Available via license: CC BY 4.0
Content may be subject to copyright.
ARCHIVES OF TRANSPORT ISSN (print): 0866-9546
Volume 71, Issue 3, 2024 e-ISSN (online): 2300-8830
DOI: 10.61089/aot2024.gk7vs246
Article is available in open access and licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0)
APPLICATION OF VIBRATION SIGNALS IN RAILWAY TRACK
DIAGNOSTICS USING A MOBILE RAILWAY PLATFORM
Roksana LICOW1, Franciszek TOMASZEWSKI2
1 Gdansk University of Technology, Faculty of Civil and Environmental Engineering, Gdansk, Poland
2 Poznan University of Technology, Faculty of Civil and Transport Engineering, Poznan, Poland
Abstract:
The article presents a comprehensive method for using vibration signals to diagnose railway tracks. The primary
objective is to gather detailed information on track conditions through a passive experiment. This involves using mo-
bile diagnostic tools and techniques to assess railway infrastructure. The article elaborates on the range of diagnostic
activities conducted in accordance with detailed railway regulations and highlights the benefits and capabilities of
mobile diagnostics in railway transport. The research includes mobile field measurements across the general railway
manager’s network, employing vibration signals to detect and evaluate track conditions. The methodology section
provides a thorough description of the mobile measurement rail platform, detailing the equipment used, the routes
taken for measurements, and the processes of data acquisition and processing. The data obtained from these meas-
urements is crucial for understanding the actual technical condition of the railway tracks. The method of obtaining
and processing data is explained in relation to the real technical condition of the railway track. This involves using
transducers with specific parameters and parametrically defined signal recording, along with dedicated analysis tech-
niques in post-processing. Vibration signals serve as the primary carrier of information in this diagnostic method. The
article details the step-by-step procedures for collecting and analyzing these signals to provide accurate assessments
of track conditions. Based on the results from the mobile measurement rail platform, the article characterizes various
areas of diagnostics where vibration signals are particularly effective for technical evaluation. These areas include
identifying track defects, monitoring track surface and railway crossing and assessing the overall structural health of
the railway infrastructure. The use of vibration signals offers a non-invasive and efficient means of track diagnostics,
providing real-time data for maintenance and repair decisions. In conclusion, the article underscores the significance
of mobile diagnostics in enhancing the safety and reliability of railway transport. By leveraging vibration signals and
advanced data processing techniques, this method provides a framework for continuous monitoring and assessment
of railway track conditions, ultimately contributing to improved maintenance strategies and operational efficiency.
Keywords: railway transport, on-board monitoring - OBM, rail defects
To cite this article:
Licow, R., Tomaszewski, F., (2024). Application of vibration signals in railway track
diagnostics using a mobile railway platform. Archives of Transport, 71(3), 127-145.
DOI: https://doi.org/10.61089/aot2024.gk7vs246
Contact:
1) roksana.licow@pg.edu.pl [https://orcid.org/0000-0002-4368-0064] – corresponding author; 2) franciszek.tomaszewski@put.poznan.pl
[https://orcid.org/0000-0003-1774-7437]
128
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
1. Introduction
Maintaining railway lines at appropriate quality
standards, in terms of their geometry and the diag-
nostics of surface elements, is crucial for infrastruc-
ture managers to meet the required standards for safe
operation and to ensure an adequate level of safety
and comfort for passengers (Sadri et al., 2020). In
Poland, according to data from the Office of Rail
Transport in Poland from 2023, railway traffic is
constantly increasing, contributing to the acceler-
ated fatigue of infrastructure elements, leading to
faster wear and potentially hazardous situations.
This article presents a novel approach to railway
track diagnostics using vibration signals collected
through a mobile railway platform. The novelty of
this research lies in its innovative and modern ap-
proach to railway track diagnostics concerning the
railway transport system currently in use in Poland.
Globally, research is being conducted on effective
track diagnostics using onboard devices (Mori et al.,
2010; Naganuma et al., 2010). This article provides
a broader perspective on the diagnostics of individ-
ual railway elements, taking into account the specif-
ics of the Polish railway transport system.
Therefore, there is a need for continuous and unam-
biguous control of track quality based on the infor-
mation received (Wang et al., 2021). A complicating
factor affecting the clarity of information obtained
about track conditions is the complexity of the rail-
way system, which includes several combinations of
track elements, significantly influencing the varia-
bility of results. These elements include varied rails,
sleepers, fastenings, terrain conditions, and the vehi-
cles in operation. Additionally, when researching the
components of surface elements, factors such as
their year of construction, degree of wear, subgrade
(Nielsen et al., 2018) and its individual layers should
be considered (Arcieri et al., 2024).
This study involves recording vibration signals us-
ing accelerometers mounted on a vehicle during nor-
mal operation. Based on the results obtained from
real conditions of the railway transport system, fur-
ther analyses were conducted. An algorithm for pro-
cessing the measurement data was developed, along
with a concept for the architecture of a rapid onboard
railway diagnostic system.
The article is divided into three main parts. The first
part is a detailed literature review on the research
and implementation of solutions in railway line di-
agnostics. It also references solutions used in other
countries and systems effectively applied in research
involving prototypes of specialized devices. Next,
the methodology of the research is described, includ-
ing details about the mobile measurement railway
platform with different sets of instruments. The arti-
cle outlines railway routes, travel speeds, and the
configuration of the measuring vehicle, taking into
consideration its movement dynamics. The pivotal
section of the article is the case study and discussion,
involving the analysis of research results aimed at
validating and exploring the feasibility of using vi-
bration signals to detect specific rail track damages
based on field tests. The article also explores the ca-
pabilities of assessing railway crossings and surface
conditions using the developed measurement rail
platform. Lastly, the article synthesizes the findings
and presents conclusions drawn from the research.
2. Literature review
In Poland, railway surface diagnostics are guided by
instruction No. 8 issued by the General Railway
Manager of PKP Polish Railway Lines. Along with
other instructions, it details the specific processes for
conducting rail inspections. Defectoscopic tests,
regulated by instruction Id-8 and further described
in instruction Id-10, are central to these diagnostics.
These tests include visual inspections, technical in-
spections measuring track geometric parameters,
and the use of specialized equipment such as hand-
cars and measuring wagons. Whether measurements
are carried out by specialized vehicles or hand-held
devices, the tested section of track must be closed
during tests or conducted during transportation
breaks (Licow, 2018).
Several researchers have made significant contribu-
tions to the field of railway diagnostics.
(Kostrzewski et al., 2021) provides a comprehensive
literature review and bibliometric analysis of rail-
way transport system monitoring (Chudzikiewicz et
al., 2013).
Recent advancements have seen the implementation
of onboard monitoring technologies for real-time as-
sessment of railway infrastructure (Hoelzl et al.,
2023). (Hoelzl et al., 2022) highlighted the use of
these systems on Federal Swiss Railways, which al-
low continuous and precise evaluation of track con-
ditions, enhancing safety and efficiency. Similarly,
(Tsunashima et al., 2023) described systems that
monitor tracks during normal operations (Westeon
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
129
et al., 2007), using innovative data analysis and sig-
nal processing algorithms to identify and locate de-
fects accurately (La Paglia et al., 2022).
A lot of research focuses on detecting specific dam-
ages to railway tracks and monitoring the dynamic
state of vehicles through acceleration measurements
(Wang et al., 2006; Barbosa et al., 2023) and subse-
quent data analysis, including lightweight vehicles
such as trams and metro systems (Firlik et al., 2012;
Faccini et al., 2023). Research in regional railways
also highlights these methods (ONO et al., 2023).
The authors of these studies emphasize the im-
portance of mathematical criteria, such as the
Nyquist criterion, which must be satisfied to obtain
undisturbed signals for further analysis and to
achieve unambiguous results (De Rosa et al., 2019).
The findings of (Nielsen et al., 2020; Stoura et al.,
2023) indicate the ability to identify wavelengths
and irregularities and roughness on the rail running
surface in accordance with European standard UNE-
EN 13848-1. High speeds (La Paglia et al., 2023) re-
sult in relatively high levels of acceleration, which
are crucial for improving the signal-to-noise ratios
used in the estimation process. When investigating
damage to the rail running surface, the condition of
the rolling stock wheels should also be considered
(Rodriguez et al., 2021). To achieve this, two inertial
modules were sequentially employed on moving
wheels, significantly enhancing the reliability of de-
fect detection by approximately 10%. This approach
is deemed justified if specific defect identification
remains unclear.
An important element in signal processing is filter-
ing methods. The authors (Hoelzl et al., 2023) pre-
sented an advanced method for assessing railway
track stiffness, utilizing the Vold-Kalman filter to
analyze accelerations of vehicle axle boxes. This
work makes a significant contribution to methods
and approaches for maintaining and diagnosing rail-
way infrastructure, offering a new approach to track
condition monitoring. Meanwhile, the work by
Muñoz et al. (2021) focuses on the application of
Kalman filtering techniques to estimate lateral track
irregularities. Geometric defects can affect the sta-
bility and safety of railway transport systems, mak-
ing their accurate detection and estimation crucial
for maintenance and operational purposes.
Advanced diagnostic systems utilize measurement
units such as IMUs (Inertial Measurement Units),
INS (Inertial Navigation Systems), satellite naviga-
tion (GNSS), railhead optical scanners, and distance
meters (odometers) (Uhl et al., 2010). IMUs and re-
lated technologies are pivotal in modern diagnostics.
(Rosano et al., 2024; Tsunashima et al., 2014) dis-
cussed the use of acceleration and contact force
measurements to monitor and improve the technical
condition of railway infrastructure. These technolo-
gies provide valuable data that aids in predicting and
identifying defects early, thus facilitating proactive
maintenance (Chrzan, 2022). For instance, (Carne-
vale et al., 2021) introduces an algorithm that signif-
icantly enhances the accuracy of location measure-
ments collected by diagnostic systems installed on
railway vehicles. This algorithm integrates data
from GPS and INS systems, improving localization
accuracy even in challenging environments such as
tunnels.
The integration of AI (Avici et al., 2021), particu-
larly deep learning and convolutional neural net-
works (CNNs), has revolutionized railway diagnos-
tics. (Di Summa et al., 2023) reviewed the use of
deep learning techniques for monitoring infrastruc-
ture, comparing them with traditional methods and
highlighting their advantages in accuracy and effi-
ciency. (Faghih-Roohi et al., 2016) demonstrated the
effectiveness of CNNs in detecting surface defects
such as cracks and corrosion, showing promising re-
sults in real-world applications. (Arcieri et al., 2024)
explored the use of Partially Observable Markov
Decision Processes (POMDP) and reinforcement
learning to optimize maintenance decisions under
uncertainty. This approach allows for better deci-
sion-making strategies by incorporating incomplete
or partially observable information, typical in rail-
way infrastructure management.
3. Research methodology
3.1. Mobile measurement platform
The measurements were planned using two mobile
measurement sets for selected routes, as discussed in
section 3.2. The first research set (A) consisted of a
platform of type Xk 03–0517 and a rail vehicle DH–
350.11 (Fig. 1). The second measurement set (B) in-
cluded the platform of type Xk 03–0517, a separat-
ing wagon, and a rail vehicle DH–350.11 (Fig. 2).
130
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
Fig. 1. Mobile measurement set in variant A
The total length of the rail platform Xk 03–0517
with buffers is 12.240 m. It is a platform built on two
two-axle bogies with flanged wheels, without its
own drive. The mass of the platform is 17,850 kg.
The second component of the mobile measuring set
is the DH-350.11 hydraulic rail vehicle. The length
of the rail vehicle is 12 m, and the power of the drive
engine is 350 kW. The maximum speed of the rail
vehicle is 80 km/h. The four-axle railway platform,
serving as a separation element in campaign B, mit-
igated the adverse effects resulting from the opera-
tion of the DH-350 rail vehicle. In both configura-
tions of the measuring sets, the DH-350.11 acted as
the driving element (pulling and pushing), while the
Xk 03-0517 platform served as the measuring ele-
ment with installed equipment.
In measurement set A, two 4504A triaxial transduc-
ers and a 3050-A-060 B&K type signal recording
cassette were used to measure and record vibration
signals. The recording cassette recorded at a sam-
pling frequency of 51.2 kHz. Three-axis vibration
transducers were mounted on the bearing housing
(axle box) of the second wheel of the first bogie (in
the direction of travel of the platform) on both sides
of the rail platform (Fig. 1). Vibration signals were
recorded in three directions: x - aligned with the
movement of the rail vehicle, y - horizontal to the
movement of the rail vehicle, and z - vertical to the
movement of the rail vehicle (perpendicular to the
axis of the wheel set and the railway track), sepa-
rately for the right and left wheels. The frequency
range for each axis of the 4504A vibration trans-
ducer was as follows – for the x-axis: 1 to 11000 Hz;
for the y-axis: 1 to 9000 Hz; and for the z-axis: 1 to
18000 Hz. The signal recording cassette was placed
on the measurement platform and controlled via the
Wi-Fi network from a mobile phone (Fig. 3a).
Measurement set B consisted of two single-axis
transducers of type 4508 and a signal recording cas-
sette of type 7533 B&K. The single-axis transducers
were mounted under the upper shelf of the side part
of the axle box on the second wheel of the first bogie
(in the direction of travel of the railway platform) on
both sides of the railway platform (Fig. 2). Vibration
signals were recorded in the z direction - vertical to
the movement of the rail vehicle, separately for the
right and left wheels. The signal recording cassette
was placed on the measurement platform and con-
trolled from a computer via LAN (Fig. 3b).
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
131
Fig. 2. Mobile measurement set in variant B
The diagram presented in Fig. 3 outlines the stages
of data acquisition, the method of obtaining them,
and the subsequent processing stages, including the
analysis of the technical condition of the railway
track during post-processing.
Measurement set B consisted of two single-axis
transducers, type 4508, and a signal recording cas-
sette, type 7533 B&K. The single-axis transducers
were mounted under the upper shelf of the side part
of the axle box on the second wheel of the first bogie
(in the direction of the railway platform's move-
ment) on both sides of the railway platform (Fig. 2).
Vibration signals were recorded in the z direction –
vertical to the movement of the rail vehicle – sepa-
rately for the right and left wheels. The cassette re-
cording the signals was placed on the measurement
platform and controlled from a computer via LAN
(Fig. 3b).
The presented diagram (Fig. 3) specifies the stages
of data acquisition, the method of obtaining the data,
and the subsequent processing stages, including the
researched technical conditions of the railway track
in post-processing analyses.
Fig. 3. Scheme of the architecture system for collecting and processing measurement data
132
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
All the signals are then sent to the created database
of source files. In this database, the recorded signals
are organized according to their timestamp and as-
signed to a specific railway line, with information
about the track being tested and the starting and end-
ing stations (stops) of the measurement. The data-
base of source files was designed to allow retrieval
of the original signal at any time.
The next stage of data acquisition involves using
mass memory to download specific signals for fur-
ther analysis. The preliminary analysis of the rec-
orded signals consists of selecting specific signal
time windows (regions) relevant to a particular anal-
ysis. For example, signal regions from km 38.000–
38.500 may be selected due to the occurrence of
squat defect No. 227. Information about defects oc-
curring in specific locations is based on knowledge
from the railway manager. Determining where to cut
these regions depends on the timestamp or listening
to the recorded audio signal with mileage indica-
tions.
In the next step, the signals prepared in this manner
are parameterized for each measured vibration direc-
tion from the two wheels of the vehicle. Signal pa-
rameterization, considering vibration directions, in-
volves defining specific characteristics relevant to
particular surface conditions. Variant (a), utilizing
six signals, allows for a broader spectrum of direc-
tional vibration analyses compared to variant (b).
The results of the analysis are stored in a backup da-
tabase, where they are categorized into specific en-
tities describing the type of each defect.
3.2. Researched railway lines
The measurements were conducted as part of two
measurement campaigns (Fig. 4). The first measure-
ment campaign (Route A) was conducted in 2019. It
involved continuous measurement of vibration sig-
nals on railway line No. 203, Tczew – Chojnice sec-
tion, and railway line No. 211, Chojnice – Brusy sec-
tion.
Fig. 4. Routes of measurement campaigns [own work based on the Interactive map of PKP PLK S.A.]
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
133
The second measurement campaign (Route B) was
conducted in 2021. This campaign involved contin-
uous measurement of vibration signals on the fol-
lowing lines: Line No. 9, Gdańsk Południe – Gdańsk
Główny section; Line No. 202, Gdańsk Główny –
Gdańsk Wrzeszcz section; Line No. 248, Gdańsk
Wrzeszcz – Gdańsk Osowa section; and Line No.
201, Gdańsk Osowa – Gdynia Port section.
Table 1 presents details related to the measurement
of vibration signals on individual railway lines
within Routes A and B. The tests were conducted on
single- and double-track lines with various railway
surface elements. The vehicle moved at variable
speeds in the range of 10 to 80 km/h. The table also
includes information about the condition of the
measurement rail platform, whether it was pulled by
a DH-350 or pushed by it on the return journey.
3.3. Data processing algorithms
The article undertakes the analysis of three cases: a)
identification of the squat defect, b) assessment of
the condition of a railway crossing based on the type
of surface used at the crossing, and c) evaluation of
the track surface condition considering the type of
fastening and railway sleeper. The analysis results
are presented in sections 4.1 – 4.3. The data pro-
cessing models are shown in Figure 6. For each
listed case, an individual model was applied, which
clearly indicated the feasibility or infeasibility of di-
agnostic identification using vibration signals.
Table 1 Description of the researched railway lines
Railway line
number
Kilometer
Track
number
Type of line (on the
research section)
Speed
[km/h]
Condition of the
measuring wagon
Route A
203
5.423 - 96.227
1
double track
40 - 80
pulled
211
0.672 - 26.598
-
single track
20 - 30
pulled
211
26.598 - 0.672
-
single track
20 - 30
pushed
203
6.227 - 5.423
2
double track
40 - 80
pushed
Route B
9
325.848 - 327.741
1
double track
80
pulled
202
-0,381 - 4.180
1
double track
40 - 80
pulled
248
-0,266 - 18.246
1
double track
40 - 80
pulled
201
188.529 - 210.763
210.763 - 188.529
1
2
double track
10 - 20
pulled / pushed
248
18.246 - 1.103
2
double track
40 - 80
pushed
202
4.180 - 1.000
2
double track
40 - 80
pushed
Fig. 5. Algorithm of data processing
134
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
Algorithms for data processing describe the equa-
tions used in the model and transformations neces-
sary to ultimately obtain information on the signal's
usability with the data processing model for railway
track diagnostic identification. The data were tested
under various terrain conditions, as each railway line
traversed by the vehicle exhibits different types of
track infrastructure, degrees of degradation, and per-
missible speeds applicable to the routes.
Identification of rail surface defects (Fig. 5a), in the
context of this article, pertained to identifying the
squat No. 227 defect from the catalog of Polish Rail-
way Lines. The recorded signal had a sampling fre-
quency of 51.2 kHz. The frequency ranges for each
axis of the 4504A vibration transducer were as fol-
lows: for the x-axis: 1 to 11000 Hz; for the y-axis: 1
to 9000 Hz; and for the z-axis: 1 to 18000 Hz. Thus,
for each axis dedicated to processing the signal using
Fast Fourier Transform, the Nyquist criterion was
satisfied.
In the case of identifying the squat defect, time-do-
main analysis proved challenging due to numerous
other occurrences such as rail joints (welds), cracks,
or conventional joints, which contribute to informa-
tional noise and lack of clarity in assessing the squat
defect. Vertical displacement analysis based on rec-
orded vibration signals over time was proposed. The
computational procedure was conducted in the ded-
icated BK Connect computing environment and in-
volved deriving vertical displacements from vibra-
tion signals during the time-domain analysis.
The first and second derivatives with respect to time
of the quantity x(t) – displacement, are called veloc-
ity (1) and acceleration (2), respectively.
(1)
(2)
where:
x(t) - displacement as a function of time,
v(t) - velocity as a function of time, the first deriva-
tive of displacement x(t) [m/s],
a(t) - acceleration as a function of time, the second
derivative of displacement x(t) [m/s2].
After performing the double integration of the vibra-
tion signal and obtaining vertical displacements, the
next step was to perform filtering to eliminate unde-
sired frequencies. Since filtering is conducted in the
frequency domain, it requires converting the data
from the time domain to the frequency domain. This
domain transformation was performed using the Fast
Fourier Transform which is expressed as:
(3)
where:
X(k) - value of the Fast Fourier Transform at fre-
quency k (k=0,1,…,N-1),
x(n) - sample of displacement signal in the time do-
main,
N - number of samples of the vibration signal,
j - imaginary unit (j2 = -1).
After computing the signal in the frequency domain,
a filtering function was implemented. A high-pass
filter is used to enhance the waveform amplitude. An
Infinite Impulse Response (IIR) filter is employed
with an accuracy of 0,1 dB. The high-pass filter is
designed to remove the DC component and low-fre-
quency drift. However, its application reduces accu-
racy below fs/3000 Hz. In the case of squat defect
observation, low frequencies up to 100 Hz are criti-
cal, thus acknowledging the accuracy reduction in
the range of 0 to 6,7 Hz. The transformed results
were analyzed in the time-amplitude-frequency do-
main to localize the defect under investigation. The
results obtained from the three measurement axes (x,
y, and z) revealed a significant identification poten-
tial in the z-direction. The final step in squat defect
analysis involved third-octave analysis using the
Constant Percentage Bandwidth module in BK Con-
nect software. Third-octave analysis was conducted
exclusively for the signal recording direction z, ver-
tical to the railway vehicle movement.
The next group of analyses focused on evaluating
the condition of a railway crossing based on the type
of surface used at the crossing (Fig. 5b). In this anal-
ysis, the original vibration acceleration signal was
initially analyzed in the time domain. Vibrations
provided clear information only when considering
two directions of their registration simultaneously.
Therefore, the analysis focused on the y-direction -
horizontal to the movement of the railway vehicle -
and the z-direction - vertical to the movement of the
railway vehicle.
The next step involved frequency domain analysis.
For this purpose, a Fast Fourier Transform was per-
formed, and similarly to the squat defect analysis,
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
135
a high-pass integrating IIR filter with 0,1 dB accu-
racy was applied in the frequency analysis. For the
railway crossing diagnostic case, the frequency anal-
ysis window ranged from 7 Hz to 10000 Hz for z
axis. In addition, a Power Spectral Density (PSD)
Analysis was applied to determine the distribution
of signal power in the frequency domain. This anal-
ysis will be expanded in future articles to address
specific aspects related to the construction of rail-
way crossings, including the use of fine aggregate
fill at the approach and departure ends of the cross-
ing plate.
The last group of analyses focused on assessing the
condition of the railway surface by considering
changes in the type of railway substructure and the
type of fastening. In this analysis, particular atten-
tion was paid to time-domain analysis, examining
vibration acceleration amplitudes from raw files in
the directions of vibration registration - y (horizontal
to the movement of the railway vehicle) and z (ver-
tical to the movement of the railway vehicle).
4. Analysis of research results
4.1. Implementation of vibration signals in the
identification of squat defects on the running
surface – case study
An example of squat defect No. 227 is presented in
part of Figure 6a. Squats occur on the running sur-
face of the rail and are initially characterized by
peeling and cracking of the material, creating a char-
acteristic "dimple" of a semicircular shape. The de-
velopment of the defect fundamentally depends on
the specific operating conditions of the line. A squat
defect is a type of point defect. Squat rail defect No.
227 can have serious consequences for vehicle
safety and dynamics. It affects the structural integ-
rity of the vehicle, reliability, and ride comfort.
However, on the railway network, there are also sec-
tional rail defects, such as wheel burns, which can
reach lengths of up to 1,5 meters.
The analysis of signal samples recorded on routes
(Table 1) was conducted according to the data pro-
cessing algorithm described in Section 3.3. In the
conducted analysis, squat defect No. 227 (Fig. 6a)
was identified using processed vibration signals.
The signal, according to the proposed algorithm,
was classified as a vertical irregularity. A double in-
tegration of the vibration signal was performed in
the vertical direction relative to the movement of the
rail vehicle to obtain vertical displacements.
Subsequently, it was necessary to filter the inte-
grated accelerations to eliminate undesired frequen-
cies and random events. The data was analyzed in
the time domain. However, for clear visualization
purposes, the x-axis is dimensionless with assigned
indices.
Figure 6 presents one of the example results ob-
tained during rail vehicle passage over a squat de-
fect. The calculated displacements during the pas-
sage over squat defects No. 1 and No. 2 are shown.
The example uses results from defects occurring on
the right rail (in the case of squat No. 1) and the left
rail (in the case of squat No. 2). The authors chose
to present this example due to the occurrence of two
squats at a similar kilometer mark but on different
rails. The analysis revealed that the results recorded
by the transducer as vibrations on the axle box,
which directly absorbs the impact from a specific
rail, directly influence and are recorded by the trans-
ducer mounted on the adjacent axle box.
Additionally, the time domain analysis showed very
similar results for both defects, even though they oc-
curred on different rails. This applies to the entire set
of squat defects recorded during passages on routes
A and B. Based on this, it can be concluded that this
analysis could be one of the components of the as-
sessment of squat defects.
The results of the spectral analysis (Fig. 6c and d)
indicate low frequencies, up to 100 Hz, at which the
defect can be observed. Additionally, a characteris-
tic feature observed during spatial visualization in
the frequency domain is the shape of the defect spec-
trum. It takes on a rectangular flat shape with a spe-
cific amplitude. Figure 6d presents the results of the
left wheel passage, initially recording the dynamic
impact caused by the presence of a squat defect on
the right rail, followed by the visualization of the
squat defect on the left rail.
The images obtained from the conducted experiment
will be verified in further analyses in the context of
other running surface defects such as spalling, trans-
verse cracking, or head checking. In future work, the
authors intend to conduct analyses aimed at distin-
guishing between types of defects and assessing
their degree of degradation.
136
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
Fig. 6. Results of the analysis of the signal recorded at the squat defect: a – squat, b – CPB analysis, c,d -
spectrum analysis for the vibration signal vertical direction for the right Rz and left Lz wheels
The presented case study of the most common squat
defect and the analysis of research results demon-
strated that the use of vibrations allows for the diag-
nostics of the rail running surface. The signals col-
lected by vibration sensors are analyzed using signal
processing algorithms. These algorithms can detect
characteristic changes in the vibration signals that
may indicate the presence of squat defects.
4.2. Analysis of the condition of the surface at
railway crossings - case study
The second broadly defined case study in the diag-
nostic assessment of the railway transport system is
the evaluation of railway crossings. Railway cross-
ings are a very complex system to assess because
there are at least a dozen possible combinations of
their construction. These surfaces can include small
sized slabs, large sized slabs, rubber slabs, or com-
binations of slabs with asphalt on the outer sides of
the track, with or without fine grained aggregate fill,
which is placed at the edge of the entry slab from the
track side. Each of these configurations requires ef-
fective diagnostics to mitigate undesirable vibration
phenomena during passage.
In this case study, vibration signals were analyzed in
both the time and frequency domains to explore the
possibility of identifying railway crossings based on
their surface type. In future work, the authors will
thoroughly consider all types of surface configura-
tions, the presence of fill, and the consideration of
degradation states and the possibility of monitoring
them using vibration signals.
The plots shown in Figure 7 pertain to the recorded
vibration signals during the passage over a railway
crossing with large-sized CBP slabs (a) and small-
sized slabs (b).
A characteristic feature of the signal recorded at a
railway crossing is the occurrence of symmetrical
peaks in both the vertical and horizontal directions
relative to the movement of rail vehicles, appearing
at specific intervals corresponding to the time of pas-
sage. Therefore, it is reasonable to analyze both di-
rections of vibrations - vertical and horizontal to the
vehicle's movement - in the diagnostic assessment of
railway crossings. In the analysis of the vibration
signal in the time domain, impulse averaging was
used, allowing for a detailed examination of short-
term signal changes that occur during the passage
over an 'obstacle,' such as a railway crossing. The
recorded vibration acceleration amplitudes were
higher in the case of small sized slabs, reaching 40
m/s², compared to the amplitudes obtained after
passing over CBP slabs, which reached up to 16
m/s².
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
137
Fig. 7. Vibration signal plot during the passage over a railway crossing with CBP slabs (a) and small-sized
slabs (b)
In the next step, a spectrum analysis was conducted,
aiming to identify the entry and exit points on the
railway crossing in the area plot. The analysis was
performed for all crossings on the route, as listed in
Table 1, without distinguishing the type of surface at
the crossing at this stage. The results of fifteen ran-
domly selected samples for the Rz direction (verti-
cal, right wheel of the vehicle) are presented in Fig-
ure 8.
The area plots on the left side of the chart were sub-
sequently filtered to retain high amplitudes over
time and remove random or out-of-range frequen-
cies of the transducer, obtaining the results shown
on the right side of Fig. 8. In this section, traces of
signals generated at specific frequencies during the
train's entry and exit from the railway crossing can
be observed.
This trace allowed for the extraction of signal sam-
ples related to selected railway crossings, taking into
account the surface construction of the railway
crossing. Fig. 9 presents example samples of vibra-
tion signals in the Lz direction (vertical direction of
vibration recording on the left wheel of the vehicle)
during passage over CBP slabs. Similarly, Fig. 10
shows the Lz signal recorded during passage over
small sized slabs.
In most samples obtained from a single wheel, the
frequency plots convey information regarding the
train's entry and exit from the crossing slabs. In case
of uncertainties, it is reasonable to retain information
from the second wheel of the rail vehicle. Combin-
ing synthesized information from both wheels of the
rail vehicle provides reliable and unambiguous in-
formation regarding the passage over the crossing
slabs.
The proposed analysis aimed at identifying the type
of railway crossing is Power Spectral Density (PSD)
Analysis. The bottom row of windows in the figures
presents the results of the PSD analysis, which was
conducted based on frequency analysis involving the
identification of frequency windows related to the
train's passage over the crossing slabs.
Fig. 8. Identification of the frequency trace resulting from the train's entry and exit from the railway crossing
138
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
Fig. 9. Frequency spectrum and power spectral density for CBP slabs in the Lz vibration signal recording
direction
Fig. 10. Frequency spectrum and power spectral density for small sized slabs in the Lz vibration signal re-
cording direction
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
139
Using Power Spectral Density (PSD) analysis, it was
possible to identify the signal power in the frequency
domain considering the passage time. The results of
the PSD analysis showed that the frequency bands
during the entry and exit from the railway crossing
had the highest power levels throughout the entire
time window of the passage. The highest power
level was recorded at a frequency of approximately
200 Hz regardless of the surface type. Subsequently,
high power levels were recorded in a narrow band of
800 Hz for CBP slabs and in the range of 700-900
Hz for small sized slabs.
Additionally, the results of the frequency analysis
indicated that the signal spectrum for the vertical (z)
and horizontal (y) directions shows characteristic vi-
bration acceleration amplitudes of 10 m/s2 and 2.5
m/s2, respectively, concentrated around a frequency
of approximately 200 Hz. In the horizontal direction,
additional amplitudes of 1 m/s2 were observed
around 2000 Hz. Further analysis of the recorded vi-
bration signals from the railway crossing with small
sized slabs inside the tracks and asphalt outside
showed similar occurrences of symmetrical peaks as
observed for CBP slabs. The spectral analysis in the
direction vertical to the vehicle's movement indi-
cated the highest amplitudes around 200 Hz, fol-
lowed by decreasing amplitudes at 500 Hz and
smaller amplitudes at 2000 Hz. In the horizontal di-
rection relative to the vehicle's movement, higher
amplitude values of around 2 m/s2 were observed in
the frequency range from 1000 to 10000 Hz, com-
pared to approximately 1 m/s2 observed for CBP
slabs.
In future studies on railway crossings, the authors
will focus on determining the presence or absence of
sand backfill at the crossing. The presence of back-
fill significantly impacts the displacement of internal
slabs during low temperatures, thereby affecting
traffic safety. Additionally, the analyses will include
assessments of the type of crossing development, its
technical condition, and other components of the
railway surface.
4.3. Analysis of changes in the type of railway
surface - case study
The analysis of changes in surface types can be uti-
lized for diagnostic assessments conducted during
railway line operations and for qualifying sections
for element replacement. This analysis evaluates vi-
bration signals recorded while traversing two differ-
ent combinations of surface types.
The first case study analysis regarding the diagnos-
tics of surface changes involves assessing the transi-
tion from wooden sleepers to concrete sleepers. This
change is quite common, particularly at railway line
junctions with turnouts. Wooden and concrete sleep-
ers are also frequently combined on routes due to
maintenance decisions to replace sections of de-
graded wooden sleepers with new concrete ones. In
this case, the analysis examined the combination of
surfaces built on wooden and concrete sleepers. In
both cases of railway surface construction, K-type
fastenings and S49 rails were used. Figure 11 illus-
trates the analysis results in both the time and fre-
quency domains.
Figure 11 depicts the results of a case study for sig-
nals analyzed during changes in railway track sur-
faces. Figure 11a shows a randomly selected analy-
sis over time. From the set of recorded signals during
the track surface change scenario, it was observed
that transitioning from wooden sleepers to concrete
sleepers results in vibration accelerations ranging
from 20 to 30 m/s² when recording in the vertical
direction relative to the motion of the railway vehi-
cle.
In the analysis of railway track surface changes, the
predominant component of diagnostic assessment is
the vertical vibrations (perpendicular to the direction
of railway vehicle motion). However, these analyses
should also consider vibrations transverse to the di-
rection of railway vehicle motion. Similar values of
transverse vibration amplitudes provide additional
informational elements regarding the change in rail-
way track surface.
For the analysis in terms of spectral content (Figure
11b), two cases of railway track surface changes
were selected and compared solely in the direction
of recording vertical to the motion of the railway ve-
hicle, separately for the left (Lz) and right (Rz)
wheels of the railway vehicle. Sample signals num-
bered 1 and 2 were used for comparison. Both sig-
nals exhibit very similar frequency characteristics
and comparable amplitude levels. Signal number 1,
on the right track (Rz_1), shows higher amplitude
levels around 800 Hz.
Analyzing changes in vibration signals in the con-
text of changes in track surface types over longer pe-
riods allows for a better understanding of amplitude
140
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
levels compared to momentary recordings. There-
fore, continuous measurement during railway vehi-
cle operation provides more insight than triggering
and recording in narrow time windows (Figure 12).
The second case to be analyzed in the context of rail-
way track surface changes involves the modification
of fastenings. In this analysis, wider time windows
of vibration signals were utilized to visualize ampli-
tudes under different surface conditions. Figure 12a)
illustrates the transition from K-type fastenings to
SB-type fastenings, while Figure 12b) depicts the re-
verse transition from SB-type fastenings back to K-
type fastenings.
Fig. 11. Spectrum signal when changing the surface with wooden sleepers to concrete sleepers, where Lz_1 -
vibration spectrum of signal No. 1 for the left wheel in the direction vertical to the vehicle movement;
Lz_2 - vibration spectrum of signal No. 2 for the left wheel in the direction vertical to the vehicle
movement; Rz_1 - vibration spectrum of signal No. 1 for the right wheel in the direction vertical to
the vehicle movement; Rz_2 - vibration spectrum of signal No. 2 for the right wheel in the direction
vertical to the vehicle movement
Fig. 12. Vibration signal when changing the type of fastening on concrete sleepers
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
141
This railway line used concrete sleepers and S49
rails. Passing over SB-type fastenings resulted in a
significant increase in amplitude levels, averaging
approximately 15 m/s2, compared to K-type fasten-
ings where amplitudes averaged around 6 m/s2. Vis-
ualization of these plots allows for the determination
of dynamic value levels during passages over the se-
lected infrastructure type in the context of fastening
type. Passages over SB fastenings, known as resili-
ent fastenings, generated amplitudes twice as high as
those observed over rigid K-type fastenings. The
analysis results emphasize a significant difference in
acceleration levels depending on the type of fasten-
ings used.
The third case of analysis in the context of track sur-
face changes involved transitioning from concrete
sleepers to wooden sleepers, which is the reverse
scenario compared to case study number 1. In this
analysis, samples were analyzed over time (Figure
13a), showing recorded amplitude levels during the
transition of approximately 30 m/s2 in the z-direc-
tion and 35 m/s2 in the y-direction. Following the
conclusions drawn from case study number 1, this
case involved analyzing samples in two vibration di-
rections – x and y. Subsequently, spectral analysis
was performed (Figure 13b).
The spectral analysis was conducted over maximum
frequency ranges tailored to the transducer recording
capabilities as per the manufacturer's specifications.
The results of this analysis revealed high vibration
amplitudes up to 14000 Hz in the z-direction. In the
y-direction, significantly greater dynamic activity
was observed compared to the z-direction. Specifi-
cally, in the numerous recorded samples, high am-
plitude values were noted only at high frequencies
ranging from 2000 to 9000 Hz. This information
constitutes another component aiding in the identifi-
cation and diagnostic assessment of railway track
surface changes.
Understanding vibration levels specific to each type
of surface can provide a basis for developing filters
that distinguish signals from surface noise, thereby
isolating signals for further defect analysis. Further
investigation of this issue should include visual in-
spections of the track or material data obtained from
cameras.
5. Conclusion
The results of the research presented in the article
lay the groundwork for further analysis in various
aspects of railway infrastructure diagnostics, includ-
ing the evaluation of rail and road crossings, as well
as monitoring changes in track conditions. The utili-
zation of vibration signals for detecting specific
track damages has already been successfully imple-
mented in systems mounted on vehicles, as exempli-
fied by practices in Japan. The findings also under-
score the effectiveness of using vibration signals in
targeted diagnostic assessments through dedicated
post-processing analyses.
The synthesis of the results from the conducted re-
search demonstrates that each case of railway area
diagnostics requires appropriate tools and analyses.
In the case of rolling surface defects, specifically
squats discussed in this article, the analysis of these
defects involves identifying displacement signals
vertical to the direction of the rail vehicle's move-
ment at low frequencies up to approximately 200
Hz. The diagnostics of railway crossings involved
the analysis of the power spectral density of vibra-
tion signals in the directions vertical and transverse
to the rail vehicle's movement at frequencies around
200 Hz and higher, approximately 800 Hz. The case
of surface changes shows the relevance of analyzing
vibration signals in the frequency domain for the di-
rections vertical and transverse to the rail vehicle's
movement at frequencies as high as 15 000 Hz.
The use of vibrations for the analysis of railway
track types is an advanced and efficient diagnostic
method that ensures the safety and efficiency of rail-
way infrastructure.
A crucial focus in future studies on vibration signals
for assessing surface conditions will be to ascertain
the dynamics of changes in diagnostic parameters,
identifying the vibration directions that provide the
most informative data. This analysis will serve as a
guiding principle for future considerations. Addi-
tionally, subsequent analyses will verify how the ar-
rangement of the mobile system set, considering the
direction of travel of the measurement platform, in-
fluences the results and mitigates disturbances aris-
ing from specific configurations.
142
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
Fig. 13. Spectrum signal when changing the surface with concrete sleepers to wooden sleepers, where Lz -
vibration spectrum for the left wheel in the direction vertical to the vehicle movement; Rz - vibration
spectrum for the right wheel in the direction vertical to the vehicle movement; Ly - vibration spec-
trum of signal for the left wheel in the direction horizontal to the vehicle movement; Ry - vibration
spectrum of signal for the right wheel in the direction horizontal to the vehicle movement
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
143
The proposed mobile measurement system offers a
solution to challenges related to railway line diag-
nostics and enhances the efficiency of diagnostic
processes. Future efforts will concentrate on devel-
oping a streamlined measurement system that is
compatible with and installable on commercial
trains. This system aims to detect damages promptly
during regular transport operations. Therefore, up-
coming steps will also focus on expanding the da-
taset with signals recorded at speeds ranging from
80 to 160 km/h.
The use of vibration signals to detect squat defects
is an efficient method because it allows for continu-
ous monitoring of rail conditions without the need to
stop railway traffic. However, to obtain accurate re-
sults, it is necessary to use advanced signal analysis
techniques. By utilizing mathematical models and
machine learning techniques, the diagnostic system
can classify identified signal changes as potential
squat defects. This may include comparing vibration
patterns with a database of known defects and their
characteristics.
Acknowledgments
Financial support of these studies from Gdansk Uni-
versity of Technology by the DEC-
3/1/2022/IDUB/I3b/Ag grant under the ARGEN-
TUM TRIGGERING RESEARCH GRANTS - ‘Ex-
cellence Initiative - Research University’ program is
gratefully acknowledged.
Acknowledgements to Prof. A. Wilk, Project Man-
ager of "InnoSatTrack," and the "InnoSatTrack"
Project Team for their technical support.
References
1. Arcieri, G., Hoelzl, C., Schwery, O., Straub, D., Papakonstantinou, K. G., & Chatzi, E. (2024). POMDP
inference and robust solution via deep reinforcement learning: An application to railway optimal
maintenance. Machine Learning, 1-29. https://doi.org/10.1007/s10994-024-06559-2.
2. Arcieri, G., Rigoni, T., Hoelzl, C., Haener, D., Chatzi, E. (2024). Zurich, E. T. H., & SBB, S. F. R.
Ground penetrating radar for moisture assessment in railway tracks: An experimental investigation.
Journal of Physics Conference Series. https://doi.org/ 10.58286/29747.
3. Avci, O., Abdeljaber, O., Kiranyaz, S., Hussein, M., Gabbouj, M., & Inman, D. J. (2021). A review of
vibration-based damage detection in civil structures: From traditional methods to Machine Learning and
Deep Learning applications. Mechanical systems and signal processing, 147, 107077.
https://doi.org/10.1016/j.ymssp.2020.107077.
4. Barbosa, B., Serrador, A., & Casaleiro, J. (2023). Condition Monitoring Maintenance in Train Bogie: A
Low-Cost Acceleration Sensor Proposal. In 2023 International Conference on Control, Automation and
Diagnosis (ICCAD) (1-4). IEEE. https://doi.org/10.1109/ICCAD57653.2023.10152337.
5. Carnevale, M., La Paglia, I., & Pennacchi, P. (2021). An algorithm for precise localization of
measurements in rolling stock-based diagnostic systems. Proceedings of the Institution of Mechanical
Engineers, Part F: Journal of Rail and Rapid Transit, 235(7), 827-839.
https://doi.org/10.1177/0954409720965798.
6. Chrzan, M. (2022). Effect of uniform time on the transmission of signals in rail open systems. Archives
of Transport, 61(1), 39-49. https://doi.org/10.5604/01.3001.0015.8150.
7. Chudzikiewicz, A., & Kostrzewski, M. (2013). Analiza sygnałów wibroakustycznych w procesie
monitorowania stanu zawieszenia pojazdów szynowych oraz toru. Pojazdy Szynowe, 1(2013), 10-17.
https://doi.org/10.53502/RAIL-139165.
8. De Rosa, A., Alfi, S., & Bruni, S. (2019). Estimation of lateral and cross alignment in a railway track
based on vehicle dynamics measurements. Mechanical Systems and Signal Processing, 116, 606-623.
https://doi.org/10.1016/j.ymssp.2018.06.041.
9. Di Summa, M., Griseta, M. E., Mosca, N., Patruno, C., Nitti, M., Renò, V., & Stella, E. (2023). A review
on deep learning techniques for railway infrastructure monitoring. IEEE Access.
https://doi.org/10.1109/ACCESS.2023.3309814.
10. Faccini, L., Karaki, J., Di Gialleonardo, E., Somaschini, C., Bocciolone, M., & Collina, A. (2023). A
methodology for continuous monitoring of rail corrugation on subway lines based on axlebox
acceleration measurements. Applied Sciences, 13(6), 3773. https://doi.org/10.3390/app13063773.
144
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 127-145, 2024
11. Faghih-Roohi, S., Hajizadeh, S., Núñez, A., Babuska, R., & De Schutter, B. (2016). Deep convolutional
neural networks for detection of rail surface defects. In 2016 International joint conference on neural
networks (IJCNN) 2584-2589. IEEE. https://doi.org/10.1109/IJCNN.2016.7727522.
12. Firlik, B., Czechyra, B., & Chudzikiewicz, A. (2012). Condition monitoring system for light rail vehicle
and track. Key Engineering Materials, 518, 66-75.
https://doi.org/10.4028/www.scientific.net/KEM.518.66.
13. Hoelzl, C., Arcieri, G., Ancu, L., Banaszak, S., Kollros, A., Dertimanis, V., & Chatzi, E. (2023). Fusing
expert knowledge with monitoring data for condition assessment of railway welds. Sensors, 23(5), 2672.
https://doi.org/10.3390/s23052672.
14. Hoelzl, C., Dertimanis, V., Ancu, L., Kollros, A., & Chatzi, E. (2023). Vold–Kalman filter order tracking
of axle box accelerations for track stiffness assessment. Mechanical Systems and Signal Processing, 204,
110817. https://doi.org/10.1016/j.ymssp.2023.110817.
15. Hoelzl, C., Dertimanis, V., Landgraf, M., Ancu, L., Zurkirchen, M., & Chatzi, E. (2022). On-board
monitoring for smart assessment of railway infrastructure: A systematic review. The Rise of Smart Cities,
223-259. https://doi.org/10.1016/B978-0-12-817784-6.00015-1
16. Hoelzl, C., Keller, L., Simpson, T., Stoura, C., Kossmann, C., & Chatzi, E. (2024). Data-Driven Railway
Vehicle Parameter Tuning using Markov-Chain Monte Carlo Bayesian updating. In Journal of Physics:
Conference Series 2647 (18), 2024. IOP Publishing. https://doi.org/10.1088/1742-
6596/2647/18/182024.
17. Kostrzewski, M., & Melnik, R. (2021). Condition monitoring of rail transport systems: A bibliometric
performance analysis and systematic literature review. Sensors, 21 (14), 4710.
https://doi.org/10.3390/s21144710.
18. La Paglia, I., Carnevale, M., Corradi, R., Di Gialleonardo, E., Facchinetti, A., & Lisi, S. (2023).
Condition monitoring of vertical track alignment by bogie acceleration measurements on commercial
high-speed vehicles. Mechanical Systems and Signal Processing, 186, 109869.
https://doi.org/10.1016/j.ymssp.2022.109869.
19. La Paglia, I., Corradi, R., & Carnevale, M. (2022). Continuous monitoring of rail vehicle dynamics by
means of acceleration measurements. In Proceedings of the Fifth International Conference on Railway
Technology: Research, Development and Maintenance, Civil-Comp Press, Edinburgh, UK, 300 (1).
https://doi.org/10.4203/ccc.1.27.4.
20. Licow, R. (2018). Ocena uszkodzeń powierzchni tocznej szyn kolejowych za pomocą zjawisk
wibroakustycznych, [Doctoral dissertation, Poznan University of Technology]. Poznan University of
Technology Repository. https://sin.put.poznan.pl/dissertations/details/d106.
21. Mori, H., Tsunashima, H., Kojima, T., Matsumoto, A., & Mizuma, T. (2010). Condition monitoring of
railway track using in-service vehicle. Journal of Mechanical Systems for Transportation and Logistics,
3(1), 154-165. https://doi.org/10.1299/jmtl.3.154.
22. Muñoz, S., Ros, J., Urda, P., & Escalona, J. L. (2021). Estimation of lateral track irregularity through
Kalman filtering techniques. IEEE Access, 9, 60010-60025.
https://doi.org/10.1109/ACCESS.2021.3073606.
23. Naganuma, Y., Kobayashi, M., & Okumura, T. (2010). Inertial measurement processing techniques for
track condition monitoring on shinkansen commercial trains. Journal of Mechanical Systems for
Transportation and Logistics, 3(1), 315-325. https://doi.org/10.1299/jmtl.3.315.
24. Nielsen, J. C., & Li, X. (2018). Railway track geometry degradation due to differential settlement of
ballast/subgrade–Numerical prediction by an iterative procedure. Journal of Sound and Vibration, 412,
441-456. https://doi.org/10.1016/j.jsv.2017.10.005.
25. Nielsen, J. C., Berggren, E. G., Hammar, A., Jansson, F., & Bolmsvik, R. (2020). Degradation of railway
track geometry–Correlation between track stiffness gradient and differential settlement. Proceedings of
the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit, 234(1), 108-119.
https://doi.org/10.1177/0954409718819581.
26. Office of Rail Transport in Poland. (2023). Data on rail transport. https://www.utk.gov.pl (Access from
15.06.2024).
Licow, R., Tomaszewski, F.,
Archives of Transport, 71(3), 107-145, 2024
145
27. ONO, H., TSUNASHIMA, H., TAKATA, T., & OGATA, S. (2023). Development and operation of a
system for diagnosing the condition of regional railways tracks. Mechanical Engineering Journal, 10(3),
22-00239. https://doi.org/10.1299/mej.22-00239.
28. Rodriguez, A., Sanudo, R., Miranda, M., Gomez, A., & Benavente, J. (2021). Smartphones and tablets
applications in railways, ride comfort and track quality. Transition zones analysis. Measurement, 182,
109644. https://doi.org/10.1016/j.measurement.2021.109644.
29. Rosano, G., Massini, D., Bocciolini, L., Zappacosta, C., Di Gialleonardo, E., Somaschini, C., ... & Pugi,
L. (2024). Diagnostics of the railway track-Possibility of development through the measurement of
accelerations and contact forces| La diagnostica dell’armamento ferroviario-Possibilità di sviluppo
attraverso la misura di accelerazioni e forze di contatto. INGEGNERIA FERROVIARIA, 79(2), 81-102.
30. Sadri, M., Lu, T., & Steenbergen, M. (2020). Railway track degradation: The contribution of a spatially
variant support stiffness-Global variation. Journal of Sound and Vibration, 464, 114992.
https://doi.org/10.1016/j.jsv.2019.114992.
31. Sadri, M., Lu, T., & Steenbergen, M. (2019). Railway track degradation: The contribution of a spatially
variant support stiffness-Local variation. Journal of Sound and Vibration, 455, 203-220.
https://doi.org/10.1016/j.jsv.2019.05.006.
32. Stoura, C. D., Dertimanis, V. K., Hoelzl, C., Kossmann, C., Cigada, A., & Chatzi, E. N. (2023). A Model‐
Based Bayesian Inference Approach for On‐Board Monitoring of Rail Roughness Profiles: Application
on Field Measurement Data of the Swiss Federal Railways Network. Structural Control and Health
Monitoring, 2023(1), 8855542. https://doi.org/10.1155/2023/8855542.
33. Tsunashima, H., Naganuma, Y., & Kobayashi, T. (2014). Track geometry estimation from car-body
vibration. Vehicle System Dynamics, 52, 207-219. https://doi.org/10.1080/00423114.2014.889836.
34. Tsunashima, H., Ono, H., Takata, T., & Ogata, S. (2023). Development and Operation of Track
Condition Monitoring System Using In-Service Train. Applied Sciences, 13(6), 3835.
https://doi.org/10.3390/app13063835.
35. Uhl, T., Mendrok, K., & Chudzikiewicz, A. (2010). Rail track and rail vehicle intelligent monitoring
system. Archives of Transport, 22, 495-510. https://doi.org/10.2478/v10174-010-0030-1.
36. UNE EN 13848-1:2020 Railway applications - Track - Track geometry quality - Part 1: Characterization
of track geometry.
37. Wang, H., Berkers, J., van den Hurk, N., & Layegh, N. F. (2021). Study of loaded versus unloaded
measurements in railway track inspection. Measurement, 169, 108556.
https://doi.org/10.1016/j.measurement.2020.108556.
38. Wang, C., Xiao, Q., Liang, H., Liu, Y., & Cai, X. (2006). On-line monitoring of railway deformation
using acceleration measurement. In 2006 6th World Congress on Intelligent Control and Automation 2,
5828-5832. IEEE. https://doi.org/10.1109/WCICA.2006.1714194.
39. Westeon, P. F., Ling, C. S., Roberts, C., Goodman, C. J., Li, P., & Goodall, R. M. (2007). Monitoring
vertical track irregularity from in-service railway vehicles. Proceedings of the institution of mechanical
engineers, Part F: Journal of Rail and Rapid Transit, 221(1), 75-88.
https://doi.org/10.1243/0954409JRRT65.
.
