null null

Rolling stock condition monitoring using emitted noise signal under car body
1
H.I. Koh, W.H. You, 2J.H. Jeon, Y.H. Kim
Korea Railroad Research Institute, Uiwang, Korea1;
Korea Advanced Institute of Science and Technology , Daejon, Korea2
Abstract
This paper concerns parametric studies and algorithms for online condition monitoring of rolling
stocks using noise signals measured under the car body. This study is primarily aimed at
detecting abnormal objects and their locations under driving circumstances.
The whole detecting system comprises microphone arrays, signal processing tools, which has
the function to detect the different signal compared to that from Rolling stocks under normal
operating condition. For the application of the developing system in the railway rolling stock
maintenance fields and for the users who are not familiar with noise signal processing, the
system produces a noise map of the under car section including information of the locations and
characteristics of components under unusual conditions. It will enable not only fast and detailed
examination of the rolling stock defects but also systematical data base system for the
maintenance works.
To do this the acoustical holography technology and beam forming technology are investigated.
These technologies have been using in various fields not only for machinery noise mapping, but
also for moving aircrafts and automobiles etc. For it is the main purpose to differentiate certain
patterns from a standard vehicle condition (and not just mapping the emitted noise), additional
parametrical studies and principles for noise mapping the under part of the moving railway
rolling stocks are required. First the noise characteristics of defected components under driving
circumstances have to be clarified. To apply this technology to the defect monitoring and
maintenance in rolling stocks the type and range of the defects has to be considered. To select
the appropriate algorithm technology the radiation and propagation has to be investigated under
various conditions such as speed, background noise and environment at the measuring
locations. Related topics are doppler effects, reflecting effects under the car body. For
optimization of the detecting and visualization technology, related parameters, their relation and
limits, boundary conditions are studied. The acoustical holography technology has been studied
based on frequency domain. The limits for usable frequency range which are dependent on
periods of the noise signal and train speed are found and studied.
At the end it was possible to draw crucial factors and optimal conditions. Under these conditions
it was also possible to find out the location of the unusual motion from the components. To
verify the applicability to the railway system model tests are performed and the results are
discussed.
Introduction
Detection systems used for fault diagnosis become more automated and efficient [1][2]. These
are aimed at enhancing the accuracy and the performance of maintenance processes but
another crucial object is the early detection in the driving condition with on-board monitoring
systems to prevent derailments of the vehicles or other accidents.
Acoustical holography technology has been using for various sound sources localisation and
visualisation in recent years. The spatial acoustical information obtained from microphone
arrays is processed for predicting and visualization of the source location. With a beam forming
method it is assumed that the sound source exists in a certain position and the sound field
caused by this virtual source is compared to the sound field generated from the real sound
source. Through repeated comparing process with changed position of the virtual source the
position of the real sound source is predicted. In acoustical holography method the information
about sound pressure distribution of the measuring plane is used to predict the sound field,
sound intensity of other planes. Recently this technology has been developing for application to
moving sources also and it was possible to extract a moving impulse source out of various
background noises. In this paper we focused on the development of the moving acoustical
holography technology to obtain an optimal algorithm and system for extracting the railway
vehicle fault patterns.
Noise characteristics of fault signals under vehicle driving circumstances
The main objects of the defect inspection in railway vehicles in recent years are cracks of
wheels, out-of-round wheels, brake detectors, defective bearings, hunting and bad behaviour of
the bogie and wheelset for examples indicating bad alignments. There are also on-board
monitoring systems in a revenue service for freight trains, which monitor bearings, wheels,
trucks, and brakes simultaneously. In general the defected objects of the moving vehicle
present periodical occurrence, which can be more or less a kind of impulsive signals. An
example of measured noise signal under a moving railway vehicle is shown in Figure 1, Figure 2
shows a noise signal in the same driving section as in Figure 1, but with a remarkable vibration
behaviour of the bogies. Along the wayside these signals are embedded in environmental and
dynamic noise. The background noise comes from friction rolling noise between wheels and the
tracks, noise from a natural environment, and noise from connectors between cars, etc. [3]. In
order to identify the target signal from the whole noise signal, the fault pattern in the driving
condition has to be clarified.
For this study we consider a periodically emitted impulsive noise signal (Figure 3) with an
impulse F0 as a first approach, which begins at the time t=t1 and passes by the microphone
arrays with a constant speed. The sound pressure measured at a microphone located at r at the
time t can be defined then as
p (r , t ) = F0 h(t ) ∗
∞
1
∑r
t = −∞ d
1
c
δ (t − (ti + iT + rd ))
Figure 1 : Example of noise signal measured under a moving vehicle
(1)
Figure 2 : Example of noise signal measured under a moving vehicle with
anomalous vibration behaviors of it’s bogies
T means the period of the impulse signal, c is sound velocity and h(t) represent the impact
response function. Since the sound source moves the distance rd between the sound source
and the measuring position varies with the time and therefore the time interval of the measured
impulse signal changes (Figure 4). The frequency spectrum of this signal can be written as
∞
r
1 j 2πf (ti +iT + cd )
e
.
i =−∞ rd
P (r , f ) = F0 H 0 ∑
Figure 3 : Emitted impulse signal
Figure 4 : Measured impulse signal of a moving sound source
(2)
H0 means the sound source considered as a monopole generated by an impulse. Therefore
moving periodical signal can be considered as a sum of the moving pure tones and this is used
for visualization of the source location by means of the moving acoustical holography method.
Condition-parameters, limits
In the moving acoustical holography spatially moving pure tones or sound with narrow band
frequency is measured by a fixed microphone array and an additional process in which the
Doppler effect is eliminated is included the. Figure 5 shows coordinate systems for a moving
holography process. Two coordinates are considered, namely a fixed coordinate, where the
microphone array is located and a hologram coordinate moves with the source plane which has
a constant moving speed u [1]. Then the sound pressure at the source plane can be visualized.
And since the sound level is pronounced at the sound source position it shows the location of
the sound source[2][3].
Figure 5 Coordinate systems for a moving holography method
As mentioned before using moving frame acoustical holography it is possible to visualize the
sound field for each frequency components with center frequency fc=i/T by means of band pass
filters. The narrow band signals have spectrums between (1-2M)fc and (1+2M) fc [1]. M means
the mach number and from it the frequency range, in which the sound field of the impact fault
signal can be visualized without error can be calculated :
fi p
1 − 2M 1
4M T
(3)
In Figure 6 two cases of results from the moving acoustical holography method are shown, the
first result is an example for a source localization within frequency range smaller than the error
limit in (3), the second (right in figure 6) one is a result from that of frequency range above the
limit value.
Figure 6 : Sound source localization by means of moving frame acoustical holography method ;
within error frequency limit (left), above error frequency limit(right)
f0 =120Hz~240Hz, with noise
z[m]
6
4
2
0
-10
-5
0
x[m]
5
10
Figure 7 : Sound source localization by means of moving frame acoustical holography method ;
measured impulse signal under a moving railway vehicle
It is clearly to see that for a optimal source localization using moving acoustical holography
method period T of the impulse fault signal and the moving speed are related crucial factors. In
figure 7 the location of the impulse signal is found that is extracted out of real vehicle rolling
noise measured under the car body.
Model test for detection of fault locations
A test model is constructed with a Microphone array including 18 small microphones (figure 9)
and two loudspeaker systems (Figure 8). The loudspeaker system moves with a constant speed
of 35cm/sec and can radiate different signal from each loudspeaker. The lower frequency limit
value for the experiment is chosen as 300 Hz and the period of the moving impulse signal is 0.1
sec. To extract the reflected sound from the floor two lines of arrays are installed with a distance
of 10 cm. In Figure 10 obtained results from a test are shown, in that one loudspeaker radiated
an impulse signal and the other loudspeaker white noise. The sound source radiated from the
Loudspeaker 2 is noticeable in Figure 10. Another experiment is done with two impulse signals.
The one has an impulse period T of 0.2 sec and the other 0.1 sec. We focused on the source 1,
Figure 11 shows that the acoustical holography method found this source location at the
frequency range between 305 Hz – 995 Hz.
Figure 8 Test model with two loudspeaker systems
Figure 9 Microphone array system
Concluding remarks
With this study the acoustical holography method is investigated and developed to localise fault
signals considered as periodically emitted impulse signal, which move with a constant speed. A
Microphone array is used and by means of moving frame acoustical holography the impulsive
sound has been separated into signals with different frequencies. For this process a condition
for the frequency limits, under which the source localization becomes possible to a certain
degree and the related parameters are studied. For verification of the developed algorithm a
model test is done, it could be shown that under the referred limit value of the frequency it is
possible to localise and visualize the moving impulse signal which is radiated with white noise or
with another impulse signal. Since this study is aimed at developing a system to detect the fault
signals of the railway vehicles this experiment is planned also for real railway vehicles. In the
future the relation between the sound to noise ratio and the accuracy of the localisation
technology will be studied and the parameters which affect the limits of the method in the real
driving condition will be investigated.
Loudspeaker 1
Loudspeaker 2
Figure 10 : Impulse source localization using hologram
(loudspeaker 1 : white noise, loudspeaker 2 : impulse signal)
Loudspeaker 1
Loudspeaker 2
Figure 11 : Impulse source localization using hologram (loudspeaker 1 : Impulse signal with a
period of 0.2 sec, loudspeaker 2 : impulse signal with a period of 0.1 sec)
References
[1]
H.-S. Kwon and Y.-H. Kim, "Moving frame technique for planar acoustic holography," J.
Acoust. Soc. Am. 103(4), 1734-1741, 1998.
[2] Michael Brandstein and Darren Ward(Eds.), Microphone arrays-Signal Processing
Techniques and Applications (Springer Verlag Berlin Heidelberg New York, 2001), chapter
1.
[3] J.D. Maynard, E.G. Williams, and Y. Lee, "Nearfield acoustic holography: I. Theory of
generalized holography and the development of NAH," J.Acoust. Soc..Am., 78, 1395-1413,
1985.

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project