Bearing fault diagnosis method, system, electronic device and storage medium
Through the OMEDA-SSD method, combining the optimal minimum entropy deconvolution and singular spectrum decomposition, highlighting the bearing characteristic frequency, solving the limitations of early weak impact signal decomposition of bearings, and achieving accurate positioning diagnosis of bearing fault parts.
Patent Information
- Application Number
- CN202210488210.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-05-06
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-05-06
AI Technical Summary
The existing time series-based bearing diagnostic methods, especially the Singular Spectral Decomposition (SSD) method, have limitations on the decomposition of weak impact signals in early bearings, making it difficult to effectively extract and identify fault characteristics.
Using the OMEDA-SSD-based method, by obtaining the original acceleration signal of the bearing, rapid spectral kurtosis analysis is performed to determine the resonant frequency band, using the optimal minimum entropy deconvolution (OMEDA) as the pre-filter, combined with SSD decomposition, to obtain Reconstruct the acceleration signal and perform envelope demodulation to highlight the bearing characteristic frequency and identify the faulty part.
It significantly enhances the impact characteristics of the bearing's weak impact signal, improves the accurate positioning ability and recognition accuracy of fault diagnosis, solves the limitations of the SSD decomposition method on early weak impact signals, and can accurately identify the bearing faulty parts.
Smart Images

Figure CN114739674B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of bearing fault diagnosis, and in particular relates to a diagnosis method, system and storage medium for weak impacts generated by early bearing faults. Background Art
[0002] As an indispensable component of the transmission system, bearings are widely used in industry, transportation and military fields. The quality of their operating status will directly affect the working condition of the entire equipment. When the bearing has an early failure, a weak impact signal will be generated, and the resonance frequency usually appears in the high-frequency band. The commonly used time-frequency analysis method has limitations for weak impact signals.
[0003] Among the existing time series-based bearing diagnosis methods, a singular spectrum decomposition (SSD) is proposed based on the singular value decomposition network SVD. SSD overcomes the defects of the singular spectrum method SSA, can adaptively select the size of the embedding dimension, and decompose several singular spectrum components (SSC) from the original time series. SSD decomposition can decompose the impact signal of the bearing fault and effectively extract the information of the bearing fault.
[0004] Although SSD decomposition can extract the characteristics of high-frequency impact signals of bearings, it has limitations for early weak impact signals of bearings. Summary of the invention
[0005] The technical problem to be solved by the present invention is to provide a bearing fault diagnosis method, system, electronic device and storage medium, which can significantly enhance the impact characteristics of weak impact signals of bearings and accurately locate, diagnose and identify weak impacts caused by early bearing failures.
[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:
[0007] A bearing fault diagnosis method based on OMEDA-SSD, characterized by comprising the following steps:
[0008] S1, obtaining the original acceleration signal of the bearing;
[0009] S2. Draw the fast spectral kurtosis according to the original acceleration signal, compare all the kurtosis, and get the maximum spectral kurtosis K max (f i , Δf m ) corresponds to the center frequency f i and the optimal bandwidth Δf m , according to the center frequency f iand the optimal bandwidth Δf m Obtaining the resonant frequency band of the original acceleration signal;
[0010] S3, setting a filtering step size according to the resonance frequency band, inputting the filtering step size into the OMEDA method as a parameter, and then applying the optimal minimum entropy deconvolution decomposition OMEDA method to process the original acceleration signal;
[0011] S4, performing SSD decomposition on the signal processed in step S3;
[0012] S5, obtaining the correlation coefficient between each layer of singular spectrum components after SSD decomposition and the signal processed in step S3, adding the corresponding values of the singular spectrum components of each layer whose correlation coefficient is greater than the set value, to obtain the reconstructed acceleration signal;
[0013] S6, performing envelope demodulation on the reconstructed acceleration signal to obtain an envelope spectrum;
[0014] S7. Calculate the characteristic frequency of a specific component or structure of the bearing according to the bearing model and / or structural parameters to be diagnosed;
[0015] S8. Check the amplitude of the bearing characteristic frequency in the envelope spectrum. If the amplitude of the characteristic frequency of a specific component or structure of the bearing exceeds the envelope threshold, the bearing component corresponding to this characteristic frequency is faulty.
[0016] In the above technical solution, in step S2, the resonant frequency of the bearing resonant frequency band is in the range of 6400 to 7600 Hz.
[0017] In the above technical solution, the OMEDA decomposition in step S3 accurately calculates the filter coefficients through non-iterative calculations, and constructs a series of FIR filters so that the D-norm of the filtered signal is maximized to obtain the impact pulse.
[0018] In the above technical solution, the SSD decomposition in step S4 decomposes the original signal into a plurality of singular spectral components arranged in descending order of frequency bands by constructing a new form of trajectory matrix.
[0019] In the above technical solution, the set value in step S5 is 3.
[0020] In the above technical solution, the characteristic frequencies of the bearing in step S8 include the characteristic frequencies of four structural components of the bearing: the inner ring, the outer ring, the cage, and the rolling elements.
[0021] A bearing fault diagnosis system based on OMEDA-SSD, characterized by at least comprising:
[0022] The OMEDA processing unit is used to obtain the original acceleration signal of the bearing, draw the fast spectrum kurtosis according to the original acceleration signal and obtain the resonance frequency band of the original acceleration signal, calculate the filter step size, and process the original acceleration signal using the OMEDA method;
[0023] An SSD decomposition unit is used to perform SSD decomposition on the signal processed by the OMEDA method, add the corresponding values of the singular spectrum components of each layer in the SSD decomposition model whose correlation coefficient with the signal processed in step S3 is greater than a set value, and obtain a reconstructed acceleration signal; perform envelope demodulation on the reconstructed acceleration signal to obtain an envelope spectrum;
[0024] The fault identification unit is used to calculate the bearing characteristic frequency of the corresponding component or structure, and compare the bearing characteristic frequency in the envelope spectrum with the bearing characteristic frequency in amplitude to identify the position where the amplitude of the bearing characteristic frequency exceeds the envelope threshold.
[0025] As an implementable carrier of the present method, the present invention also provides an electronic device, comprising: a memory, a processor, and a computer program on the memory and executable on the processor, wherein the processor implements the steps of any of the above methods when executing the computer program.
[0026] The above-mentioned bearing fault diagnosis system based on OMEDA-SSD may or may not include a sensor device for acquiring the original acceleration signal of the bearing.
[0027] In addition, as an alternative implementation of the present invention, the present invention also provides a transient or non-transient computer-readable storage medium, on which a computer program is stored, characterized in that when the computer program is executed, the steps of the method described in any one of the above items are implemented.
[0028] Therefore, the method, system, electronic device and storage medium of the present invention decompose the bearing acceleration signal by combining the optimal minimum entropy deconvolution (OMED) as a pre-filter and a post-SSD decomposition unit; obtain the correlation coefficient between each layer of singular spectrum components after SSD decomposition and the signal after OMEDA processing, add the corresponding values of the singular spectrum components of each layer whose correlation coefficient is greater than the set value, and obtain the reconstructed acceleration signal; by performing envelope demodulation on the reconstructed signal and obtaining the demodulated spectrum, the bearing characteristic frequency is brought into the envelope spectrum to find the amplitude of the corresponding frequency, and the bearing position classification with a relatively prominent amplitude close to the characteristic frequency is found, so as to preliminarily determine the specific location of the bearing failure.
[0029] Compared with the prior art, the present invention has the following obvious beneficial effects:
[0030] (1) The optimal minimum entropy deconvolution (OMED) is used as a pre-filter to enhance the impact characteristics of the weak impact signal of the bearing, making it easier for SSD to quickly decompose the impact signal for analysis and improve the decomposition efficiency.
[0031] (2) The optimal minimum entropy deconvolution OMEDA method in the present invention can highlight the impact components in the signal and suppress the noise in the signal, which can effectively solve the limitations of relying solely on SSD decomposition of the signal. Compared with SSD decomposition, it can highlight the decomposition of the early weak impact signals of the bearing and effectively and accurately identify the faulty parts of the bearing, thus solving the blind spots and limitations of the SSD decomposition method in identifying the early weak impact signals of the bearing.
[0032] (3) The specific location of the early weak impact signal fault of the bearing can be obtained simply by extracting or detecting the acceleration signal and importing it into the present method or system. The detection principle is simple and easy to implement, which is convenient for manual assistance or unmanned operation on site. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:
[0034] Figure 1 It is a flow chart of the bearing fault diagnosis method based on OMEDA-SSD of the present invention.
[0035] Figure 2 It is a fast spectral kurtosis diagram obtained by the method of the present invention.
[0036] Figure 3 It is a comparison diagram between the original signal and the signal after being processed by the method, system or storage medium of the present invention.
[0037] Figure 4 It is a signal envelope spectrum obtained according to the method, system or storage medium of the present invention. DETAILED DESCRIPTION
[0038] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0039] like Figure 1 As shown, Figure 1 The following is a flow chart of the analysis method of the present invention, a bearing fault diagnosis method based on OMEDA-SSD, which specifically includes the following steps:
[0040] S1, obtaining the original acceleration signal of the bearing;
[0041] S2. Draw the fast spectral kurtosis according to the original acceleration signal, compare all the kurtosis, and get the maximum spectral kurtosis K max (f i , Δf m ) corresponds to the center frequency f i and the optimal bandwidth Δf m , according to the center frequency f i and the optimal bandwidth Δf m Obtaining the resonant frequency band of the original acceleration signal;
[0042] S3, setting a filtering step size according to the resonance frequency band, inputting the filtering step size into the OMEDA method as a parameter, and then using the OMEDA method to process the original acceleration signal;
[0043] S4, performing SSD decomposition on the signal processed in step S3 to obtain the singular spectrum components of each layer;
[0044] S5, obtaining the correlation coefficient between each layer of singular spectrum components and the signal processed in step S3, adding the corresponding values of the singular spectrum components of each layer with a correlation coefficient greater than 0.3 to obtain a reconstructed acceleration signal;
[0045] S6, performing envelope demodulation on the reconstructed acceleration signal to obtain an envelope spectrum;
[0046] S7, calculate the bearing characteristic frequency;
[0047] S8. Check the amplitude of the bearing characteristic frequency in the envelope spectrum. If the amplitude of a certain characteristic frequency exceeds the envelope threshold, the bearing component corresponding to this characteristic frequency is faulty.
[0048] In step S2, the fast spectral kurtosis is plotted according to the original acceleration signal of the bearing, such as Figure 2 As shown, wherein the horizontal axis is frequency and the vertical axis is kurtosis value, the present invention preferably obtains a resonant frequency of the bearing resonant frequency band in the range of 6400 to 7600 Hz.
[0049] In step S3, the filter step size is set to 1200, and the OMEDA method is used to highlight the impact component in the original acceleration signal, such as Figure 3 As shown in the figure, the upper part is the original signal, and the lower part is the signal after the original acceleration signal is processed using the OMEDA method. The horizontal axis in the upper and lower figures is (unit: time / s) and the vertical axis is (unit: amplitude / m / s 2 ), where several obvious impact components are enhanced. The OMEDA algorithm of the present invention calculates the filter coefficients accurately through non-iteration, constructs a series of FIR filters, and maximizes the D-norm of the filtered signal, thereby obtaining the impact pulse and extracting the fault characteristics. The OMEDA has the advantages of few input parameters and does not require any prior knowledge.
[0050] In step S4, the signal processed in step S3 is subjected to SSD decomposition to obtain the correlation coefficient between each layer of singular spectrum components and the signal processed in step S3, as shown in Table 1, which shows 7 correlation coefficients obtained according to each layer of singular spectrum components in 7 layers. SSD decomposition decomposes the original signal into multiple singular spectrum components arranged in descending order of frequency bands by constructing a new form of trajectory matrix.
[0051] Table 1 Correlation coefficients of singular spectral components at each layer
[0052] Layer Correlation coefficient SSC1 0.3542 SSC2 0.6625 SSC3 0.4524 SSC4 0.2687 SSC5 0.1739 SSC6 0.6321 SSC7 0.1293
[0053] It can be seen from Table 1 that among the 7 correlation coefficients, the correlation coefficients of the 1st to 3rd components SSC1-SSC3 and the 6th component SSC6 are greater than 0.3. According to step S5, these 4 components are added together to form a reconstructed signal.
[0054] Select bearing model 6314 deep groove ball bearing as a specific diagnostic example for analysis:
[0055] Bearing parameters: number of rolling elements Z = 8, rolling element diameter d = 25.4 mm, bearing median diameter D = 110 mm, contact angle α = 0. The bearing failure frequency at a speed of 1700 r / min is shown in Table 2, including the characteristic frequencies of the four structures of the bearing: inner ring, outer ring, cage, and rolling element.
[0056] Table 2 Bearing characteristic frequency (Hz)
[0057] structure Inner Circle Outer ring Cage Rolling element Characteristic frequency 139.5 87.2 11.3 116.2
[0058] According to step S6, the reconstructed signal is envelope demodulated and a demodulated spectrum is obtained, such as Figure 4 The signal envelope spectrum shown in the figure, where the horizontal axis is the characteristic frequency amplitude (unit: frequency / Hz) and the vertical axis is the spectrum value (unit: amplitude / m / s 2 ).
[0059] According to step S8, the characteristic frequency of the bearing is brought into the envelope spectrum to find the amplitude of the corresponding frequency. According to experience, the amplitude of 139Hz is relatively prominent, so it can be preliminarily judged that the inner ring of the bearing is faulty. The verification results show that this method can effectively identify the faulty part of the bearing, thereby effectively diagnosing the bearing fault.
[0060] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.
Claims
1. A bearing fault diagnosis method based on OMEDA-SSD, characterized in that The following steps are involved: S1, obtaining the original acceleration signal of the bearing; S2. Draw the fast spectral kurtosis according to the original acceleration signal, compare all the kurtosis, and get the maximum spectral kurtosis K max (f i , Δf m ) corresponds to the center frequency f i and the optimal bandwidth Δf m , according to the center frequency f i and the optimal bandwidth Δf m Obtaining the resonant frequency band of the original acceleration signal; S3, setting a filtering step size according to the resonance frequency band, inputting the filtering step size into the OMEDA method as a parameter, and then applying the optimal minimum entropy deconvolution OMEDA method to process the original acceleration signal; S4, performing SSD decomposition on the signal processed in step S3; S5, obtaining the correlation coefficient between each layer of singular spectrum components after SSD decomposition and the signal processed in step S3, adding the corresponding values of the singular spectrum components of each layer whose correlation coefficient is greater than the set value, to obtain the reconstructed acceleration signal; S6, performing envelope demodulation on the reconstructed acceleration signal to obtain an envelope spectrum; S7. Calculate the bearing characteristic frequency according to the bearing model and / or structural parameters to be diagnosed; S8. Check the amplitude of the bearing characteristic frequency in the envelope spectrum. If the amplitude of a bearing characteristic frequency exceeds the envelope threshold, the bearing component corresponding to this characteristic frequency is faulty.
2. The bearing fault diagnosis method based on OMEDA-SSD according to claim 1 is characterized in that In step S2, the resonant frequency of the bearing resonant frequency band is in the range of 6400-7600 Hz.
3. The bearing fault diagnosis method based on OMEDA-SSD according to claim 1 is characterized in that The OMEDA decomposition in step S3 accurately calculates filter coefficients through non-iterative calculations and constructs a series of FIR filters so that the D-norm of the filtered signal is maximized to obtain the impact pulse.
4. The bearing fault diagnosis method based on OMEDA-SSD according to claim 1 is characterized in that The SSD decomposition in step S4 decomposes the original signal into a plurality of singular spectral components arranged in descending order of frequency bands by constructing a new form of trajectory matrix.
5. The bearing fault diagnosis method based on OMEDA-SSD according to claim 1 is characterized in that The set value in step S5 is 3.
6. The bearing fault diagnosis method based on OMEDA-SSD according to claim 1 is characterized in that The characteristic frequencies of the bearing in step S8 include the characteristic frequencies of four structural components of the bearing: the inner ring, the outer ring, the cage, and the rolling elements.
7. A bearing fault diagnosis system based on OMEDA-SSD, characterized in that The bearing fault diagnosis method based on OMEDA-SSD for implementing any one of claims 1 to 6 comprises at least: The OMEDA processing unit is used to obtain the original acceleration signal of the bearing, draw the fast spectrum kurtosis according to the original acceleration signal and obtain the resonance frequency band of the original acceleration signal, calculate the filter step size, and process the original acceleration signal using the OMEDA method; An SSD decomposition unit is used to perform SSD decomposition on the signal processed by the OMEDA method, add the corresponding values of the singular spectrum components of each layer in the SSD decomposition model whose correlation coefficient with the signal processed in step S3 is greater than a set value, and obtain a reconstructed acceleration signal; perform envelope demodulation on the reconstructed acceleration signal to obtain an envelope spectrum; The fault identification unit is used to calculate the bearing characteristic frequency of the corresponding component or structure, and compare the bearing characteristic frequency in the envelope spectrum with the bearing characteristic frequency in amplitude to identify the position where the amplitude of the bearing characteristic frequency exceeds the envelope threshold.
8. An electronic device, comprising: A memory, a processor, and a computer program on the memory and executable on the processor, wherein the processor implements the steps of the method as described in any one of claims 1 to 6 when executing the computer program.
9. The electronic device according to claim 8, characterized in that Includes or does not include a sensor device for acquiring the raw acceleration signal of the bearing.
10. A transient or non-transient computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the steps of the method according to any one of claims 1 to 6 are implemented.
