2
DESCRIPTION
Title of Invention: ABNORMALITY DETECTION APPARATUS FOR RAIL
VEHICLES
Technical Field
[0001]
The present invention relates to an abnormality detection
apparatus for rail vehicles.
Background Art
[0002]
Although there is increasing demand to improve performance
of maintenance of railway vehicles, frequent maintenance decreases
train operation efficiency. Therefore, in the field of railway
vehicles, a technical concept of carrying out maintenance at
required timing, which is called CBM (Condition Based
Maintenance), attracts attention for the purpose of, inter alia,
curtailing downtime during which systems and services in operation
will stop and reducing maintenance cost. In order to realize CBM,
there is a need for an abnormality detection apparatus capable of
early detecting abnormality regarding vibratory comfort and safety
of railway vehicles and deciding on a cause of the abnormality.
[0003]
As such an abnormality detection apparatus, for example,
there is a technology disclosed in Patent Literature 1.
Citation List
Patent Literature
[0004]
3
Patent Literature 1: Japanese Unexamined Patent Application
Publication No. 2004-170080
Summary of Invention
Technical Problem
[0005]
In Patent Literature 1, disclosed is an abnormality
detection apparatus that detects abnormality of a bogie, using
acceleration of a bogie frame, which is acquired by an
acceleration sensor provided in a bogie frame, and acceleration of
a wheel axle, which is acquired by an acceleration sensor provided
in a wheel axle. According to the abnormality detection apparatus
disclosed in Patent Literature 1, acceleration of a bogie is
estimated with a physical model to which acceleration of a wheel
axle acquired by an acceleration sensor is input and which outputs
acceleration of a bogie in normal operation. Then, depending on
whether or not estimated and measured values of the acceleration
are close, judgment can be made as to whether or not abnormality
occurs in the bogie.
[0006]
Here, in the abnormality detection apparatus disclosed in
Patent Literature 1, to detect abnormality of a bogie, it is
required to provide acceleration sensors respectively in both a
wheel axle and a bogie frame with respect to one bogie. However,
because larger vibration acceleration occurs in a wheel axle and a
bogie frame than in a vehicle body, acceleration sensors having a
high measurable range need to be used, which has brought about an
increase in installation cost. Besides, because a wheel axle and a
bogie frame have less space for installing an acceleration sensor
as compared with a vehicle body, also, there has arisen a problem
4
in which the degree of freedom of installing an acceleration
sensor is limited. Furthermore, because a great number of
acceleration sensors are required in the abnormality detection
apparatus disclosed in Patent Literature 1, also, there has
arisen, inter alia, a problem in which time and cost for
maintenance of these sensors increases.
[0007]
The present invention has been developed in view of the
problems noted above and is intended to provide an abnormality
detection apparatus for rail vehicles capable of deciding on
abnormality of what portion of a vehicle causing abnormal
vibration of a vehicle, when occurring, based on information from
sensors installed in vehicle bodies, while curbing costs for
installation and maintenance.
Solution to Problem
[0008]
To solve the problems noted above, one representative
abnormality detection apparatus for rail vehicles pertaining to
the present invention is achieved by comprising a vibration data
acquisition unit which acquires vibration data of a rail vehicle,
an operation data acquisition unit which acquires operation data
of a rail vehicle, a data analysis unit which distinguishes
whether each of rail vehicles traveling on a same track is placed
in normal or abnormal condition, based on the vibration data and
the operation data, and extracts vibration data of rail vehicles
in association with normal or abnormal condition, a comparison
processing unit which compares the vibration data of a rail
vehicle placed in normal condition and the vibration data of a
rail vehicle placed in abnormal condition and calculates a result
5
of comparison, and a diagnosis unit which decides on a cause of
abnormality of vibration in a rail vehicle distinguished as being
placed in abnormal condition, based on the result of comparison.
Advantageous Effects of Invention
[0009]
According to the present invention, it is possible to
provide an abnormality detection apparatus for rail vehicles
capable of deciding on abnormality of what portion of a vehicle
causing abnormal vibration of a vehicle, when occurring, based on
information from sensors installed in vehicle bodies, while
curbing costs for installation and maintenance. Thereby, it is
also enabled to curtail downtime of vehicles and reduce
maintenance cost.
Problems, configurations, and advantageous effects other
than described above will be made apparent from the following
description of embodiments.
Brief Description of Drawings
[0010]
Fig. 1 is a system configuration diagram of an abnormality
detection apparatus for railway vehicles, depicting a first
embodiment of the present invention.
Fig. 2 is a diagram depicting time historical waveforms of
vehicle body’s transverse acceleration in a case where abnormal
vibration has occurred in a vehicle B in the abnormality detection
apparatus for railway vehicles of the first embodiment.
Fig. 3 is a flowchart illustrating a processing procedure of
a data analysis unit 100 in the abnormality detection apparatus
for railway vehicles of the first embodiment.
6
Fig. 4 is a flowchart illustrating a processing procedure of
a comparison processing unit 200 in the abnormality detection
apparatus for railway vehicles of the first embodiment.
Fig. 5A is a diagram representing a relationship between a
measured value of vehicle body’s transverse acceleration PSD of a
vehicle A placed in normal condition and frequency.
Fig. 5B is a diagram representing a relationship between a
measured value of vehicle body’s transverse acceleration PSD of
the vehicle B placed in abnormal condition and frequency.
Fig. 6 is a diagram representing a PSD ratio between vehicle
body’s transverse acceleration PSD of the vehicle A placed in
normal condition and vehicle body’s transverse acceleration PSD of
the vehicle B placed in abnormal condition in the comparison
processing unit in the first embodiment.
Fig. 7 is a flowchart illustrating a processing procedure of
a parameter estimation unit 400 in the abnormality detection
apparatus for railway vehicles of the first embodiment.
Fig. 8A is a characteristic diagram representing both a
measured value (a solid line) and an analytic value (a dotted
line) before estimation by the parameter estimation unit in the
first embodiment.
Fig. 8B is a characteristic diagram representing both a
measured value (a solid line) and an analytic value (a dotted
line) after estimation by the parameter estimation unit in the
first embodiment.
Fig. 9 is a flowchart illustrating a processing procedure of
an abnormality cause estimation unit 400 in the abnormality
detection apparatus for railway vehicles of the first embodiment.
7
Fig. 10 is a diagram representing a result of estimating a
cause of abnormality by the abnormality cause estimation unit in
the first embodiment.
Fig. 11 is diagram illustrating a processing flow of a
parameter estimation unit in an abnormality detection apparatus
for railway vehicles of a second embodiment.
Fig. 12 is diagram illustrating a processing flow of a
parameter estimation unit in an abnormality detection apparatus
for railway vehicles of a third embodiment.
Description of Embodiments
[0011]
In the following, an abnormality detection apparatus for
rail vehicles of the present invention will be described with
reference to the drawings. Rail vehicles are vehicles that are
operated to travel along rail tracks laid and include railway
vehicles, monorail vehicles, streetcars, new transportation
vehicles, etc. Railway vehicles are taken as a typical example of
the rail vehicles and embodiments of the present invention are
described.
[0012]
[First Embodiment]
First, a configuration of an abnormality detection apparatus
for railway vehicles is described with reference to Fig. 1.
Fig. 1 is a functional block diagram depicting a system
configuration of the abnormality detection apparatus for railway
vehicles.
[0013]
Rail vehicles A and B (hereinafter referred to as vehicles A
and B) that travel on a rail track 20 vehicle each have a vehicle
8
body 1 and bogies 16, respectively. The vehicle body 1 is mounted
on the bogies 16 via air suspension spring 8. A bogie 16 is
comprised of, inter alia, a bogie frame 11, an air suspension
spring 8, a yaw damper 4, an axle box body 12, a wheel axle 13, an
axle spring device 14, and an axle box support rubber 15 which
serves as a housing of bearings for the wheel axle 13.
[0014]
The wheel axle 13 is rotatably held with axle box bodies 12,
elastic suspension in a vertical direction is provided by the axle
spring device 14 between an axle box body 12 and a bogie frame 11,
and elastic suspension in a transverse direction is provided by
the axle box support rubber 15. The air suspension springs 8 are
disposed between the vehicle body 1 and the bogie frames 11 and,
through these air suspension springs 8, the vehicle body 1 is
elastically supported by the bogie frames 11.
[0015]
On the floor of the vehicle body 1, for example, a vibration
data detection unit (a vibration data acquisition unit) 60 is
installed to acquire data on vibration of the vehicle body 1.
It should be noted that a physical quantity that the
vibration data detection unit 60 detects to acquire data on
vibration is not limited to acceleration; speed, displacement,
distortion, sound, etc. may be detected. However, description in
the present embodiment assumes that the vibration data detection
unit 60 detects acceleration.
[0016]
Besides, although the vibration data detection unit 60 is
installed in the vehicle body 1 having relatively plenty of space
for installation, the same may also be installed on, inter alia, a
bogie frame 11 and an axle box body 12, to enable it to detect
9
vibration of these parts. Description in the present embodiment
assumes that the vibration data detection unit 60 was installed in
the vehicle body 1.
[0017]
Besides, although multiple vibration data detectors 60 may
be installed in one vehicle, description in the present embodiment
assumes that a single vibration data detection unit 60 was
installed. Moreover, a position in which the vibration data
detection unit 60 is installed may be a central position of the
vehicle body 1, a position directly above a bogie 16, an end
portion of the vehicle body 1, such as a position where a vehicle
equipment compartment is present.
[0018]
Vibration data acquisition by the vibration data detection
unit 60 is applicable in any of the following directions: in
longitudinal, transverse, and vertical directions with respect to
a traveling direction of the railway vehicle, translational
(longitudinal translational, transverse translational, and
vertical translational) directions and a rotational (roll pitch
yaw) direction. Description in the present embodiment assumes that
the vibration data detection unit 60 detects vibration in a
transverse translational direction.
[0019]
The vehicle body 1 is equipped with an operation data
detection unit (an operation data acquisition unit) 50 having a
function to acquire operation data such as traveling speed, a
traveling position, and a vehicle occupancy rate from, e.g., an
operational administrative system which administrates operation
data.
[0020]
10
Additionally, in Fig. 1, an example is presented in which
two rail vehicles (hereinafter simply referred to as vehicles) are
each only equipped with the vibration data detection unit 60 and
the operation data detection unit 50; however, three or more
multiple vehicles may be equipped with them.
Besides, the vibration data detection unit 60 and the
operation data detection unit 50 may be provided in multiple
vehicles in one train set or can also be provided in each vehicle
in different train sets traveling on the same rail track.
[0021]
An abnormality cause detection device 30 includes a data
analysis unit 100, a comparison processing unit 200, a parameter
estimation unit 300, an abnormality cause decision unit 400, and a
decision result output unit 500 and these units constitute the
abnormality detection apparatus to detect a cause of abnormal
vibration of a railway vehicle. Additionally, the parameter
estimation unit 300 and the abnormality cause decision unit 400
constitute a diagnosis unit.
[0022]
Additionally, it is desirable to install the abnormality
cause detection device 30 in a ground facility such as a rail yard
where inspection and maintenance of vehicles are performed;
however, that device may be installed as on-vehicle equipment.
[0023]
The data analysis unit 100, comparison processing unit 200,
parameter estimation unit 300, abnormality cause decision unit
400, and decision result output unit 500 are, for example,
configured as those of a computing device to perform processing
tasks, which will be described later, according to programs stored
within the abnormality cause detection device 30.
11
[0024]
The data analysis unit 100 has a function to acquire, from
multiple vehicles, vibration data detected by the vibration data
detection unit 60 and operation data detected by the operation
data detection unit 50, distinguish whether a subject vehicle is
placed in normal condition or abnormal condition, and output
vibration data and operation data of the vehicles. A concrete
processing flow that is performed by the data analysis unit 100
will be described later.
[0025]
The comparison processing unit 200 has a function to acquire
vibration data and operation data of the vehicles placed in either
of abnormal and normal conditions, which has been output by the
data analysis unit 100, and output a result of comparison between
a vehicle in abnormal condition and a vehicle in normal condition,
based on the acquired data. A concrete processing flow that is
performed by the comparison processing unit 200 will be described
later.
[0026]
The parameter estimation unit 300 has a function to estimate
parameters regarding vehicle internal conditions from the result
of comparison between a vehicle in abnormal condition and a
vehicle in normal condition, which has been output by the
comparison processing unit 200.
[0027]
Additionally, the parameters regarding a vehicle internal
condition are those that are likely to be a cause of abnormality
of vibration of a railway vehicle and refer to, for example,
rigidity of an axle box support rubber 15 of a box support device
installed in a railway vehicle, an attenuation coefficient of a
12
yaw damper 4, and a wheel thread inclination of a wheel axle 13
among others. A concrete processing flow that is performed by the
parameter estimation unit 300 will be described later.
[0028]
The abnormality cause decision unit 400 has a function to
decide on a cause of abnormality from the estimated values of
parameters which has been output by the parameter estimation unit
300. A concrete processing flow that is performed by the
abnormality cause decision unit 400 will be described later.
[0029]
The decision result output unit 500 has a function to
notify, inter alia, a driver of the vehicle body 1 and a ground
operations manager or maintenance personnel of the cause of
abnormality which has been output by the abnormality cause
decision unit 400 via a monitor, a speaker, etc.
[0030]
Then, using Figs. 2 thru 10, descriptions are provided about
concrete processing flows by which the abnormality detection
apparatus for railway vehicles of the present embodiment decides
on a cause of abnormality.
[0031]
Here, the processing flows for abnormality detection are
described, using an example case where abnormal vibration has
occurred, attributed to a cause that is a decrease in the
attenuation coefficient because of oil leakage in the yaw damper 4
(see a vibration waveform b in Fig. 2).
Additionally, description in the present embodiment assumes
a case where an acceleration sensor in a transverse translational
direction (hereinafter referred to as a transverse acceleration
13
sensor) installed directly above the bogie’s position was used as
the vibration data detection unit 60.
[0032]
A vibration waveform a in Fig. 2 is a chart presenting
vehicle body’s transverse acceleration 621 measured by a
transverse acceleration sensor installed in a vehicle A in place
corresponding to the rail track position and a vibration waveform
b in Fig. 2 is a chart presenting vehicle body’s transverse
acceleration 622 measured by a transverse acceleration sensor
installed in a vehicle B in place corresponding to the rail track
position. In the vibration waveforms a and b, it is seen that the
higher the measured acceleration, the larger will be waveform
amplitude.
[0033]
With the assumption that yaw dampers 4 in the vehicle A are
normal, when oil leakage has occurred in a yaw damper 4 in the
vehicle B, attributed to a cause that is a decrease in the
attenuation coefficient of the yaw damper, the amplitude of
vehicle body’s transverse acceleration of the vehicle B becomes
larger, as compared with that of the vehicle A, and abnormal
vibration occurs in the vehicle B. This condition is revealed by
the fact that amplitude of the vibration waveform b is
significantly higher than amplitude of the vibration waveform a
(e.g., an average), as presented in Fig. 2.
[0034]
Fig. 4 is a flowchart to explain a processing procedure of
the data analysis unit 100 in the abnormality detection apparatus
for railway vehicles in Fig. 1. Operation according to the
flowchart in Fig. 3 is as described below.
[0035]
14
First, at step S111, the data analysis unit 100 acquires
vehicle body’s transverse acceleration measurements from the
transverse acceleration sensors in the multiple vehicles and
simultaneously acquires operation data such as a traveling
position, traveling speed, and a vehicle occupancy rate from the
operation data detection unit 50.
[0036]
Then, at step S112, the data analysis unit 100 performs
filtering on the vehicle body’s transverse acceleration
measurements acquired at step S111 using a commonly known
filtering technique.
[0037]
As the filtering, inter alia, preprocessing for analysis
processing to be performed at step S113 is performed, such as
processing to extract only a frequency band that is easily
perceivable by human senses from the vehicle body’s transverse
acceleration measurements and processing to calculate a value of
RMS (Root Mean Square) or a maximum of the acceleration
measurements.
[0038]
Moreover, at step S113, the data analysis unit 100 sets a
traveling section or the like where vibration is especially large
as a representative section from the filtered acceleration
measurements and performs processing to extract vehicle body’s
transverse acceleration data and operation data such as traveling
speed and a vehicle occupancy rate of each vehicle of, inter alia,
the vehicles A and B in this representative section.
[0039]
15
Through the processing at step S113, with regard to the
vehicles A and B, vibration data in the same section can be
extracted.
[0040]
It should be noted that a representative section can be
determined optionally, not limited to a traveling section where
vibration is large. Also, there may be one representative section
or two or more representative sections.
[0041]
Additionally, as for a representative section, it is
expedient to set a section for which, for each vehicle of, inter
alia, the vehicles A and B, data pieces are present, measured
under relatively similar measurement conditions including, inter
alia, the following: traveling speed of each vehicle is equal or
similar, the vehicle occupancy rates of each vehicle are similar,
and measurement dates and time of each vehicle are close.
[0042]
At step S114, the data analysis unit 100 compares vehicle
body’s transverse acceleration data pieces of the respective
vehicles in the representative section and performs processing to
distinguish whether each vehicle is placed in normal condition or
abnormal condition. For distinguishing whether a vehicle is placed
in normal condition or abnormal condition, a threshold based
processing may be performed, i.e., a threshold of vehicle body’s
acceleration is set beforehand and a determination is made as to
whether or not the threshold is exceeded; alternatively, a
commonly known method may be used, i.e., distinguishing is
performed using a method such as Malanobis-Taguchi System.
[0043]
16
In the case of the present embodiment, through the
threshold-based processing or the like, the vehicle B producing
large vibration that exceeds a threshold 922 which is presented in
Fig. 2 should be judged to be placed in “abnormal condition” and
the vehicle A whose vibration does not exceed the threshold 922 in
“normal condition”.
[0044]
At step S115, the data analysis unit 100 outputs the vehicle
body’s transverse acceleration data and operation data in the
representative section, extracted at step S113, in addition to
distinguishing information as to whether each vehicle is placed in
“normal condition” or “abnormal condition” to the comparison
processing unit 200.
[0045]
Fig. 4 is a flowchart to explain a processing procedure of
the comparison processing unit 200 in the abnormality detection
apparatus for railway vehicles in Fig. 1. Operation according to
the flowchart in Fig. 4 is as described below.
[0046]
First, at step S211, the comparison processing unit 200
acquires the distinguishing information, inter alia, indicating
that the vehicle A is placed in normal condition and the vehicle B
is placed in abnormal condition and the vehicle body’s transverse
acceleration data and operation data in the representative
section.
[0047]
Then, at step S212, from the acquired vehicle body’s
transverse acceleration data, the comparison processing unit 200
calculates vehicle body’s transverse acceleration PSD (Power
Spectrum Density) with respect to vehicle body’s transverse
17
acceleration of each vehicle. The PSD is a spectral function that
gives expression as a power value per unit frequency width (a
width of 1 Hz).
[0048]
In the case of the present embodiment, for the vehicle A,
the calculation gives a result as vehicle body’s transverse
acceleration PSD 721 of a vehicle placed in normal condition, as
is presented in Fig. 5A; for the vehicle B, the calculation gives
a result as vehicle body’s transverse acceleration PSD 722 of a
vehicle placed in abnormal condition, as is presented in Fig. 5B.
[0049]
As presented in Figs. 5A and 5B, in a given frequency band
900, the vehicle body’s transverse acceleration PSD 722 of the
vehicle B becomes larger than the vehicle body’s transverse
acceleration PSD 721 of the vehicle A.
[0050]
At step S213, the comparison processing unit 200 a
calculates a PSD ratio 821 that is a ratio between the vehicle
body’s transverse acceleration PSD 721 of the vehicle A placed in
normal condition and the vehicle body’s transverse acceleration
PSD 722 of the vehicle B placed in abnormal condition, as in Fig.
6.
[0051]
This PSD ratio 821 indicates a frequency band in which PSD
in abnormal condition differs from that in normal condition and a
degree of the difference. Here, the PSD ratio 821 increases
significantly in a frequency band 900, as is evident from Fig. 6.
[0052]
Additionally, after ensuring that there is little influence
of difference in response due to, inter alia, difference in rail
18
track conditions between sections, because vibration data in the
same section is extracted by the data analysis unit 100
beforehand, a change in response due to a change in transfer
characteristics between the normal and abnormal conditions of the
vehicles can be extracted as the PSD ratio 821.
[0053]
And now, an example where the PSD ratio 821 is used was
described in steps S212 and S213. However, any value indicating a
characteristic quantity regarding a change between the abnormal
and normal conditions can be used instead of the PSD ratio; as an
example, inter alia, a ratio between an RMS value of vehicle
body’s transverse acceleration in abnormal condition and that in
normal condition can be used.
[0054]
At step S214, the comparison processing unit 200 outputs
this PSD ratio 821 to the parameter estimation unit 300.
[0055]
Next, a processing flow of the parameter estimation unit 300
is described with Fig. 7. Operation according to the flowchart in
Fig. 7 is as described below.
[0056]
Additionally, description in the present embodiment assumes
a case where the parameters to be estimated to identify a cause of
abnormality are the spring constant of an axle box support rubber
15, the attenuation coefficient of a yaw damper 4, and the wheel
thread inclination of a wheel axle 13.
[0057]
First, at step S311, the parameter estimation unit 300
acquires the PSD ratio 821 which has been output by the comparison
19
processing unit 200. This is referred to as the measured PSD ratio
821.
[0058]
Then, at step S312, using a dynamic model, the parameter
estimation unit 300 calculates an analytic value of vehicle body’s
transverse acceleration PSD, when the parameters regarding vehicle
internal conditions have values under normal condition.
[0059]
This dynamic model is a model for predicting vibration
characteristics of a railway vehicle. Diverse dynamic models can
be used according to a type of vibration that one wants to
predict, such as a vehicle transverse dynamic model to predict
vehicle body’s transverse vibration and a vehicle vertical dynamic
model to predict vehicle body’s vertical vibration, when a virtual
vehicle in which the vehicle body 1, bogies 16, and wheel axles 13
are simulated with rigid bodies, axle box support rubbers 15, yaw
dampers 4, and others are simulated with springs and dampers, and
the respective rigid bodies are connected with the springs and
dampers travels on a virtual rail track that simulates track
irregularities, e.g., such as an alignment irregularity of a rail
track 20 (a degree of irregularity of the lateral surfaces of
rails in a longitudinal direction) and a gauge irregularity (an
error in a fundamental dimension of an interval between left and
right rails).
[0060]
Additionally, to define a virtual rail track that simulates
track irregularities, if there is actually measured track
irregularity data, the track irregularity data may be used.
However, if there is no track irregularity data measured, inter
alia, a method may be adopted in which track irregularity data
20
sets typical of track states (such as “good”, “ordinary”, and
“poor” states) are prepared beforehand and a virtual rail track is
defined by selecting and using one of these sets according to a
train line or the like of a vehicle in service.
[0061]
Besides, by using conditions such as traveling speed and a
vehicle occupancy rate detected by the operation data detection
unit 50 as the conditions for analysis with a dynamic model,
estimation with high accuracy reflecting actual traveling
conditions is possible.
[0062]
In the present embodiment, with the parameters such as the
spring constant of an axle box support rubber 15, the attenuation
coefficient of a yaw damper 4 and the wheel thread inclination of
a wheel axle 13 being set to normal values, an analytic value of
vehicle body’s transverse acceleration PSD is calculated using a
dynamic model.
[0063]
Additionally, as for the normal values, a design value or an
element test value is used for the spring constant of an axle box
support rubber 15 and the attenuation coefficient of a yaw damper
4 and, inter alia, a value in new product condition or a control
value in maintenance is used for the wheel thread inclination of a
wheel axle 13.
[0064]
At step S313 and subsequent steps, the parameter estimation
unit 300 estimates the parameters for the vehicle placed in
abnormal condition.
[0065]
21
At step S313, first, the parameter estimation unit 300 sets
initially estimated values of the parameters. The initially
estimated values of the parameters may be either normal values of
the parameters or randomly generated values.
[0066]
In the present embodiment, the parameter estimation unit 300
sets the initially estimated values of the spring constant of an
axle box support rubber 15, the attenuation coefficient of a yaw
damper 4, and the wheel thread inclination of a wheel axle 13
among others.
[0067]
At step S314, the parameter estimation unit 300 estimates
the parameters in abnormal condition using the measured PSD ratio
821 processed by the comparison processing unit 200.
[0068]
Using the initially estimated values of the parameters set
at step S313, first, the parameter estimation unit 300 calculates
an analytic value of vehicle body’s transverse acceleration PSD
with the parameters having the initially estimated values by using
the dynamic model.
[0069]
At step S314, then, the parameter estimation unit 300
calculates an analytic PSD ratio 822 that is a ratio between the
analytic value of vehicle body’s transverse acceleration PSD with
the parameters having the initially estimated values and the
analytic value of vehicle body’s transverse acceleration PSD with
the parameters having values under normal condition, calculated at
step S312.
[0070]
22
Then, the parameter estimation unit 300 determines estimated
values of the parameters so that the analytic PSD ratio 822 will
match the measured PSD ratio 821, i.e., an error will be
minimized, as is presented in Figs. 8A and 8B.
[0071]
As an index of error evaluation, any index that makes it
possible to evaluate the error can be used without restrictions;
nevertheless, in the present embodiment, a value is used that is
obtained by integrating difference between measured PSD ratio 821
and the analytic PSD ratio 822 per frequency of PSD ratio (this
value will be referred to as an integrated difference value,
hereinafter), as is presented in Figs. 8A and 8B.
[0072]
In the present embodiment, the estimated values of the
parameters are obtained to make matching between the measured PSD
ratio 821 which was calculated and the analytic PSD ratio 822, as
is presented in Fig. 8B, with regard to the spring constant of an
axle box support rubber 15, the attenuation coefficient of a yaw
damper 4, and the wheel thread inclination of a wheel axle 13
among others.
[0073]
Fig. 8A is a characteristic diagram with frequency being
plotted on the abscissa and PSD ratio plotted on the ordinate,
presenting the measured PSD ratio 821 represented in a sold line
and the analytic PSD ratio 822 represented in a dotted line before
parameter estimation and Fig. 8B is a characteristic diagram with
frequency being plotted on the abscissa and PSD ratio plotted on
the ordinate, presenting the measured PSD ratio 821 represented in
a sold line and the analytic PSD ratio 822 represented in a dotted
line after parameter estimation.
23
[0074]
Before parameter estimation, as in Fig. 8A, a large
difference between the measured and analytic values in some
frequency band in which abnormal vibration has occurred; whereas,
after parameter estimation, as in Fig. 8B, the measured and
analytic values almost match with little difference.
[0075]
After the termination of estimation, at step S315, the
parameter estimation unit 300 outputs the estimated values of the
parameters to the abnormality cause decision unit 400.
[0076]
And now, in the present embodiment, the description has been
provided, assuming that one transverse acceleration sensor was
installed in each vehicle body 1. On the other hand, assuming a
case where transverse acceleration sensors were installed in,
inter alia, the vehicle body 1, bogies 16, and axle box bodies 12
respectively, by updating the parameters so that an analytic value
will match multiple measured values of transverse acceleration
detected from these sensors, it is possible to make the accuracy
of parameter estimation higher.
[0077]
Next, a processing flow of the abnormality cause decision
unit 400 is described with Fig. 9. Operation according to the
flowchart in Fig. 9 is as described below.
[0078]
First, at step S411, the abnormality cause decision unit 400
acquires the estimated values of the parameters which have been
output by the parameter estimation unit 300.
In the present embodiment, the abnormality cause decision
unit 400 acquires the estimated values of the following: the
24
spring constant of an axle box support rubber 15, the attenuation
coefficient of a yaw damper 4, and the wheel thread inclination of
a wheel axle 13.
[0079]
Then, at step S412, the abnormality cause decision unit 400
calculates a ratio between the estimated and normal values of each
parameter.
[0080]
Then, at step S413, the abnormality cause decision unit 400
compares this ratio with a preset threshold with respect to each
parameter and judges a parameter as normal or abnormal.
[0081]
In particular, for a parameter of which this ratio is less
than or equal to the threshold, because it can be regarded as
close to normal, the abnormality cause decision unit 400 judges
the parameter as normal (step S414). On the other hand, for a
parameter of which this ratio exceeds the threshold, the
abnormality cause decision unit 400 judges the parameter as
abnormal (step S415).
[0082]
Then, at step S416, the abnormality cause decision unit 400
outputs a parameter judged as abnormal as a cause of abnormality
to the decision result output unit 500.
[0083]
Fig. 10 is a schematic diagram which schematically
represents an example of judging a parameter as normal or abnormal
in step S413 and represents a relationship between the ratio
between the estimated and normal values of a parameter and the
threshold with regard to elements that may be a possible cause of
abnormality, such as spring rigidity of an axle box support rubber
25
15, the attenuation coefficient of a yaw damper 4 and the wheel
thread inclination.
[0084]
In the example of Fig. 10, the attenuation coefficient of a
yaw damper 4 is judged as abnormal, as the ratio between the
estimated and normal values of this parameter exceeds the
threshold 91. Then, a result of decision that a decrease in the
attenuation coefficient of a yaw damper 4 is a cause of
abnormality is output from the abnormality cause decision unit 400
to the decision result output unit 500.
[0085]
Additionally, the larger the ratio between the estimated and
normal values of a parameter, the parameter deviates from its
normal value to a larger extent; therefore, the abnormality extent
of the parameter can be evaluated on how large the ratio is.
[0086]
The decision result output unit 500 notifies, inter alia, a
vehicle driver and a ground operations manager or maintenance
personnel of this result of estimating the cause of abnormality
through the use of a commonly known communication technology.
[0087]
As described hereinbefore, the abnormality detection
apparatus for railway vehicles of the present embodiment is
capable of estimating a cause of abnormality, like a decrease in
the attenuation coefficient of a yaw damper 4.
[0088]
[Second Embodiment]
An abnormality detection apparatus for vehicles pertaining
to the present embodiment has the same configuration as described
for the foregoing first embodiment; however, its parameter
26
estimation unit 300 performs processing which will be described
below with a flowchart in Fig. 11. And now, only details that
differ from the first embodiment are mainly described below.
[0089]
In the present embodiment, a step S312a is added as
processing that is performed by the parameter estimation unit 300
described in the first embodiment without using track irregularity
data prepared beforehand, such as measured data, with regard to
track irregularity data which is used for analysis with a dynamic
model.
[0090]
It should be noted that processing steps that are the same
as those described in the first embodiment are assigned identical
reference signs and their detail description is omitted.
[0091]
First, at step S311, the parameter estimation unit 300
acquires the measured PSD ratio 821 processed by the comparison
processing unit 200, as is the case for the first embodiment.
[0092]
Then, at step S312a, the parameter estimation unit 300
performs processing to identify track irregularities, based on a
measured value of vehicle body’s transverse acceleration PSD of a
vehicle placed in normal condition.
[0093]
Here, the parameter estimation unit 300 performs the
processing to identify track irregularity data so that the
measured value of vehicle body’s transverse acceleration PSD of
the vehicle will match an analytic value of vehicle body’s
transverse acceleration PSD of the vehicle calculated with a
dynamic model.
27
[0094]
In the case of the present embodiment, track irregularity
data is to be identified from vehicle body’s transverse
acceleration PSD of the vehicle A placed in normal condition.
[0095]
In steps S313 thru S315 of processing, the parameter
estimation unit 300 performs the same processing as in the first
embodiment using the track irregularity data identified at step
S312a.
[0096]
It should be noted that, although an example of using
vehicle body’s transverse acceleration PSD was described for the
present embodiment, no limitation to vehicle body’s transverse
acceleration PSD is intended, provided that track irregularity
data can be identified.
[0097]
As described hereinbefore, the abnormality detection
apparatus for railway vehicles of the present embodiment is
capable of deciding on a cause of abnormality without preparing
track irregularity data beforehand.
[0098]
[Third Embodiment]
An abnormality detection apparatus for vehicles pertaining
to the present embodiment has the same configuration as described
for the foregoing first embodiment; however, its parameter
estimation unit 300 performs processing which will be described
below with a flowchart in Fig. 12. And now, only details that
differ from the first embodiment are mainly described below.
[0099]
28
In the abnormality detection apparatus of the present
embodiment, step S314a and step S315a are added as processing that
is performed by the parameter estimation unit 300 in a case where
accuracy of estimating the parameters is low; these steps are
added to the abnormality detection apparatus of the foregoing
embodiments.
[0100]
As described for the first embodiment, at step S314,
parameter estimation unit 300 estimates the parameters so that an
integrated difference value between the measured PSD ratio 821 and
the analytic PSD ratio 822 of PSD ratio will be minimized. Here,
the integrated difference value becomes an index indicating
estimation accuracy.
[0101]
At step S314a, if this integrated difference value is less
than or equal to a preset threshold, the parameter estimation unit
300 judges that the estimation accuracy is more than or equal to a
reference level and outputs the estimated values of the parameters
to the abnormality cause decision unit 400 at step S315, as is the
case for the first embodiment.
[0102]
However, if the integrated difference value is not less than
or equal to the preset threshold, the parameter estimation unit
300 judges that the estimation accuracy is lower than the
reference level and outputs only information that the vehicle is
abnormal, but does not output the estimated values of the
parameters (including a parameter estimated to be a cause of
abnormality)
[0103]
29
As described hereinbefore, the abnormality detection
apparatus for railway vehicles of the present embodiment is
capable of outputting at least abnormality occurring in the
vehicle without stopping abnormality detection, even in a case
where the accuracy of estimating the parameters is not sufficient.
List of Reference Signs
[0104]
1 vehicle body, 4 yaw damper, 8 air suspension spring, 11 bogie
frame, 12 axle box body, 13 wheel axle, 14 axle spring device, 15
axle box support rubber, 16 bogie, 20 rail track, 30 abnormality
cause detection device, 50 operation data detection unit, 60
vibration data detection unit, 100 data analysis unit, 200
comparison processing unit, 300 parameter estimation unit, 400
abnormality cause decision unit, 500 decision result output unit.

We Claim:
[Claim 1]
An abnormality detection apparatus for rail vehicles,
characterized by comprising:
a vibration data acquisition unit which acquires vibration
data of a rail vehicle;
an operation data acquisition unit which acquires operation
data of a rail vehicle;
a data analysis unit which distinguishes whether each of
rail vehicles traveling on a same track is placed in normal or
abnormal condition, based on the vibration data and the operation
data, and extracts vibration data of rail vehicles in association
with normal or abnormal condition.
a comparison processing unit which compares the vibration
data of a rail vehicle placed in normal condition and the
vibration data of a rail vehicle placed in abnormal condition and
calculates a result of comparison; and
a diagnosis unit which decides on a cause of abnormality of
vibration in a rail vehicle distinguished as being placed in
abnormal condition, based on the result of comparison.
[Claim 2]
The abnormality detection apparatus for rail vehicles
according to claim 1, characterized in that:
the diagnosis unit comprises
a parameter estimation units which estimates parameters
regarding internal conditions of a rail vehicle, based on the
result of comparison calculated by the comparison processing unit;
and
31
an abnormality cause decision unit which decides on a cause
of abnormality of vibration in the rail vehicle, based on the
estimated parameters.
[Claim 3]
The abnormality detection apparatus for rail vehicles
according to claim 2, characterized in that:
the parameter estimation unit estimates the parameters
regarding internal conditions of a rail vehicle, based on a PSD
ratio of a rail vehicle placed in normal condition and a rail
vehicle placed in abnormal condition.
[Claim 4]
The abnormality detection apparatus for rail vehicles
according to claim 2 or 3, characterized in that:
the parameter estimation unit estimates the parameters
regarding internal conditions of a rail vehicle, based on a
dynamic model for predicting vibration characteristics of a rail
vehicle.
[Claim 5]
The abnormality detection apparatus for rail vehicles
according to claim 4, characterized in that:
the dynamic model is a rail vehicle transverse dynamic model
or a rail vehicle vertical dynamic model.
[Claim 6]
The abnormality detection apparatus for rail vehicles
according to any one of claims 2 to 5, characterized in that:
32
the abnormality cause decision unit decides on a cause of
abnormality of vibration in the rail vehicle, based on a result of
comparison between estimated values of the parameters regarding
internal conditions of a rail vehicle, estimated by the parameter
estimation unit, and normal values thereof.
[Claim 7]
The abnormality detection apparatus for rail vehicles
according to any one of claims 2 to 6, characterized in that:
the parameter estimation unit identifies track
irregularities based on vibration data of the rail vehicle.
[Claim 8]
The abnormality detection apparatus for rail vehicles
according to any one of claims 2 to 7, characterized in that:
the parameter estimation unit determines estimation accuracy
of estimated values of the parameters regarding internal
conditions of a rail vehicle, outputs a parameter estimated to be
a cause of abnormality, if the determined estimation accuracy is
more than or equal to a reference level, and outputs only
information indicating that a rail vehicle is abnormal, if the
determined estimation accuracy is lower than the reference level.
[Claim 9]
The abnormality detection apparatus for rail vehicles
according to any one of claims 1 to 8, characterized in that:
the vibration data acquisition unit detects vibration of a
vehicle body, a bogie frame, or an axle box body of the rail
vehicle.
33
[Claim 10]
The abnormality detection apparatus for rail vehicles
according to any one of claims 1 to 9, characterized in that:
the vibration data acquisition unit and the operation data
acquisition unit are provided in multiple rail vehicles in a
single train set or in each rail vehicle in different train sets
traveling on the same rail track.
[Claim 11]
The abnormality detection apparatus for rail vehicles
according to any one of claims 1 to 10, characterized in that:
the data analysis unit extracts vibration data of each rail
vehicle in a representative section which was set from among rail
vehicle traveling sections.
[Claim 12]
The abnormality detection apparatus for rail vehicles
according to any one of claims 1 to 11, characterized in that:
The comparison processing unit, if vibration data of each
rail vehicle exceeds a threshold, judges that the rail vehicle is
placed in abnormal condition and, if the vibration data does not
exceed the threshold, judges that the rail vehicle is placed in
normal condition.