation complexity,
potential for false alarms.
Early detection of
component degradation,
preventive maintenance
scheduling.
Periodic inspections
Thorough examination, allows
for visual and manual checks,
identication of wear and tear
beyond sensor capabilities.
Time-consuming, disruptive to
operations, relies on inspector
expertise.
Compliance with
regulations, detection of
advanced failures not
captured by sensors.
Vibration analysis
Non-intrusive, identies issues
within bearings, gears, and other
rotating components.
Requires specialized
equipment and expertise for
data interpretation, may not
pinpoint the exact location of
the fault.
Monitoring component
health trends, detection of
developing problems.

379
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
was activated when any combination of these
thresholds was exceeded [72]
Samuels and Palesano [73] presented the
statistical analyses and assessed the ecacy of a
computerized detection system for determining
wheel impacts on a primary rail line implement-
ed by Conrail in 1986. Information collected
from the detector was utilized to analyze the sta-
tistical characteristics of wheel impacts for vari-
ous types of trains, including trailer vans, coal
transports and mixed freight trains. Instances
of wheel impacts surpassing specic threshold
levels for dynamic incremental loads and peak
impact loads were employed to identify wheel
sets on railcars in need of inspection. Inspection
data revealed that the utilization of dynamically
measured wheel impact loads is an exception-
ally ecient method for pinpointing inspection
eorts toward wheel sets with a high likelihood
of having condemnable defects.
In 1999, Lechowicz and Hunt [74] described a
wheel impact load detector (WILD) that allowed
rail and infrastructure owners to track the per-
formance of vehicles through precise in-motion
weighing, thorough analysis of load distribution
and the detection and categorization of defects at
the wheel, bogie, wagon and train levels.
Over 200 instances of wheel ats were inten-
tionally created under controlled conditions and
investigated by Jergéus et al. [75] in the form of
on-site tests involving a moving train. The study
included varying wheel loads, train speeds, slid-
ing durations and adjustments to the friction coef-
cient between the wheel and rail. Samples were
extracted from the aected wheel surfaces and
subjected to detailed metallographic analysis to in-
vestigate phase transformations and the presence
of cracks. Additionally, a numerical model for
predicting wheel ats was both qualitatively vali-
dated and quantitatively ne-tuned through these
experiments. The ndings revealed the consistent
presence of martensite beneath all ats, and cracks
were observed in most cases. Consequently, it is
recommended that the potential for future spalling
be considered for all wheelsets with ats.
In 2001, Danneskiold-Samsøe and Ramkow-
Pedersen [76] presented a wheel prole diagnos-
tic system based on an array of accelerometers,
laser scanners and cameras which, coupled with
calculation algorithms, enabled the diagnosis
of various defects such as wheel ats, out-of-
round wheels and corrugated wheels. Substantial
cost savings and potential enhancements, wheel
maintenance expenses, wheel inspection costs,
track maintenance expenditures, component re-
pair costs, as well as expenses associated with
noise and vibration reduction were identied.
In 2003, Johansson and Nielsen [77] explored
the impact of various forms of railway wheel ir-
regularities on the vertical dynamic forces be-
tween the wheels and the tracks, as well as the
overall track response. Their investigation in-
volved a wide-ranging set of experiments and
numerical simulations. Analysis of freight trains
equipped with dierent severe wheel tread dam-
ages, covering wheelats, local spalls resulting
from rolling contact fatigue cracking, extended
local defects, and polygonal wheels was con-
ducted. Some of the eld tests considered mea-
surements of vertical wheel-rail contact forces
through a strain gauge-based wheel impact load
detector. The track response was recorded using
strain gauges and accelerometers on rails and
sleepers. The collected data were used for cali-
bration and validation of numerical models used
to simulate the interaction between trains and
tracks. The study also includes a comparison of
results between a linear track model and a state-
dependent track model.
In 2007, Attivissimo et al. [19] assessed real-
time wheel-rail interaction quality by employing
a railway measurement system. The aim of this
study was to evaluate the equivalent conicity, a
parameter dened by the international Union
of Railways (UIC) 518 Standard. The proposed
measurement system processes geometric data
acquired through a contactless optical unit. The
verication of this measurement system has been
conducted according to the procedures outlined in
the UIC 519 Standard, particularly with respect to
the necessary measurement precision.
In 2011, Thakkar et al. [78] presented the in-
vestigation of acoustic emission (AE) generated
during the interaction between rails and wheels.
They simulated wheel defects with the aim of
developing methods for diagnosing wheel ats
in situ utilizing sensors mounted on the rails. A
series of experiments was conducted on a scaled
testing apparatus, featuring a single wheel moving
along a circular track. During these experiments,
AE was recorded at a xed location on the track.
The wheel had three unevenly distributed ats
machined onto it, and the rolling speed and axle
load varied. A previously established time-based
analytical model was employed to reconstruct the
typical pattern of a wheel traveling around the

380
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
track. A frequency-based pulse-train model was
employed to establish a method for matching the
observed spectra with the expected pulse train
spectra, which resembled the concept of defect
frequencies in bearings. Across the full range
of conditions, all the enveloped time-series data
exhibited a discernible pattern of ats, which
was reasonably reproducible between consecu-
tive rotations. The authors concluded that, with
appropriate calibration, the proposed method
could be employed to diagnose wheel defects
in actual railway wheels using either track- or
wheel-mounted sensors.
Also in 2011, Wei et al. [79] presented a real-
time diagnostic system that uses Fiber Bragg grat-
ing sensors to monitor wheel defects. It measures
and processes the track strain response during
wheel–rail interactions to generate a condition in-
dex that directly indicates the state of the wheels.
This approach has been veried via extensive
eld tests, and the initial results indicate that
this system, which is immune to electromagnetic
interference, oers an eective alternative for
wheel defect detection. It signicantly improves
maintenance management eciency, reduces de-
tection costs and, most importantly, helps prevent
derailments in a timely manner.
In their 2014 study, Asplund et al. [80] indi-
cated that wayside monitoring systems along the
rails identify issues such as wheels exceeding de-
ned safety limits, but can only identify such prob-
lems at specic points along the track. As a result,
damage may occur on the track before the system
detects faults at its location. The research team ex-
amined the wheel prole parameters measured us-
ing a wayside wheel prole measurement system
installed along the rail and correlated these mea-
surements with warning and alarm signals from
a wheel defect detector installed on the same rail
line. The research demonstrates that an increase in
wheel wear, which can be detected through chang-
es in the wheel prole parameters, could help re-
duce the risk of capacity-limiting wheel defect fail-
ures and the subsequent reactive measures.
In 2014, Papaelias et al. [81] presented a nov-
el condition monitoring system centered around
high-frequency acoustic emission and vibration
analysis, which has been integrated onto a train.
The diagnostic system leverages cost-eective and
durable acoustic emission sensors and accelerom-
eters, ensuring straightforward installation on the
axle bearing box with minimal disruption. Em-
pirical testing conducted in re-al-world conditions
established that the developed system can detect
defects associated with wheels and axle bearings.
In 2015, Asplund et al. [82] introduced a tech-
nique for evaluating the quality of data collected
from a condition monitoring system designed to
assess the condition of rolling stock wheels. The
focus of this research was to determine whether
the data meet the necessary standards for further
analysis. Simultaneously, variations between
measurement units within the same system were
investigated and potential correlations between
dierent measurements of wheel parameters,
speed and time were identied. The evaluation of
data quality was achieved through the dimension
of freedom from error. Two sources of data were
analyzed: an automated wheel prole measure-
ment system and a manually operated wheel pro-
le measurement device. The manual measure-
ments of wheel proles validated the accuracy of
the automated wheel prole measurements. The
results reveal certain inconsistencies, suggesting
that this system has poor accuracy. This indicates
a need for internal calibration or self-adjustment
to enhance its quality.
In their 2016 study, Alemi et al. [83] proposed
a wheel diameter measuring system based on
WILD. With knowledge of the wheel diameter,
recorded impacts were correlated with specic
positions on the wheel’s circumference. A novel
conguration of strain sensors was proposed
alongside an algorithm for data ltering and pro-
cessing. A series of simulations were conducted to
explore the impact of various parameters, includ-
ing the number of sensors, lter thresholds, defect
sizes and sensor noise. The principal outcome of
this research conrms the WILD’s capability for
monitoring wheel diameter and its eectiveness
in monitoring multiple aspects of wheel condition
within a single system.
In 2018, Song and Sun [84] investigated the
inuence of polygonal wheels on vehicle dynam-
ics and monitoring methods to enhance the safety
and service quality of high-speed railway transpor-
tation. A measurement system that identies wheel
polygons based on wheel–rail (W/R) contact force
measurements using PVDF strain sensors was
established in the railroad network, demonstrat-
ing long-distance capabilities and high stability.
Utilizing the W/R contact forces resulting from
wheel polygon impacts, the data from the PVDF
strain sensors were processed to create an indica-
tor of wheel polygon. An automatic remote con-
dition monitoring system was developed for use

381
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
under GSM-R transmission lines. Preliminary ex-
perimental results indicated that this system is well-
suited to the complex electromagnetic environment
and stability requirements of high-speed railways.
Xu et al. [85] 2018 introduced a data-driven
approach for monitoring the wear and tear of
high-speed train wheels by utilizing onboard vi-
bration sensors. Their method was tested using
real operational data gathered from high-speed
trains in China over a six-month period. The ini-
tial results demonstrated the accuracy and practi-
cal applicability of the method in real-world sce-
narios. A real-time wear predictive model, which
was based on short-time fourier transform (STFT)
and principal component analysis (PCA) using
onboard vibration signals, was proposed and vali-
dated during on-tracking test trials. Strong corre-
lation between vibration signals and the dynamic
performance of the railway system was indicated,
emphasizing the sensitivity of these signals with
regard to predicting wear.
In 2019, Ni and Zhang [86] introduced a
Bayesian statistical method for assessing wheel
conditions with track-side monitoring. Data from
monitoring were used to extract wheel quality-
related components, and their Fourier amplitude
spectra were adjusted to create a series of cumula-
tive distribution functions that represented wheel
quality characteristics. Subsequently, a data-driven
reference model was established through Bayes-
ian learning to model characteristic functions for
wheels in good condition. The Bayes factor was
then employed to dierentiate new observations
from the reference model, enabling a real-time
quantitative assessment of wheel conditions. To
test the feasibility and eectiveness of this ap-
proach, strain monitoring data from rail bending,
collected via an optical ber sensor-based track-
side monitoring system, was implemented for
validation. In 2020, Gao et al. [87] developed a
diagnostic system based on a reective optical po-
sition sensor which enabled dynamic and quanti-
tative at detection during high-speed train travel.
The system incorporates two sensors positioned
alongside the rails to assess the wheel–rail impact
force throughout the wheel’s entire circumference
by monitoring the displacement of a collimated
laser spot. To establish a quantitative relationship
between the sensor readings and at wheel length,
a vehicle-track coupling dynamics analysis model
was developed using nite element and multi-body
dynamics methods. The model considered various
factors such as train speed, load, at wheel lengths,
and impact positions to simulate their eects on im-
pact forces, with measured data normalized based
on simulation results. The system’s performance
was evaluated through simulations, laboratory
tests, and real eld trials, conrming its accuracy
and practicality. This system not only identies at
wheels but also quanties the extent of the detected
at, oering a wide range of potential applications.
In 2021, Mosleh et al. [88] focused on wheel
ats wayside monitoring system sensors with
the aim of determining their optimal placement.
Specic measurement points for shear and ac-
celeration were established to explore how dif-
ferent sensor types (strain gauge and accelerom-
eter) and their installation locations impact lay-
out schemes. Using shear and acceleration data
collected from 19 track positions as inputs, the
presence of wheel ats was identied through
the envelope spectrum approach using spectral
kurtosis analysis. The research delves into the
impact of sensor types and their positions on the
accuracy of the wheel at detection system.
In 2024 Chung and Lin [89] a study on wheel
condition monitoring during train operation
highlighted its importance in preventing unex-
pected events. The research involved installing
piezoelectric sensors on railway tracks to col-
lect dynamic voltage-and-strain signals gener-
ated when train wheels passed over them. These
one-dimensional time series signals were trans-
formed into two-dimensional Recurrence Plots
images, which served as input data sets for two
deep learning models: Xception and Ecient-
Net-B7. The models performed binary classi-
cation to indicate the health state of train wheels
as either Normal or Faulty. The performance of
the models was evaluated using ve metrics: Ac-
curacy, Precision, Recall, Miss Rate, and AUC.
The ndings demonstrated high classication
accuracy (91.1%) for both models. However,
EcientNet-B7 outperformed Xception in terms
of Recall, Miss Rate, and AUC, making it a
more suitable classier for identifying defective
wheels. This research signicantly contributes
to train wheel condition monitoring and health
management by providing eective diagnostic
information for maintenance decisions, thereby
reducing the occurrence of unexpected events.
Bogies
Selinger et al. [90] presented a novel bogie de-
signed for freight wagons with an integrated life

382
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
cycle unit (LCU). This LCU enables the collection
of operational data and the measurement of stress
on wear components. These data can then be uti-
lized to assess the actual lifespan of the bogie’s
components and streamline relevant maintenance
procedures. The approach was veried with dier-
ent stress cases and implemented in a testbed. In
2005, Goda and Goodall [91] presented a railway
vehicle bogie suspension faults detection system
based on a model estimation approach involving a
Kalman–Bucy lter and an isolation scheme. The
authors reported that the main advantage of the
system is its ability to detect numerous faults with
only a limited number of sensors. The simulations
demonstrated that the Kalman–Bucy lter’s re-
sidual can eectively detect suspension faults and
changes in the system for railway vehicle bogies.
In 2007, Li et al. [92] presented the estima-
tion of parameters for railway vehicle suspen-
sions with the aim of facilitating condition-based
maintenance. A simplied plan-view model of
railway vehicle dynamics was presented and a
Rao-Blackwellized particle ltering (RBPF) tech-
nique was employed for parameter estimation.
Computer simulations were conducted to evalu-
ate the accuracy of parameter estimation across
various sensor congurations and its resilience to
uncertainties related to the statistical properties of
random track inputs. Their method was validated
using real test data obtained from a Coradia Class
175 railway vehicle equipped solely with bogie
and body-mounted sensors.
In 2009, Mei and Ding [93] presented an in-
novative approach to detect faults and monitor
the condition of vehicle suspensions. Their meth-
od, which is based on cross-correlations between
the measurements from bogie-mounted cost-ef-
fective inertial sensors, utilizes the dynamic in-
teractions between various vehicle modes result-
ing from component failures. It has proven to be
highly sensitive when it comes to distinguishing
dierent fault conditions and is capable of han-
dling complex dynamic and nonlinear systems.
Although the vertical primary suspensions are the
subject of this study, the authors indicated that the
technique may be also applied to detect faults in
lateral primary suspensions and in secondary sus-
pensions and could possibly be extended to moni-
tor the conditions in other dynamic systems with
symmetrical congurations.
In 2010, Ward et al. [94] assessed the utili-
zation of condition monitoring to identify sus-
pension component status, detect low-adhesion
conditions and evaluate the condition of the
wheel–rail interface. The main aims of the pre-
sented research focused on generally employed
techniques, inexpensive sensors and advanced
ltering, which could feasibly be applied to ev-
ery bogie and wheelset in a train formation. The
authors investigated the possibility of develop-
ing real-time condition monitoring algorithms for
safety-critical components within the rail vehicle
bogie system by monitoring suspension damp-
ers using a Rao Blackwellized particle lter and
validation through network data. The authors also
discussed the necessity of developing ecient
algorithms which can be applied in real time, as
well as the challenges related to sensor placement.
In 2010, Tsunashima and Mori [95] discussed
the possibility of identifying suspension failures
in railway vehicles by employing a multi-model
approach with on-board measurement data. The
model of the railway vehicle incorporates lateral
and yaw movements of the wheelsets and bogie,
as well as the lateral motion of the body. Inertial
sensors were used to measure the lateral accelera-
tion and yaw rate of the bogie, as well as the lat-
eral acceleration of the service vehicle body. The
detection algorithm was devised based on the in-
teracting multiple-model (IMM) approach, which
integrates a method for updating the estimation
model. The IMM method has been employed in
a simulation study for detecting faults in vehicle
suspension systems. It estimates the mode prob-
abilities and states of the vehicle suspension sys-
tems through a Kalman lter (KF). The perfor-
mance of this algorithm is assessed through simu-
lation examples, and the results demonstrate its
ability to detect on-board faults in railway vehicle
suspension systems under realistic conditions.
Daadbin et al. [96] (2012) focused on the
monitoring of critical components in rail trans-
portation, such as axle vibration and pantograph
load. A system was designed for monitoring the
transmitted torque and bending moments in the
rail vehicle axle, serving as a means to identify
abnormal loading. Advancement of strain gauge
instrumentation, telemetry systems, and data log-
gers for demanding transportation applications
were discussed. A rail axle stress monitoring
(RASM) consisting of six data loggers, synchro-
nized across all channels, was used, and each data
logger was congured to perform both Rainow
Count (which counts fatigue or acceleration cy-
cles) and to record the 100 highest events in the
time domain. By collecting synchronized data

383
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
from axle stress and axle box acceleration, stress
occurrences can be associated with vehicle opera-
tion, allowing the investigation of dynamic inter-
action between the wheels and the rail.
In 2016, Amini et al. [97] presented the out-
comes of high-frequency acoustic emission mea-
surements conducted on freight rolling stock in
Long Marston UK. Intentional damage was in-
duced in axle bearings and acoustic signals were
recorded during lab and eld testing. Passive
high-frequency resonant piezoelectric acoustic
emission transducers and time spectral kurtosis
were employed for acquisition and analysis of
the acoustic emission data. The results demon-
strated that time spectral kurtosis can eectively
distinguish axle bearing defects from the back-
ground noise generated by various sources, such
as the wheel–rail interaction, braking and chang-
es in train speed. Extraneous noises, e.g., noises
caused by the braking system, can increase the
RMS value and cause a slight increase in kurto-
sis, potentially leading to false defect identica-
tions. The authors demonstrated that using time
spectral kurtosis, which incorporates time and
frequency domain information, signicant en-
hancement in the ability to detect bearing defects
was observed in such scenarios.
Los Santos et al. [98] (2017) investigated the
rate at which surface defects in railroad bearings,
specically spalls, grow per mile of full-load op-
eration. The presented data were gathered from
defective bearings subjected to various load and
speed conditions using specialized railroad bear-
ing dynamic test rigs operated by the University
Transportation Center for Railway Safety (UT-
CRS) at the University of Texas Rio Grande Val-
ley (UTRGV). Periodic removal and disassembly
of these railroad bearings was carried out to in-
spect and measure the size of defects at the outer
ring (cup). Castings of spalls was introduced us-
ing low-melting, zero-shrinkage Bismuth-based
alloys to record and investigate spall geometry.
Spalls were measured using optical techniques
combined with digital image analysis, as well as
manual coordinate measuring instruments. The
study determined the spall growth rate in terms
of area per mile of full-load operation. Initially,
the size of spalls was randomly distributed, de-
pending on the originating defect’s depth, size,
and location on the rolling raceway. The surface
spalls had two distinct growth regimes, with an
initial slower growth rate that accelerated once
spalls reached a critical size. While there was
signicant variability, upper and lower bounds
for spall growth rates were proposed, and the
critical dimension for the transition to rapid
spall growth was estimated. A preliminary mod-
el for spall growth was dened for the purpose
of initial detection of bearing spalls with condi-
tion monitoring tools.
In 2018, Li et al. [99] presented a concept of
an onboard health monitoring system tailored for
heavy haul wagons. Their system encompasses a
signal-based fault detection and isolation (FDI)
method along with a real-time fault diagnosis ap-
proach. The proposed monitoring system imple-
ments two accelerometers, which are positioned
on the front left and right rear of each car-body
within a heavy haul train to detect faults in the
bolster springs. The authors investigated the
impact of these faults and their detectability by
conducting simulations on straight and curved
tracks, utilizing a model of a heavy 40 ton axle
load haul wagon. The simulation results were
analyzed and compared using cross-correlation
techniques, leading to the proposal of a fault de-
tection and isolation system which introduces
ve potential fault indicators, demonstrating the
feasibility of detecting changes in bolster sti-
ness within a range of ± 25%.
In 2019, Chudzikiewicz et al. [100] presented
the process of selecting specic points on a rail
vehicle to record signals intended for monitoring
the vehicle’s condition, especially the rst and
second-degree suspension system elements. Ini-
tially, seven possible points for signal registration
on a rail vehicle were dened and subsequently
reduced to four. The statistical measures used to
evaluate the suitability of recorded signals were
dened. Practical implementation of a prototype
system on the ED74 type rail vehicle conrmed
the validity of the initial assumptions and enabled
an assessment of the dened diagnostic indicators.
In 2019, Tarawneh et al. [101] indicated that
rolling contact fatigue (RCF) was a signicant
factor behind the failure of railroad bearings used
in freight service. The authors’ primary aim was
create dependable prognostic models for spall
growth in railroad bearings based on actual ser-
vice life testing. The data used for these models
came from laboratory and eld tests. Empirically
observed spall growth patterns in both the inner
and outer rings of bearings, even when subjected
to varying conditions such as speed, load, and
temperature, indicate that a straightforward em-
pirical model for spall area expansion can serve

384
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
as the basis for predicting the remaining useful
life of a bearing when a spall is detected.
In 2020, Sánchez et al. [102] presented a
method for evaluation of condition indicators for
railway axles crack detection in condition-based
monitoring. Authors assessed the vibration sig-
nals captured by accelerometers placed along
the longitudinal direction and implemented data
fusion technique, involving the evaluation of six
accelerometers and the merging of condition in-
dicators based on sensor placement. Fifty-four
condition indicators were calculated for each
vibration signal, with the best features selected
using the Mean Decrease Accuracy method of
Random Forest. The chosen indicators were
tested using a K-Nearest Neighbor classier. A
real bogie test bench was utilized to simulate
crack faults in railway axles, with vibration sig-
nals measured on both the left and right sides
of the axle. Tests highlighted the performance of
condition indicators and demonstrated the eec-
tiveness of fusing condition indicators for crack
fault detection in railway axles.
In 2022, Zanelli et al. [103] proposed a wire-
less monitoring system designed for diagnosing
freight train brake systems. Their system incorpo-
rates a low-power architecture that focuses on en-
ergy harvesting and wireless communication and
allows for the collection of brake pressure data
at critical points to verify the system’s operation
parameters. The study also presents experimen-
tal results obtained during a ve-month eld test
over a distance of more than 24.000 km on sig-
nicant rail routes across Europe.
A 2024 Guo et al. [104] study aimed to im-
prove evaluation criteria for hunting stability
in high-speed trains by developing a method to
identify small-amplitude bogie hunting (SABH)
motion. Using labeled eld-measured bogie lat-
eral acceleration data categorized as normal and
SABH, the study identied harmonic character-
related features, like the autocorrelation coef-
cient and approximate entropy, as crucial for
SABH identication. The decision tree classi-
er outperformed others, including support vec-
tor machine and naive Bayes. Sensitivity analy-
sis conrmed the method’s eectiveness with a
sampling frequency of 50–200 Hz and a window
length of 5–10 seconds. This research provides
valuable insights for developing improved moni-
toring and control systems to address hunting in-
stability in high-speed trains.
Locomotive systems
In 1989, Appun and Daum [105] discussed
the problem of high-power and high-speed three-
phase traction drive systems’ external and inter-
nal diagnostic methods. The authors described
external diagnostic methods and established sev-
eral less expensive methods that did not require
installing equipment on the vehicle. The authors
stated that external diagnostic methods are pri-
marily suitable for maintenance rather than op-
erational diagnosis, and employed mainly for
specic cases, such as testing objects with limited
scope and complexity. Further attention has been
directed towards internal diagnostic equipment
integrated into the vehicle, as well as potential
self-diagnostic capabilities. They also presented
typical applications of internal diagnostic sys-
tems that have already been eld-tested.
In 1995, Daley et al. [106] presented an over-
view of several rail vehicle traction and braking
systems fault diagnosis methods spanning from
classical condition monitoring to advanced mod-
el-based techniques. According to the authors,
FDI techniques promise to yield several benets,
such as decreased maintenance needs, improved
service availability and enhanced safety. Schemes
for fault-tolerant traction motor torque control,
which rely on innovative state variable observers,
were proposed. The Implementation of the bilin-
ear state space observer was reported as a feasi-
ble way to simultaneously estimate unmeasured
states and isolate faults in up to two sensors with
minimal computational overhead. Simulations of
fault detection schemes illustrated the potential
benets of the proposed approach.
In 1997, Winterling et al. [107] analyzed the
dynamic load conditions for traction drives. Fault
conditions within the electrical drive were re-
ported in peak torques or trigger oscillations that
might potentially harm the mechanical drive. To
address this concern, models of traction drives,
encompassing components such as power con-
verters, motors, mechanical drives and wheel-rail
contact, were devised to replicate drive operation
under fault conditions. The authors reported that
simulating fault scenarios in electrical drives is
essential for designing an electromechanical trac-
tion system that can operate without damaging its
mechanical components. The most challenging
conditions were observed when the stator oper-
ated at its highest frequency with rated ux and
torque while running on dry rails.

385
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
Deuszkiewicz and Radkowski [108] (2002)
investigated diagnostic parameters and methods
for classifying the condition of a power transmis-
sion unit within a commuter train. The axle box of
the vibrations and track vibrations were analyzed,
both before and after undergoing repairs. The di-
agnostic method combined diagnostic parameters
derived from the signal itself and its envelope
and proved to be suitable for online procedures.
Despite the relatively minor wear and tear on the
rolling bearings, the produced signals provided
valuable insights into the overall unit’s condi-
tion. Experiments revealed that dierent power
transmission units exhibited distinct individual
characteristics, and the proposed algorithm could
eectively dierentiate between them. An algo-
rithm used to identify an object’s technical condi-
tion using neural networks and online diagnostic
system was presented.
In 2008, Kia et al. [109] introduced a scaled
experimental platform designed for the advance-
ment of railway traction monitoring methods. The
proposed sensor system was intended to establish a
clear understanding of electromechanical interac-
tions, facilitating the investigation of mechanical
monitoring techniques using Motor current signa-
ture analysis (MCSA). A downsized experimental
setup in the form of a test-bed traction bogie served
as an initial evaluation stage for studying the typi-
cal electromechanical properties of a traction sys-
tem and establishing the fundamental relationships
between mechanical and electrical phenomena.
Due to the non-stationary nature of railway trac-
tion systems, conventional signal processing
methods were found not suitable for intensive use.
Therefore, proposed time-frequency signal pro-
cessing techniques should oer a more dependable
approach to analyzing crucial mechanical com-
ponents within a traction system. Later this year,
the same authors proposed the application of non-
invasive measurements for estimating electromag-
netic torque as a means of monitoring mechanical
torsional stresses in an induction machine drive
system operating under non-steady conditions
[110]. An experimental setup involving a 5.5 kW
squirrel-cage induction machine connected to a
one-stage gearbox was employed to validate the
approach. The authors stated that estimation of
electromagnetic torque can provide valuable in-
sights into the operational health and eectiveness
of an electromechanical system. This study imple-
mented this method to analyze torsional vibra-
tions in a gearbox under nonstationary condition,
as well as the longevity of shafts, bearings and
gearboxes in electromechanical systems. The de-
sign, development and test results of the rst fully
proven Pantograph Monitoring System, which was
regularly used on trains in the UK, was presented
in 2012 alongside a Rail Axle Stress Monitoring
system created by Daadbin et al. [94]. The Panto-
graph Damage Assessment System (PANDAS) de-
scribed in this paper is in regular use on the Class
390 “Pendolino” tilting trains operating on the
West Coast Main Line (WCML) in the UK. This
system has received Network Rail approval and is
being rolled out nationwide.
CONCLUSIONS
In conclusion, the eld of railway diagnos-
tics continues to evolve with advances in sensor
technology, data analytics, and communication
systems. Each diagnostic method—whether it’s
track geometry measurement systems, in-service
vehicle monitoring, ber optic sensors for infra-
structure, or onboard monitoring systems for roll-
ing stock—brings unique advantages and chal-
lenges to ensuring the safety and eciency of
railway operations. Track geometry measurement
systems oer high accuracy and detailed data for
preventive maintenance planning but come with
signicant cost and operational disruption consid-
erations. In-service vehicle monitoring systems
leverage rolling stock for cost-eective, real-time
monitoring, yet require sophisticated algorithms
for data interpretation. Fiber optic sensors provide
sensitive, long-range monitoring capabilities but
entail high initial costs and complex data analysis.
Similarly, onboard monitoring systems facilitate
continuous, real-time monitoring of rolling stock
health, enabling early fault detection and mainte-
nance scheduling, while periodic inspections re-
main essential for comprehensive checks and reg-
ulatory compliance, despite their time-consuming
nature and potential operational disruptions. Vi-
bration analysis oers non-intrusive insights into
component health trends but requires specialized
expertise for accurate interpretation. Looking
forward, the integration of multiple sensor tech-
nologies, advanced data analytics, and wireless
communication systems holds promise for the re-
liability, safety, and cost-eectiveness of railway
operations. By addressing current challenges and
leveraging emerging technologies, railway op-
erators can achieve more ecient maintenance

386
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
practices, predictive capabilities, and improved
overall performance, ensuring sustainable and op-
timized railway transport systems for the future.
Research on bogies, suspension systems, and
other less researched components is crucial for en-
hancing the safety and eciency of railway sys-
tems. These components play a fundamental role
in the overall performance and reliability of trains.
Bogies, for instance, are essential for ensuring
smooth and stable rides by housing the wheels and
axles, and facilitating the train’s maneuverability
on tracks. However, due to wear and tear, they are
susceptible to issues like axle fractures and bearing
failures, which can lead to derailments and signi-
cant safety hazards. Similarly, suspension systems
are vital for absorbing shocks and vibrations, pro-
viding a comfortable ride for passengers and re-
ducing stress on the track infrastructure. Failures in
the suspension system can result in uncomfortable
rides, increased wear on the tracks, and even de-
railments in severe cases. Despite their importance,
these components often receive less research atten-
tion compared to more visible elements like loco-
motives and power supply systems.
Our literature review demonstrates that rail-
way diagnostic and fault detection research is
gaining momentum. There has been a clear in-
crease in the number of research papers on this
topic since 2012. The greater part of the research
on diagnostic systems and fault detection in rail-
way systems focuses on track or wheels, with
track accounting for approximately 58% and
wheels for approximately 17% of the analyzed
works. Boogies and suspension systems of loco-
motives and fright wagons, brake systems, rail
vehicle axles and bearings are the subject of some
research papers, but not to the same extent, ren-
dering this an underexplored topic with many po-
tential research opportunities.
New research agendas focus on sophisticated
digital signal processing techniques, simplica-
tion of data acquisition systems in regards of sen-
sor count and placement. as well as the reduction
of continuously recorded data weight. Accelerom-
eters and inertial sensors are commonly used in
railway diagnostic systems as they provide robust
data that are easy to process. Visual techniques,
alongside acoustic and beroptic measurements,
are mentioned in the literature, but not the same
extent as the previously mentioned inertial sen-
sors. Big data and machine learning techniques,
coupled with properly chosen sensor and sensor
placement, can collect vast amounts of useful
data; thus, they have been identied as the opti-
mal approach to diagnostic problems in railway
systems, for both infrastructure and rolling stock.
In the realm of railway diagnostics, signicant
progress has been made through various method-
ologies and technologies aimed at enhancing the
safety, reliability, and eciency of railway sys-
tems. However, several gaps remain that present
opportunities for future research and development.
One major area for future research is the inte-
gration of advanced data analytics and machine
learning algorithms into railway diagnostic sys-
tems. Current methods generate vast amounts of
data that are often underutilized. By employing
sophisticated data analytics and machine learn-
ing techniques, it is possible to enhance the ac-
curacy and predictive capabilities of fault de-
tection systems. This can lead to more eective
predictive maintenance strategies, reducing the
occurrence of unexpected failures and optimiz-
ing maintenance schedules.
Another key direction is the development of
wireless sensor networks (WSNs). While tradi-
tional wired sensors provide accurate data, they
are often limited by their physical connectivity
requirements. WSNs oer a exible and cost-
eective solution, especially useful in monitor-
ing remote or hard-to-access areas of the rail-
way infrastructure. These networks can enhance
real-time monitoring capabilities, allowing for
continuous data collection and analysis without
signicant disruption to railway operations. Fur-
thermore, research into robust cybersecurity mea-
sures is crucial, given the increasing digitization
and interconnectivity of diagnostic systems. En-
suring the security of these systems from cyber
threats is essential to maintain the integrity and
reliability of railway operations.
REFERENCES
1. U.S. Department Of Transportation, Federal Rail-
road Administration Oce of Safety Analysis, Ten
Year Accident/Incident Overview. [Online]. Dostęp-
ne na: https://safetydata.fra.dot.gov/OceofSafety/
publicsite/Query/TenYearAccidentIncidentOve-
rview.aspx
2. Strano S., Terzo M. Review on model-based meth-
ods for on-board condition monitoring in railway
vehicle dynamics. Advances in Mechanical En-
gineering 2019; 11(2): 1687814019826795. doi:
10.1177/1687814019826795.

387
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
3.
Bernal E., Spiryagin M., Cole C. Onboard condi-
tion monitoring sensors, systems and techniques
for freight railway vehicles: a review. IEEE Sen-
sors Journal 2019, 19(1): 4–24, doi: 10.1109/
JSEN.2018.2875160.
4.
Muthukumar M., Nallathambi S. Remote sensor
networks for condition monitoring: An application
on railway industry. In IEEE International Confer-
ence on Electrical, Instrumentation and Communi-
cation Engineering (ICEICE), Karur: IEEE, 2017,
s. 1–5. doi: 10.1109/ICEICE.2017.8191894.
5.
Tsunashima H. Condition monitoring of railway
tracks from car-body vibration using a machine
learning technique. Applied Sciences 2019; 9(13),
doi: 10.3390/app9132734.
6.
Chia L., Bhardwaj B., Lu P., Bridgelall R. Rail-
road track condition monitoring using inertial
sensors and digital signal processing: a review.
IEEE Sensors J. 2019; 19(1): 25–33. doi: 10.1109/
JSEN.2018.2875600.
7.
Hamadache M., Dutta S., Olaby O., Ambur R.,
Stewart E., Dixon R. On the fault detection and di-
agnosis of railway switch and crossing systems: an
overview. Applied Sciences 2019; 9(23): 5129, doi:
10.3390/app9235129.
8.
Renfrew A.C. Condition monitoring of railway
electric point machines. IEE Proceedings - Electric
Power Applications 2002; 149(6): 465–473(8).
9. Oyebande B.O., Renfrew A.C. The design and im-
plementation of a simulated trackside environment
test system for an electric point machine. Proceed-
ings of the Institution of Mechanical Engineers, Part
F: Journal of Rail and Rapid Transit 2007; 221(4):
477–485, doi: 10.1243/09544097F02605.
10. Network Rail. Broken rail explained. https://www.
networkrail.co.uk/stories/broken-rail-explained/
11.
Network Rail. Buckled rail. https://www.
networkrail.co.uk/running-the-railway/
looking-after-the-railway/delays-explained/
buckled-rail-and-summer-heat/
12.
Bogue R.W. University spin‐off company
launches novel structural stress monitoring sys-
tem. Sensor Review 2003; 23(3): 207–208. doi:
10.1108/02602280310481823.
13. Kojima T., Tsunashima H., Matsumoto A. Fault de-
tection of railway track by multi-resolution analysis.
In 4th International Conference on Wavelet Analy-
sis and Its Applications, Macau, China 2005. IEEE
Macau Section: Macau.
14.
Kojima T., Tsunashima H., Matsumoto A. Fault
detection of railway track by multi-resolution
analysis. Computers in Railways X 2006; 88.
doi:10.2495/CR060931.
15.
Weston P.F., Ling C.S., Goodman C.J., Roberts
C., Li P., Goodall R.M. Monitoring lateral track
irregularity from in-service railway vehicles. Pro-
ceedings of the Institution of Mechanical Engineers,
Part F: Journal of Rail and Rapid Transit 2007;
221(1): 89–100, doi: 10.1243/0954409JRRT64.
16.
Naderi H., Mirabadi A. Railway track condition
monitoring using FBG and FPI ber optic sensors.
In IET International Conference on Railway Con-
dition Monitoring, Birmingham, UK: IEE, 2006,
198–203. doi: 10.1049/ic:20060069.
17. Hayashi Y., Kojima T., Tsunashima H., Marumo Y.
Real time fault detection of railway vehicles and
tracks. In IET International Conference on Rail-
way Condition Monitoring, Birmingham, UK: IEE,
2006, 20–25. doi: 10.1049/ic:20060037.
18. Westeon P.F., Ling C.S., Roberts C., Goodman C.J.,
Li P., Goodall R.M. Monitoring vertical track irreg-
ularity from in-service railway vehicles. Proceed-
ings of the Institution of Mechanical Engineers, Part
F: Journal of Rail and Rapid Transit2007; 221(1),
75–88, doi: 10.1243/0954409JRRT65.
19. Attivissimo F., Danese A., Giaquinto N., Sforza P.
A Railway Measurement System to Evaluate the
Wheel–Rail Interaction Quality. IEEE Trans. In-
strum. Meas 2007; 56(5): 1583–1589, doi: 10.1109/
TIM.2007.903583.
20.
Ho S.L. Futuristic railway condition monitoring
system. In IET International Conference on Railway
Engineering 2008 (ICRE 2008), Hong Kong, China:
IEE, 271–279. doi: 10.1049/ic:20080049.
21.
Mizuno Y., Fujino Y., Kataoka K., Matsumoto
Y. Development of a mobile sensing unit and
its prototype implementation. Tinshhua Sci.
Technol 2008; 13(S1): 223–227. doi: 10.1016/
S1007-0214(08)70153-6.
22.
Mori H., Tsunashima H., Kojima T., Matsumoto
A., Mizuma T. Condition monitoring of railway
track using in-service vehicle. JMTL 2010; 3(1):
154–165, doi: 10.1299/jmtl.3.154.
23.
Ward C.P. Condition monitoring opportunities using
vehicle-based sensors. Proceedings of the Institu-
tion of Mechanical Engineers, Part F: Journal of
Rail and Rapid Transit 2011; 225(2): 202–218. doi:
10.1177/09544097JRRT406.
24. Lee J.S., Choi S., Kim S.S, Kim Y.G., Kim S.W.,
Park C. Track condition monitoring by in-service
trains: a comparison between axle-box and bogie
accelerometers. In 5th IET Conference on Railway
Condition Monitoring and Non-Destructive Testing
(RCM 2011), Derby, UK: IET, 2011, 3A1-3A1. doi:
10.1049/cp.2011.0586.
25. Lee J.S., Choi S., Kim S.S., Park C., Kim Y.G. A
mixed ltering approach for track condition moni-
toring using accelerometers on the axle box and
bogie. IEEE Trans. Instrum. Meas 2012; 61(3):
749–758, doi: 10.1109/TIM.2011.2170377.
26.
Tesfa B., Horler G., Thobiani F.A., Gu F., Ball A.D. A

388
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
clamping force measurement system for monitoring
the condition of bolted joints on railway track joints
and points. J. Phys. Conf. Ser. 2012; 364(012021).
doi:10.1088/1742-6596/364/1/012021.
27. Ngigi R.W., Pislaru C., Ball A., Gu F. Modern tech-
niques for condition monitoring of railway vehicle
dynamics. J. Phys.: Conf. Ser 2012; 364: 012016.
doi: 10.1088/1742-6596/364/1/012016.
28. Chellaswamy C., Akila V., Dinesh Babu A., Arasan
N.K. Fuzzy logic based railway track condition
monitoring system. In IEEE International Con-
ference ON Emerging Trends in Computing,
Communication and Nanotechnology (ICECCN)
2013, Tirunelveli: IEEE, 250–255. doi: 10.1109/
ICE-CCN.2013.6528503.
29.
Bagshawe M.J. Investigating inertial measurements
using MEMS devices in trainborne automatic track
condition monitoring applications. In IET Confer-
ence on Control and Automation 2013: Uniting
Problems and Solutions, Birmingham, UK: Institu-
tion of Engineering and Technology, s. 33–33. doi:
10.1049/cp.2013.0010.
30. Chellaswamy C., Aishwarya R., Murthy S.L.N., Vi-
gneshwar C. Green controller for rail track condition
monitoring and information system. In International
Conference on Green Computing, Communication
and Conservation of Energy (ICGCE) 2013, CHEN-
NAI, India: IEEE, 2013, 562–567. doi: 10.1109/
ICGCE.2013.6823499.
31.
Qin L., Wei D., Sun Y.B. Design of onboard de-
vice to diagnose track fault Online. AMM 2013;
483: 465–470. doi: 10.4028/www.scientic.net/
AMM.483.465.
32. Tsunashima H., Naganuma Y., Kobayashi T. Track
geometry estimation from car-body vibration. Ve-
hicle System Dynamics 2014; 52(1): 207–219. doi:
10.1080/00423114.2014.889836.
33. Reiterer A. Railway Measurement Techniques: Op-
portunities and Challenges. In The Second Inter-
national Conference on Railway Technology: Re-
search, Development and Maintenance, Ajaccio,
Corsica, France. 86. doi: 10.4203/ccp.104.86.
34.
Yeo G.J., Weston P.F., Roberts C. The utility of
continual monitoring of track geometry from an in-
service vehicle. In 6th IET Conference on Railway
Condition Monitoring (RCM 2014), Birmingham,
UK: Institution of Engineering and Technology,
2014(2.2.3–2.2.3). doi: 10.1049/cp.2014.0997.
35. Roveri N., Carcaterra A., Sestieri A. Remote condi-
tion monitoring of railway track using FBG sensors.
In Proceedings of the ISMA, 2014.
36.
Hodge V.J., O’Keefe S., Weeks M., Moulds A.
Wireless sensor networks for condition monitoring
in the railway industry: A Survey. IEEE Trans. In-
tell. Transport. Syst 2015; 16(3), 1088–1106, doi:
10.1109/TITS.2014.2366512.
37. Weston P., Roberts C., Yeo G., Stewart E. Perspec-
tives on railway track geometry condition moni-
toring from in-service railway vehicles. Vehicle
System Dynamics 2015; 53(7): 1063–1091, doi:
10.1080/00423114.2015.1034730.
38. Tsunashima H., Mori H., Ogino M., Azami S., Asano
A. Development of track condition monitoring sys-
tem for conventional railways. Mechanical Engineer-
ing 2016; 1115–1124. doi: 10.1201/b21185-119.
39. Tsunashima H., Mori H., Ogino M., Azami S., Asa-
no A. The application of an on-board track condition
monitoring system for local railways. In The Third
International Conference on Railway Technology:
Research, Development and Maintenance, Cagliari,
Sardinia, Italy, 65. doi: 10.4203/ccp.110.65.
40.
Ilie G., Stancalie A.A general approach for real-time
rail condition monitoring using optical ber sensors.
UPB Scientic Bulletin, Series A: Applied Math-
ematics and Physics 2016; 78, 295–306,
41. Maddison S., Smith B. Wireless remote condition
monitoring opens new monitoring applications
in railway tunnel deformation and trackbed. In
Transforming the Future of Infrastructure through
Smarter Information: Proceedings of the Inter-
national Conference on Smart Infrastructure and
Construction 2016, 27–29(161–166). doi: 10.1680/
ttsi.61279.161.
42. Lienhart W., Wiesmeyr C., Wagner R., Klug F., Lit-
zenberger M., Maicz D. Condition monitoring of
railway tracks and vehicles using bre optic sensing
techniques. In Transforming the Future of Infrastruc-
ture through Smarter Information - Proceedings of
the International Conference on Smart Infrastruc-
ture and Construction, ICSIC 2016, ICE Publishing,
2016, 45–50. doi: 10.1680/ttsi.61279.45.
43. [43] Xu X., Zheng J., Yang J., Xu D., Sun X. Track
irregularity fault identication Based on Evidence
Reasoning rule. In 2016 IEEE International Confer-
ence on Intelligent Rail Transportation (ICIRT), Bir-
mingham, United Kingdom: IEEE, 2016; 98–306.
doi: 10.1109/ICIRT.2016.7588747.
44. Balouchi F., Bevan A., Formston R. Detecting rail-
way under-track voids using multi-train in-service
vehicle accelerometer. in 7th iet conference on
railway condition Monitoring 2016 (RCM 2016),
Birmingham, UK: Institution of Engineering and
Technology 2016; 7(6). doi: 10.1049/cp.2016.1194.
45.
Chellaswamy C., Duraichamy S., Glaretsubin P.,
Vanathi A. Optimized vehicle acceleration mea-
surement for rail track condition monitoring. In
2017 2nd International Conference on Comput-
ing and Communications Technologies (ICCCT),
Chennai, India: IEEE, 2017, 155–160. doi: 10.1109/
ICCCT2.2017.7972265.
46.
Seraj F., Meratnia N., Havinga P.J.M. RoVi: Contin-
uous transport infrastructure monitoring framework

389
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
for preventive maintenance. In 2017 IEEE Inter-
national Conference on Pervasive Computing and
Communications (PerCom), Kona, Big Island,
HI, USA: IEEE, 2017, 217–226. doi: 10.1109/
PERCOM.2017.7917868.
47.
Groos J.C., Havrila P., Schubert L.A. In-service railway
track condition monitoring by analysis of axle box ac-
celerations for small to mid-size infrastructure opera-
tors. In 1st World Congress on Condition Monitoring,
WCCM 2017. [Online] https://elib.dlr.de/123161/
48.
Chudzikiewicz A., Bogacz R., Kostrzewski M., Ko-
nowrocki R. Condition monitoring of railway track
systems by using acceleration signals on wheelset
axle-boxes. Transport 2017; 33(2): 555–566, doi:
10.3846/16484142.2017.1342101.
49. Zhang Y., Liu, F., Jing Y., Li W. Application of FBG
sensing technique for monitoring and early warn-
ing system of high-speed railway track conditions.
In 25th International Conference on Optical Fiber
Sensors, Y. Chung, W. Jin, B. Lee, J. Canning, K.
Nakamura, i L. Yuan, Red., Jeju, , Republic of Korea
2017, 103237X. doi: 10.1117/12.2265412.
50.
Ngamkhanong C., Kaewunruen S., Costa B.
State-of-the-Art Review of Railway Track Resil-
ience Monitoring Infrastructures2018; 3(1): 3 doi:
10.3390/infrastructures3010003.
51.
Känsälä K., Rantala S., Kauppila O., Leviäkangas P.
Acceleration sensor technology for rail track asset
condition monitoring. Proceedings of the Institu-
tion of Civil Engineers - Management, Procure-
ment and Law 2018, 171(1): 32–40, doi:10.1680/
jmapl.17.00040.
52.
Milne D., Pen L.L., Thompson D., Powrie W.
Automated processing of railway track deec-
tion signals obtained from velocity and accelera-
tion measurements. Proceedings of the Institution
of Mechanical Engineers, Part F: Journal of Rail
and Rapid Transit 2018; 232(8): 2097–2110. doi:
10.1177/0954409718762172.
53. Tešić P., Jovanović S., Dick M. Analysis of vehicle/
track interaction measurement data using the V/TI
Monitor system. JCE 2018; 70(02): 105–119. doi:
10.14256/JCE.2067.2017.
54.
Roth M., Baasch B., Havrila P., Groos J. Map-
supported positioning enables in-service condition
monitoring of railway tracks. In 21st International
Conference on Information Fusion (FUSION),
Cambridge 2018, United Kingdom: IEEE, 2346–
2353. doi:10.23919/ICIF.2018.8455377.
55. Tam H., Lee K., Liu S., Cho L., Cheng K. Intelli-
gent Optical Fibre Sensing Networks Facilitate Shift
to Predictive Maintenance in Railway Systems. In
International Conference on Intelligent Rail Trans-
portation (ICIRT), Singapore 2018: IEEE, 1–4. doi:
10.1109/ICIRT.2018.8641602.
56. Lu J., Gao M., Wang Y., Wang P. Health monitoring
of urban rail corrugation by wireless rechargeable
sensor nodes. Structural Health Monitoring 2019;
18(3): 838–852. doi: 10.1177/1475921718782395.
57.
Chandran P., Rantatalo M., Odelius J., Lind H.,
Famurewa S.M. Train-based dierential eddy
current sensor system for rail fastener detection.
Meas. Sci. Technol 2019; 30(12): 125105, doi:
10.1088/1361-6501/ab2b24.
58. Farkas A. Measurement of railway track geometry:
a state-of-the-art review. Period. Polytech. Transp.
Eng 2019; 48(1): 76–88. doi:10.3311/PPtr.14145.
59. Bhardwaj B., Bridgelall R., Chia L., Lu P., Dhingra
N. Signal lter cut-o frequency determination to
enhance the accuracy of rail track irregularity detec-
tion and localization. IEEE Sensors J 2020; 20(3):
1393–1399, doi:10.1109/JSEN.2019.2947656.
60. Aung H.T. Condition monitoring of yangon circu-
lar railway and yangon–mandalay railway based
on car-body acceleration response using a portable
device. JDR 2020; 15(3), 353–359, doi: 10.20965/
jdr.2020.p0353.
61. Liu X., Markine V.L. Train hunting related fast deg-
radation of a railway crossing—condition monitor-
ing and numerical verication. Sensors 2020; 20(8):
2278. doi: 10.3390/s20082278.
62.
Wilk A. Digital ltering of railway track coordinates
in mobile multi–receiver gnss measurements. Sen-
sors 2020; 20(18): 5018. doi:10.3390/s20185018.
63.
Shah A.A., Bhatti N.A., Dev K., Chowdhry B.S.
MUHAFIZ: IoT-Based track recording vehicle for
the damage analysis of the railway track. IEEE Inter-
net Things J 2021; 8(11), 9397–9406, doi:10.1109/
JIOT.2021.3057835.
64. Kukulski J., Gołębiowski P., Makowski J., Jacyna-
Gołda I., Żak J. Eective method for diagnosing
continuous welded track condition based on experi-
mental research. Energies 2021; 14(10): 2889. doi:
10.3390/en14102889.
65.
Zvolenský P., Leštinský L., Ďungel J., Grenčík J.
Acoustic diagnostics of railway vehicles. Trans-
portation Research Procedia 2021; 55: 667–672,
doi:0.1016/j.trpro.2021.07.033.
66. Balci E., Yalçin E., Yelce T.U., BezgıN N. Smart di-
agnosis and maintenance systems for railway tracks.
Akıllı Ulaşım Sistemleri ve Uygulamaları Dergisi
2021; 4(2), 134–147, doi: 10.51513/jitsa.951322.
67.
Daniyan I., Mpofu K., Nwankwo S. Design of a
robot for inspection and diagnostic operations of
rail track facilities. IJQRM 2023; 40(3), 653–673.
doi:10.1108/IJQRM-03-2020-0083.
68. Bolshakova A.V., Boronakhin A.M., Klionsky D.M.,
Yu Larionov D., Tkachenko A.N., Shalymov R.V. Rail-
way Track Diagnostics by Combined Kinematic and
Vibroacoustic Analysis. In International Conference on
Quality Management, Transport and Information Se-
curity, Information Technologies (IT&QM&IS), Saint

390
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
Petersburg 2022, Russian Federation: IEEE. 188–192.
doi: 10.1109/ITQMIS56172.2022.9976711.
69. Tsunashima H., Ono H., Takata T., Ogata S. Devel-
opment and operation of track condition monitor-
ing system using in-service train. Applied Sciences
2023; 13(6): 3835. doi: 10.3390/app13063835.
70.
Zanelli F., La Paglia I., Debattisti N., Mauri M.,
Tarsitano D., Sabbioni E.„Monitoring Railway In-
frastructure Through a Freight Wagon Equipped
with Smart Sensors. In Experimental Vibration
Analysis for Civil Engineering Structures, 432, M.P.
Limongelli, P.F. Giordano, S. Quqa, C. Gentile, i A.
Cigada, Red., In Lecture Notes in Civil Engineer-
ing, 432. Cham: Springer Nature Switzerland, 2023,
193–202. doi: 10.1007/978-3-031-39109-5_20.
71. Pal A., Datta A.K. Development of smart real-time
fault detection approach in railway track deploying
a single acoustic emission sensor data. J. Vib. Eng.
Technol., 2024, doi: 10.1007/s42417-024-01374-4.
72.
Partington W. Wheel impact load monitoring.
Proceedings of the Institution of Civil Engineers
- Transport 1993; 100(4): 243–245. doi:10.1680/
itran.1993.25451.
73. Samuels J.M., Palesano J.J. Using Mainline Wheel
Impact Detectors to Increase the Productivity of Car
Inspection Functions. Conference on Railway En-
gineering 1987: Preprints of Papers. Barton, ACT.
[Online]. https://search.informit.org/doi/10.3316/
informit.465991385409489 https://search.informit.
org/doi/pdf/10.3316/informit.465991385409489
74.
Lechowicz S., Hunt C. Monitoring and Manag-
ing Wheel Condition and Loading. 1999. [On-
line]. Dostępne na: https://api.semanticscholar.org/
CorpusID:107721014
75.
Jergéus J., Odenmarck C., Lundén R., Sotkovsz-
ki P., Karlsson B., Gullers P. Full-scale railway
wheel at experiments. Proceedings of the Insti-
tution of Mechanical Engineers, Part F: Journal
of Rail and Rapid Transit 1999; 213(1): 1–13.
doi:10.1243/0954409991530985.
76.
Danneskiold-Samsøe U., Ramkow-Pedersen O.
New on-line monitoring system of surface defects
and prole wear. 2001; 125: 429–433,
77. Johansson A.N., Nielsen J.C.O. Out-of-round rail-
way wheels—wheel-rail contact forces and track re-
sponse derived from eld tests and numerical simu-
lations. Proceedings of the Institution of Mechanical
Engineers, Part F: Journal of Rail and Rapid Transit
2003; 217, 135–146,
78. Thakkar N., Steel J., Reuben R. Rail-wheel contact
stress assessment using acoustic emission: A labora-
tory study of the eects of wheel ats. Proceedings
of the Institution of Mechanical Engineers, Part F:
Journal of Rail and Rapid Transit 2012; 226, 3–13.
doi:10.1177/0954409711402138.
79. Wei C., Xin Q., Chung W.H., Liu S., Tam H., Ho
S.L. Real-time train wheel condition monitoring by
ber bragg grating sensors. International Journal of
Distributed Sensor Networks 2011, 8(1): 409048.
doi: 10.1155/2012/409048.
80. Asplund M., Palo M., Famurewa S., Rantatalo M.
A study of railway wheel prole parameters used as
indicators of an increased risk of wheel defects. Pro-
ceedings of the Institution of Mechanical Engineers,
Part F: Journal of Rail and Rapid Transit 2016,,
230(2): 323–334, doi: 10.1177/0954409714541953.
81. Papaelias M., Amini A., Huang Z., Vallely P., Dias
D.C., Kerkyras S. Online condition monitoring of
rolling stock wheels and axle bearings. Proceedings
of the Institution of Mechanical Engineers, Part F:
Journal of Rail and Rapid Transit 2014; 230(3):
709–723. doi: 10.1177/0954409714559758.
82.
Asplund M., Famurewa S.M., Rantatalo M. Data
Quality Assessment of Automatic Wheel Prole
Measurement Systems. In Current Trends in Reliabil-
ity, Availability, Maintainability and Safety, U. Ku-
mar, A. Ahmadi, A K. Verma, P. Varde, Red., Cham:
Springer International Publishing, 2016, 717–738.
83. Alemi A., Corman F., Lodewijks G. Monitoring of
railway wheel diameters by means of a Wheel Im-
pact Load Detector. In Proceedings of the 3rd In-
ternational Conference on Railway Technology, J.
Pombo, Red., w Civil-Comp Proceedings. United
Kingdom: Civil-Comp Press, 2016. doi: 10.4203/
ccp.110.237.
84.
Song Y., Sun B. Recognition of wheel polygon
based on w/r force measurement by piezoelectric
sensors in GSM-R network. Wireless Personal
Communications 2018, 102(2): 1283–1291. doi:
10.1007/s11277-017-5194-z.
85. Xu P. Condition monitoring of wheel wear for high-
speed trains: A data-driven approach. In 2018 IEEE
International Conference on Prognostics and Health
Management (ICPHM) 2018, 1–8. doi: 10.1109/
ICPHM.2018.8448864.
86. Ni Y.Q., Zhang Q.H.A. Bayesian Machine Learn-
ing Approach for Online Wheel Condition De-
tection Using Track-side Monitoring. In 2018
International Conference on Intelligent Rail
Transportation (ICIRT), 2018, 1–6. doi:10.1109/
ICIRT.2018.8641663.
87.
Gao R., He Q., Feng Q., Cui J. In-service detec-
tion and quantication of railway wheel at by the
reective optical position sensor. Sensors 2020;
20(17). doi:10.3390/s20174969.
88.
Mosleh A., Montenegro P.A., Costa P.A., Calçada R.
Railway Vehicle Wheel Flat Detection with Multiple
Records Using Spectral Kurtosis Analysis. Applied
Sciences 2021; 11(9): doi: 10.3390/app11094002.
89.
Chung K.J., Lin C.W. Condition monitoring for
fault diagnosis of railway wheels using recurrence

391
Advances in Science and Technology Research Journal 2024, 18(6), 361–391
plots and convolutional neural networks (RP-CNN)
models. Measurement and Control 2024; 57(3):
330–338. doi: 10.1177/00202940231201376.
90. Seliger G., Kross U., Buchholz A. Ecient main-
tenance approach by on-board monitoring of in-
novative freight wagon bogie. In Proceedings
2003 IEEE/ASME International Conference on
Advanced Intelligent Mechatronics (AIM 2003),
Kobe, Japan: IEEE, 2003, 407–411. doi: 10.1109/
AIM.2003.1225130.
91. Goda K., Goodall R. Fault-detection-and-isolation
system for a railway vehicle bogie. In The Dynam-
ics of Vehicles on Roads and on Tracks Supplement
to Vehicle System Dynamics, M. Abe, Red., w Ve-
hicle System Dynamics, vol. 41. United States: CRC
Press, 2005, s. 468–476. [Online]. https://www.tand-
fonline.com/doi/abs/10.1076/vesd.38.5.391.8281
92.
Li P., Goodall R., Weston P., Seng Ling C., Goodman
C., Roberts C. Estimation of railway vehicle suspen-
sion parameters for condition monitoring. Control
Engineering Practice 2007, 15(1): s. 43–55, doi:
https://doi.org/10.1016/j.conengprac.2006.02.021.
93.
Mei T.X., Ding X.J. Condition monitoring of rail vehi-
cle suspensions based on changes in system dynamic
interactions. Vehicle System Dynamics 2009, 47(9):
1167–1181, doi: 10.1080/00423110802553087.
94.
Ward C.P., Goodall R.M., Dixon R., Charles G.
Condition monitoring of rail vehicle bogies. In
UKACC International Conference on Control 2010,
1–6. doi:10.1049/ic.2010.0447.
95.
Tsunashima H., Mori H. Condition monitoring of
railway vehicle suspension using adaptive multiple
model approach. In ICCAS 2010, 584–589. doi:
10.1109/ICCAS.2010.5669915.
96.
Daadbin A., Rosinski J., Smurthwaite D. Online
monitoring of essential components helps urban
transport management and increases the safety of
rail transport. Urban and Maritime Transport 2012;
541–552. doi:10.2495/UT120461.
97. Amini A., Entezami M., Huang Z., Rowshandel H.,
Papaelias M. Wayside detection of faults in railway
axle bearings using time spectral kurtosis analysis
on high frequency acoustic emission signals. Ad-
vances in Mechanical Engineering 2016; 8. doi:
10.1177/1687814016676000.
98. De Los Santos N., Jones R., Tarawneh C.M., Fuen-
tes A., Villarreal A. Development of prognostic
techniques for surface defect growth in railroad
bearing rolling elements. Mechanical Engineering
Faculty Publications and Presentations 2017; 4. doi:
10.1115/JRC2017-2262.
99. Li C., Luo S., Cole C., Spiryagin M. Bolster spring
fault detection strategy for heavy haul wagons. Ve-
hicle System Dynamics 2018; 56(10): 1604–1621.
doi: 10.1080/00423114.2017.1423090.
100.
Chudzikiewicz A., Stelmach A., Wawrzyński
W. Selection of signal registration points in the
monitoring process and diagnosing the condition
of the rail vehicle. Journal of Civil Engineer-
ing and Transport 2019; 1, 7–13, doi:10.24136/
tren.2019.001.
101.
Tarawneh C., Lima J.D., Santos N.D.L., Jones
R.E. Prognostics Models for Railroad Tapered
Roller Bearings with Spall Defects on Inner or
Outer Rings. Tribology Transactions 2019; 62(5):
897–906. doi: 10.1080/10402004.2019.1634228.
102. Sánchez R.V. Evaluation of Time and Frequency
Condition Indicators from Vibration Signals for
Crack Detection in Railway Axles. Applied Sci
-
ences 2020; 10(12). doi: 10.3390/app10124367.
103. Zanelli F., Mauri M., Castelli-Dezza F., Sabbioni
E., Tarsitano D., Debattisti N. Energy autonomous
wireless sensor nodes for freight train braking
systems monitoring. Sensors 2022; 22(5). doi:
10.3390/s22051876.
104.
Guo J., Zhang G., Shi H., Zeng J. Small-ampli-
tude bogie hunting identication method for high-
speed trains based on machine learning. Vehicle
System Dynamics 2024; 62(5): 1253–1267, doi:
10.1080/00423114.2023.2224906.
105.
Appun P., Daum M. Fault diagnosis on modern
three-phase traction vehicles. In International
Conference on Main Line Railway Electrication
1989, 207–211.
106. Daley S., Newton D.A., Bennett S.M., Patton R.J.
Methods for fault diagnosis in rail vehicle trac-
tion and braking systems. In IEE Colloquium on
Qualitative and Quantitative Modelling Methods
for Fault Diagnosis, 1995, 5: 1–513. doi: 10.1049/
ic:19950513.
107.
Winterling M.W., Tuinman E., Deleroi W. Fault
analysis of electromechanical traction drives. In
Eighth International Conference on Electrical Ma-
chines and Drives (Conf. Publ. No. 444), 1997,
248–252. doi:10.1049/cp:19971077.
108.
Deuszkiewicz P., Radkowski S. On-line condi-
tion monitoring of a power transmission unit of a
rail vehicle. Mechanical Systems and Signal Pro-
cessing 2003; 17(6), 1321–1334, doi: https://doi.
org/10.1006/mssp.2002.1578.
109. Hedayati Kia S., Henao H., Capolino G.-A. Me-
chanical health assessment of a railway traction
system. In MELECON 2008. The 14th IEEE
Mediterranean Electrotechnical Conference, Ajac-
cio, France: IEEE, 2008, 453–458. doi: 10.1109/
MELCON.2008.4618477.
110. Sh. H., Kia, H., Henao, G.A. Capolino. Non-Sta-
tionary Condition Torsional Vibration Monitoring
Using Induction Machine Electromagnetic Torque
Estimation. In 2008 18th International Conference
on Electrical Machines, Vilamoura: IEEE, 2008,
1–7. doi: 10.1109/ICELMACH.2008.4800056.