Rolling bearing fault diagnosis method, system and device based on AR-MED and SSD and medium

Abstract

The invention discloses a rolling bearing fault diagnosis method, system and device based on AR-MED and SSD, and a medium, and the method comprises the following steps: carrying out the preprocessing of a bearing fault signal through AR-MED to reduce the influence of noise, and obtaining a noise reduction signal; performing SSD decomposition on the noise reduction signal to obtain the number of singular spectrum components when the energy difference is minimum; calculating the kurtosis value of each singular spectrum component, and screening out the singular spectrum component with the most obvious fault impact characteristic as the optimal component according to the principle of maximum kurtosis; and performing envelope demodulation analysis on the optimal component, and diagnosing the fault type of the bearing. According to the method, the AR-MED method and the SSD method are effectively combined, and the optimal component is screened by using the kurtosis index, so that the screened result is more accurate, and the extraction of the fault features of the rolling bearing is facilitated.

Description

Technical Field

[0001] The present invention relates to the technical field of rolling bearings, and in particular to a rolling bearing fault diagnosis method, system, equipment and medium based on AR-MED and SSD. Background Art

[0002] Rotating machinery often uses rolling bearings as support. Its working performance and reliability are affected by the bearing damage state. Therefore, it is of great significance to diagnose rolling bearing faults. Common rolling bearing faults include inner ring faults, outer ring faults and rolling element faults, which will form periodic impacts. Many parts of rotating machinery will produce coupling phenomena during operation. The vibration signal includes a large amount of background noise and vibration interference caused by gear meshing, which greatly affects the resolution of bearing fault pulse signals. Therefore, extracting bearing fault characteristics from complex vibration signals is an important part of fault diagnosis.

[0003] The singular spectrum decomposition (SSD) method can effectively suppress modal aliasing and pseudo components in vibration signals and has a good multi-component signal decomposition effect. It can effectively extract bearing fault feature information and improve the accuracy of traditional bearing fault diagnosis methods. Summary of the invention

[0004] The purpose of the present invention is to provide a rolling bearing fault diagnosis method, system, equipment and medium based on AR-MED and SSD in order to solve the above problems.

[0005] To achieve the above purpose, the technical solution provided by the present invention is:

[0006] The first aspect of the present application provides a rolling bearing fault diagnosis method based on AR-MED and SSD, comprising the following steps:

[0007] The bearing fault signal is preprocessed using AR-MED to reduce the influence of noise and obtain a noise-reduced signal.

[0008] Perform SSD decomposition on the denoised signal to obtain the number of singular spectral components with the minimum energy difference;

[0009] Calculate the kurtosis value of each singular spectrum component, and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle;

[0010] Perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type.

[0011] To optimize the above technical solutions, the specific measures taken also include:

[0012] The process of using AR-MED to preprocess the bearing fault signal to reduce the influence of noise and obtain a noise reduction signal is as follows:

[0013] According to the vibration signal x in the bearing fault signal = [x1, x2, ..., x N ] T and the length of the filter p, construct the filter signal set X = [x p+1 ,x p+2 ,...,x N ], and the filtered signal matrix The AR filter error signal ε=X-YF is obtained;

[0014] Where F = (Y Τ Y) -1 Y Τ X, construct the error signal matrix According to the set inverse filter f(n) = [0,0,...,1,...0,0], the first filtering result is calculated to obtain the noise reduction signal sequence Then update the inverse filter where y i for The i-th element of Update the sequence estimation results again Repeat the above steps according to the set number of iterations to save the denoised signal sequence obtained from the last iteration Replace the original vibration signal as the vibration signal after noise reduction.

[0015] Furthermore, the SSD decomposition of the denoised signal x to obtain the number of singular spectral components with the minimum energy difference includes:

[0016] Create an M×N trajectory matrix X t , the mth row of the matrix is ​​expressed as:

[0017] x m =(x(i),x(i+1),...,x(N),x(1),x(2),...,x(i-1));

[0018] where i = 1, ..., M, x(i) is the i-th element in x, and the matrix

[0019] Adaptively select the embedding dimension M. The singular spectrum decomposition uses an adaptive rule to select the embedding dimension M at the jth iteration. First, calculate the residual component v at the jth iteration. j The power spectral density of (n), v j (n) The iteration formula is x(n) represents the vibration signal after noise reduction; then v jThe frequency f corresponding to the maximum peak of the power spectrum density of (n) max An estimate is made to reconstruct the singular spectral components.

[0020] The described j The frequency f corresponding to the maximum peak of the power spectrum density of (n) max For the first iteration, if f max / f s <10 -3 , f s is the sampling rate, M is set to N / 3; when j>1, M=1.2×(f s / f max ).

[0021] Furthermore, the singular spectral components are reconstructed in order from high frequency to low frequency, and X is obtained according to the singular value decomposition method. t =σμv Τ , where σ is the left singular matrix, v is the right singular matrix, and μ is the diagonal matrix. The average value of the diagonal of this matrix is ​​defined as the diagonal average value g(n), and the diagonal average value obtained at the jth iteration is recorded as g j (n), as the reconstruction component;

[0022] Find the residual v of the denoised vibration signal x(n) after removing all reconstructed component signals after the jth iteration j (n), and then calculate the normalized mean square error with the original denoised vibration signal and perform SSD decomposition.

[0023] Furthermore, the normalized mean square error is expressed as:

[0024]

[0025] When NMSE (j) When <1%, the decomposition is terminated; otherwise, the residual continues to be decomposed until the stop condition is met; after m iterations, m decomposition components are obtained; the final decomposition result of SSD is v m+1 (n) is the residual after the iteration.

[0026] Furthermore, the kurtosis value of each singular spectral component is calculated, and according to the maximum kurtosis principle, the singular spectral component with the most obvious fault impact characteristics is screened out as the best component, so as to screen out the singular spectral component with the largest kurtosis value of the vibration signal x(n) after noise reduction as the best component.

[0027] The second aspect of the present application provides a rolling bearing fault diagnosis system based on AR-MED and SSD, comprising:

[0028] A noise reduction module is used to pre-process the bearing fault signal through AR-MED to reduce the influence of noise and obtain a noise reduction signal;

[0029] An SSD decomposition module is used to perform SSD decomposition on the noise reduction signal to obtain the number of singular spectral components with the minimum energy difference;

[0030] The kurtosis calculation module is used to calculate the kurtosis value of each singular spectrum component and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle;

[0031] The analysis and diagnosis module is used to perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type.

[0032] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the rolling bearing fault diagnosis method based on AR-MED and SSD as described above is implemented.

[0033] A fourth aspect of the present application provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the rolling bearing fault diagnosis method based on AR-MED and SSD as described above.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] In the early stage of the rolling bearing failure of the present invention, the impact part is not obvious and is seriously disturbed by noise. At this time, it is difficult to remove noise and vibration interference and extract the fault component using only the SSD method, so noise reduction preprocessing is required. The Auto Regressive Minimum Entropy Deconvolution (AR-MED) method can filter out noise and enhance pulse components through the optimal filter, thereby improving the signal-to-noise ratio of the bearing vibration signal. Therefore, AR-MED is combined with SSD to propose a bearing fault diagnosis method, and its effectiveness is verified.

[0036] The present invention effectively combines the AR-MED and SSD methods, and uses the kurtosis index to screen the optimal component, so that the screened result is more accurate, which is beneficial to the extraction of rolling bearing fault characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 : Schematic diagram of the flow chart of the rolling bearing fault diagnosis method based on AR-MED and SSD of the present invention.

[0038] Figure 2 : Envelope spectrum obtained based on the traditional SSD method.

[0039] Figure 3 : Envelope spectrum obtained based on the method of combining AR-MED with SSD of the present invention. DETAILED DESCRIPTION

[0040] The above contents of the present invention are further described in detail below in the form of embodiments, but this should not be understood as the scope of the above subject matter of the present invention being limited to the following embodiments, and all technologies realized based on the above contents of the present invention belong to the scope of the present invention.

[0041] This application provides a rolling bearing fault diagnosis method based on AR-MED and SSD, such as Figure 1 As shown, the following steps are included:

[0042] The bearing fault signal is preprocessed using AR-MED to reduce the influence of noise and obtain a noise-reduced signal.

[0043] Perform SSD decomposition on the denoised signal to obtain the number of singular spectral components with the minimum energy difference;

[0044] Calculate the kurtosis value of each singular spectrum component, and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle;

[0045] Perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type.

[0046] In some implementations, AR-MED is used to pre-process the bearing fault signal to reduce the impact of noise and obtain a noise-reduced signal, and the process is as follows:

[0047] According to the vibration signal x in the bearing fault signal = [x1, x2, ..., x N ] T and the length of the filter p, construct the filter signal set X = [x p+1 ,x p+2 ,...,x N ], and the filtered signal matrix The AR filter error signal ε=X-YF is obtained;

[0048] Where F = (Y Τ Y) -1 Y Τ X, construct the error signal matrix According to the set inverse filter f(n) = [0,0,...,1,...0,0], the first filtering result is calculated to obtain the noise reduction signal sequence Then update the inverse filter where y i for The i-th element of Update the sequence estimation results again Repeat the above steps according to the set number of iterations to save the denoised signal sequence obtained from the last iteration Replace the original vibration signal as the vibration signal after noise reduction.

[0049] In some implementations, performing SSD decomposition on the denoised signal x to obtain the number of singular spectral components with the minimum energy difference includes:

[0050] Create an M×N trajectory matrix X t , the mth row of the matrix is ​​expressed as:

[0051] x m =(x(i),x(i+1),...,x(N),x(1),x(2),...,x(i-1));

[0052] where i = 1, ..., M, x(i) is the i-th element in x, and the matrix

[0053] Adaptively select the embedding dimension M. The singular spectrum decomposition uses an adaptive rule to select the embedding dimension M at the jth iteration. First, calculate the residual component v at the jth iteration. j The power spectral density of (n), v j (n) The iteration formula is x(n) represents the vibration signal after noise reduction; then v j The frequency f corresponding to the maximum peak of the power spectrum density of (n) max An estimate is made to reconstruct the singular spectral components.

[0054] v j The frequency f corresponding to the maximum peak of the power spectrum density of (n) max For the first iteration, if f max / f s <10 -3 , f s is the sampling rate, M is set to N / 3; when j>1, M=1.2×(f s / f max ).

[0055] Reconstruct the singular spectrum components in order from high frequency to low frequency, and find X by singular value decomposition method t =σμv Τ , where σ is the left singular matrix, v is the right singular matrix, and μ is the diagonal matrix. The average value of the diagonal of this matrix is ​​defined as the diagonal average value g(n), and the diagonal average value obtained at the jth iteration is recorded as g j (n), as the reconstruction component;

[0056] Find the residual v of the denoised vibration signal x(n) after removing all reconstructed component signals after the jth iteration j (n), and then calculate the normalized mean square error with the original denoised vibration signal and perform SSD decomposition.

[0057] The normalized mean square error is expressed as:

[0058]

[0059] When NMSE (j) When <1%, the decomposition is terminated; otherwise, the residual continues to be decomposed until the stop condition is met; after m iterations, m decomposition components are obtained; the final decomposition result of SSD is v m+1 (n) is the residual after the iteration.

[0060] Calculate the kurtosis value of each singular spectral component, and according to the maximum kurtosis principle, screen out the singular spectral component with the most obvious fault impact characteristics as the best component, and screen out the singular spectral component with the largest kurtosis value of the vibration signal x(n) after denoising as the best component.

[0061] The present invention effectively combines the AR-MED and SSD methods, and uses the kurtosis index to filter the optimal component, so that the results after screening are more accurate, which is beneficial to the extraction of rolling bearing fault characteristics. The technical solution of the present invention is further described in detail in combination with specific simulation experiments:

[0062] In the simulation environment, the bearing speed is set to 2951r / min, the inner ring fault frequency is 152Hz, and the sampling frequency is 50kHz. According to the simulated rolling bearing vibration acceleration signal, the envelope analysis results obtained based on different methods are as follows: Figure 2 , Figure 3 shown. Figure 2 This is the envelope spectrum obtained based on the traditional SSD method. The 1st frequency and 3rd frequency can be obtained in the result. At the same time, it can be seen that the amplitude is small, it is not easy to identify the fault frequency, and the diagnostic effect is not ideal. Figure 3 The envelope spectrum obtained by combining AR-MED with SSD can extract 1 to 5 times the frequency, and the amplitude is significantly enhanced, and the diagnostic effect is very good. It can be seen that the present invention uses AR-MED combined with SSD to pre-process the acceleration of the rolling bearing, while reducing noise, enhancing the impact component of the signal, which can increase the amplitude of the fault frequency point on the spectrum and make the fault easier to identify.

[0063] In another embodiment, the present invention proposes a rolling bearing fault diagnosis system based on AR-MED and SSD, comprising:

[0064] A noise reduction module is used to pre-process the bearing fault signal through AR-MED to reduce the influence of noise and obtain a noise reduction signal;

[0065] An SSD decomposition module is used to perform SSD decomposition on the noise reduction signal to obtain the number of singular spectral components with the minimum energy difference;

[0066] The kurtosis calculation module is used to calculate the kurtosis value of each singular spectrum component and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle;

[0067] The analysis and diagnosis module is used to perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type.

[0068] In another embodiment, the present invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the rolling bearing fault diagnosis method based on AR-MED and SSD as described above is implemented.

[0069] In another embodiment, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the rolling bearing fault diagnosis method based on AR-MED and SSD as described above.

[0070] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0071] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any technician familiar with the profession, without departing from the scope of the technical solution of the present invention, according to the technical essence of the present invention, any simple modification, equivalent replacement and improvement made to the above embodiment still falls within the protection scope of the technical solution of the present invention.

Claims

1. A rolling bearing fault diagnosis method based on AR-MED and SSD, characterized in that: The following steps are involved: The bearing fault signal is preprocessed using AR-MED to reduce the influence of noise and obtain a noise-reduced signal. Perform SSD decomposition on the denoised signal to obtain the number of singular spectral components with the minimum energy difference; Calculate the kurtosis value of each singular spectrum component, and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle; Perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type.

2. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 1, characterized in that: The process of using AR-MED to preprocess the bearing fault signal to reduce the influence of noise and obtain a noise reduction signal is as follows: According to the vibration signal x in the bearing fault signal = [x1, x2, ..., x N ] T and the length of the filter p, construct the filter signal set X = [x p+1 ,x p+2 ,...,x N ], and the filtered signal matrix The AR filter error signal ε=X-YF is obtained; Where F = (Y Τ Y) -1 Y Τ X, construct the error signal matrix According to the set inverse filter f(n) = [0,0,...,1,...0,0], the first filtering result is calculated to obtain the noise reduction signal sequence Then update the inverse filter where y i for The i-th element of Update the sequence estimation results again Repeat the above steps according to the set number of iterations to save the denoised signal sequence obtained from the last iteration Replace the original vibration signal as the vibration signal after noise reduction.

3. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 2 is characterized in that: The SSD decomposition of the denoised signal x to obtain the number of singular spectral components with the minimum energy difference includes: Create an M×N trajectory matrix X t , the mth row of the matrix is ​​expressed as: x m =(x(i),x(i+1),...,x(N),x(1),x(2),...,x(i-1)); where i = 1, ..., M, x(i) is the i-th element in x, and the matrix Adaptively select the embedding dimension M. The singular spectrum decomposition uses an adaptive rule to select the embedding dimension M at the jth iteration. First, calculate the residual component v at the jth iteration. j The power spectral density of (n), v j (n) The iteration formula is x(n) represents the vibration signal after noise reduction; then v j The frequency f corresponding to the maximum peak of the power spectrum density of (n) max An estimate is made to reconstruct the singular spectral components.

4. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 3 is characterized in that: The pair v j The frequency f corresponding to the maximum peak of the power spectrum density of (n) max For the first iteration, if f max / f s <10 -3 , f s is the sampling rate, M is set to N / 3; when j>1, M=1.2×(f s / f max ).

5. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 3 is characterized in that: Reconstruct the singular spectrum components in order from high frequency to low frequency, and find X by singular value decomposition method t =σμv Τ , where σ is the left singular matrix, v is the right singular matrix, and μ is the diagonal matrix. The average value of the diagonal of this matrix is ​​defined as the diagonal average value g(n), and the diagonal average value obtained at the jth iteration is recorded as g j (n), as the reconstruction component; Find the residual v of the denoised vibration signal x(n) after removing all reconstructed component signals after the jth iteration j (n), and then calculate the normalized mean square error with the original denoised vibration signal and perform SSD decomposition.

6. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 5 is characterized in that: The normalized mean square error is expressed as: When NMSE (j) When <1%, the decomposition is terminated; otherwise, the residual continues to be decomposed until the stop condition is met; after m iterations, m decomposition components are obtained; The final decomposition result of SSD is v m+1 (n) is the residual after the iteration.

7. The rolling bearing fault diagnosis method based on AR-MED and SSD according to claim 3 is characterized in that: The kurtosis value of each singular spectral component is calculated, and according to the maximum kurtosis principle, the singular spectral component with the most obvious fault impact characteristics is screened out as the best component, so as to screen out the singular spectral component with the largest kurtosis value of the vibration signal x(n) after noise reduction as the best component.

8. A rolling bearing fault diagnosis system based on AR-MED and SSD, characterized in that: include: A noise reduction module is used to pre-process the bearing fault signal through AR-MED to reduce the influence of noise and obtain a noise reduction signal; An SSD decomposition module is used to perform SSD decomposition on the noise reduction signal to obtain the number of singular spectral components with the minimum energy difference; The kurtosis calculation module is used to calculate the kurtosis value of each singular spectrum component and select the singular spectrum component with the most obvious fault impact characteristics as the optimal component according to the maximum kurtosis principle; The analysis and diagnosis module is used to perform envelope demodulation analysis on the optimal component and diagnose the bearing fault type. The system is used to execute the rolling bearing fault diagnosis method based on AR-MED and SSD as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the rolling bearing fault diagnosis method based on AR-MED and SSD as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the rolling bearing fault diagnosis method based on AR-MED and SSD according to any one of claims 1 to 7.

Images