Bearing peeling size prediction processing method and application

Abstract

The invention relates to the technical field of bearing detection, and relates to a bearing peeling size prediction processing method and application, and the method comprises the steps: collecting a signal, determining a fault type according to the signal, processing the signal according to the fault type to obtain a WVS spectrum, and predicting the fault size according to the WVS spectrum. Secondary processing is carried out on the signals, a WVS spectrum is calculated, an inherent frequency fluctuation curve is calculated according to the WVS spectrum, according to the fluctuation width in the inherent frequency, the size of a predicted bearing fault is calculated, the bearing state is evaluated, the fault severity degree of the bearing is judged more accurately, and the bearing fault severity degree can be judged more accurately according to the working standard and requirement of the bearing. And detailed maintenance and replacement plan making are carried out, and obvious advantages are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing detection, and in particular to a bearing peeling size prediction and processing method and application. Background Art

[0002] Rolling bearings are a common mechanical component. Their main function is to support the weight of the bearing shaft and bear axial and radial loads, while reducing friction and energy loss to ensure the stable operation of the bearing shaft. Rolling bearings are widely used in mechanical equipment, such as industrial machinery, aerospace equipment, automobiles, railways, and wind turbines. Since they are usually used in applications that need to support high-speed rotation and bear heavy loads, the rolling elements may suffer severe wear under high speed and high load, thus affecting the safety of rotating machinery. Therefore, timely and accurate identification of rolling bearing fault types, as well as tracking and predicting the degree of fault development are crucial to ensure the normal operation of mechanical equipment.

[0003] A Chinese patent with publication number "CN115791169A" discloses a rolling bearing fault diagnosis method, device and electronic equipment, and specifically discloses that the fault characteristic parameter value is obtained and calculated by analyzing the model and structural parameters of each rolling bearing, and preliminary fault diagnosis is performed based on the obtained time domain and frequency domain.

[0004] However, although the above scheme can preliminarily determine that the bearing is faulty at an early stage, in actual applications, the time domain characteristics and frequency domain characteristics are not completely linearly related to the degree of bearing fault, and the fault size of the bearing cannot be predicted. Therefore, how to accurately predict the bearing fault and the fault size to determine the bearing status and fault severity is still a technical problem.

[0005] It should be noted that the information disclosed in this background technology section is only intended to increase the understanding of the overall background of the present invention, and should not be regarded as acknowledging or suggesting in any form that the information constitutes prior art already known to a person skilled in the art. Summary of the invention

[0006] In order to solve the above-mentioned problems in the background technology, the present invention proposes a bearing spalling size prediction and processing method, which comprises the following steps:

[0007] Collect the original vibration signal of the test bearing, filter and reduce noise on the original vibration signal to obtain a valid signal;

[0008] Extract the time domain features, frequency domain features and bearing fault feature amplitude from the effective signal to determine the fault type;

[0009] According to the bearing fault type, the effective signal is periodically segmented to obtain a segmented signal;

[0010] Calculate the WVS spectrum based on the segmented signal;

[0011] The natural frequency fluctuation curve is obtained by longitudinally averaging the WVS spectrum to predict the bearing fault size.

[0012] Furthermore, before filtering the original vibration signal, the auto-power spectral density of the collected original vibration signal of the faulty bearing and the original signal of the healthy bearing is compared, and the frequency band range in which the bearing fault is more sensitive to the vibration signal is found and then filtered.

[0013] Further, determining the fault type comprises the following steps:

[0014] Extract time domain features from valid signals, including at least RMS and kurtosis;

[0015] Perform fast Fourier transform on the effective signal to obtain the frequency domain spectrum;

[0016] Perform Hilbert transform on the effective signal to obtain the envelope spectrum;

[0017] The characteristic frequency of each bearing fault type is calculated by the bearing size and shaft speed. The peak value of the spectrum and envelope spectrum of the effective signal is selected within 10% before and after the characteristic frequency of each fault type to obtain the characteristic amplitude of each fault type.

[0018] The bearing fault type is determined based on the comparison results of the time domain characteristics, frequency spectrum and envelope spectrum with the characteristic amplitudes of each fault type.

[0019] Furthermore, the bearing fault types include cage fault, rolling element rotation fault, outer ring fault and inner ring fault, and the cage fault frequency FTF is expressed as:

[0020]

[0021] The rolling element rotation fault frequency BSF is expressed as:

[0022]

[0023] The outer race fault frequency BPFO is expressed as:

[0024]

[0025] The inner race fault frequency BPFI is expressed as:

[0026]

[0027] Among them, N is the bearing rotation frequency, n is the number of rolling elements, d is the rolling element diameter, D is the bearing median diameter, and α is the rolling element contact angle.

[0028] Furthermore, the periodic segmentation of the effective signal comprises the following steps:

[0029] According to the fault type, extract the envelope spectrum of the corresponding fault type, filter within the range of 10%-15% before and after the envelope spectrum, and perform order tracking to obtain the phase;

[0030] Resampling the time domain feature according to the phase so that the number of sampling points of the effective signal in each fault cycle is consistent;

[0031] The signal is segmented according to the fault cycle to obtain a segmented signal.

[0032] Furthermore, the WVS spectrum is calculated based on the segmented signal x(t) as follows:

[0033]

[0034] Among them, R x (t,τ) is the autocorrelation function of the segmentation signal, expressed as

[0035]

[0036] E[] is the expected function, t is time, τ is the time difference, and f is the frequency.

[0037] Furthermore, the WVS spectrum is vertically averaged to obtain the frequency average value f avg , we can get the natural frequency fluctuation curve, then f avg Expressed as:

[0038]

[0039] Among them, A i is the WVS spectrum amplitude, f i is the frequency.

[0040] Furthermore, the duration of the fluctuation region in the natural frequency fluctuation curve is set to t exit , then the fault size l is expressed as:

[0041]

[0042] Among them, D p is the average diameter of the inner and outer rings of the bearing, d is the rolling element diameter, and f r is the shaft speed / Hz.

[0043] The present invention also provides an application of a bearing spalling size prediction and processing method in formulating a bearing maintenance and replacement plan, which uses the bearing fault size obtained by any of the bearing spalling size prediction and processing methods described above to formulate a detailed bearing maintenance and replacement plan.

[0044] Based on the above, the bearing spalling size prediction and processing method provided by the present invention, compared with the prior art, detects the bearing fault type based on the conventional time domain and frequency domain, performs secondary processing on the signal, calculates the WVS spectrum, and obtains the natural frequency fluctuation curve based on the WVS spectrum. According to the width of the fluctuation in the natural frequency, the size of the bearing fault is calculated and predicted, the bearing status is evaluated, and the severity of the bearing fault is judged more accurately.

[0045] Other features and beneficial effects of the present invention will be described in the following description, and some of the features and beneficial effects can be understood by implementing the present invention. The purpose and other beneficial effects of the present invention can be achieved and obtained by the structures specifically pointed out in the description and other contents. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. The positional relationships described in the drawings in the following description are based on the directions in which the components are drawn in the diagrams, unless otherwise specified.

[0047] Figure 1 It is a flow chart of a bearing spalling size prediction and processing method provided by an embodiment of the present invention;

[0048] Figure 2 It is a schematic diagram of a flow chart for confirming the fault type in a bearing spalling size prediction and processing method provided by an embodiment of the present invention;

[0049] Figure 3 It is a schematic flow chart of signal period segmentation in a bearing spalling size prediction and processing method provided by an embodiment of the present invention;

[0050] Figure 4 It is a WVS spectrum of a faulty bearing with an outer ring fault and a corresponding natural frequency fluctuation curve provided by an embodiment of the present invention;

[0051] Figure 5 It is a mapping diagram of a bearing degradation experiment provided by an embodiment of the present invention; DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments; the technical features designed in different implementation modes of the present invention described below can be combined with each other as long as they do not conflict with each other; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0053] In the description of the present invention, it should be noted that all terms used in the present invention (including technical terms and scientific terms) have the same meanings as those generally understood by ordinary technicians in the field to which the present invention belongs, and cannot be understood as limiting the present invention; it should be further understood that the terms used in the present invention should be understood to have the same meanings as these terms in the context of this specification and in the relevant fields, and should not be understood in an idealized or overly formal sense, unless explicitly defined in the present invention.

[0054] Embodiment 1

[0055] In order to solve the above-mentioned technical problem of how to predict bearing faults and fault sizes, the present invention provides a bearing spalling size prediction and processing method, such as Figure 1 As shown, the following steps are included:

[0056] Collect the original vibration signal of the test bearing, filter and reduce noise on the original vibration signal to obtain a valid signal;

[0057] Extract the time domain features, frequency domain features and bearing fault feature amplitude from the effective signal to determine the fault type;

[0058] According to the bearing fault type, the effective signal is periodically segmented to obtain a segmented signal;

[0059] Calculate the WVS (Wi gner-Vi lle Spectrum) spectrum based on the split signal;

[0060] The natural frequency fluctuation curve is obtained by longitudinally averaging the WVS spectrum to predict the bearing fault size.

[0061] In specific implementation, the collected original vibration signal often contains a lot of noise and interference, which affects the subsequent signal processing and analysis. Therefore, it needs to be preprocessed, including filtering and noise reduction.

[0062] The purpose of filtering is to remove unnecessary frequency components in the signal to extract useful signal information. Commonly used filters include low-pass filters, band-pass filters, etc. Low-pass filters can remove high-frequency noise, and band-pass filters can remove low-frequency and high-frequency noise. In practical applications, multi-stage filters are often used for filtering to gradually remove noise in different frequency ranges. Preferably, before filtering the original vibration signal, the present invention compares the collected original vibration signal of the faulty bearing with the original vibration signal of the healthy bearing. The self-power spectral density (PSD) is compared to find the frequency band range where the bearing fault is more sensitive to the acoustic emission signal, and then the original vibration signal is filtered using a band-pass filter. However, the present invention is not limited to this, and other methods such as fast kurtosis spectrum (FK) can also be used for filtering, and there is no limitation here.

[0063] Noise reduction is to remove noise interference in the signal and retain useful signal information. Commonly used noise reduction algorithms include wavelet transform, adaptive filtering, wavelet denoising, etc. Wavelet transform can decompose the signal into different frequency sub-bands, and realize denoising by threshold processing of each sub-band. Adaptive filtering can perform filtering processing according to the inherent characteristics of the signal, and can automatically adjust the parameters of the filter to adapt to different signal noise interference conditions. Wavelet denoising can decompose the signal into multiple frequency sub-bands, and decide whether to retain the information of the sub-band according to the signal-to-noise ratio of each sub-band. Preferably, the present invention uses wavelet denoising, deterministic random signal separation (DRS), minimum entropy inverse convolution (MED) and other methods to perform noise reduction processing on the original vibration signal.

[0064] In some preferred embodiments, Figure 2 As shown, the steps to determine the fault type are as follows:

[0065] Extract time domain features from valid signals, including at least RMS (root mean square) and kurtosis;

[0066] Perform fast Fourier transform on the effective signal to obtain the frequency domain spectrum;

[0067] Perform Hilbert transform on the effective signal to obtain the envelope spectrum (SES);

[0068] The characteristic frequency of each bearing fault type is calculated by the bearing size and shaft speed. The peak value of the spectrum and envelope spectrum of the effective signal is selected within 10% before and after the characteristic frequency of each fault type to obtain the characteristic amplitude of each fault type.

[0069] The bearing fault type is determined based on the comparison results of the time domain characteristics, frequency spectrum and envelope spectrum with the characteristic amplitudes of each fault type.

[0070] In the specific implementation, the time domain features of the effective signal are extracted, including at least RMS and kurtosis. RMS is one of the important features of the vibration signal, which can reflect the overall energy of the signal. When diagnosing bearing faults, RMS can be used to measure the strength and degree of change of the signal. For example, when the bearing is faulty, the RMS value of the vibration signal will increase. Kurtosis is an indicator that describes the sharpness of the signal waveform. By comparing the kurtosis index, it can be used to determine whether the signal has abnormal conditions such as sharp changes or irregular waveforms. For example, when the rolling element of the bearing passes through the fault area, the two collide to generate an impact signal, and the kurtosis value reflects the amplitude and sharpness of the impact.

[0071] Time domain features and frequency domain features can reflect the severity of bearing faults to a certain extent, but neither of them can quantitatively analyze the specific size of the fault, and the characteristic signals of the two are not completely linearly related to the degree of bearing fault, especially when the bearing fault size is large. Therefore, it is necessary to further process and predict the effective signals.

[0072] Extract the frequency domain features of the effective signal, perform fast Fourier transform on the effective signal to obtain the spectrum of the bearing model, and perform Hilbert transform on the effective signal to obtain the envelope spectrum (SES) of the bearing signal. It is generally believed that the envelope spectrum is more effective than the spectrum, because the bearing fault signal is usually reflected in the high frequency band of the signal, and the envelope spectrum can effectively convert the high frequency signal to the low frequency area.

[0073] Furthermore, the characteristic frequency of each bearing fault type is calculated by the bearing size and shaft speed. Specifically, each bearing fault type includes cage fault, rolling element rotation fault, outer ring fault and inner ring fault. The cage fault frequency FTF is expressed as:

[0074]

[0075] The rolling element rotation fault frequency BSF is expressed as:

[0076]

[0077] The outer race fault frequency BPFO is expressed as:

[0078]

[0079] The inner race fault frequency BPFI is expressed as:

[0080]

[0081] In specific implementation, there will be some errors between the calculated characteristic frequencies of each fault type and the actual fault characteristic frequencies. Therefore, within the range of 10%-15% before and after the characteristic frequencies of each bearing fault type calculated corresponding to the effective signal, preferably 10%, the high point peak is selected. The frequency corresponding to the high point peak is the actual characteristic frequency of each fault type, and the amplitude of the actual characteristic frequency is the characteristic amplitude of each fault type. When the characteristic amplitude of a bearing fault is significantly higher than that of a normal bearing, it can be determined that the bearing has this fault type.

[0082] Finally, the effective signal is periodically segmented according to the fault type to obtain the segmented signal, the WVS spectrum is calculated based on the segmented signal, the WVS spectrum is longitudinally averaged to obtain the natural frequency fluctuation curve, and the bearing fault size is predicted.

[0083] Compared with the existing technology, the signal is further processed on the basis of time domain and frequency domain to calculate the WVS spectrum. The natural frequency fluctuation curve is calculated based on the WVS spectrum. According to the width of the fluctuation in the natural frequency, the size of the bearing fault is calculated and predicted, so as to more accurately evaluate the bearing condition.

[0084] Embodiment 2

[0085] The second embodiment refines the signal segmentation process based on the first embodiment. Figure 3 As shown, the following steps are included:

[0086] According to the fault type, extract the envelope spectrum of the corresponding type, filter 10% before and after the envelope spectrum, and perform order tracking to obtain the phase;

[0087] Resampling the time domain feature according to the phase so that the number of sampling points of the effective signal in each fault cycle is consistent;

[0088] The signal is segmented according to the fault cycle to obtain a segmented signal.

[0089] In the specific implementation, taking the bearing with outer ring fault BPFO as an example, the envelope spectrum characteristic amplitude of the outer ring fault BPFO is extracted, filtered within the range of 10% before and after the characteristic amplitude, and inverse Fourier transform is performed on it to obtain the phase, so that each 2π interval corresponds to a period of a rolling body passing through the outer ring fault (i.e., order tracking), and then the time domain characteristic signal is resampled in the angular domain according to the phase, so that the number of sampling points of the original signal in each fault cycle is consistent, equal to the set angular domain sampling frequency. After order tracking and resampling, each fault cycle of the signal has the same sampling length, so it can be easily divided, and each section corresponds to a period of a rolling body passing through the outer ring fault, which can effectively eliminate the influence of small-scale speed fluctuations and random sliding of the rolling body.

[0090] In some embodiments, the WVS spectrum of the effective signal calculated according to the segmented signal is expressed as:

[0091]

[0092] Among them, R x (t,τ) is the autocorrelation function of the segmented signal, expressed as:

[0093]

[0094] E[] is the expected function, t is time, τ is the time difference, and f is the frequency.

[0095] WVS is a time-frequency transform analysis algorithm. It has been proven to have higher time-frequency resolution than traditional time-frequency algorithms such as fast Fourier transform (STFT), and can therefore track changes in weak signals more clearly. Compared with algorithms such as WVD (Wigner-Vi l le Distr ibution), WVS can effectively eliminate the mutual influence between frequency components.

[0096] The problem with WVS is that it requires averaging of multiple periodic signals, so this method performs signal segmentation before WVS; in addition, WVS requires large computing resources, so the present invention uses the signal self-power density spectrum (PSD) to select a certain order of natural frequency at the measured bearing, and performs filtering and down-sampling with a bandwidth of about 1000Hz around this frequency, which greatly reduces the computing resource requirements and operation time.

[0097] Preferably, the WVS spectrum is vertically averaged to obtain the frequency average value f avg , we can get the natural frequency fluctuation curve, then f avg Expressed as:

[0098]

[0099] Among them, A i is the WVS spectrum amplitude, f i is the frequency.

[0100] like Figure 4 As shown, Figure 4 (a) is the WVS spectrum of a faulty bearing with an outer ring fault. Figure 4 (b) is the natural frequency fluctuation curve of the faulty bearing.

[0101] Set the duration of the fluctuation region in the natural frequency fluctuation curve to t exit , then the fault size l is expressed as:

[0102]

[0103] Among them, Dp is the average diameter of the inner and outer rings of the bearing, d is the rolling element diameter, and f r is the shaft speed / Hz.

[0104] According to the above method, the fault size of the faulty bearing can be predicted by calculation.

[0105] Embodiment 3

[0106] In order to demonstrate the effectiveness of this method, this embodiment provides a bearing degradation experiment. A destructive experiment of long-term decay is conducted on a pre-treated bearing with an initial fault. During the experiment, the sampling frequency, sampling length and acquisition cycle interval are controlled by self-compiled software. The bearing test bench is shut down every 200,000 revolutions, the bearing under test is removed, the bearing size is measured, and the bearing under test is reinstalled and the experiment is continued. The experimental results are as follows: Figure 5 As shown, the lateral fluctuation curve is the natural frequency fluctuation curve calculated by this method, the area between the vertical straight lines is the fault area size predicted by this method, and the shaded part is the fault area length actually measured by disassembling the bearing.

[0107] As shown in the figure, the length of the fault area predicted by this method is consistent with the actual measured size (as shown in the 24th, 26th and 28th hours, the lengths are 2.2, 4.9 and 5.5 mm respectively), which proves the effectiveness and reliability of this method.

[0108] In a preferred embodiment, a detailed maintenance and replacement plan can be formulated for the faulty bearing based on the above calculation results and specific bearing operating standards and requirements to ensure the normal operation of the bearing.

[0109] In summary, compared with traditional prediction methods, the bearing spalling size prediction and processing method and application provided by the present invention performs secondary processing on the signal on the basis of detecting the bearing fault type in the conventional time domain and frequency domain, calculates the WVS spectrum, and obtains the natural frequency fluctuation curve based on the WVS spectrum. According to the width of the fluctuation in the natural frequency, the size of the bearing fault is calculated and predicted, the bearing status is evaluated, and the severity of the bearing fault is judged more accurately, so as to facilitate the formulation of detailed maintenance and replacement plans according to the working standards and requirements of the bearing, which has obvious advantages.

[0110] In addition, those skilled in the art should understand that, although there are many problems in the prior art, each embodiment or technical solution of the present invention can be improved in only one or several aspects, without having to solve all the technical problems listed in the prior art or background technology at the same time. Those skilled in the art should understand that the content not mentioned in a claim should not be used as a limitation on the claim.

[0111] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting and processing the spalling size of a bearing, characterized in that: The following steps are involved: Collect the original vibration signal of the test bearing, filter and reduce noise on the original vibration signal to obtain a valid signal; Extract the time domain features, frequency domain features and bearing fault feature amplitude from the effective signal to determine the fault type; According to the bearing fault type, the effective signal is periodically segmented to obtain a segmented signal; Calculate the WVS spectrum based on the segmented signal; The natural frequency fluctuation curve is obtained by longitudinally averaging the WVS spectrum to predict the bearing fault size.

2. The bearing spalling size prediction and processing method according to claim 1 is characterized in that: Before filtering the original vibration signal, the autopower spectral density of the collected original vibration signal of the faulty bearing and the original signal of the healthy bearing is compared to find the frequency band range in which the bearing fault is more sensitive to the vibration signal and then filter it.

3. The bearing peeling size prediction and processing method according to claim 1 is characterized in that: Determining the fault type comprises the following steps: Extract time domain features from valid signals, including at least RMS and kurtosis; Perform fast Fourier transform on the effective signal to obtain the frequency domain spectrum; Perform Hilbert transform on the effective signal to obtain the envelope spectrum; The characteristic frequency of each bearing fault type is calculated by bearing size and shaft speed. The peak value of the spectrum and envelope spectrum of the effective signal is selected within 10% before and after the characteristic frequency of each fault type to obtain the characteristic amplitude of each fault type. The bearing fault type is determined based on the comparison results of the time domain characteristics, frequency spectrum and envelope spectrum with the characteristic amplitudes of each fault type.

4. The bearing peeling size prediction and processing method according to claim 3 is characterized in that: The bearing fault types include cage fault, rolling element rotation fault, outer ring fault and inner ring fault, and the cage fault frequency FTF is expressed as: The rolling element rotation fault frequency BSF is expressed as: The outer race fault frequency BPFO is expressed as: The inner race fault frequency BPFI is expressed as: Among them, N is the bearing rotation frequency, n is the number of rolling elements, d is the rolling element diameter, D is the bearing median diameter, and α is the rolling element contact angle.

5. The bearing peeling size prediction and processing method according to claim 1 is characterized in that: The periodic segmentation of the effective signal comprises the following steps: According to the fault type, extract the envelope spectrum of the corresponding fault type, filter within the range of 10%-15% before and after the envelope spectrum, and perform order tracking to obtain the phase; Resampling the time domain feature according to the phase so that the number of sampling points of the effective signal in each fault cycle is consistent; The signal is segmented according to the fault cycle to obtain a segmented signal.

6. The bearing peeling size prediction and processing method according to claim 1 is characterized in that: The WVS spectrum is calculated based on the segmented signal x(t) as follows: Among them, R x (t,τ) is the autocorrelation function of the segmented signal, expressed as E[] is the expected function, t is time, τ is the time difference, and f is the frequency.

7. The bearing peeling size prediction and processing method according to claim 1 is characterized in that: The WVS spectrum is averaged longitudinally to obtain the frequency average value f avg , we get the natural frequency fluctuation curve, then f avg Expressed as: Among them, A i is the WVS spectrum amplitude, f i is the frequency.

8. The bearing spalling size prediction and processing method according to claim 1 is characterized in that: Set the duration of the fluctuation region in the natural frequency fluctuation curve to t exit , then the fault size l is expressed as: Among them, D p is the average diameter of the inner and outer rings of the bearing, d is the rolling element diameter, and f r is the shaft speed / Hz.

9. An application of a bearing spalling size prediction and processing method, characterized in that: The bearing fault size is obtained according to the bearing spalling size prediction processing method according to any one of claims 1-8, so as to formulate a detailed maintenance and replacement plan for the bearing.

Images