Device and method for processing a digital signal

Abstract

[Device and method for processing a digital signal] The device for processing a digital signal comprises: - a Farrow structure (14) configured to apply to the digital signal a time-variable sampling frequency conversion from a fixed sampling frequency to a time-variable sampling frequency, the digital signal sampled at the time-variable sampling frequency being a resulting signal, and the Farrow structure (14) being configured to be controlled from a control variable, - spectral analysis means (15) configured to perform a spectral analysis of the resulting signal in order to determine the frequency values ​​of the resulting signal, - determination means (16) configured to determine a scattering parameter for the frequency values ​​of the resulting signal,and - control means (17) configured to modify the control variable according to the value of the scattering parameter. Reference: Figure 2,

Description

Title of the invention: Device and method for processing a digital signal Technical field of the invention

[0001] The present invention relates to devices for processing a digital signal and methods for processing a digital signal. Prior art

[0002] Condition-based maintenance algorithms, for example for detecting a fault in a bearing such as a fault in an internal raceway of the bearing, require a constant rotational speed of the bearing to function correctly.

[0003] In general, condition-based maintenance algorithms perform a spectral analysis of the rotating bearing to detect defects from specific tones and harmonics.

[0004] Variations in bearing rotational speed during signal measurement spread the spectral tones, which reduces the ability to identify frequencies and harmonics that may be associated with defects, for example defects in an inner raceway of the bearing or defects in the outer raceway of the bearing.

[0005] Therefore, the present invention aims to correct the deleterious effects of changes in rotational speed to perform spectral analysis. Summary of the invention

[0006] According to one aspect, a method for processing a digital signal is proposed comprising samples of a continuous signal sampled at a fixed sampling frequency, the continuous signal being representative of vibrations of a bearing undergoing changes in rotational speed.

[0007] The process comprises:

[0008] - the application of a sampling frequency conversion to the digital signal variable in time by a Farrow structure, from fixed sampling frequency to time-variable sampling, the digital signal sampled at time-variable sampling being a resultant signal, the Farrow structure being controlled from a control variable.

[0009] - performing a spectral analysis of the resulting signal to determine the values of the resulting signal frequency,

[0010] - the determination of a parameter for scattering the frequency values ​​of the signal resulting, and

[0011] - the modification of the control variable according to the value of the parameter of scattering.

[0012] The method allows the use of existing algorithms with constant rotation speed under conditions of variable rotation speed, so that it is not necessary to make repetitive attempts to acquire data at constant speed.

[0013] Eliminating repetitive attempts to acquire data saves the power supply of a device implementing the method, for example a wireless power supply including a power source such as a battery.

[0014] Advantageously, performing a spectral analysis of the resulting signal includes performing a fast Fourier transform of the resulting signal.

[0015] Preferably, the statistical parameter of the frequency values ​​is the asymmetry of the frequency values.

[0016] Advantageously, the modification of the intermediate sampling includes the implementation of a line search algorithm to modify the control variable.

[0017] According to one aspect, a digital signal processing device is proposed comprising samples of a continuous signal sampled at a fixed sampling frequency, the continuous signal being representative of vibrations of a bearing undergoing changes in rotational speed.

[0018] The device comprises:

[0019] - a Farrow structure configured to apply to the digital signal a time-variable sampling rate conversion from a fixed sampling rate to a time-variable sampling rate, the digital signal sampled at the time-variable sampling rate being a resultant signal, and the Farrow structure being configured to be controlled from a control variable,

[0020] - spectral analysis means configured to perform a spectral analysis of the resulting signal in order to determine the frequency values ​​of the resulting signal,

[0021] - means of determination configured to determine a parameter scattering of the frequency values ​​of the resulting signal, and

[0022] - control means configured to modify the control variable in function of the value of the scattering parameter.

[0023] Preferably, the spectral analysis means are configured to implement a fast Fourier transform algorithm.

[0024] Advantageously, the statistical parameter of the frequency values ​​is the asymmetry of the frequency values.

[0025] Preferably, the control means are configured to implement an al- line search algorithm in order to modify the command variable.

[0026] According to one aspect, a rolling device is proposed.

[0027] The bearing device comprises:

[0028] - a bearing equipped with an inner ring and an outer ring capable of rotate concentrically relative to each other,

[0029] - a sensor configured to measure the vibrations of said inner ring or ex external and configured to deliver a continuous signal

[0030] - a sampler configured to sample the continuous signal at a frequency fixed sampling and configured to deliver the digital signal comprising the samples, and

[0031] - a device as defined above, configured to process the digital signal. Brief description of the figures

[0032] Other advantages and features of the invention will become apparent upon examination of the detailed description of embodiments, which is in no way limiting, and the accompanying drawings in which:

[0033] Fig. 1 schematically illustrates a rotating machine according to the invention;

[0034] Figure 2 schematically illustrates an example of a processing device for a digital signal according to the invention;

[0035] Figure [Fig. 3] schematically illustrates an example of a Farrow structure;

[0036] Figure 4 schematically illustrates a method for determining the coefficients of the Farrow structure;

[0037] Figure 5 schematically illustrates a method for processing a digital signal according to the invention; and

[0038] Fig. 6 schematically illustrates examples of the spectrum provided by spectral analysis means for a constant rotational speed of a bearing, when the speed of the bearing is reduced during the measurement according to the prior art and according to the invention. Detailed description of the invention

[0039] Reference is made to [Fig.1] which schematically represents a partial longitudinal cross-section of a machine 1.

[0040] The machine 1 comprises a housing 2 and a shaft 3 supported in the housing 2 by a bearing 4 (for example, a roller bearing or a ball bearing).

[0041] The bearing 4 is provided with an inner ring 5 mounted on the shaft 3 and an outer ring 6 mounted in the bore of the housing 2. The outer ring 6 radially surrounds the inner ring 5. The inner and outer rings 5 ​​and 6 rotate concentrically with respect to each other.

[0042] The bearing 4 is further provided with a row of interposed rolling elements 7 ra- The bearing is located between the inner and outer raceways of the inner and outer rings 5 ​​and 6. In the example shown, the rolling elements 7 are balls. Alternatively, the bearing may include other types of rolling elements 7, for example, rollers. In the example shown, the bearing comprises one row of rolling elements 7. Alternatively, the bearing may comprise several rows of rolling elements.

[0043] A sensor 8 is mounted in the housing 2 to measure the vibrations of the bearing 4 subjected to changes in rotational speed.

[0044] The sensor 8 can be mounted on a bore of the housing 2.

[0045] Alternatively, the sensor 8 can be mounted elsewhere on the machine, near the outer ring 6 or close to the housing 2, for example.

[0046] The sensor 8 delivers a continuous signal S8 representative of the vibrations of the bearing 4 to an input of a sampler 9.

[0047] The sampler 9 delivers a digital signal S9 comprising multiple sequential samples xp of the continuous signal S 8 sampled at a fixed sampling frequency to an input 101 of a device 10 for processing the digital signal S9, p being an integer.

[0048] The bearing 4, the sensor 8, the sampler 9 and the device 10 form a bearing device.

[0049] A memory (not shown) can store the output signal S9 and deliver the output signal S9 to the device 10.

[0050] A first output 102 of the device 10 is connected to first implementation means 11 implementing at least one constant-speed algorithm in the time domain from a first output signal S102 delivered by the device 10 on the first output 102, for example to implement an algorithm for detecting enveloping defects.

[0051] A second output 103 of the device 10 is connected to second implementation means 12 implementing at least one constant-speed spectrally oriented algorithm from a second output signal S103 delivered by the device 10 on the second output 103, for example to implement a fast Fourier transform coupled with a fault frequency detection method.

[0052] The first and second means of implementation 11,12 are for example each made up of a processing unit implementing said algorithm.

[0053] A processing unit 13 implements the sensor 8, the sampler 9 and the device 10.

[0054] Fig. 2 schematically illustrates an example of device 10.

[0055] The device 10 comprises a Farrow structure 14 known from US patent 4,866,647, spectral analysis means 15, determination means 16 and control methods 17.

[0056] The Farrow structure 14 includes an input 141 connected to the input 101 of the device 10, an output 142 connected to the first output 102 of the device 10 and to an input 151 of the spectral analysis means 15, and includes a control input 143 connected to an output 172 of the control means 17.

[0057] An output 152 of the spectral analysis means 15 is connected to the second output 103 of the device 10 and to an input 161 of the determination means 16.

[0058] An output 162 of the determination means 16 is connected to an input of the control means 17.

[0059] The spectral analysis means 15 include a spectral analysis algorithm ALGO1, for example a fast Fourier transform algorithm.

[0060] The means of determination 16 include an ALGO2 algorithm for determining a scattering parameter.

[0061] The scattering parameter includes, for example, asymmetry or flattening.

[0062] It is assumed that the ALGO2 algorithm determines the asymmetry in such a way that the output determination means 16 emit the asymmetry of the data received at the input 161 of the determination means 16.

[0063] The control means 17 include a line search algorithm ALGO3, for example a fixed step size, a speed change, a golden section or a gradient estimation.

[0064] The Farrow 14 structure iteratively adjusts the inter-sample delays (resampling) of the measurement data.

[0065] The Farrow 14 structure is based on a finite impulse response FIR filter of order N whose coefficients h(n, A) can be modified by means of a control variable A equal to the inter-sample position or the delay of the Farrow structure, n being an integer between 0 and N.

[0066] The coefficients of the filter h(n, A) are formed from a polynomial of the control variable A.

[0067] The coefficient h(n, A) is equal to: 100681

[0069] where M is the order of a set of polynomials whose value is chosen according to performance requirements. The matrix C represents the collection of each coefficient Cmn of each polynomial of order M for each filter coefficient h(n, A). Each coefficient Cmn implements the (n+1) hole filter implementing a delay of A denoted by h(n, A).

[0070] The coefficients Cmn can be represented in the form of a matrix of coefficients C of dimension (M+l)x(N+l).

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078]

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086]

[0087]

[0088] oo ON (2) -MO MN The transfer function H(z, A) of the Farrow 14 structure is given by: with CC for m varying between 0 and M, and n varying between 0 and N. The term Cm(z) denotes a sub-filter of the Farrow 14 structure, the Farrow 14 structure comprising M+l sub-filters. Figure [Fig. 3] schematically illustrates an example of the Farrow 14 structure. The Farrow 14 structure includes M+l sub-filters denoted • ••, C] (z), Co( z), M+l memory banks 19, 20, 21, M multipliers 22, 23 each having a variable gain G22, G23, and M adders 24, 25. Each multiplier 22, 23 includes an input, an output delivering a signal received at the input multiplied by the variable gain G22, G23, and a control input receiving the value of the variable gain. Each adder 24, 25, includes a first and a second input, and an output delivering the sum of the first and second inputs. Each subfilter CM(z), • • •, (z), Co(z) includes an input 26, 27, 28 connected to input 141 of the Farrow structure and an output 29, 30, 31 connected to an input of a different memory bank 19, 20, 21. An output of the Me memory bank 19 is connected to the input of the Me multiplier 22. Each output of the M-le to the first memory banks 20, 21 is connected to the first input of a different adder 24, 25. The output of the multiplier Me is connected to the input of the adder in the next stage. For example, as shown in [Fig. 3], the output of the multiplier 23 is connected to the second input of the adder 25. The output of the final adder 25 is connected to the output 142 of the Farrow structure 14. Each memory bank 19, 20, 21 is connected to the control input 143 to select the memory data element from each memory bank 19, 20, 21 which is passed to the multiplier G22, G23 and the adders 24, 25. The control input of the variable gains G22, G23 is connected to the control input 143 of the Farrow structure 14 to control the value of the gains variables each having the same value.

[0089] Since the structure of the sub-filters CM(z), • • •, Cj(z), (z) is identical, only The structure of the CM(z) sub-filters is detailed.

[0090] The subfilter CM( z ) comprises a chain of N delay elements D 33, 34, N second multipliers 35, 36, 37, 38 and a second summing 39.

[0091] The N second multipliers 35, 36, 37, 38 multiply N+l sequential samples xn of the signal received at the input 26 of the sub-filter C^(z) Pæ" the N+l coefficients of the filter CM0 to CMN and deliver the multiplied sequential samples xn to the second summing 39.

[0092] The second summer 39 adds the N+l sequential samples xn multiplied by the N+l second multipliers 35, 36, 37, 38 and provides the sum at the input of the memory bank 19.

[0093] Figure 4 illustrates an example of a method known in the prior art for determining the coefficients Cmn of the matrix of coefficients C of dimension (M+l)x(N+l).

[0094] In a step 40, the order N of the Farrow structure 14 and the order M of the polynomial are defined as a function of the accuracy required for the device 10.

[0095] At each instant, the control value A of each multiplier G22, G23 is the same.

[0096] In another embodiment, the control value A of each multiplier G22, G23 can vary from one sample to another and is denoted An. The values ​​of the fractional delay A„ are controlled by a control unit of the device 10 (not shown) and are chosen according to the fractional delay required to be applied to each input sample xn. The control unit also selects the data sample extracted from the memories 19, 20, 21 according to the fractional delay An in order to obtain the necessary integer component of the required delay.

[0097] To allow the implementation of a continuous range of delays, the Farrow structure relies on a polynomial curve fitting based on a set of fixed-delay reference filters. For example, assuming a bank of 8 reference filters, each implementing a fixed delay, the fixed delay A for each reference filter could be chosen between -0.5 and 0.5 in increments of 0.125 such that the integer j varies between 0 and 7 with do = -0.5, . J] = -0.375, ... / 17= +0.375.

[0098] In a step 41, a set of functions ^ / n, dp is calculated for each value j and a given value n.

[0099] The function Sj can, for example, be equal to:

[0100] q (n j.) (5)

[0101] In step 42, the coefficients Cmn of the coefficient matrix C are determined such that, for a given value n and the desired delay value A, the filter coefficient h(n, A) corresponds to the polynomial interpolation of the functions ^ / (n, dj) defined by Cmn for all values ​​of A between -0.5 and +0.5.

[0102] Fig. 5 illustrates an example of the implementation of device 10.

[0103] We assume that the coefficient matrix C is defined and that the sub-filters CZ)> • • •, C} ( z ), Co( z ) are parameterized as a function of the coefficient matrix C.

[0104] In a step 50, the sampler 9 delivers the digital signal S9 comprising the samples xp of the continuous signal S8 delivered by the sensor 8.

[0105] In a step 51, the Farrow device 14 applies to the digital signal S9 comprising the samples xp a conversion of the time-variable sampling frequency from the fixed sampling frequency to a time-variable sampling interval, and delivers on the output 142 of the Farrow device 14 a resulting signal comprising the digital signal sampled at the time-variable sampling interval achieving the compensation of the change in speed.

[0106] The resulting signal is the first output signal S102 delivered on the first output 102 of the device 10.

[0107] The temporal variation of the resulting sampling depends on the value of the control value A delivered by the control means 17.

[0108] At step 52, the spectral analysis means 15 perform a spectral analysis of the resulting signal S102 to determine the frequency values ​​of the resulting signal S102 using the ALGO1 algorithm and output the second output signal S103 on the output 152 of the spectral analysis means 15.

[0109] At step 53, the determination means 16 determine the asymmetry of the frequency values ​​of the spectral analysis using the ALGO2 algorithm and output the determined asymmetry value on the output 162 of the determination means 16.

[0110] At step 54, the control means 17 implement the ALGO3 line search algorithm on the asymmetry value and provide a control value of the control variable A on the control input 143 of the device 10 in order to reduce the dispersion of spurious frequencies in the spectral analysis due to changes in machine speed.

[0111] The process returns to step 50 with the following samples xp supplied by the sampler 9.

[0112] Figure 6 illustrates the spectrum of vibrations delivered by the sensor 8 for a constant rotational speed of the bearing 4 represented by graph SP1, the spectrum re presented by graph SP2 when the rotational speed of bearing 4 underwent a speed change of 3% during data collection compared to the constant rotational speed when sensor 8 measures the rotational speed.

[0113] An SP3 graph is plotted and represents the spectrum delivered on the second output 103 of the device 10 when the rotational speed of the bearing 4 underwent a 3% change during data collection after application of the speed change compensation. This value is very close to the constant rotational speed.

[0114] In graph SP1, three tones T10, T20, T30 are easily identifiable at the respective frequencies Fl, F2, F3.

[0115] The tone T10 is the fundamental and the tones T20 and T30 are two harmonics. The asymmetry value of the graph SP1 is the reference asymmetry value.

[0116] In the SP2 graph, three tones T1, T21, and T31 with reduced peaks are difficult to identify. It is not possible to determine their precise location, such that the constant-speed time-domain algorithm implemented by the first implementation means 11 and the constant-speed spectral-oriented algorithm implemented by the second implementation means 12 may not be able to determine whether the three tones T1, T21, and T31 are of interest. The skewness value of the SP1 graph is lower than the reference skewness value.

[0117] In the SP3 graph, three tones T12, T22, T33 are easily identifiable at the respective frequencies Fl, F2, F3.

[0118] The asymmetry value of the SP3 graph has been increased compared to the asymmetry value of SP2 and is similar to the reference asymmetry value.

[0119] Device 10 restores the high peaks of tones T12, T22, T32 and their frequencies Fl, F2, F3.

[0120] The device 10 allows the use of existing algorithms at constant rotational speed when the rotational speed varies, so that it is not necessary to make repetitive attempts to acquire data due to changes in speed.

[0121] Eliminating repetitive attempts to acquire data saves the power supply of a device comprising the sensor 8, the sampler 9 and the device 10, for example a wireless power supply including a power source such as a battery.

Claims

Demands

1. A method for processing a digital signal (S9) comprising samples (xp) of a continuous signal (S8) sampled at a fixed sampling frequency, the continuous signal being representative of vibrations of a bearing (4) undergoing changes in rotational speed, the method comprising: - applying to the digital signal a time-variable sampling frequency conversion by a Farrow structure (14) from the fixed sampling frequency to a time-variable sampling, the digital signal sampled at the time-variable sampling being a resultant signal, the Farrow structure being controlled from a control variable (4;4), - performing a spectral analysis of the resulting signal to determine the frequency values ​​of the resulting signal, - determining a scattering parameter for the frequency values ​​of the resulting signal, and - modifying the control variable (4 4û) as a function of the value of the scattering parameter.

2. A method according to claim 1, wherein the performance of a spectral analysis of the resulting signal includes the performance of a fast Fourier transform (ALGO1) of the resulting signal.

3. A method according to claim 1 or 2, wherein the frequency value scattering parameter is the frequency value asymmetry.

4. A method according to any one of claims 1 to 3, wherein the modification of the variable sampling includes the implementation of a line search algorithm (ALGO3) to modify the control variable.

5. A device (10) for processing a digital signal (S9) comprising samples of a continuous signal (S8) sampled at a fixed sampling frequency, the continuous signal being representative of vibrations of a bearing undergoing changes in rotational speed, the device comprising: - a Farrow structure (14) configured to apply to the digital signal (S9) a time-varying sampling frequency conversion from the fixed sampling frequency to a sample- time-variable sampling, the digital signal sampled at time-variable sampling being a resultant signal, and the Farrow structure (14) being configured to be controlled from a control variable (A, An), - spectral analysis means (15) configured to perform a spectral analysis of the resultant signal in order to determine the frequency values ​​of the resultant signal, - determination means (16) configured to determine a scatter parameter of the frequency values ​​of the resultant signal, and - control means (17) configured to modify the control variable according to the value of the scatter parameter.

6. Device according to claim 5, wherein the spectral analysis means (15) are configured to implement a fast Fourier transform algorithm (ALG01).

7. Device according to claim 5 or 6, wherein the frequency value scattering parameter is the frequency value asymmetry.

8. Device according to any one of claims 5 to 7, wherein the control means (17) are configured to implement a line search algorithm (ALG03) to modify the control variable.

9. Bearing device comprising: - a bearing (4) having an inner ring (5) and an outer ring (6) capable of rotating concentrically with respect to each other, - a sensor (8) configured to measure the vibrations of said inner ring (5) or outer ring (6) and configured to deliver a continuous signal (S8), - a sampler (9) configured to sample the continuous signal (S8) at a fixed sampling frequency and configured to deliver the digital signal (S9) comprising the samples, and - a device (10) according to any one of claims 5 to 8 configured to process the digital signal (S9).

Images