Apparatus and method for processing digital signal

Abstract

It is proposed a device for processing a continuous signal sampled at a fixed sampling rate to obtain a digital signal, the continuous signal comprising state monitoring data of a rotating element, the rotating element experiencing a change in rotational speed. Said means comprise:-segmentation means (15),-spectral analysis means (16),-filtering means (17),-first processing means (18),-memory (19),-second processing means (20), and-sampling means (21).

Description

Devices and methods for processing digital signals Technical Field

[0001] The present invention relates to apparatus for processing digital signals and methods for processing digital signals. Background Technology

[0002] For example, condition monitoring algorithms used to detect bearing faults (such as faults in the bearing's inner raceway) require a constant rotational speed for the bearing to function properly.

[0003] Typically, condition monitoring algorithms perform spectral analysis on rotating bearings to detect faults from specific spectral waves and harmonics.

[0004] During signal measurement, changes in the bearing's rotational speed obscure the spectral wave, thereby reducing the ability to identify frequencies and harmonics that may be associated with faults, such as a fault in the bearing's inner raceway or a fault in the bearing's outer raceway.

[0005] Without the use of a tachometer or other direct speed measurement equipment, the difficulty of correcting for speed changes can be further exacerbated by complex speed change curves (such as curves containing several acceleration and deceleration phases as well as time-varying spectral amplitude changes). Summary of the Invention

[0006] Therefore, this invention is used to correct the harmful effects of speed variation.

[0007] According to one aspect, a method for processing a continuous signal sampled at a fixed sampling rate to obtain a digital signal, the continuous signal including state monitoring data of a rotating element undergoing a change in rotational speed.

[0008] The method includes:

[0009] - (a) Divide the timestamped samples of the digital signal into multiple frames.

[0010] - (b) Transform each of the plurality of frames to the frequency domain and provide an array including the absolute values ​​of the amplitude and frequency of each transformed frame.

[0011] - (c) For each transformed frame, determine the significant spectral peaks and their center frequencies from the background noise.

[0012] -For each pair of frames in one set of the plurality of frames:

[0013] (d) Determine all feasible frequency ratios of each significant spectral peak in the first frame of the pair relative to each significant spectral peak in the second frame of the pair.

[0014] (e) Cluster the frequency ratios into clusters of similar frequency ratios.

[0015] (f) Determine the rate of change coefficient of the frequency ratio from the cluster with the largest number of frequency ratios.

[0016] (g) Assigning a timestamp value to the velocity change coefficient determined in step (f), wherein the timestamp value is assigned to the velocity change coefficient determined based on at least one timestamp value of the first or second frame in the pair.

[0017] -(h) Store the velocity variation coefficient and its timestamp value in an array.

[0018] - (i) Generate velocity curves from the velocity variation coefficients and their timestamp values ​​stored in the array, and

[0019] - (j) Resample the digital signal in the radian domain according to the velocity curve.

[0020] This method allows for machine health assessment of systems and / or sensors that do not have devices for direct rotational speed measurement.

[0021] This method allows the use of any existing algorithm that has been designed for constant velocity conditions and is industrially accepted, trusted, and well understood.

[0022] This method also avoids repeated attempts to obtain data that must be discarded simply because of changes in velocity.

[0023] Instead, such data can now be used efficiently. Suppression of repeated attempts to acquire data allows for savings in the power consumption of systems implementing this method, such as wireless systems powered by batteries.

[0024] Advantageously, step (i) includes:

[0025] - For each timestamp value, select a representative velocity change coefficient stored in the array.

[0026] - Multiply each representative velocity variation coefficient by the average velocity of the rotating element to obtain the velocity value.

[0027] - Obtain an array of representative velocity change coefficients based on the timestamp values ​​of the representative velocity change coefficients, and

[0028] - Apply interpolation to the selected velocity variation coefficient to obtain the velocity curve.

[0029] Preferably, selecting a representative velocity change coefficient for each timestamp value includes:

[0030] - Cluster the velocity change coefficients associated with the timestamp values, and

[0031] - Statistical methods are applied to cluster the data to determine the representative velocity change coefficient.

[0032] Advantageously, the method includes applying a Hanning window to each of the plurality of frames before transforming each of the plurality of frames to the frequency domain.

[0033] Preferably, the method includes padding each of the plurality of frames with zeros before transforming each of the plurality of frames to the frequency domain.

[0034] Advantageously, step (d) further includes eliminating each frequency ratio not included in a predetermined interval, which includes feasible frequency ratios.

[0035] According to another aspect, an apparatus is proposed for processing a continuous signal sampled at a fixed sampling rate to obtain a digital signal, the continuous signal including state monitoring data of a rotating element undergoing a change in rotational speed.

[0036] The device includes:

[0037] - Segmentation devices are configured to segment timestamped samples of a digital signal into multiple frames.

[0038] - A spectrum analyzer configured to transform each of a plurality of frames to the frequency domain and provide an array including the absolute values ​​of the amplitude and frequency of each transformed frame.

[0039] - Filtering devices, configured to determine significant spectral peaks and their center frequencies from the background noise for each transformed frame.

[0040] - A first processing device is configured for each pair of frames in a set of frames among the plurality of frames:

[0041] • Determine all feasible frequency ratios of each significant spectral peak in the first frame of the pair relative to each significant spectral peak in the second frame of the pair.

[0042] • Cluster the frequency ratios into clusters of similar frequency ratios.

[0043] • Determine the cluster with the largest number of frequency ratios and the rate of change coefficient of the frequency ratios from the cluster with the largest number of frequency ratios.

[0044] • Assign a timestamp value to the velocity change coefficient determined in step (f), wherein the timestamp value is assigned to the velocity change coefficient determined based on at least one timestamp value of the first or second frame in the pair.

[0045] - A memory configured to store the velocity change coefficient and its timestamp value in an array.

[0046] - A second processing device, configured to generate a velocity curve from velocity variation coefficients and their timestamp values ​​stored in the array, and

[0047] - A sampling device is configured to resample the digital signal in the radian domain according to the velocity curve.

[0048] Preferably, the first processing device is further configured to apply a Hanning window to each of the multiple frames before transforming each of the multiple frames into the frequency domain.

[0049] Advantageously, the first processing device is also configured to zero-padding each of the multiple frames before transforming each of the multiple frames to the frequency domain.

[0050] Based on this, a bearing device is proposed.

[0051] The bearing equipment includes:

[0052] - A bearing having an inner ring and an outer ring capable of rotating concentrically relative to each other.

[0053] - A sensor configured to measure the vibration of the inner or outer ring and configured to transmit a continuous signal.

[0054] A sampler configured to sample the continuous signal at a fixed sampling rate and configured to transmit the digital signal including the sequential samples, and

[0055] - The device as defined above is configured to process digital signals. Attached Figure Description

[0056] Other advantages and features of the invention will become apparent upon review of the detailed description of the embodiments (in no way limiting) and the accompanying drawings, wherein:

[0057] Figure 1 schematically illustrates the machine according to the present invention.

[0058] Figure 2 schematically illustrates an example of a device for processing digital signals according to the present invention.

[0059] Figure 3 schematically illustrates a method for processing digital signals according to the present invention.

[0060] Figure 4 schematically illustrates an example of clustering of the velocity variation coefficient according to the present invention.

[0061] Figure 5 schematically illustrates an example of a velocity curve according to the present invention, and

[0062] Figure 6 schematically illustrates an example of the spectrum transmitted by a spectrum analysis device for a bearing undergoing a change in rotational speed during measurement, according to prior art and the present invention. Detailed Implementation

[0063] Referring to Figure 1, a partial longitudinal cross-section of machine 1 is schematically shown.

[0064] Machine 1 includes housing 2 and shaft 3 supported in housing 2 by rolling bearings 4 (e.g., roller bearings or ball bearings).

[0065] The rolling bearing 4 has an inner ring 5 mounted on the shaft 3 and an outer ring 6 mounted in a hole in the housing 2. The outer ring 6 radially surrounds the inner ring 5. The inner ring 5 and the outer ring 6 rotate concentrically relative to each other.

[0066] The rolling bearing 4 also includes a row of rolling elements 7 radially positioned between the inner and outer raceways of the inner ring 5 and the outer ring 6. In the illustrated example, the rolling element 7 is a ball. Alternatively, the rolling bearing may include other types of rolling elements 7, such as rollers. In the illustrated example, the rolling bearing includes a row of rolling elements 7. Alternatively, the rolling bearing may include several rows of rolling elements.

[0067] Sensor 8 is installed in housing 2 to measure the vibration of bearing 4 as it experiences changes in rotational speed.

[0068] Sensor 8 can be installed in the hole of housing 2.

[0069] In a variant, sensor 8 can be mounted in other locations on the machine, such as near the outer ring 6 or near the housing 2.

[0070] Sensor 8 transmits a continuous signal S8 indicating the operation of the rotating element.

[0071] The rotating element can be a bearing 4, and the sensor 8 transmits a continuous signal S8 representing the vibration of the bearing 4 to the input of the sampler 9.

[0072] Sampler 9 will include timestamped samples of the continuous signal S8 sampled at a fixed sampling rate. The digital signal S9 is transmitted to the input section 101 of the device 10 for processing the digital signal S9, where p is an integer.

[0073] Bearing 4, sensor 8, sampler 9 and device 10 form the bearing device.

[0074] The memory (not shown) can store the output signal S9 and transmit the output signal S9 to the device 10.

[0075] The output section 102 of the device 10 can be connected to the implementation device 11, which implements at least one constant-rate time-domain algorithm based on the first output signal S102 transmitted by the device 10 on the first output section 102, such as implementing an envelope fault detection algorithm.

[0076] The output unit 102 of the device 10 can be further connected to the second implementation device 12 to perform spectrum analysis of the output of the device 10.

[0077] The second implementing device 12 implements, for example, a fast Fourier transform.

[0078] The first and second implementing devices 11 and 12 are, for example, each configured as a processing unit that implements the algorithm.

[0079] The processing unit 13 implements the sensor 8, the sampler 9, and the device 10.

[0080] Figure 2 schematically illustrates an example of device 10.

[0081] The device 10 includes a first memory 14, a segmenting means 15, a spectrum analysis device 16, a filtering device 17, a first processing device 18, a second memory 19, a second processing device 20, and a sampling device 21.

[0082] The first memory 14 is used to store digital signal S9, which includes timestamped samples received at input 101 of device 10. .

[0083] The first memory 14 is connected to the input section of the segmentation device 15.

[0084] The segmentation device 15 also includes a first output connected to the input of the spectrum analyzer 16.

[0085] Segmentation device 15 is used to segment sequential samples of digital signals. Segmented into multiple frames F1, F2...F k -1, F k k is an integer. The frames have the same size.

[0086] The segmentation device 15 can be further used to segment frames F1, F2...F... k-1 F k Zero-padding is performed, and the Hanning window is applied to each frame F1, F2…F in multiple frames. k-1 F k .

[0087] continuous samples They are grouped together in p frames.

[0088] The integer p is determined based on the resolution at which the machine's speed changes in each frame are expected to be captured.

[0089] Segmentation device 15 is used to transmit frames F1, F2...F on the input of spectrum analyzer 16. k-1 F k .

[0090] The output of the spectrum analyzer 16 is connected to the input of the filter 17.

[0091] Spectrum analyzer 16 is used to execute frames F1, F2…F k-1 F k Spectral analysis to provide information including each frame's F1, F2...F k-1 F k An array of absolute values ​​of frequency and amplitude.

[0092] Frames F1, F2...F k -1, F k Zero padding and / or the application of Hanning windows improve spectral resolution.

[0093] The spectrum analysis device 16, for example, implements a fast Fourier transform algorithm.

[0094] The output of the filter device 17 is connected to the input of the first processing device 18.

[0095] Filtering device 17 is used to determine significant spectral peaks from background noise, and from frames F1, F2…F… k -1, F k The transform frames F1, F2…F are determined from an array of absolute values ​​of frequency and amplitude. k -1, F k The peak center frequency.

[0096] Filtering component 17, for example, implements a noise carpet filter to filter frames F1, F2…F… k -1, F k The noise in the spectrum is filtered to retain only those components that may be, for example, +10 dB higher than the local spectral carpet level. Significant spectral peaks are components that may be +10 dB higher than the local spectral carpet level.

[0097] The first processing device 18 includes a memory 22 for storing an array comprising frames F1, F2…F1 along a plurality of frames. k -1, F k The peak center frequency of each significant spectral peak.

[0098] The first processing device 18 also includes a first determining device 23, a clustering device 24, and an allocation device 25.

[0099] The output of the first processing device 18 is connected to the second memory 19.

[0100] The input section of the second processing device 20 is connected to the second memory 19, and the output section of the second processing device 20 is connected to the input section of the sampling device 21.

[0101] The output of sampling device 21 is connected to the output of device 102.

[0102] Figure 3 shows an example of a method for processing digital signal S9 implemented with device 10.

[0103] In step 30, sampler 9 transmits digital signal S9, which includes timestamped samples of the continuous signal S8 transmitted from sensor 8. .

[0104] Samples with timestamps It is stored in memory 14.

[0105] In step 31, the segmentation device 15 determines frames F1, F2, ..., F... k -1, F k .

[0106] In step 32, the spectrum analysis device 16 divides multiple frames F1, F2, ..., F... k -1, F k Each frame in the transform is converted to the frequency domain, and an array including the absolute frequency and amplitude of each transformed frame is provided.

[0107] In step 33, the filter device 17 is based on frames F1, F2…F… k -1, F k An array of absolute values ​​of frequency and amplitude is used to determine the transform frames F1, F2…F k -1, F k The significant spectral peaks and their center frequencies are transmitted along multiple frames, including frames F1, F2…F… k -1, F k An array of the peak center frequencies of each significant spectral peak.

[0108] The array transmitted by the filter device 17 is stored in the memory 22 of the first processing device 18.

[0109] In step 34, multiple frames F1, F2...F are defined. k -1, F k A group of frames. This group may include multiple frames F1, F2…Fk -1, F k One or more frames F1, F2...F k -1, F k All frames.

[0110] In step 35, some frame pairs in the group of frames are identified.

[0111] Each pair of frames includes a first frame Fi and a second frame F. j i and j are different integers.

[0112] For each pair of frames F i F j In step 36, the first determining device 23 determines each significant spectral peak of the first frame Fi of the pair relative to the second frame F of the pair. j All feasible frequency ratios for each significant spectral peak.

[0113] The frequency ratio represents the ratio of the rotating element in the first frame Fi to the second frame F. j Changes in velocity (acceleration and deceleration) between them.

[0114] It is unrealistic to use some frequency ratios to represent changes in velocity.

[0115] The first determining device 23 eliminates each frequency ratio not included in a predetermined interval, which includes feasible frequency ratios.

[0116] Clustering device 24 clusters frequency ratios into clusters with similar frequency ratios.

[0117] Each cluster has the same width centered at different frequency values.

[0118] Clustering device 24 determines the cluster with the largest number of frequency ratios.

[0119] The clustering device 24 also determines the velocity variation coefficient based on the frequency ratio of the cluster, wherein the frequency ratio of the cluster has the maximum number of frequency ratios.

[0120] Various static methods are used to determine the velocity variation coefficient, such as physical center, centroid, weighted average frequency ratio, and average frequency of clusters with the largest number of frequency ratios.

[0121] The allocation device 25 assigns the timestamp value to the velocity change coefficient.

[0122] According to the first frame Fi or the second frame F in the pair j The assigned timestamp value is determined by at least one timestamp value.

[0123] The assigned timestamp value is, for example, equal to the second frame F. j The midpoint timestamp.

[0124] The velocity variation coefficients and timestamp values ​​of the paired frames in the frame group are stored in an array stored in the second memory 19.

[0125] In step 37, the second processing device 20 generates a velocity curve from the velocity variation coefficients and their timestamp values ​​stored in the array.

[0126] For each timestamp value, select a representative velocity change coefficient from the velocity change coefficients stored in the array.

[0127] The second processing device 20 can select a first velocity change coefficient β1 stored in the array representing the velocity change between the first frame F1 and the second frame F2, a second velocity change coefficient β2 stored in the array representing the velocity change between the second frame F2 and the third frame F3..., and a second velocity change coefficient β2 stored in the array representing the velocity change between the (k-1)th frame F1... k-1 and frame k F k The (k-1)th velocity change coefficient βk-1 between the two consecutive frames containing consecutive samples.

[0128] The representative velocity variation coefficient is the velocity variation coefficient β1…βk-1.

[0129] In another embodiment, the second processing device 20 can determine and select velocity change coefficients that can represent the velocity change between two non-contiguous frames, such that, for example, velocity change coefficients β1,3 stored in an array representing the velocity change between the first frame F1 and the third frame F3, velocity change coefficients β1,3... stored in an array representing the velocity change between the first frame F1 and the third frame F3, and velocity change coefficients β1,3... stored in an array representing the velocity change between frame F2 and frame F3, are stored in an array. k The coefficient of change of velocity between them is β2,k.

[0130] In another embodiment, the second processing device 20 determines and stores the velocity variation coefficient matrix that compares all frames sequentially with all other frames in the second memory 19.

[0131] Multiple velocity variation coefficients are associated with each timestamp value.

[0132] The second processing device 20 clusters the velocity change coefficients associated with each timestamp value.

[0133] Figure 4 shows an example of clustering of velocity change coefficients.

[0134] Each point represents the velocity change coefficient of the matrix.

[0135] The velocity change coefficient associated with time t1 is clustered in cluster C1, and similarly, the velocity change coefficient associated with time tn is clustered in cluster Cn, where n is, for example, 2 to 15.

[0136] Of course, the velocity change coefficient can be clustered in more than fifteen clusters or less than fifteen clusters.

[0137] For each cluster C1 to C15 (timestamp values), the representative velocity change coefficient is determined by various statistical methods, such as the average velocity change coefficient value for each cluster, curve fitting of the velocity change coefficient value for each cluster, or other statistical methods.

[0138] When determining the representative speed variation coefficient, the second processing device 20 multiplies each representative speed variation coefficient by the average speed of the rotating element to obtain the speed value.

[0139] The second processing device 20 determines an array of representative velocity variation coefficients based on the timestamp values ​​of the representative velocity variation coefficients, and applies interpolation to the selected velocity variation coefficients to obtain a velocity curve. Depending on the density of the representative velocity variation coefficients, the interpolation can be, for example, linear, polynomial, or spline.

[0140] Figure 5 shows an example of a velocity curve.

[0141] The point represents the representative velocity variation coefficient associated with timestamp values ​​t1 to t15.

[0142] In step 38, the sampling device 21 resamples the digital signal S9 in the radian domain according to the velocity curve determined in step 37.

[0143] Figure 6 shows examples of vibration spectra transmitted by sensor 8 with and without velocity compensation (dashed line), where digital signal S9 is resampled by sampling device 21.

[0144] Spectral waves are easily identifiable on the spectrum of vibrations with velocity compensation, but are blurred on the spectrum of vibrations without velocity compensation.

[0145] Device 10 allows for machine health assessment of systems and / or sensors that do not have devices for direct rotational speed measurement.

[0146] Device 10 allows the use of any existing algorithm that has been designed for constant speed conditions and is industrially accepted, trusted, and well understood.

[0147] Device 10 also avoids repeated attempts to obtain data that must be discarded simply due to speed changes.

[0148] Instead, such data can now be used efficiently. Suppression of repeated attempts to acquire data allows for savings in the power supply of systems including sensor 8, sampler 9, and device 10, such as wireless systems powered by batteries.

Claims

1. A method for processing a continuous signal (S8) sampled at a fixed sampling rate to obtain a digital signal, the continuous signal including state monitoring data of a rotating element (4) undergoing a change in rotational speed, the method comprising: - (a) Divide the timestamped samples of the digital signal (S9) into multiple frames (F1, F2, F3). k (b) Each of the plurality of frames (F1, F2) k (c) Transform to the frequency domain and provide an array including the absolute values ​​of the amplitude and frequency of each transformed frame; (d) For each transformed frame, determine significant spectral peaks and their center frequencies from the background noise, for the plurality of frames (F1, F2). k For each pair of frames in a set of frames: (d) determine all feasible frequency ratios of each significant spectral peak of the first frame in the pair relative to each significant spectral peak of the second frame in the pair; (e) cluster the frequency ratios into clusters of similar frequency ratios; (f) determine the cluster with the largest number of frequency ratios and the velocity variation coefficient of the frequency ratios from the cluster with the largest number of frequency ratios; (g) assign a timestamp value to the velocity variation coefficient determined in step (f), assigning the timestamp value to the velocity variation coefficient determined based on at least one timestamp value of the first or second frame in the pair; (h) store the velocity variation coefficients and their timestamp values ​​in an array; (i) generate a velocity curve from the velocity variation coefficients and their timestamp values ​​stored in the array; and (j) resample the digital signal in the radian domain based on the velocity curve.

2. The method according to claim 1, wherein step (i) comprises: - For each timestamp value, select a representative velocity change coefficient stored in the array, - multiply each representative velocity change coefficient by the average velocity of the rotating element to obtain a velocity value, - obtain an array of representative velocity change coefficients based on the timestamp values ​​of the representative velocity change coefficients, and - apply interpolation to the selected velocity change coefficients to obtain a velocity curve.

3. The method according to claim 2, wherein, Selecting a representative velocity change coefficient for each timestamp value includes: - clustering the velocity change coefficients associated with the timestamp value, and - applying statistical methods to cluster the coefficients to determine the representative velocity change coefficient.

4. The method according to any one of claims 1 to 3, comprising: Before transforming each of the plurality of frames to the frequency domain, a Hanning window is applied to each of the plurality of frames.

5. The method according to any one of claims 1 to 4, comprising: Before transforming each of the plurality of frames to the frequency domain, each of the plurality of frames is padded with zeros.

6. The method according to any one of claims 1 to 5, wherein, Step (d) further includes eliminating each frequency ratio that is not included in a predetermined interval, which includes feasible frequency ratios.

7. An apparatus for processing a continuous signal (S8) sampled at a fixed sampling rate to obtain a digital signal (S9), said continuous signal including state monitoring data of a rotating element (4) undergoing a change in rotational speed, said apparatus comprising: - Segmentation device (15) is configured to segment timestamped samples of the digital signal (S9) into multiple frames (F1, F2, F3). k ), - Spectrum analysis device (16), is configured to analyze each of the plurality of frames (F1, F2, F3, F4, F5, F6, F7, F8, F9, F1, F1, F1, F1, F2 ... k The transform is performed in the frequency domain and provides an array including the absolute values ​​of the amplitude and frequency of each transform frame; a filtering device (17) is configured to determine significant spectral peaks and their center frequencies from the background noise for each transform frame; and a first processing device (18) is configured to process the plurality of frames (F1, F2, F3, F4, F5, F6, F7, F8, F9, F1, F1, F1, F2 ... k For each pair of frames in a set of frames: • determine all feasible frequency ratios of each significant spectral peak of the first frame in the pair relative to each significant spectral peak of the second frame in the pair, • cluster the frequency ratios into clusters of similar frequency ratios, • determine the cluster with the largest number of frequency ratios and the velocity change coefficient of the frequency ratios from the cluster with the largest number of frequency ratios, • assign a timestamp value to the velocity change coefficient determined in step (f), assigning the timestamp value to the velocity change coefficient determined based on at least one timestamp value of the first or second frame in the pair, - memory (19) configured to store the velocity change coefficients and their timestamp values ​​in an array, - second processing device (20) configured to generate a velocity curve from the velocity change coefficients and their timestamp values ​​stored in the array, and - sampling device (21) configured to resample the digital signal in the radian domain based on the velocity curve.

8. The device according to claim 7, wherein, The first processing device (18) is also configured to apply a Hanning window to each of the plurality of frames before transforming each of the plurality of frames to the frequency domain.

9. The device according to claim 7 or 8, wherein, The first processing device (18) is also configured to zero-padding each of the plurality of frames before transforming each of the plurality of frames to the frequency domain.

10. A bearing device, comprising: - Bearing (4), which is provided with an inner ring (5) and an outer ring (6) capable of rotating concentrically relative to each other - Sensor (8), which is configured to measure the vibration of the inner ring (5) or the outer ring (6) and is configured to transmit a continuous signal (S8) - Sampler (9), which is configured to sample the continuous signal (S8) at a fixed sampling rate and is configured to transmit the digital signal including the sequential sample (S9) - Device (10) according to any one of claims 7 to 9, which is configured to process the digital signal (S9).

Images