Bearing fault diagnosis method based on improved time-varying mathematical morphology filtering

Abstract

The invention discloses a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering. The method comprises the following steps: collecting a vibration acceleration signal of a bearing; constructing improved time-varying structure elements with different orders; determining a mathematical product morphological operator with excellent performance; performing mathematical morphology filtering processing on the collected vibration acceleration signals to obtain a series of mathematical morphology filtering signals; selecting an optimal mathematical morphology filtering signal from the obtained mathematical morphology filtering signals; and performing power spectrum analysis on the optimal mathematical morphology filtering signal to judge whether the bearing has a fault and the fault type. The problem that in the prior art, under the strong background noise interference condition, time-varying structural elements are prone to noise interference, pulse features are difficult to accurately match and extract, and fault diagnosis of a bearing in the noisy industrial environment is not facilitated is solved. The improved time-varying structural element capable of eliminating interference of noise on the shape and length of the structural element is provided for bearing fault diagnosis, and the method has the advantages of being excellent in anti-noise performance and sensitive to fault pulse characteristics.

Description

Technical Field

[0001] The present invention belongs to the technical field of bearing fault diagnosis, and in particular relates to a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering. Background Art

[0002] Bearings are essential components in rotating machinery, supporting rotating parts, reducing friction, and ensuring smooth operation. Bearing health is crucial to the proper functioning of rotating machinery, and failures can lead to significant economic losses or serious personal injury. Therefore, fault diagnosis of bearings in key areas of rotating machinery is crucial.

[0003] Mathematical morphology filtering is a typical nonlinear signal processing method widely used in bearing fault diagnosis due to its simple principle, high computational efficiency, and minimal parameter requirements. Mathematical morphology filtering matches and extracts hidden pulse features through morphological operations (i.e., morphological operators) between specified structuring elements and vibration signals. Among them, time-varying structuring elements (TVEs) offer excellent pulse feature matching performance and computational efficiency, making them widely used in bearing fault diagnosis based on TVEs. However, actual bearing vibration acceleration signals are often contaminated by various fault-independent interferences. Research has shown that when processing vibration signals contaminated by strong background noise, TVEs based on local extreme points of the original vibration signal have difficulty accurately matching and extracting the repetitive transient pulse features contaminated by noise. Consequently, in low signal-to-noise ratio conditions, traditional TVEs exhibit insufficient feature extraction and poor noise reduction performance, making them unsuitable for bearing fault diagnosis in complex industrial environments. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering to solve the problem that bearing fault features are difficult to accurately extract under strong background noise interference.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: A bearing fault diagnosis method based on improved time-varying mathematical morphology filtering, the steps are as follows: S1. Collect the vibration acceleration signal of the bearing; S2. constructing improved time-varying structural elements of different orders according to the vibration acceleration signal collected in step S1; S3. Determine a mathematical product morphological operator with excellent performance; S4. Performing mathematical morphological filtering on the vibration acceleration signal based on the improved time-varying structuring elements of different orders constructed in step S2 and the mathematical product morphological operator with excellent performance determined in step S3 to obtain mathematical morphological filtered signals of different orders; S5, selecting an optimal mathematical morphology filtering signal from the mathematical morphology filtering signals of different orders obtained in step S4; S6. Perform power spectrum analysis on the optimal mathematical morphology filtering signal to determine the bearing fault. Preferably, the sub-steps of step S2 are: S21. Calculate the original vibration signal through k Median filtered signal after order median filtering , the median filter formula is: ; ; in, represents the median operator, Represents a data point n Nearby datasets, k is the order of the median filter, k The initial value is 2 ， express k remainder when divided by 2; S22, identifying median filter signal All local minimum points of ; S23, select the median filter signal The narrow waveform between two adjacent minimum points in the is used as the shape of the structural element; S24, adjusting each narrow waveform so that its minimum amplitude is zero; S25, the adjusted narrow waveform is used as k Improve the time-varying structure element by 10-order to extract the pulse features hidden in the corresponding narrow waveform of the vibration signal; S26, in turn k Assign values ​​to 2, 3, 4, 5, 6, 7, 8, 9, 10 , Repeat steps S21 to S25 to obtain improved time-varying structural elements of different orders.

[0006] Preferably, the sub-steps of step S3 are: S31. Select two combined morphological operators with excellent feature extraction and noise reduction performance, and they have similar filtering performance, that is, the same type of combined morphological operators, which are specifically defined as follows: ; ; in and are two combined morphological operators with excellent performance. represents the original vibration signal, represents the structural element in the morphological filtering method, represents the dilation operation in the morphological filtering method, Represents the corrosion operation in the morphological filtering method, represents the closing operation in the morphological filtering method, Represents the opening operation in the morphological filtering method.

[0007] S32. A mathematical product morphological operator is constructed using the product operation with theoretical advantages and the two combined morphological operators obtained in S31. The specific definition is as follows: ; in represents the product operation, It is a mathematical product morphological operator constructed based on these two combined morphological operators.

[0008] Preferably, step S5 specifically includes calculating the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filter signal obtained in step S4, and selecting the mathematical morphology filter signal with the largest characteristic frequency intensity coefficient as the optimal mathematical morphology filter signal.

[0009] Preferably, in step S5, the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filtered signal is calculated. I CFIC The calculation formula is: , ; in Represents the amplitude of different frequencies in the power spectrum; f m Indicates a certain specified fault characteristic frequency; f x express[( -0.02) f m , ( +0.02) f m ]Different frequencies in the interval; f m Indicates a certain specified fault characteristic frequency (such as inner ring, outer ring, rolling element or cage). By inputting different fault characteristic frequencies, you can get different fault characteristic frequencies. I CFIC ; f k represents the spectrum frequency, ;generally f 1=0; H is the order of the fault characteristic frequency involved in the calculation in the power spectrum, Kis the number of discrete spectral frequencies involved in the power spectrum calculation; where ±0.02 f m Used to eliminate the effects of slight speed fluctuations. Preferably, the sub-steps of step S6 are: S61. Calculate the fault characteristic frequencies of the inner ring, outer ring, rolling element, and cage of the bearing respectively. The calculation formula is as follows: The calculation formula for the inner race fault characteristic frequency (BPFI) is: ; The calculation formula for the outer race fault characteristic frequency (BPFO) is: ; The calculation formula for rolling element fault characteristic frequency (BSF) is: ; The calculation formula for cage fault characteristic frequency (FTF) is: ; in Z is the number of rolling elements, f r is the shaft speed, d is the rolling element diameter, D is the pitch diameter of the rolling bearing, is the contact angle; S62, calculate the power spectrum of each characteristic frequency in turn according to the power spectrum of the optimal mathematical morphology filtering signal and the different fault characteristic frequencies of step S61 I CFIC , the largest of which I CFIC The corresponding fault is a bearing fault.

[0010] Preferably, the content of step S1 includes: Select an acceleration sensor that is suitable for the working environment and measurement requirements, install the sensor on the bearing housing close to the bearing, and install multiple acceleration sensors in different directions to obtain comprehensive information.

[0011] A bearing fault diagnosis system based on improved time-varying mathematical morphology filtering adopts the bearing fault diagnosis method based on improved time-varying mathematical morphology filtering.

[0012] A computer device comprising: one or more processors; The processor is configured to store one or more programs; When the one or more programs are executed by the one or more processors, a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering as described in any one of claims 1 to 7 is implemented.

[0013] A computer-readable storage medium stores a computer program, which, when executed, implements a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering.

[0014] The present invention can achieve the following beneficial effects: The present invention only needs to perform power spectrum analysis on the optimal mathematical morphology filter signal to determine whether the bearing has a fault and the fault type according to the bearing fault characteristic frequency and its harmonic components in the power spectrum. The judgment method is simple and fast, and the calculation efficiency is high.

[0015] The proposed method, based on improved time-varying mathematical morphological filtering, employs an improved time-varying structuring element and the high-performance mathematical product morphological operator. This improved time-varying structuring element effectively mitigates the effects of interfering noise on the shape and length selection of the structuring element. This allows for more accurate matching and extraction of local pulse features hidden within noisy vibration signals even under strong background noise. This method addresses the inadequate feature extraction and poor noise reduction performance of traditional time-varying structuring elements in low signal-to-noise ratio conditions, resulting in superior feature extraction and noise reduction performance.

[0016] The present invention utilizes the mathematical product morphological operator, which not only has the excellent feature extraction performance of the combination morphological operator, but also further improves the extraction and enhancement effect of bearing fault features due to the unique feature enhancement advantage of the product operation, which helps to diagnose faults more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below with reference to the accompanying drawings and examples: Figure 1 It is a schematic diagram of the process of the present invention; Figure 2 Schematic diagram of signals corresponding to each step in an embodiment of the present invention; Figure 3 Schematic diagram of rolling element failure signal of rolling bearing in an embodiment of the present invention; (a) is a schematic diagram of the vibration acceleration signal, and (b) is a schematic diagram of the envelope spectrum; Figure 4 Schematic diagram of the filtered signal and its power spectrum obtained by processing the bearing rolling element fault signal using the proposed method ITVMMF in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the power spectrum; Figure 5Schematic diagram of the filtered signal and its diagonal slice spectrum obtained by processing the bearing rolling element fault signal using the comparative method ATVMF-DSS in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the diagonal slice spectrum; Figure 6 Schematic diagram of the filtered signal and its diagonal slice spectrum obtained by processing the bearing rolling element fault signal using the comparative method ESMHPF in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the diagonal slice spectrum; Figure 7 Schematic diagram of the filtered signal and its spectrum obtained by processing the bearing rolling element fault signal using the comparative method MMF in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the spectrum; Figure 8 Schematic diagram of a rolling bearing outer ring fault signal according to an embodiment of the present invention; (a) is a schematic diagram of the vibration acceleration signal, and (b) is a schematic diagram of the envelope spectrum; Figure 9 2 is a schematic diagram of a filtered signal and its power spectrum obtained by processing a bearing outer race fault signal using the proposed ITVMMF method in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the power spectrum; Figure 10 2 is a schematic diagram of a filter signal and a diagonal slice spectrum thereof obtained by processing a bearing outer race fault signal using the comparative method ATVMF-DSS in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the diagonal slice spectrum; Figure 11 1 is a schematic diagram of a filtered signal and its diagonal slice spectrum obtained by processing a bearing outer race fault signal using the comparative method ESMHPF in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the diagonal slice spectrum; Figure 12 1 is a schematic diagram of a filtered signal and its spectrum obtained by processing a bearing outer race fault signal using the comparative method MMF in an embodiment of the present invention; Where (a) is a schematic diagram of the filtered signal, and (b) is a schematic diagram of the spectrum. DETAILED DESCRIPTION

[0018] The preferred solution is Figures 1 to 12 As shown, a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering includes the following steps: S1. Collect the vibration acceleration signal of the bearing; In the embodiment of the present invention, step S1 specifically includes using a data acquisition device and a vibration acceleration sensor to acquire a vibration signal of the bearing.

[0019] S2. constructing improved time-varying structural elements of different orders according to the vibration acceleration signal collected in step S1; In the embodiment of the present invention, step S2 includes the following sub-steps: S21. Calculate the original vibration signal through k Median filtered signal after order median filtering , the median filter formula is: ; ; in, represents the median operator, Represents a data point n Nearby datasets, k is the order of the median filter, k The initial value is 2 ， express k remainder when divided by 2; S22, identifying median filter signal All local minimum points of ; S23, select the median filter signal The narrow waveform between two adjacent minimum points in the is used as the shape of the structural element; S24, adjusting each narrow waveform so that its minimum amplitude is zero; S25, the adjusted narrow waveform is used as k Improve the time-varying structure element by 10-order to extract the pulse features hidden in the corresponding narrow waveform of the vibration signal; S26, in turn k Assign values ​​to 2, 3, 4, 5, 6, 7, 8, 9, 10 , Repeat steps S21 to S25 to obtain improved time-varying structural elements of different orders.

[0020] S3. Determine a mathematical product morphological operator with excellent performance; After determining the structural element g and the morphological operator, the vibration signal can be processed according to the formula of the morphological operator. In the embodiment of the present invention, step S3 includes the following sub-steps: S31. Select two combined morphological operators with excellent feature extraction and noise reduction performance, and they have similar filtering performance, that is, the same type of combined morphological operators, which are specifically defined as follows: ; ; in and are two combined morphological operators with excellent performance. represents the original vibration signal, represents the structural element in the morphological filtering method, represents the dilation operation in the morphological filtering method, Represents the corrosion operation in the morphological filtering method, represents the closing operation in the morphological filtering method, Represents the opening operation in the morphological filtering method.

[0021] S32. A mathematical product morphological operator is constructed using the product operation with theoretical advantages and the two combined morphological operators obtained in S31. The specific definition is as follows: ; in represents the product operation, It is a mathematical product morphological operator constructed based on these two combined morphological operators.

[0022] S4. Based on the improved time-varying structuring elements of different orders constructed in step S2 and the excellent mathematical product morphological operator determined in step S3, mathematical morphological filtering is performed on the vibration acceleration signal collected in step S1 to obtain mathematical morphological filtered signals of different orders, the number of which is equal to the number of improved time-varying structuring elements. (S2 obtains a series of narrow waveforms, i.e., the designed improved time-varying structuring elements, which, when combined in sequence, form a window function of the same length as the vibration signal. The power spectrum of the mathematical morphological filtered signals of different orders obtained in S4 is calculated to calculate the characteristic frequency intensity coefficient.) S5, selecting an optimal mathematical morphology filtering signal from the mathematical morphology filtering signals of different orders obtained in step S4; In the embodiment of the present invention, step S5 specifically includes calculating the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filter signal obtained in step S4, and selecting the mathematical morphology filter signal with the largest characteristic frequency intensity coefficient as the optimal mathematical morphology filter signal.

[0023] In the embodiment of the present invention, in step S5, the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filtered signal is calculated. I CFIC The calculation formula is: , ; in Represents the amplitude of different frequencies in the power spectrum;f m Indicates a certain specified fault characteristic frequency; f x express[( -0.02) f m , ( +0.02) f m ]Different frequencies in the interval, f m Indicates a certain specified fault characteristic frequency (such as inner ring, outer ring, rolling element or cage). By inputting different fault characteristic frequencies, you can get different fault characteristic frequencies. I CFIC ; f k represents the spectrum frequency, ; H is the order of the fault characteristic frequency involved in the calculation in the power spectrum, K is the number of discrete spectral frequencies involved in the power spectrum calculation. f m Used to eliminate the effects of slight speed fluctuations.

[0024] S6. Perform power spectrum analysis on the optimal mathematical morphology filtering signal to determine the bearing fault.

[0025] In the embodiment of the present invention, step S6 includes the following sub-steps: S61. Calculate the fault characteristic frequencies of the inner ring, outer ring, rolling element, and cage of the bearing respectively. The calculation formula is as follows: The calculation formula for the inner race fault characteristic frequency (BPFI) is:

[0026] The calculation formula for the outer race fault characteristic frequency (BPFO) is:

[0027] The calculation formula for rolling element fault characteristic frequency (BSF) is:

[0028] The calculation formula for cage fault characteristic frequency (FTF) is:

[0029] in Z is the number of rolling elements, f r is the bearing inner ring speed, d is the rolling element diameter, D is the pitch diameter of the rolling bearing, is the contact angle; S62, calculate the power spectrum of each characteristic frequency in turn according to the power spectrum of the optimal mathematical morphology filtering signal and the different fault characteristic frequencies of step S61 I CFIC , the largest of which I CFIC The corresponding fault is a bearing fault.

[0030] Next, two specific rolling bearing fault diagnosis examples are used to further illustrate the present invention.

[0031] Example 1: Rolling element fault diagnosis of rolling bearings.

[0032] To validate the bearing fault detection performance of the ITVMMF method, a test analysis was first conducted using bearing rolling element fault data. Furthermore, the adaptive time-varying mathematical morphological filtering and diagonal slice spectrum method (ATVMF-DSS), enhanced scale morphological hat product filtering (ESMHPF), and multiscale morphological filtering (MMF) were used for comparison to highlight the advantages of the proposed method.

[0033] ATVMF-DSS references are: CHEN BY, SONG DL, ZHANG WH, et al. Aperformance enhanced time-varying morphological filtering method for bearingfault diagnosis[J]. Measurement, 2021, 176: 109163.1-109163.19. ESMHPF references are: YAN XA, LIU Y, JIA M P. Research on an enhancedscale morphological-hat product filtering in incipient fault detection of rolling element bearings[J]. Measurement, 2019, 147: 106856.1-106856.14. MMF references are: ZHANG LJ, XU JW, YANG JH, et al. Multiscalemorphology analysis and its application to fault diagnosis[J]. MechanicalSystems and Signal Processing, 2008, 22(3): 597–610. Figure 3 The figure below shows the bearing rolling element fault signal and its envelope spectrum. The sampling frequency of the bearing vibration signal is 12.8 kHz, and the analyzed signal length is 8192 sampling points. The periodic pulse characteristics in the time domain signal are submerged in the interference components; due to the interference of the fault-irrelevant frequency, Figure 3 Only 2 can be identified in the envelope spectrum shown in (b) f b spectral lines. Figure 4 The results obtained by analyzing the bearing test signal using the proposed ITVMMF. Figure 4 As shown in (a), compared with the original signal, the repetitive transient pulses in the filtered signal are effectively enhanced. Figure 4 In (b), we can identify f b and its first three harmonic frequencies, indicating that the ITVMMF method accurately detects the rolling element fault of the bearing. Figures 5-7 The rolling element fault detection results of ATVMF-DSS, ESMHPF and MMF methods are shown respectively. The three comparison methods have achieved relatively close filtering results: a large amount of interference noise remains in the time domain signal, and the fault related information cannot be observed; only the fault related information can be identified in its corresponding spectrum. f b or 2 f b These results show that the fault detection performance of the three comparison methods is not as good as that of the ITVMMF method.

[0034] Example 2: Rolling bearing outer ring fault diagnosis Furthermore, the bearing outer race fault data was used to verify the bearing fault diagnosis performance of the proposed method. The pitch diameter, rolling element diameter, contact angle, and number of rolling elements of the test bearing were 176.29 mm, 24.74 mm, 8.83°, and 20, respectively. The sampling frequency and sampling time were 12.8 kHz and 0.64 s, respectively. The fault characteristic frequency of the bearing outer race was calculated to be f o About 66.75 Hz.

[0035] Figure 8(a) and (b) are the vibration signal and envelope spectrum of the bearing outer race fault, respectively. In the time domain signal, the repetitive pulse characteristics caused by the bearing outer race fault are severely contaminated by background noise and are difficult to observe directly. Figure 8 In the envelope spectrum shown in (b), the characteristic frequency of the outer race fault f o The ITVMMF method is used to process the bearing outer ring fault signal, and the results are as follows: Figure 9 As shown in Figure 2. In the time domain signal, the transient pulse characteristics are effectively enhanced and the interference noise is suppressed; Figure 9 The fault characteristic frequency can be clearly observed in the power spectrum shown in (b) f o The results of ATVMF-DSS, ESMHPF and MMF methods for analyzing the fault signal of the bearing outer ring are as follows: Figures 10-12 shown. Figure 10 (b) can only detect f o The spectrum of the fault features shows that the fault feature extraction effect of ATVMF-DSS is poor. ESMHPF has achieved filtering results close to those of the ITVMMF method, and the interference noise in the time domain signal has been effectively filtered out. Figure 11 (b) The diagonal slice spectrum shows that f o and its harmonic 2 f o , 3 f o , and 4 f o However, there are more interference frequencies and the amplitude of the fault-related frequency is lower than that of ITVMMF, indicating that the feature extraction effect of ESMHPF is not as good as that of ITVMMF. The MMF method shows a certain feature extraction effect, which can be observed in its spectrum. f o and its harmonic 2 f o spectral lines, but there are more interference frequencies, such as Figure 12 (b) These results further confirm the advantages of the ITVMMF method in detecting bearing faults.

[0036] The above embodiments are merely preferred technical solutions of the present invention and should not be construed as limiting the present invention. The scope of protection of the present invention shall be the technical solutions set forth in the claims, including equivalent alternatives to the technical features of the technical solutions set forth in the claims. In other words, equivalent alternatives and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A bearing fault diagnosis method based on improved time-varying mathematical morphology filtering, characterized in that The following steps are involved: S1. Collect the vibration acceleration signal of the bearing; S2. constructing improved time-varying structural elements of different orders according to the vibration acceleration signal collected in step S1; S3. Determine a mathematical product morphological operator with excellent performance; S4. Performing mathematical morphological filtering on the vibration acceleration signal based on the improved time-varying structuring elements of different orders constructed in step S2 and the mathematical product morphological operator with excellent performance determined in step S3 to obtain mathematical morphological filtered signals of different orders; S5, selecting the optimal mathematical morphology filtering signal from the mathematical morphology filtering signals of different orders obtained in step S4; S6. Perform power spectrum analysis on the optimal mathematical morphology filtering signal to determine the bearing fault.

2. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 1 is characterized in that: The sub-steps of step S2 are: S21. Calculate the original vibration signal through k Median filtered signal after order median filtering , the median filter formula is: ； ； in, represents the median operator, Represents data points n Nearby datasets, k is the order of the median filter, k The initial value is 2 ， express k remainder when divided by 2; S22, identifying median filter signal All local minimum points of ; S23, select the median filter signal The narrow waveform between two adjacent minimum points in the is used as the shape of the structural element; S24, adjusting each narrow waveform so that its minimum amplitude is zero; S25, the adjusted narrow waveform is used as k Improve the time-varying structure element by 10-order to extract the pulse features hidden in the corresponding narrow waveform of the vibration signal; S26, in turn k Assign values ​​to 2, 3, 4, 5, 6, 7, 8, 9, 10 , Repeat steps S21 to S25 to obtain improved time-varying structural elements of different orders.

3. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 1 is characterized in that: The sub-steps of step S3 are: S31. Select two combined morphological operators with excellent feature extraction and noise reduction performance, and they have similar filtering performance, that is, the same type of combined morphological operators, which are specifically defined as follows: ； ； in and are two combined morphological operators with excellent performance. represents the original vibration signal, represents the structural element in the morphological filtering method, represents the dilation operation in the morphological filtering method, Represents the corrosion operation in the morphological filtering method, represents the closing operation in the morphological filtering method, Represents the opening operation in the morphological filtering method; S32. A mathematical product morphological operator is constructed using the product operation with theoretical advantages and the two combined morphological operators obtained in S31. The specific definition is as follows: ； in represents the product operation, It is a mathematical product morphological operator constructed based on these two combined morphological operators.

4. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 1 is characterized in that: Step S5 specifically includes calculating the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filter signal obtained in step S4, and selecting the mathematical morphology filter signal with the largest characteristic frequency intensity coefficient as the optimal mathematical morphology filter signal.

5. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 4 is characterized in that: In step S5, the characteristic frequency intensity coefficient of the power spectrum of each mathematical morphology filtered signal is calculated. I CFIC The calculation formula is: ， ； in Represents the amplitude of different frequencies in the power spectrum; f m Indicates a certain specified fault characteristic frequency; f x express[( -0.02) f m , ( +0.02) f m ]Different frequencies in the interval; f k represents the spectrum frequency, ; H is the order of the fault characteristic frequency involved in the calculation in the power spectrum, K is the number of discrete spectral frequencies involved in the calculation of the power spectrum.

6. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 1 is characterized in that: The sub-steps of step S6 are: S61. Calculate the fault characteristic frequencies of the inner ring, outer ring, rolling element, and cage of the bearing respectively. The calculation formula is as follows: The calculation formula for the inner race fault characteristic frequency (BPFI) is: ； The calculation formula for the outer race fault characteristic frequency (BPFO) is: ； The calculation formula for rolling element fault characteristic frequency (BSF) is: ； The calculation formula for cage fault characteristic frequency (FTF) is: ； in Z is the number of rolling elements, f r is the shaft speed, d is the rolling element diameter, D is the pitch diameter of the rolling bearing, is the contact angle; S62, calculate the power spectrum of each characteristic frequency in turn according to the power spectrum of the optimal mathematical morphology filtering signal and the different fault characteristic frequencies of step S61 I CFIC , the largest of which I CFIC The corresponding fault is a bearing fault.

7. The bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to claim 1 is characterized in that: The contents of step S1 include: Select an acceleration sensor that is suitable for the working environment and measurement requirements, install the sensor on the bearing housing close to the bearing, and install multiple acceleration sensors in different directions to obtain comprehensive information.

8. A bearing fault diagnosis system based on improved time-varying mathematical morphology filtering, characterized by: A bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to any one of claims 1 to 7 is adopted.

9. A computer device, characterized in that: include: one or more processors; The processor is configured to store one or more programs; When the one or more programs are executed by the one or more processors, a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering as described in any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, a bearing fault diagnosis method based on improved time-varying mathematical morphology filtering according to any one of claims 1 to 7 is implemented.

Images