A bearing fault diagnosis method, device, equipment and storage medium

Abstract

The present invention discloses a bearing fault diagnosis method, device, equipment and storage medium. First, a bearing vibration signal is subjected to variational modal decomposition to obtain a first preset number of eigenmodes; then, a second preset number of target modes are screened out from the eigenmodes according to the phase angle of each eigenmode; wherein the second preset number is less than the first preset number; then, a reconstructed signal is constructed according to the target mode; finally, an envelope spectrum analysis is performed on the reconstructed signal to obtain the bearing fault type. The embodiment of the present invention selects a number of target modes based on the phase angle of the eigenmode as an indicator, then constructs a reconstructed signal based on the target mode, and then performs bearing fault diagnosis based on the reconstructed signal, thereby realizing early fault feature extraction and fault type diagnosis of the bearing.

Description

Technical Field

[0001] The present invention relates to the field of bearing fault diagnosis, and in particular to a bearing fault diagnosis method, device, equipment and storage medium. Background Art

[0002] Bearings are important components in many mechanical equipment. When bearings are damaged or fail, they often cause mechanical equipment failure or shutdown. In order to ensure the long-term reliable operation of mechanical equipment, it is necessary to perform early fault detection on bearings to assess their fatigue life so that they can be maintained and replaced in time. However, in the early failure stage of bearings, the fault sign signal is relatively weak and easily buried in the noise, making it difficult to extract and diagnose early fault features. Summary of the invention

[0003] The purpose of the embodiments of the present invention is to provide a bearing fault diagnosis method, device, equipment and storage medium, which are used to solve the problems of difficulty in extracting early bearing fault features and difficulty in fault diagnosis.

[0004] To achieve the above object, an embodiment of the present invention provides a bearing fault diagnosis method, comprising:

[0005] Performing variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes;

[0006] Filtering a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number;

[0007] constructing a reconstructed signal according to the target modality;

[0008] The reconstructed signal is subjected to envelope spectrum analysis to obtain the bearing fault type.

[0009] As an improvement of the above solution, the step of constructing a reconstructed signal according to the target modality includes:

[0010] Acquire the signal components of each target mode;

[0011] The signal components are time-aligned and averaged to obtain the reconstructed signal.

[0012] As an improvement of the above solution, the step of obtaining the signal components of each target mode specifically includes:

[0013] A pre-constructed sliding window is used to intercept data from each of the target modes to obtain signal components of each of the target modes.

[0014] As an improvement of the above solution, the sliding window is constructed in the following manner:

[0015] Obtaining the maximum value and index of each target mode;

[0016] Selecting a maximum value from the maximum values ​​of each target mode and using its index as the center of the sliding window;

[0017] A sliding window is constructed according to the center of the sliding window and a preset sliding window size.

[0018] As an improvement of the above solution, the step of selecting a second preset number of target modes from the eigenmodes according to the phase angle of each eigenmode specifically includes:

[0019] Mapping the phase angle of each eigenmode to a preset range, and obtaining the initial phase angle of each eigenmode;

[0020] The initial phase angles with the same sign and the same initial change trend are grouped together to obtain four groups of initial phase angles to be grouped;

[0021] Selecting a plurality of groups of candidate initial phase angles from each group of initial phase angles to be grouped; wherein each group of candidate initial phase angles includes a second preset amount of initial phase angles;

[0022] Calculate the standard deviation of each set of candidate initial phase angles, and select a set of candidate initial phase angles with the smallest standard deviation as the target initial phase angle;

[0023] The eigenmode corresponding to each initial phase angle in the target initial phase angle is used as the target mode.

[0024] As an improvement of the above solution, the bearing vibration signal is a horizontal vibration acceleration signal of the bearing.

[0025] As an improvement of the above solution, the bearing vibration signal is a pre-warning signal.

[0026] To achieve the above object, an embodiment of the present invention further provides a bearing fault diagnosis device, comprising:

[0027] An eigenmode acquisition module, used for performing variational mode decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes;

[0028] A target mode screening module, used for screening a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number;

[0029] A signal reconstruction module, used for constructing a reconstructed signal according to the target modality;

[0030] The fault diagnosis module is used to perform frequency domain analysis on the reconstructed signal to obtain an envelope spectrum, and determine the fault type of the bearing according to the envelope spectrum.

[0031] To achieve the above objectives, an embodiment of the present invention further provides a bearing fault diagnosis device, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the bearing fault diagnosis method as described in any of the above embodiments is implemented.

[0032] To achieve the above objectives, an embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the bearing fault diagnosis method described in any of the above embodiments.

[0033] Compared with the prior art, the bearing fault diagnosis method, device, equipment and storage medium provided in the embodiments of the present invention firstly perform variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes; then, according to the phase angle of each of the eigenmodes, a second preset number of target modes are screened out from the eigenmodes; wherein the second preset number is less than the first preset number; then, a reconstructed signal is constructed according to the target mode; finally, an envelope spectrum analysis is performed on the reconstructed signal to obtain the bearing fault type. The embodiments of the present invention can realize early fault feature extraction and fault type diagnosis of the bearing by screening out several target modes based on the phase angle of the eigenmode as an indicator, then constructing a reconstructed signal based on the target mode, and then performing bearing fault diagnosis based on the reconstructed signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a flow chart of a bearing fault diagnosis method provided by one embodiment of the present invention;

[0035] Figure 2 It is a schematic diagram of the experimental results of the test set Bear i ng1_1-70;

[0036] Figure 3 It is a schematic diagram of the experimental results of the test set Bearing1_2-64;

[0037] Figure 4 It is a schematic diagram of the experimental results of the test set Bearing2_1-460;

[0038] Figure 5 It is a schematic diagram of the experimental results of the test set Bear i ng2_2-75;

[0039] Figure 6It is a schematic diagram of the experimental results of the test set Bearing3_1-2450;

[0040] Figure 7 It is a schematic diagram of the experimental results of the test set Bearing3_2-1000;

[0041] Figure 8 It is a structural schematic diagram of a bearing fault diagnosis device provided by one embodiment of the present invention;

[0042] Fig. 9 It is a structural schematic diagram of a bearing fault diagnosis device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] See also Figure 1 , is a flow chart of a bearing fault diagnosis method provided by an embodiment of the present invention, the bearing fault diagnosis method comprising steps S1 to S4:

[0045] S1. performing variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes;

[0046] S2. Filtering out a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number;

[0047] S3, constructing a reconstructed signal according to the target modality;

[0048] S4. Perform envelope spectrum analysis on the reconstructed signal to obtain the fault type of the bearing.

[0049] Exemplarily, before step S1, it also includes setting appropriate variational mode decomposition parameters, such as the decomposition level K, the regularization parameter λ, and the bandwidth limit a, etc. Further, due to the gravity effect in the vertical direction, in order to reduce the error caused by gravity, the horizontal vibration acceleration signal of the bearing can be selected as the bearing vibration signal.

[0050] It is understandable that variational mode decomposition (VMD) modulates the original signal within a certain number of frequency bands, and then separates the modulated components within each frequency band to obtain a series of modal functions, and these modal functions have different frequency and amplitude characteristics. By decomposing the signal into multiple modal components, variational mode decomposition can retain components with a higher signal-to-noise ratio and filter out components with a lower signal-to-noise ratio, thereby improving signal quality. In step S1, after performing variational mode decomposition on the bearing vibration signal, a first preset number of intrinsic modes, that is, a first preset number of intrinsic mode functions (I ntri nsic Mode Function, I MF) will be obtained.

[0051] Furthermore, in step S2, several target modes are selected using the phase angle of the eigenmode as an indicator; and in step S3, a reconstructed signal is constructed according to the target mode. By selecting the target mode and constructing the reconstructed signal, early fault features can be extracted from the original signal for early bearing fault diagnosis, and by selecting a limited number of components (target modes) from the eigenmode, unnecessary components can be removed, noise interference can be reduced, the accuracy of the baseline signal can be improved, and the calculation efficiency can be optimized.

[0052] Furthermore, in step S4, envelope spectrum analysis is performed on the reconstructed signal to implement bearing fault diagnosis, thereby effectively improving the accuracy of bearing fault diagnosis.

[0053] Compared with the prior art, the bearing fault diagnosis method provided in the embodiment of the present invention can accurately extract the early fault characteristics of the bearing by reconstructing the different modes decomposed by variational mode using the phase angle as an indicator to obtain a reconstructed signal; further, bearing fault diagnosis is performed based on the reconstructed signal, so that the bearing fault type can be diagnosed timely and accurately in the early fault stage of the bearing.

[0054] As an optional embodiment, the step of selecting a second preset number of target modes from the eigenmodes according to the phase angle of each eigenmode specifically includes:

[0055] Mapping the phase angle of each eigenmode to a preset range, and obtaining the initial phase angle of each eigenmode;

[0056] The initial phase angles with the same sign and the same initial change trend are grouped together to obtain four groups of initial phase angles to be grouped;

[0057] Selecting a plurality of groups of candidate initial phase angles from each group of initial phase angles to be grouped; wherein each group of candidate initial phase angles includes a second preset amount of initial phase angles;

[0058] Calculate the standard deviation of each set of candidate initial phase angles, and select a set of candidate initial phase angles with the smallest standard deviation as the target initial phase angle;

[0059] The eigenmode corresponding to each initial phase angle in the target initial phase angle is used as the target mode.

[0060] It is understandable that before the eigenmodes are screened according to the phase angle, the phase angles of each eigenmode need to be mapped to a uniform range (preset range). Exemplarily, the preset range can be [-180°, 180°]. Furthermore, after the phase angles of each eigenmode are mapped to the preset range, the initial phase angles of each eigenmode need to be obtained to screen out the target mode according to each initial phase angle. Specifically, a second preset number of initial phase angles with consistent initial change trends, consistent positive and negative values ​​and the closest values ​​are screened out from each initial phase angle as the target initial phase angle, that is, a group of indicators (initial change trends, signs and sizes) that are closest are selected from all combinations of initial phase angles as the target initial phase angle. Exemplarily, the initial phase angles can be divided into four groups according to the signs and initial change trends, namely, four groups of initial phase angles to be grouped, namely, "positive sign, rising trend", "positive sign, falling trend", "negative sign, rising trend" and "negative sign, falling trend". Then, the initial phase angles in each group of initial phase angles to be grouped have the same signs and initial change trends; then, all possible combinations are selected from each group of initial phase angles to be grouped according to a second preset number to form several groups of alternative initial phase angles, wherein the second preset number is less than the first preset number. For example, when the first preset number is 11, the second preset number can be 5; finally, since the initial phase angles in each group of alternative initial phase angles already have the same sign and initial change trend, it is only necessary to select the group with the closest value from each group of alternative initial phase angles to obtain the target initial phase angle. Exemplarily, the standard deviation of each group of alternative initial phase angles can be calculated and the group with the smallest standard deviation can be selected. This is because the smaller the standard deviation, the smaller the difference in the initial phase angles. It is worth noting that the above description of obtaining the target initial phase angle is only an example, and the target initial phase angle can also be obtained by other methods. Further, the eigenmode corresponding to each initial phase angle in the target initial phase angle is the target mode.

[0061] As an optional implementation manner, constructing a reconstructed signal according to the target modality includes:

[0062] Acquire the signal components of each target mode;

[0063] The signal components are time-aligned and averaged to obtain the reconstructed signal.

[0064] It is worth noting that there are often a lot of noises in the process of signal acquisition and transmission, which reduce the quality and reliability of the signal. The embodiment of the present invention can effectively reduce the interference of noise and improve the ratio of signal to noise through signal reconstruction, making the signal clearer and more accurate, thereby more accurately extracting the early fault characteristics of the bearing.

[0065] As one of the optional embodiments, the acquiring the signal components of each target modality specifically includes:

[0066] A pre-constructed sliding window is used to intercept data from each of the target modes to obtain signal components of each of the target modes.

[0067] It is worth noting that the embodiment of the present invention intercepts data through a sliding window, which can extract local features of the signal and retain timing information, which is conducive to more accurate extraction of early fault characteristics of the bearing; further, by intercepting signal components for signal reconstruction, the amount of data processed can be reduced, the computational complexity can be reduced, and the efficiency of bearing fault diagnosis can be improved.

[0068] As an optional embodiment, the sliding window is constructed in the following manner:

[0069] Obtaining the maximum value and index of each target mode;

[0070] Selecting a maximum value from the maximum values ​​of each target mode and using its index as the center of the sliding window;

[0071] A sliding window is constructed according to the center of the sliding window and a preset sliding window size.

[0072] Specifically, the maximum value and index of each target mode are calculated, and the maximum values ​​are compared and the largest maximum value is selected from them, and the index of the largest maximum value is used as the center of the sliding window. Further, the window size can be set to X% of the bearing vibration signal. For example, X can be 20. Then, if the length of the bearing vibration signal is 32768, the size of the sliding window is 6554. Further, a sliding window is used to intercept data packets: based on the index position (center position), 10% of the data is intercepted to the left and right. If the left data length is insufficient, all sliding windows are translated to the right by the missing length unit. On the contrary, if the right data length is insufficient, all sliding windows are translated to the left by the missing length unit.

[0073] The embodiment of the present invention uses the position corresponding to the largest maximum value as the center of the sliding window, and can intercept the signal with the most obvious characteristics. Therefore, even in the early stage of bearing failure, early fault characteristics can be extracted from the vibration signal with only weak fluctuations, so as to achieve more timely and accurate bearing fault diagnosis.

[0074] It is understandable that after the signal components of each target mode are intercepted by a sliding window, it is also necessary to align and average the signal components in time, that is, to perform time-domain synchronous averaging on each signal component to obtain a reconstructed signal. Time-domain synchronous averaging can reduce the impact of noise on the signal. By averaging the signals collected repeatedly, the characteristics of the uniform distribution of random noise in time can be offset, thereby increasing the signal-to-noise ratio. Since random noise will be eliminated or weakened during the averaging process, time-domain synchronous averaging can significantly improve the signal-to-noise ratio of the signal. This is very important for applications that need to extract weaker signals or observe subtle changes. At the same time, time-domain synchronous averaging can enhance the signal components of interest by superimposing and averaging multiple repeated signals. When the recurring signal has stable frequency, phase or amplitude characteristics, the averaged signal will better reflect these characteristics.

[0075] Furthermore, it is necessary to perform envelope spectrum analysis on the reconstructed signal to obtain the bearing fault type. The basic principle of envelope spectrum analysis is to use signal filtering and Hilbert transform technology to remove the high-frequency oscillation components and noise in the original signal, extract the envelope information in the signal, and then calculate the spectrum amplitude of the envelope signal through Fourier transform, so as to obtain the frequency characteristics and change trend of the signal.

[0076] Specifically, first, the reconstructed signal is bandpass filtered to limit the frequency range of the reconstructed signal to a narrow bandwidth, thereby reducing the impact of high-frequency noise and retaining the main frequency components of the reconstructed signal. Then, the filtered signal is Hilbert transformed to obtain the analytical signal, where the real part of the analytical signal is the original signal and the imaginary part is the Hilbert transform pair. The analytical signal is then modulo-calculated to obtain the envelope of the reconstructed signal. The formulas used are as follows:

[0077]

[0078] Among them, x(t) is the reconstructed signal, * is the convolution, is the Hilbert transform of the reconstructed signal, z(t) is the analytical signal,

[0079] It is worth mentioning that the envelope spectrum can not only reflect the energy distribution and frequency characteristics of the signal, but also has better resistance to low-frequency vibration interference, higher signal-to-noise ratio, and can better highlight the fault characteristic frequency compared with the amplitude spectrum and power spectrum. Therefore, the embodiment of the present invention further improves the accuracy of bearing fault diagnosis by performing envelope spectrum analysis on the reconstructed signal.

[0080] Furthermore, the obtained envelope spectrum is compared with the theoretical fault characteristic frequency of the bearing to obtain the fault type of the bearing. Taking rolling bearings as an example, the bearing fault types include outer ring fault, inner ring fault, rolling element fault and cage fault, and the corresponding fault characteristic frequencies are as shown in equations (3) to (6):

[0081]

[0082]

[0083] Among them, f1 is the characteristic frequency of outer ring fault, f2 is the characteristic frequency of inner ring fault, f3 is the characteristic frequency of rolling element fault, f4 is the characteristic frequency of cage fault, d is the diameter of rolling element, D is the pitch diameter of bearing, α is the contact angle of rolling element bearing, z is the number of rolling elements, f r It is the rotation frequency of the inner ring of the bearing.

[0084] As an optional embodiment, the bearing vibration signal is a horizontal vibration acceleration signal of the bearing.

[0085] As one of the optional embodiments, the bearing vibration signal is a pre-warning signal.

[0086] For example, the warning time can be set according to the guiding numerical range and suggestions provided in GBT 6075.1-2012 Mechanical vibration - Measurement and evaluation of vibration of machines on non-rotating parts - Part 1 - General in the national standard (machine condition monitoring and diagnosis). That is, the vibration signal before the warning time can be used to perform bearing fault diagnosis and analysis, thereby achieving early fault diagnosis.

[0087] It is worth noting that the early warning can only detect the existence of a bearing fault, but cannot determine the type of fault. The bearing fault diagnosis method provided by the embodiment of the present invention can accurately identify the type of fault. Moreover, it can be understood that the bearing fault diagnosis method provided by the present invention can not only be used for the analysis of the signal before the early warning, but also for the analysis of the signal after the early warning to diagnose the type of bearing fault.

[0088] Compared with the prior art, the bearing fault diagnosis method provided by the embodiment of the present invention first performs variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes; then, based on the phase angle of each of the eigenmodes, a second preset number of target modes are screened out from the eigenmodes; wherein the second preset number is less than the first preset number; then, a reconstructed signal is constructed based on the target mode; finally, an envelope spectrum analysis is performed on the reconstructed signal to obtain the bearing fault type. The embodiment of the present invention selects a number of target modes based on the phase angle of the eigenmode as an indicator, then constructs a reconstructed signal based on the target mode, and then performs bearing fault diagnosis based on the reconstructed signal, thereby realizing early fault feature extraction and fault type diagnosis of the bearing.

[0089] In order to further illustrate the beneficial effect of the embodiment of the present invention in accurately identifying the type of bearing fault, an experiment is described below.

[0090] This experiment uses the XJTU-SY rolling bearing accelerated life test dataset. The XJTU-SY (Xi'an Jiaotong University-Changxing Sumyoung Technology Co., Ltd.) rolling bearing accelerated life test dataset (hereinafter referred to as the XJTU-SY dataset) is a dataset released by Professor Lei Yaguo's team from the School of Mechanical Engineering of Xi'an Jiaotong University and Zhejiang Changxing Shengyang Technology Co., Ltd. By selecting rolling bearings, which are typical key components in industrial scenarios, as test objects and conducting accelerated life tests on rolling bearings for two years, the acquired test data are finally integrated into the XJTU-SY dataset. The XJTU-SY dataset contains data from 15 rolling bearings from operation to failure. The sampling frequency of the XJTU-SY dataset is 25.6kHz, the sampling interval is 1min, and the sampling time for each sampling is 1.28s. The data acquired in each sampling is saved as a CSV file, in which the first column is the horizontal vibration acceleration signal and the second column is the vertical vibration acceleration signal. See Table 1, which lists the detailed information of each test bearing in the XJTU-SY dataset, where the samples are the full life cycle vibration signals of the rolling bearing.

[0091] Table 1 XJTU-SY dataset information list

[0092]

[0093]

[0094] Specifically, referring to Table 2, this experiment selects two data sets in each working condition shown in Table 1, and selects the pre-warning samples of each data set, and only extracts the horizontal vibration acceleration signal of each data set. Afterwards, the selected data sets are divided into training sets and test sets.

[0095] Table 2 Training set and test set

[0096]

[0097] This experiment first uses the training set to conduct experiments to determine the parameters of the variational mode, including the number of decomposition layers K, regularization parameter λ, and bandwidth limit a. Specifically, when it is observed that there are too many useless or redundant modes in the decomposition results, you can try to increase a to enhance sparsity, so as to obtain a clearer and more accurate mode decomposition; if the decomposed mode is not rich enough or there is information loss, you can try to reduce a to reduce the constraint strength, increase the number of IMFs and the degree of information retention, so as to better capture the details or small fluctuations of the signal and facilitate the observation of the fault characteristic frequency; if the signal structure is complex, has multiple details or is under-decomposed, K can be increased to provide a more detailed and comprehensive mode description; if the signal is relatively simple or contains only a small number of characteristic modes or there is over-decomposition, K can be reduced to reduce redundant modal functions and simplify the model.

[0098] After setting the VMD parameters, the bearing vibration signals in the test set are obtained for variational mode decomposition. Specifically, this experiment decomposes each bearing vibration signal into 11 eigenmodes; then, the phase angle of each eigenmode is mapped to [-180°, 180°], the phase diagram of each eigenmode is plotted and the initial phase angle is obtained; then, the 5 closest ones are selected from the initial phase angles as the target initial phase angles, and the corresponding eigenmodes are used as the target modes; further, the maximum value of each target mode is calculated, and the maximum value of each target mode is obtained from the initial phase angles.

[0099] The largest maximum value is selected and its index is used as the center of the sliding window to construct a sliding window; the sliding window is used to intercept each target mode to obtain the signal component; the signal component is synchronously averaged in the time domain to obtain the reconstructed signal; the reconstructed signal is analyzed by envelope spectrum; the above steps are repeated until the bearing fault diagnosis of 6 test sets is completed, that is, the bearing fault diagnosis of Bearing 1_1-70, Bearing 1_2-64, Bearing 2_1-460, Bearing 2_2-75, Bearing 3_1-2450 and Bearing 3_2-1000.

[0100] See Table 3, which shows the fault characteristic frequencies under various working conditions of the XJTU-SY dataset. Figure 2 to Figure 7They are the fault diagnosis results of this experiment on the test sets Bearing1_1-70, Bearing1_2-64, Bearing2_1-460, Bearing2_2-75, Bearing3_1-2450 and Bearing3_2-1000, respectively. Figure 2 to Figure 7 ,Apart from Figure 7 The second vertical line represents the rotation frequency, and the remaining vertical lines are used to indicate the theoretical fault characteristic frequency. When the peak of the envelope spectrum of the reconstructed signal appears near the vertical line, it is determined that this experiment can accurately identify the bearing fault type.

[0101] Specifically, see Figure 2 , the corresponding outer race fault characteristic frequency is about 107.91 Hz, the peak near 13.49 Hz corresponds to the cage, the peak near 35 Hz corresponds to the rotation frequency, and the peak after 200 Hz has no corresponding fault frequency and can be ignored. Therefore, Bearing1_1-70 is judged to be an outer race fault. The diagnosis result is the same as the actual fault type of the XJTU-SY data set in Table 1.

[0102] See also Figure 3 , the corresponding outer race fault characteristic frequency is about 107.91 Hz, and the peak near 35 Hz corresponds to the rotation frequency. Therefore, Bearing1_2-64 is judged to be an outer race fault. The diagnosis result is the same as the actual fault type of the XJTU-SY data set in Table 1.

[0103] See also Figure 4 , the corresponding inner race fault characteristic frequency is about 184.38 Hz, and the peak near 75 Hz is the rotation frequency. Therefore, Bearing2_1-460 is judged to be an outer race fault. The diagnosis result is the same as the actual fault type of the XJTU-SY data set in Table 1.

[0104] See also Figure 5 , the corresponding outer race fault characteristic frequency is about 115.62 Hz, so Bearing2_2-75 is judged to be an inner race fault, and the diagnosis result is the same as the actual fault type of the XJTU-SY data set in Table 1.

[0105] See also Figure 6 , the corresponding fault characteristic frequency is about 123.32 Hz, so Bearing3_1-2450 is judged to be an outer race fault, and the diagnosis result is the same as the actual fault type of the XJTU-SY data set in Table 1.

[0106] See also Figure 7, the corresponding characteristic frequency of inner ring fault is about 196.68Hz, the characteristic frequency of rolling element fault is about 82.66Hz, the characteristic frequency of cage fault is about 15.42Hz, and the characteristic frequency of outer ring fault is about 123.32Hz. The second vertical line corresponds to a rotation frequency of about 40Hz. Therefore, Bearing3_2-1000 is judged to have inner ring, rolling element, cage and outer ring faults. The diagnostic results are the same as the actual fault types of the XJTU-SY data set in Table 1.

[0107] Table 3 Fault characteristic frequencies under various working conditions of the XJTU-SY dataset

[0108]

[0109] It can be seen from this experiment that the bearing fault diagnosis method provided by the embodiment of the present invention can extract early fault characteristics from pre-warning signals and accurately detect the bearing fault type, and has strong practicality.

[0110] See also Figure 8 The embodiment of the present invention further provides a bearing fault diagnosis device 10, comprising:

[0111] An eigenmode acquisition module 11 is used to perform variational mode decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes;

[0112] A target mode screening module 12 is used to screen out a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number;

[0113] A signal reconstruction module 13, used to construct a reconstructed signal according to the target modality;

[0114] The fault diagnosis module 14 is used to perform frequency domain analysis on the reconstructed signal to obtain an envelope spectrum, and determine the fault type of the bearing according to the envelope spectrum.

[0115] It is worth noting that the bearing fault diagnosis device provided in the embodiment of the present invention can implement all the process steps of the bearing fault diagnosis method described in the above embodiment. The functions of each module and unit in the device and the technical effects achieved are respectively the same as the functions and technical effects achieved by the bearing fault diagnosis method described in the above embodiment. The specific implementation method will not be repeated here.

[0116] See also Fig. 9 The embodiment of the present invention further provides a bearing fault diagnosis device 20, comprising a processor 21, a memory 22, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the above-mentioned bearing fault diagnosis method embodiment are implemented, for example Figure 1 or, the processor implements the functions of each module in the above-mentioned device embodiments when executing the computer program.

[0117] An embodiment of the present invention further provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the bearing fault diagnosis method as described in any of the above embodiments.

[0118] Compared with the prior art, the bearing fault diagnosis device, equipment and storage medium provided in the embodiment of the present invention firstly performs variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes; then, according to the phase angle of each of the eigenmodes, a second preset number of target modes are screened out from the eigenmodes; wherein the second preset number is less than the first preset number; then, a reconstructed signal is constructed according to the target mode; finally, an envelope spectrum analysis is performed on the reconstructed signal to obtain the bearing fault type. The embodiment of the present invention selects a number of target modes based on the phase angle of the eigenmode as an indicator, then constructs a reconstructed signal based on the target mode, and then performs bearing fault diagnosis based on the reconstructed signal, thereby realizing early fault feature extraction and fault type diagnosis of the bearing.

[0119] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A bearing fault diagnosis method, characterized in that: include: Performing variational modal decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes; Filtering a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number; constructing a reconstructed signal according to the target modality; Performing envelope spectrum analysis on the reconstructed signal to obtain the fault type of the bearing; The step of selecting a second preset number of target modes from the eigenmodes according to the phase angle of each eigenmode specifically includes: Mapping the phase angle of each eigenmode to a preset range, and obtaining the initial phase angle of each eigenmode; The initial phase angles with the same sign and the same initial change trend are grouped together to obtain four groups of initial phase angles to be grouped; Selecting a plurality of groups of candidate initial phase angles from each group of initial phase angles to be grouped; wherein each group of candidate initial phase angles includes a second preset amount of initial phase angles; Calculate the standard deviation of each set of candidate initial phase angles, and select a set of candidate initial phase angles with the smallest standard deviation as the target initial phase angle; The eigenmode corresponding to each initial phase angle in the target initial phase angle is used as the target mode.

2. The bearing fault diagnosis method according to claim 1, characterized in that: The step of constructing a reconstructed signal according to the target modality comprises: Acquire the signal components of each target mode; The signal components are time-aligned and averaged to obtain the reconstructed signal.

3. The bearing fault diagnosis method according to claim 2, characterized in that: The acquiring of the signal components of each target mode specifically includes: A pre-constructed sliding window is used to intercept data from each of the target modes to obtain signal components of each of the target modes.

4. The bearing fault diagnosis method according to claim 3, characterized in that: The sliding window is constructed in the following way: Obtaining the maximum value and index of each target mode; Selecting a maximum value from the maximum values ​​of each target mode and using its index as the center of the sliding window; A sliding window is constructed according to the center of the sliding window and a preset sliding window size.

5. The bearing fault diagnosis method according to claim 1, characterized in that: The bearing vibration signal is a horizontal vibration acceleration signal of the bearing.

6. The bearing fault diagnosis method according to claim 1, characterized in that: The bearing vibration signal is a pre-warning signal.

7. A bearing fault diagnosis device, characterized in that: include: An eigenmode acquisition module, used for performing variational mode decomposition on the bearing vibration signal to obtain a first preset number of eigenmodes; A target mode screening module, used for screening a second preset number of target modes from the eigenmodes according to the phase angle of each of the eigenmodes; wherein the second preset number is less than the first preset number; A signal reconstruction module, used for constructing a reconstructed signal according to the target modality; A fault diagnosis module, used for performing frequency domain analysis on the reconstructed signal to obtain an envelope spectrum, and determining the fault type of the bearing according to the envelope spectrum; The step of selecting a second preset number of target modes from the eigenmodes according to the phase angle of each eigenmode specifically includes: Mapping the phase angle of each eigenmode to a preset range, and obtaining the initial phase angle of each eigenmode; The initial phase angles with the same sign and the same initial change trend are grouped together to obtain four groups of initial phase angles to be grouped; Selecting a plurality of groups of candidate initial phase angles from each group of initial phase angles to be grouped; wherein each group of candidate initial phase angles includes a second preset amount of initial phase angles; Calculate the standard deviation of each set of candidate initial phase angles, and select a set of candidate initial phase angles with the smallest standard deviation as the target initial phase angle; The eigenmode corresponding to each initial phase angle in the target initial phase angle is used as the target mode.

8. A bearing fault diagnosis device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the bearing fault diagnosis method according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the bearing fault diagnosis method according to any one of claims 1 to 6.

Images