AT522937A2 - Method for the detection of technical irregularities of railway vehicles from the analysis of sound and vibration data - Google Patents

Abstract

The invention relates to a method for analyzing sound or vibration waves emitted by one or more rail vehicles while driving with regard to the detection of possible technical irregularities on one or more wheels of the vehicles, which thereby pose a potential risk to the wheel and rail and it can lead to axle and / or rail breakage. The method is based on a sequence of automatically recorded and digitized sound emission values that the rail vehicle (s) emits while driving past the measuring point. As a result, one or more indicators are provided within the period under consideration, the value (s) of which serve as an indication of the existence of the irregularities sought and can be used for further interpretation.

Description

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

"Procedure for the detection of technical irregularities of railway vehicles from the analysis of sound and vibration data"

A procedure for the analysis of sound or vibration waves emitted by one or more rail vehicles ("train") during the journey is registered with regard to the detection of possible technical irregularities on one or more wheels of the vehicles, which thus represent a potential hazard to wheel and rail and axle and / or rail breakage can occur.

Such irregularities are often caused by so-called "flat spots" on the running surface of the wheel of a rail vehicle. These are usually caused by a temporary blockage of the brakes of a wheel during braking or by stones being thrown out. As a result, the wheel grinds on the rail and part of the running surface can be abraded, whereby the wheel loses its ideal circular circumference. After the blockade has been lifted, this flat spot (s) hit the rail with every revolution of the wheel during the further journey (s) and generate a characteristic knocking noise (“knocking” / “hammering”). Such situations are recognized according to the state of the art

(a) By visual inspection of the stationary train ("wagon master control"),

(b) By "eavesdropping" on the train while it is passing by an employee of the railway company (e.g. station master in train stations) or

(c) By measuring the wheel geometry with complex and high-priced trackside measuring systems (so-called “train monitoring systems”) in the route network.

This is cumbersome, not very accurate, costly, and not very efficient, if only because of the low "density" of test points in the railway network!

The method is based on a sequence of automatically recorded and digitized sound emission values that the rail vehicle (s) emits while driving past the measuring point.

As a result, one or more indicators are provided within the period under consideration, the value (s) of which serve as an indication of the existence of the irregularities sought and can be used for further interpretation (“flat spot indicator” / FS1). The process described minimizes the effort involved in setting up and operating the measuring equipment and improves the quality and quantity of the monitoring. This means that possible negative side effects (axle / rail breakage, failure of the vehicle, etc.) can be reacted to in good time and the process contributes to safe, efficient and highly available (rail) operation.

FSI_prio_V7otal.docx 1/14 Page 1 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Obtaining the output data:

The derivation of the basic data is shown in sketch 1 and is not part of the registration!

The method according to the invention is based on the evaluation of the sound or vibration emissions ("train noise") emitted by the rail vehicles of a train during travel, which is primarily generated by the rolling noise from the wheel / rail contact up to speeds of 200 km / h.

The (sound) emission emitted by the railway vehicle (s) of a train (1) while driving past the measurement cross-section Q (2) is picked up by a stationary electroacoustic pick-up (microphone) (3) positioned on the measurement cross-section Q at right angles to the track axis and recorded in converted into a continuous electrical signal "A" (4a).

The following description is based on the assumption that the sound emission is recorded by an electroacoustic sensor. However, other signal sources representing the sound emission, such as acceleration sensors, fiber-optic sensors, strain gauges, and others sensors, can also serve as signal sources.

The analog signal is then converted by an analog / digital converter (4) with an adjustable but fixed sampling interval taptası IN temporally equidistant point digital measured values (4a). The sampling interval tapıası is chosen in practice at least so that the original analog signal can be correctly reconstructed up to frequencies of at least 20kHz (Shannon's theorem).

The signal is recorded at least over a time range tyess, which shows the passage of the train or the railway vehicle to be measured - represented by the first and last axles of the vehicle / train - at the measurement cross-section Q plus an adjustable leading (gate) and lag time (tnacn ) Crosses over. This time range is referred to below as the “measurement time” (tyess) (3a). The sampling of the analog signal now results, depending on the sampling interval taptası and the measuring time tmess, a sequence of measured values that are equidistant in time with tavtası along the time axis.

This sequence of values “measured value (to)” to “measured value (t“) ”(5) forms the basis for the method according to the invention.

FSI_prio_V7otal.docx 2/14 Page 2 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Analytical basis:

The analytical basis of the described method for analyzing and calculating the “flat spot indicator” (“FSI”) is based on the task of identifying repetitive - that is, repetitive - step functions of the amplitude in the individual frequency bands in the frequency mix of the recorded (sound) signal . These jump functions in their entirety form the described "knocking noise" of a flat spot. The signal level does not necessarily have to significantly exceed the general level of noise emissions.

It can be represented by various mathematical models.

One of them is the “Dirac push”. (see relevant literature). A Dirac shock represents a single, very short and very large (sound) pulse, i.e. a strong increase in amplitude and a subsequent decrease in the (sound) signal in the time domain. A Dirac impact has the frequency distribution 1, that is, all frequencies are represented in it in equal parts. It follows that the 'hammering' sound of a flat spot should theoretically be recognizable in all frequency bands. In this approach, the recorded signal image is therefore examined for the presence of Dirac collisions.

In addition to the approach based on modeling using a “Dirac impact”, other approaches for modeling jump functions can also be used. In the following, the analysis is followed by modeling using a Dirac impact.

Another characteristic is a repetitive repetition of the occurrence of Dirac impacts over time due to the rotation of the wheel while driving. This repetition frequency depends on the wheel circumference and the current speed. The intended procedure takes this into account.

The observed frequency spectrum can - but does not have to - be restricted in practice to the range from 100 Hz to 20 kHz. This means that the signals can be recorded using common technical means, such as a weatherproof standard measuring microphone or a corresponding sensor system.

Likewise, the observed time period tmess can include the duration of the entire train's passage or also only a subset of the recorded and digitized measured values.

FSI_prio_V7otal.docx 3/14 Page 3 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Description of the method steps according to the invention

The process is based on 2 process steps, the sequence of which is shown in sketch 2: Process step 1 - "Mathematical analysis":

From the sequence of values measured value (to) .. measured value (t ") of the recorded and digitized (sound) signal, a 3-dimensional result matrix is formed with 4 consecutive steps. In the time domain, this contains information on possible repetitive jump functions via the measurement time.

Process step 2 - "FSI decision maker":

This step adopts the result matrix from step 1 and uses defined criteria to decide whether and, if so, where a possible technical irregularity could be identified in the sound signal.

As a result, the process identifies time ranges within the measurement time in which possible technical irregularities were detected - the "flat spot indicator" (fsi /)

FSI_prio_V7otal.docx 4/14 Page 4 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Step 1 - "Mathematical Analysis":

The 'mathematical analysis' carried out in the first step is a string of different mathematical / technical analysis methods in order to identify a repetitive 'hammering' noise.

To do this, the following steps are applied to the value sequence of the sound signal:

1) Division of the sampled signal curve into frequency bands 2) Time derivative

3) Autocorrelation of time intervals

4) Calculation of the repetition frequency

The result is a 3-dimensional vector fsi3 (t, fpand, Frep), in which t (x-axis) covers the measuring time, foana (y-axis) the frequency bands and fie (z-axis) the repetition frequencies per fbana and which can now contain an indication of possible repetitive jump functions. This vector is then passed on to the 2nd process step (“decision maker”) as input.

procedure

1) Division into frequency bands

A first method is to split the signal into frequency bands, which are logarithmic

are divided. The signal is subjected to a "Fast Fourier Transformation" (FFT). Then the powers of the FFT "bands" are added by adding the frequency bands

grouped according to the logarithm.

A second method is the division into 'fractional octave bands'. Bandpass filters with a fraction n of the bandwidth of an octave (factor 2 / n of the cutoff frequencies) filter the audio signal in order to divide it into the different frequency bands, whereby the following formulas apply:

FSI_prio_V7otal.docx 5/14 Page 5 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

k {1} felk) = fcent * 2 {2} fuck) = fe (k) * 22u {3} filk) = fo (k) / 22n The number of bands per octave Korean index of a 'fractional octave band' feent- ....- Center frequency of the frequency band k = 0 f- (k) ..... Center frequency of the frequency band k fu (k) ..... Upper limit frequency of the frequency band k fi (k) ..... Lower limit frequency of the frequency band k fmax -..... Maximum occurring frequency fmin ---. Minimum occurring frequency Kmax ..... index of the highest frequency band Kine ... index of the lowest frequency band

The center frequency of a frequency band is given as a reference. Here it is common to use the value 1000 Hz. Based on this, the center frequency, the upper and the lower limit frequency are calculated for all frequency bands.

Transforming the formulas {1} + {3} results in the formulas {4} and {5} for the calculation of the maximum and minimum index k. The function 'ld' is the / ogarithmus dualis with the base 2, 'floor' rounds down to the next whole number and 'ceil' rounds up to the next whole number.

{4} Kmax = floor (n x Id (Az) _ 3)

fcent

_ cei fmin \ + 1 {5} Kmin = ceil (n * 1d (Zen) + z) The maximum and minimum frequency fmax and fmin are parameters that must be specified for the calculation. fmax can be a maximum of half the sampling frequency of the microphone and fmin is sensibly chosen not to be lower than the smallest frequency that the microphone can still pick up. A good approach is fimin = 200 Hz and fmax = 20 kHz.

FSI_prio_V7otal.docx 6/14 Page 6 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Method:

(1) Fixing the number of bands per octave ("1") - can be freely selected (2) First the minimum and maximum value of the index k (kmax and kmin) is calculated. (3) Every integer from kmin to kmax is an index k for a 'fractional octave band'.

The following steps (4-8) are now carried out for each band:

(4) For each of these bands, the index k is used to calculate the center frequency f. (K), the upper limit frequency fu (k) and the lower limit frequency fi (k).

(5) The filter coefficients for a bandpass filter are now calculated using the upper and lower limit frequency of a fractional octave band (fu and fi) and the sampling frequency of the sound recording.

(6) The signal is then filtered through a digital filter with the calculated filter coefficients.

(7) The filtered signal is now divided into sequence blocks with a length of n, where: n> 0 and the form 2 ", ideally in the application n = 256, samples, whereby the blocks overlap half and in the time domain can be multiplied by a Hann window.

(8) Now the energy of the sound in each of these windows is calculated by squaring each sample value for itself, adding all values in a window and then calculating the logarithm of them.

The result is a 2-dimensional matrix which provides the signal energy over time in different frequency bands.

2) Temporal derivation In the next step, each frequency band is temporally derived. To get a flat spot too

recognize, the difference is taken over at least 8 samples

Since the characteristic of a flat spot is the rapid increase, only positive values are important. Negative results of the derivation are set to 0. This process is carried out individually for all frequency bands.

A 2-dimensional matrix is again provided as output, which provides the derived time course of the signal energy in different frequency bands.

FSI_prio_V7otal.docx 7/14 Page 7 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

3) Autocorrelation of time intervals

Autocorrelation is now used to examine whether peaks are repeated in the derived time signal. This step is carried out individually for each frequency band, so that the signal to be processed is only dependent on time [d (t)].

The signal is divided into blocks of an adjustable length (e.g. approx. 1.5 seconds), whereby the number of blocks should be a natural power of two (2, 4, 8, ...). With the parameters of the previous steps, blocks with e.g. 512 values are formed, which corresponds to about 1.56 seconds at a sampling frequency of 42 kHz. For the next block, the time is shifted by approx. 0.2 seconds, which in this example corresponds to a shift by 65 values.

Each block is then auto-correlated. Since the autocorrelation is symmetrical, only displacement values> = 0 are considered. This step again creates a vector with the same number of values as before the autocorrelation.

This vector is now normalized by dividing each value of the autocorrelation by the value at the shift = 0.

The result of this step is a 3-dimensional matrix. For each frequency band there is a vector every 0.2 seconds that contains the results of the autocorrelation as a function of the shift (lag).

FSI_prio_V7otal.docx 8/14 Page 8 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

4) Calculation of the repetition frequency

It is then analyzed whether there is a regularly repeating component in the autocorrelated signal. This step is carried out individually for each frequency band and for each time segment. Therefore, the only input here is the vector of the autocorrelation, which begins with the shift equal to 0 and, due to the normalization, with the value 1.

The FFT is now calculated from this vector in order to be able to consider regular signal parts. The first value of the autocorrelation is one. If the signal is shifted by the distance between two 'hammer blows', the autocorrelation is high again, with close to zero in between. Because the waveform begins with a maximum, the signal you are looking for corresponds to a cosine.

According to the FFT, the sine component in the time signal is represented by the imaginary part and the cosine component by the real part. Since only the cosine is of interest, only the real part of the result of the FFT - and only positive values - is used. All negative values are set to 0.

Since the input signal of the FFT is not complex (imaginary part = 0), the output vector of the FFT is symmetrical. Therefore, only the lower half of the vector is considered further after the FFT (only positive frequencies, not negative).

This frequency vector is now normalized by dividing it by its length.

The result is a frequency vector. The "frequency" is the rate / second at which a knocking noise is repeated and the value represents the "significance" or "signal strength" for these repetitions.

FSI_prio_V7otal.docx 9/14 Page 9 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Result of the mathematical analysis

The mathematical analysis provides a 3-dimensional matrix fs / 3 as output. The last step was carried out individually for each frequency band and individually for each point in time within each frequency band. The result is, in turn, a vector which supplies values depending on the repetition frequency. This is how three dimensions (axes) come about.

Dimension #ime: Distance between the value groups on the time axis Example: If time blocks with a shift of 0.2 seconds are created in step 3, there are separate values on the time axis every 0.2 seconds.

Dimension repetition frequency: vector with repetition frequencies

At each point in time on the t-axis there is a vector with different repetition frequencies for each frequency band. If a value in this vector is significantly high, this means that a “noise” is regularly repeated at this point in time with the corresponding repetition frequency.

Dimension frequency band

The complete analysis was carried out individually for each frequency band. If a high value appears at a certain point in time with a certain repetition frequency, the dimension of the frequency band corresponds to the frequency of the repetitive ('hammering') sound.

FSI_prio_V7otal.docx 10/14 Page 10 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Process step 2 - "FSIL decision maker":

In the second process step, the analysis of the result of the first process step now takes place. The “flat spot indicator” fsi depends on the frequency band (fband), time (t) and repetition frequency (frep).

In a further 3 consecutive steps, the decision (indication) is made as to whether possible flat spots (number> = 0) have been detected:

1. Elimination of the dimension fang - "foMerge"

The individual frequency bands are merged. A corresponding approach based on the median from the statistics is to adopt the second largest value of the dimension fiandg. This ensures that, on the one hand, a single “outlier” in a frequency band is not noticed and, on the other hand, several low values do not “pull the mean value down”. This results in a 2-dimensional matrix fsi2 (t, frep).

2. Identification of repetitions - "frGet"

The time dimension is now recorded in the 2-dimensional matrix fsi2 (t, frep) and each point in time is viewed as a vector depending on the repetition frequency: fSi (fep).

The maximum is searched for in this vector and used as the basis for a threshold value. The threshold value can vary according to requirements. Now neighboring values which exceed the threshold value are combined into groups with 1-n elements. The repetition frequency of the element with the highest fsi value is now adopted from each group. Now the factor of the respective repetition frequency to that of the group element "1" is calculated for the groups 2-m, the relative deviation to the next whole number is calculated for this factor and the absolute value is taken from this.

In this step, 5 one-dimensional result vectors were generated from the 2-dimensional input vector, which are only dependent on the time dimension and extend over the measurement time:

FSI_prio_V7otal.docx 11/14 Page 11 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Numgroups: number of groups

following: the largest number of consecutive elements of a group deviation: the maximum relative deviation (dimensionless, value range 0.0 to 0.5) fr: the fundamental oscillation (= the% ep of the 1st group)

fsi: The maximum flat spot indication value (dimensionless)

3. Indication of possible flat spots - "Detector"

The third step is the evaluation of the 5 vectors from step 2. The following criteria are used:

K1: maximum deviation not too large (e.g. deviation <0.1) K2: "narrow frequency groups" (e.g. following <4)

K3: repetition frequency not too high (e.g. fep <15)

If - depending on the selected parameters - the criteria K1, K2 and K3 apply, then it can be a possible flat spot.

As a possible additional sharpening of the decision, further criteria such as

K4: minimum consecutive number of values which K1 to K3 satisfy and

K5: a minimum threshold value which the maximum value fsS / imax must at least exceed

come into use.

{E8} [fsi = E / (K4 + K5)] = [fsi ‚(tbeg, tend, indicatorvalue)]

FSI_prio_V7otal.docx 12/14 Page 12 of 14

Priority registration FSI analysis Prosoft SUD

Sonsulting GmbH www.prosoftconsult.at

Example:

Spectoagram - Sit, Ab)

2

sis Fiat Spot Indicator [fbMenge] + Isat, fr}

+

4 a0 0.08 Sr “GO> & U OR> 40 DR

G

Ems is

X

Fiat Spot Indicator frGet] - fs} & fat}

ST x

S KA

Figure 1 (top) shows the frequency spectrum of a train over the duration of the measurement. F £ are the individual frequency bands.

Figure 2 (middle) shows the result after step 2 (fbpbMerge) with repetitive jump functions.

Figure 3 (bottom) shows the result of step 3 (frGet) for 2 of the result vectors fsi and frep.

After: applying: the: criteria K1: to: K5, the result is the assumption area of one or more flat spots on a wheel or several: wheels in a train (C. Flat spot indicator ‘), shown with the: orange line. In this example, the inspection even showed 2 flat spots on each wheel of two neighboring cars.

By mapping these results onto a train axis pattern recorded in parallel over the measurement time and / or automatic wagon identification reader, conclusions can be drawn about the wagons concerned and, if necessary, appropriate further measures can be initiated.

FSI_prio_V7otal.docx 13/14 Page 13 of 14

Priority registration FSI analysis Prosoft SUD

Zonsulting GmbH www.prosoftconsult.at

(a

Sketch1 a 2

Sketch 2 p

X

ASEISIMBTESER ANNIVESIS FSHEHSORSISE

N

Ze 1 listing In 1, AMerge frequency bands Eimuinalion dar Limensh

OR foanı

{RSS Spagener Kal ga) 3: Rn

Aettermalisıh analysis of FSHE niche

3. Temporal 2, irGet derivation idenfification of Wisder hopes

2 {ESHren

devinkan, FRE

Se

AIGERSAHSCHE ANRÖNR FS-Enfscheiger

&, Autgkorrelalion 3, Detector of Zeilintenvaßen 3 possible "Fachstefen" KZNI 3SENS

RA

, Schästung NANS> Enbschuntng

AEEEENEÄSCHE ANSO <

4, calculation of the repetition sequence

FSI_prio_V7otal.docx 14/14 Page 14 of 14