JPS6353490B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a wear detection method and apparatus for reliably identifying flat wear of wheels using correlation.

[Conventional technology]

In general, if flat wear occurs on the wheel treads of a railway vehicle, the noise generated during running will increase significantly, and the vibration impact on the vehicle itself and the track will also increase, which will affect safety, durability, etc. Therefore, it cannot be said that it is preferable. Therefore, wheels with flat wear exceeding a certain tolerance level need to be ground down.

Conventionally, as a method for detecting wheel flat wear, as described in Japanese Patent Application Laid-Open No. 131301/1982, the difference in arrival time of vibration waveforms due to flat wear to vibration meters installed at two points in the longitudinal direction of the rail (approximately 1 ms) was used. A wheel with flat wear is detected using a vibration meter, while another vibration meter installed between these vibration meters detects the magnitude of vibration.

[Problem that the invention seeks to solve]

However, according to the method described above, it is difficult to extract the time difference when the vibration waveform is complex, and it is difficult to extract the time difference when the vibration waveform is complex. If vibrations due to flat wear occur at almost the same time at the far end of a car, it is difficult to determine which vibration waveforms should have a time difference between them. The disadvantage is that it is difficult to identify. As explained above,
Wheels with flat wear should be reliably identified and ground down, but conventional methods have made it difficult to reliably identify them.

The purpose of the present invention is to prevent which wheels from generating vibrations, even when adjacent wheels with flat wear or wheels with large flat wear in the distance generate impact vibrations almost simultaneously. The object of the present invention is to provide a method for reliably detecting the presence of a substance and a device therefor.

[Means for solving problems]

The present invention provides at least three flat detectors in the longitudinal direction of the rail at intervals equal to the circumferential length of the wheel with the largest diameter, and also provides the position of the flat detector where the wheel in the longitudinal direction of the rail passes first. A wheel passing detector is installed at a position closer to the front, and a speed calculator is installed to calculate the running speed of the vehicle using the two wheel passing detectors. the difference in the arrival time of each vibration to the flat detector where the vibration first arrives and the flat detector where the vibration arrives next, and the difference in the arrival time of each vibration to the flat detector where the vibration first arrives and yet another flat detector. It consists of a correlator that calculates the difference in arrival times.

[Effect]

In the configuration, the difference in the arrival time of the vibrations generated in the rail due to flat wear of the wheels obtained by the device reaching each of the flat detectors, and The difference in distance between the passing flat detector and other flat detectors from the point of contact between each moving wheel and the rail to each flat detector is divided by the vibration propagation velocity in the rail, and the value is calculated over time. By determining the correspondence with the expected time difference curve of each wheel expressed in accordance with the above, the correspondence between the vibration detected by each flat detector and a wheel having flat wear is identified. That is, due to the difference in the time it takes for vibrations detected by at least three flat detectors to arrive at each flat detector, a wheel with flat wear is detected on the rail. The distance between each wheel and the installation interval of each flat detector,
From the running speed of the vehicle, the wheel is identified by determining the correspondence between the expected time difference curve, which is a diagram of the time period in which vibration can occur due to flat wear in each wheel, and the flat vibration detection time difference actually detected. This is what I decided to do.

〔Example〕

Hereinafter, one embodiment of the present invention will be described with reference to FIGS. 1 to 4. FIG. 1 shows a block diagram of one embodiment of the present invention. In the figure, 1 represents a rail, n and n+1 represent a front wheel and a rear wheel having an interval l in the vehicle traveling direction, and points D and E represent a wheel n and a wheel n+1.
This is the point of contact with the rail 1. 2d is a wheel passing detector installed at point A on rail 1 to detect the passage of a wheel, and 2e is a wheel passing detector installed at point A on rail 1 at an interval δ.
This is a wheel passing detector installed at point F, which is set apart. 2a, 2b, and 2c are shocks caused by flat wear of wheels provided on the rail 1 at points A, B, and C at intervals L that are approximately equal to the circumferential length of the wheel with the largest diameter among the various wheels that pass. This is a flat detector that detects vibrations. 3e, 3d are amplifiers for amplifying the outputs of the wheel passing detectors 2e, 2d; 3a, 3b, 3c are flat detectors 2a, 2;
This is an amplifier that amplifies the outputs of b and 2c. Reference numeral 4 indicates a level determination device for determining the level of the output 3'a of the amplifier 3a, and 5 receives the output pulse waveform 3'd of the amplifier 3d, and from the level determination device 4, wheel n and wheel n+1 approach point A. In response to the judgment signal when the vibration level is judged to be above a certain allowable level, a transmission start command is sent to transmit various outputs to the subsequent stage from the time when the passing pulse from the wheel n is input during the output pulse waveform 3'd. It is a vessel. 7 is a speed calculator to which the output pulse waveforms 3'd and 3'e are input and calculates the speed V of the wheel n from these outputs; 8 is the speed V of the speed calculator 7 and the wheel n stored in advance;
This is a wheel period calculator that calculates the wheel period from the circumference of the wheel. Reference numeral 9 denotes a recording time command device that stores various signals using the output of the wheel cycle calculation period 8, outputs a time length, that is, a recording time T, etc., and instructs to stop recording after the recording time T has elapsed. be. 1
0 counts the pulses of output pulse waveform 3'd,
This is a counter that outputs a number n which means that wheel n passes point A. Here, 6 is the transmission start command device 5
This is a switch circuit that opens in response to a command from the recording time command device 9 and closes in response to a command from the recording time command unit 9 after the recording time T has elapsed. Reference numerals 11 indicate the speed V which is the output of the speed calculator 7, the number n which is the output of the counter 10, the time such as the recording time T which is the output of the recording time command unit 9, and the wheel passing detection which is the output of the level determination device 5. Output pulse waveform 3'd of detector 2d, flat detector 2a, 2
This is a recording device into which the outputs of the amplifiers 3a, 3b, 3c, ie, the flat vibration signals 3'a, 3'b, 3'c, are input and recorded, and 12 is the output 11e, 11d of the horizontal writing device 11. , 11a, 11b,
11c is the recorded value 12e, 12d, 12a, 12
b, 12c, and the recorded value 12e includes a speed V, a number n, and a recording time T.
etc. are recorded, and the wheel pulse waveform is recorded in 12d.
12a, 12b, 12c have flat detectors 2.
Vibration waveforms caused by wheels with flat wear detected at points a, 2b, and 2c are recorded. 13 is a storage device 1 corresponding to the outputs of the flat detectors 2a and 2b;
1, that is, the vibration waveform outputs 11a and 11b are input, and a cross-correlation function 13ab between the vibration waveform outputs 11a and 11b is obtained.
1a and 11c are input, and the cross-correlation function 13
Each is a correlator that calculates ac. 14 is the number n
is recorded as the output 14e, and the cross-correlation functions 13ab, 13ac are recorded as the vibration waveforms 12a, 12
This is a correlation recording device that records cross-correlation function waveforms 14ab and 14ac at inter-interrogation positions corresponding to b and 12c.

Next, the operation of the above configuration will be explained. wheel n
passes point F on rail 1, a pulse is output from wheel passing detector 2e, and then wheel n
When the point A reaches just above the point A, a pulse is output from the wheel passage detector 2d. Therefore, the speed calculation unit 7 calculates the speed V= from the time interval T〓 of both pulses and the interval δ between both wheel passage detectors 2e and 2d.
δ/T〓 is calculated and output. The counter 10 counts the pulse waveform 3'd caused by the wheel n reaching point A, that is, directly above the wheel passing detector 2d, and outputs a number n. At the same time, the wheel period calculator 8 is given the diameter of each wheel of the passing train in advance, and uses the speed V, which is the output of the speed calculator 7, to calculate the rotation period Tn (Tn=πDn/
V and Dn are the diameters of wheel n) are calculated. Also,
Flat detectors 2a, 2b, 2 stored in advance
A time T _L =L/V is calculated using the interval L between wheels c and the speed V of the speed calculator 7, and a time T _lo =l/V is calculated from the interval l between wheels n and wheel n+1 and the speed V. In the recording time command device 9, the output of the wheel period calculator 8
Using T _o , T _L , and T _lo , if the recording time T for recording various signals as the start when the passing pulse of wheel n occurs is T _o <T _lo , then T = T _o , T _o > T When _lo , T=T _lo is set, and times T, T _L , T _o , and T _lo are output. However, T _o ≦T _L. On the other hand, until the wheel n approaches point A, the level of the flat vibration signal 3'a of the flat detector 2a is determined by the level determining device 4, and if it is above a certain level, the transmission start command device 5 transmits the flat vibration signal 3'a to the wheel passing detector 2d. A command is issued to start transmitting various signals starting from the time when the pulse of wheel n is generated in the output pulse waveform 3'd of , and the switch circuit 6 is closed. Therefore, the speed V, the number n, the times T, T _L , T _lo , and the pulse waveform 3'd of the wheel n are stored in the storage device 11, and the flat vibration signals 3'a, 3'
b, 3'c are stored. Incidentally, after the recording time T has elapsed, a command to stop recording is output from the recording time command device 9, and the switch circuit 6 is opened. Waveform recording device 1
2, the speed V, number n, times T, T _L , T _o , T _lo which are the outputs of the recording device 11 are recorded as the recorded value 12e, and the passing pulse waveform 3'd of the wheel n is recorded as the recorded value 12d. The flat vibration waveforms 3'a, 3'b, and 3'c are recorded as recorded values, that is, vibration waveforms 12a, 12b, and 12c, respectively. Further, flat vibration signals 3'a and 3'b and flat vibration signals 3'a and 3'c are inputted to the correlator 13 as vibration waveform outputs 11a and 11b and vibration waveform outputs 11a and 11c, respectively, and the vibration waveform Between the outputs 11a and 11b, the vibration waveform output 11
Cross-correlation functions 13 _ab and 13 _ac between a and 11c are calculated and output to the correlation recording device 14. In the correlation recording device 14, the count number n of wheel passing pulses is set as a recorded value 14e, and a cross-correlation function 13 _a
b , 13 _ac is the recorded value, that is, cross-correlation function waveform 1
Recorded as 4 _ab , 14 _ac . Examples of the recording are shown in FIGS. 2 to 4.

In FIG. 2, flat wear exists on wheel n, the rotation period T _o of wheel n is smaller than T _L , and
Assume that it is smaller than the time T _lo due to the wheel spacing. Therefore, the recording time T of the signal waveform is T=T _o .
In the same figure, V, n, T, T _L of the recorded value 12e,
T _o and T _lo are recorded numerically, and the wheel pulse waveform 12
d, 12a, 12b, 12c are time elapsed on the slave axis,
It is recorded in a diagram with the magnitude of the signal waveform on the horizontal axis. In addition, a numerical value is recorded for n in the recorded value 14e, and the cross-correlation function waveforms 14 _ab and 14 _ac represent the passage of time on the vertical axis, and the two-signal vibration waveforms 12a and 12b on the horizontal axis.
arrival time difference t, vibration waveforms 12a, 12 of the two signals
It is recorded in a diagram in which the arrival time difference t of c is taken. Then, the cross-correlation function waveform 14 is parallel to the horizontal axis.
_ab , 14 _ac are drawn. For supplementary explanation, the positions of wheel n and wheel n+1 in the rail longitudinal direction and the flat wear position of each wheel at a specific time are shown on the left side of the figure. Regarding cross-correlation function waveform 14 _ab , vibration waveforms 12a and 12 that may occur when wheel n has flat wear
The time difference between the two signals b, that is, the expected time difference curve 15 _o of wheel n, and the vibration waveforms 12a and 12b that may occur if there is flat wear on wheel n+1.
The time difference between the two signals, that is, the expected time difference curve 15o ₊₁ for wheel n+1 is also shown. Note that the expected time difference curve _15o or 15o ₊₁ is based on the flat detectors 2a and 2b of wheel n or wheel n+1.
This is a graph of the difference in the time it takes for the vibration to reach the rail surface, and shows that the flat wear of the wheel changes depending on the position where it contacts the rail surface. Therefore, the wheel where the maximum value of the cross-correlation function waveform 14 _ab matches the expected time difference curve is located between the flat detectors 2a and 2b, and its position on the rail can be determined from the matching point. be. Similarly, regarding the cross-correlation function waveform 14 _ac , the expected time difference curve 15' _o between the vibration waveforms 12a and 12c when wheel n has flat wear, and the vibration waveform 12a, when wheel n+1 has flat wear. The expected time difference curve 15'o ₊₁ between the two signals of 12c is also shown. Note that the solid line is the part that is actually recorded.

H _o in the wheel pulse waveform 12d is wheel n
The positions of wheel n and wheel n+1 on the rail 1 at that time are shown on the left.
Starting from Hn, the waveform is shown as a solid line for recording time T
recorded only during Here, before the above H _o
Vibration waveforms 12a, 12b, 12 at T ₁ hour
Vibration is occurring at c. Considering that the information at this time is after the recording time for wheel n-1, it is not actually recorded, but in the cross-correlation function waveform 14 _ab , at this time T ₁ , point A due to wheel n and wheel n+1 is recorded. , the maximum value or peak value of the cross-correlation function between the vibration waveforms 12a and 12b detected at point B occurs at a time difference _tL . In cross-correlation function waveform 14 _ac , the maximum value is at a time difference of 2·t _L. In other words, when the wheel is on the left side of point A (or on the right side of point C), the maximum value of the cross-correlation function of the vibration caused by any wheel is always the same as the cross-correlation function waveform 14 _ab and the time difference t _L , and the cross-correlation function waveform At 14 _ac , the time difference is 2·t _L. Therefore, when the wheel is on the left side of point A, it is impossible to determine which wheel is causing the vibration. However, the maximum value P of the cross-correlation function waveform _14ab of the vibration waveforms 12a and 12b generated after time _T2 with _Ho as the starting point occurs at the time difference _to , and the expected time difference curve _15o of the wheel n Since it occurs at a point that coincides with , it can be seen that the vibration is caused by wheel n. Even if wheel n+1 has flat wear and generates vibration,
Since the maximum value of the cross-correlation function waveform occurs at the time difference _tL , it can be seen that this is not caused by the wheel n.
The expected time difference curve 15 _o is the vibration waveform 12b relative to the vibration waveform 12a at time 0 when H _o occurs.
The time difference t is t _L , and as the wheel n moves from point A to point B, the time difference t decreases, and the time difference t decreases when wheel n moves to the center between point A and point B, that is, T _L /2 in time.
When the wheel n approaches the point B side and finally reaches the point B, the time difference t between the vibration waveform 12a and the vibration waveform 12b becomes 0 again.
t _L. The maximum value P of cross-correlation function waveform 14 _ab is determined by the time difference t at time T ₂ after time T _L /2.
Therefore, the vibration of the vibration waveform 12 _a(o) delayed by _the time difference t _o with respect to the vibration waveform 12 _b(o) is the vibration caused by the wheel n. (n) is wheel n
This means that it was caused by flat wear. On the other hand, at the same time T ₂ , vibration waveform 1
Cross correlation function waveform 14 of 2a and 12c Maximum value of _ac
P′ occurs at a time difference t′ _o , and t′ _o occurs at a point that coincides with the expected time difference curve 15′ _o of wheel n, and from now on, the vibration waveforms 12a and 12c are due to wheel n. It is determined that Vibration waveform 12 _a(o
) is determined to _be caused by the wheel _n . Recording of the waveform ends at T=T _o .

Then, when the vibration level of the vibration waveform 12a is equal to or higher than the allowable level, starting from the passing pulse H _o+1 of the wheel n+1 in the wheel pulse waveform 12d,
After time _T3 , vibrations shown in vibration waveforms 12a, 12b, and 12c are recorded.
2b cross-correlation function waveform 14 The maximum value of _ab is the time difference
Expected time difference curve 1 of wheel n+1 located at t _L
5. Since it is not on _o+1 , it is determined that the cause is not due to wheel n+1 located between points A and B, but due to wheel n located outside of points A and B. In addition, the maximum value of the cross-correlation function waveform 14 _ac is the expected time difference curve 1
The vibration waveforms 12c and 12a are determined to be caused by the wheel n, since they occur at _t''o , which coincides with _5'o , and do not occur at a location that coincides with the expected time difference curve 15'o ₊₁ . Then, vibration waveform 1 with time difference t _L delay with respect to vibration waveform 12 _b(o)
2 _a(o) vibration and the vibration waveform 12 It is determined that the vibration waveform 12 _{c(o), which} is a time difference t'' _o before the a(o) vibration, is the vibration caused by the wheel n. It can be said that the vibration waveform on the waveform recording device 12 at 2 is not caused by the wheel n.

In Fig. 3, both wheel n and wheel n+1 have flat wear, and the rotation period T _o of wheel n is equal to the time T _L
If the time is less than or greater than the time T _lo , then the recording time of the waveform for wheel n is T = T _lo ;
Assuming that the rotation period T _{o+1 of wheel n+1} is smaller than time T _L and smaller than time T _{l (o+1)} , recording time T=T _o+1 for wheel n. When the vibration level is higher than the vibration level, no vibration is recorded for a period of time T=T _lo (<Tn) in the vibration waveform starting from the passing pulse H _o by the wheel n at the recorded value 12d. Next, since a passing pulse of wheel n+1 is generated, the waveform is further recorded for time T=T _{o+1 starting from pulse H o+} 1.
Vibration waveform 12a, 1 after time T ₄ starting from H _o+1
2b and 12c are occurring. At this time, the maximum value Q ₁ of the cross-correlation function waveform 14 _ab of the vibration waveforms 12a and 12b occurs at a point that coincides with the expected time difference curve 15 _o , and the time difference is t _o , so the vibration waveform 12 _b(o) On the other hand, the vibration waveform 12a _(o) with a time delay t _o is due to the wheel n, and the maximum value Q' ₁ of the cross-correlation function waveform 14 _ac of the vibration waveforms 12a and 12c is the expected time difference curve 1
5' _o , and the time difference is t' _o , so the time difference t' _o for vibration waveform 12 _a(o) vibration.
Vibration waveform 12c _(o) vibration is caused by wheel n. Furthermore, the maximum value _{Q 2} of the vibration cross-correlation function waveform 14 ab of the vibration waveforms 12a and 12b occurs at a point that coincides with the expected time difference curve 15 _o+1 , and the time difference is t _o+1 , so the vibration waveform 12 _a Vibration waveform 12 _{b (o+1)} with time difference t _o+1 with respect to (o+1) is due to wheel n+1,
Cross-correlation function waveform 1 of each vibration waveform 12a, 12c
4 The maximum value of _ac occurs where it matches the expected time difference curve 15' _o+1 , and the time difference is t' _o+1 , so the time difference t' _o+1 with respect to vibration waveform 12 _a(o+1) The vibration waveform _12c(o+1) at is due to wheel n+1. Vibration waveforms with other relationships are wheel n,
It is not due to wheel n+1.

Next, in Fig. 4, the difference from Fig. 3 is that the time elapsed after the generation of the passing pulse H _o+1 of wheel n+1.
At T ₅ , the maximum value R of the cross-correlation function waveform 14 _ab of the vibration waveforms 12a and 12b is the expected time difference curve 1
5 _o and the expected time difference curve 15 _o+1 . In the cross-correlation function waveform 14 _ab , it cannot be determined whether wheel n or wheel n+1 is vibrating. However, vibration waveform 1
Maximum value R′ ₁ of the cross-correlation function of vibrations 2a and 12c
occurs on the expected time difference curve 15′ _o , and the maximum value R′ ₂
Since this occurs on the expected time difference curve 15'o ₊₁ , it can be seen that vibration occurs in both wheel n and wheel n+1. It can also be seen that the flat worn parts of wheel n and wheel n+1 are in contact at symmetrical positions between point A and point B.

〔Effect of the invention〕

As explained above, according to the present invention, even if the wheel diameter, wheel spacing, and flat impact generation time vary, it is possible to determine which wheel has generated how much impact, that is, how much flat wear has occurred. This has the effect of being able to reliably and quickly determine whether the

[Brief explanation of drawings]

FIG. 1 is a circuit diagram of an embodiment of a flat wear detection device according to the present invention, and FIGS. 2 to 4 are diagrams showing vibration waveform recording and cross-correlation function waveform recording in the embodiment. DESCRIPTION OF SYMBOLS 1...Rail, 2a-2c...Flat detector, 2d, 2e...Wheel passing detector, 4...Level judgment device, 5...Transmission start command device, 7...Speed leak counter, 8... Wheel cycle calculator, 9... Recording time command unit, 10... Wheel counter, 11... Storage device, 12... Waveform recording device, 13... Correlation meter, 1
4...Correlation recording device.

Claims (1)

[Scope of Claims] 1. At least three flat detectors are provided in the longitudinal direction of the rail at intervals equal to the circumferential length of the wheel with the largest diameter, and among the flat detectors, the flat detector through which the wheel passes first is A wheel passing detector is provided at the position and a position on the near side in the wheel running direction from that position, and the output of the flat detector provided at the position where the wheel first passes, and the output of the flat detector provided at the same position as the flat detector. Record the pulse output of the wheel passing detector and the output of each other flat detector, and find a cross-correlation function between the output of the first flat detector used as a reference and the output of each other flat detector. , the occurrence time of the maximum value of the cross-correlation function is defined as the time difference between the outputs of the respective flat detectors, and the value of the time difference is calculated from the contact between the wheel and the rail moving between the two flat detectors. The present invention is characterized by identifying the correspondence between a flat-worn wheel and vibration by determining whether or not the arrival time of the vibration to the device is on the expected time difference curve of the wheel shown in accordance with the movement of the contact point. A method for detecting flat wear on wheels. 2. At least three flat detectors arranged in the longitudinal direction of the rail at intervals equal to the circumferential length of the wheel with the largest diameter, one of which is located at the installation position of the flat detector through which the wheel first passes. , and two wheel passing detectors, the other of which is disposed on the front side in the wheel running direction from the installation position,
a level determination device that determines the impact level due to flat wear; a transmission start command device that operates based on the determination result of the level determination device; a speed calculator that calculates the vehicle speed from the output of the wheel passing detector; a recording time command device that determines the time length of signal storage based on the cycle; a counter that counts each passing of a wheel; a storage device that stores outputs from the flat detector, wheel passing detector, etc.; A wheel flat comprising: a waveform recording device that records the output waveform of the storage device; a correlator that calculates a cross-correlation function from the output from the storage device; and a correlation recording device that records the output of the correlator. Wear detection device.