Bearing fault sparse feature extraction method based on local feature online learning

Abstract

The invention relates to the technical field of bearing fault detection, in particular to a bearing fault sparse feature extraction method based on local feature online learning, and the method comprises the steps: collecting a bearing vibration original signal of warehouse logistics equipment, and carrying out the normalization processing; performing sliding window segmentation on the processed bearing vibration signal based on a kurtosis index, and selecting a preset number of signal segments with the maximum kurtosis as an initial atom set; carrying out orthogonalization operation on atoms in the initial atom set in sequence to form an online learning dictionary based on local feature learning; carrying out convolution operation on atoms in the online learning dictionary and the normalized bearing vibration signal to obtain a sparse coefficient matrix, and executing soft threshold operation; kurtosis values of sparse coefficient vectors in the sparse coefficient matrix are calculated respectively, the sparse vector with the maximum kurtosis value is selected as a target feature vector for envelope spectrum analysis, and finally whether the bearing of the warehouse logistics equipment breaks down or not is judged through an envelope spectrum.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of bearing fault detection, and in particular to a bearing fault sparse feature extraction method based on local feature online learning. BACKGROUND

[0002] With the development of intelligent warehouse logistics equipment towards digitization, the traditional operation and maintenance mode of "periodic repair" and "fault repair" is gradually upgraded to the mode of "condition repair" and "predictive repair". Bearings, as one of the key components of warehouse equipment, reliable operation is the cornerstone of operational efficiency. However, there are still many challenges in bearing fault diagnosis, mainly including: Weak early fault warning capability: traditional vibration spectrum analysis is sensitive to noise. When the bearing is in the early fault stage, the fault characteristic signal is weak and easy to be overwhelmed by background noise, resulting in a significant decrease in diagnostic accuracy; Subject to artificial experience and rules: simple diagnostic methods (such as amplitude value and kurtosis value) require pre-set decision criteria. Different types of bearings require different threshold values, and they lack adaptability to intermittent faults or new types of faults; Low efficiency of traditional signal analysis: traditional signal processing methods such as signal decomposition and wavelet transform have high computational complexity, which is difficult to meet the real-time monitoring demand, and the feature extraction capability for data with high redundant components is limited; Poor adaptability to multi-working condition scenarios: traditional methods (such as auscultation and temperature monitoring) are greatly disturbed by environmental factors. In complex working conditions with variable loads and multi-energy domain coupling, the fault feature separability is significantly reduced. SUMMARY

[0003] To at least partially overcome the problems of difficulty in extracting fault features, poor adaptability of traditional signal analysis methods, and experience dependence in bearing fault detection, the application provides a bearing fault sparse feature extraction method based on local feature online learning.

[0004] The scheme of the application is as follows: A bearing fault sparse feature extraction method based on local feature online learning, comprising: Collecting bearing vibration original signals of warehouse logistics equipment; Normalizing the bearing vibration original signals; Performing sliding window segmentation on the processed bearing vibration signals based on the kurtosis index, and selecting a preset number of signal segments with the maximum kurtosis as an initial atom set; Orthogonalizing the atoms in the initial atom set in sequence, removing the repeated signal segments from the atoms according to the orthogonalization operation structure, and forming an online learning dictionary based on local feature learning; convolve atoms in the online learning dictionary with the normalized bearing vibration signal to obtain a sparse coefficient matrix, perform soft threshold operation on the sparse coefficient matrix to suppress noise components of the sparse coefficient matrix and enhance effective features of the bearing fault signal; calculate kurtosis values of sparse coefficient vectors in the sparse coefficient matrix respectively, and select a sparse vector with the largest kurtosis value as a target feature vector; perform envelope spectrum analysis on the target feature vector to obtain an envelope spectrum; compare peak frequencies and high-order harmonic frequencies in the envelope spectrum with bearing theoretical fault frequencies under the same working condition, and calculate error values; if the error values are within a preset range, it is judged that the bearing of the warehouse logistics equipment has a fault.

[0005] Preferably, the processed bearing vibration signal is subjected to sliding window segmentation based on the kurtosis index, including: the processed bearing vibration signal is subjected to sliding window segmentation based on a set length; the mean values of the segmented bearing vibration signals are calculated; the kurtosis values of the segmented bearing vibration signals are calculated according to the segmented bearing vibration signals and the mean values thereof.

[0006] Preferably, the atoms in the initial atom set are subjected to orthogonalization operation in sequence, and repeated signal segments in the atoms are removed according to the orthogonalization operation structure to form an online learning dictionary based on local feature learning, including: a dictionary matrix is constructed; a current atom in the initial atom set is taken, the current atom is projected with existing atoms in the dictionary matrix, and a projection residual is calculated; the projection residual is normalized and added to the dictionary matrix as a newly generated atom; after the initial atom set is traversed, the dictionary matrix is taken as an online learning dictionary based on local feature learning.

[0007] Preferably, before the soft threshold operation is performed on the sparse coefficient matrix, the method further includes: the sparse coefficient vectors in the sparse coefficient matrix are normalized.

[0008] Preferably, the kurtosis values of the sparse coefficient vectors in the sparse coefficient matrix are calculated, including: the mean value and the standard deviation of the sparse coefficient vectors in the sparse coefficient matrix are calculated; the kurtosis values of the sparse coefficient vectors in the sparse coefficient matrix are calculated according to the mean value and the standard deviation of the sparse coefficient vectors in the sparse coefficient matrix.

[0009] Preferably, the envelope spectrum analysis comprises: band-pass filtering, centering processing, envelope signal, envelope calculation, feature extraction and demodulation.

[0010] Preferably, the cutoff frequency band in the envelope spectrum is set as [0, Fr / 3] Hz.

[0011] Preferably, the method further comprises: The bearing theoretical fault frequency is calculated through the rotating speed condition of the bearing and the bearing parameters.

[0012] The technical scheme provided in the application can include the following beneficial effects: The signal segment containing the impact feature is screened out through the sliding window segmentation based on the kurtosis index, and the online learning dictionary and the sparse feature extraction technology are combined, so that the noise interference is effectively suppressed, the significance of the early weak fault signal of the bearing is enhanced, and the sensitive early warning of the early fault is realized.

[0013] Through the local feature online learning dictionary, the atomic set can be adaptively updated under different operating conditions, and the sparse feature vector best representing the fault can be extracted without fixed artificial threshold. The dependence on artificial experience is reduced, the adaptability to different types and new types of faults is improved, and the intelligent diagnosis capability is stronger.

[0014] Through the envelope spectrum analysis of the target sparse vector, combined with the comparison of the theoretical fault frequency, high fault separability can be maintained in a complex noise environment, and the monitoring under multi-condition conditions is suitable.

[0015] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the application. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings incorporated in the specification and forming a part thereof illustrate embodiments consistent with the application and, together with the description, serve to explain the principles of the application.

[0017] Figure 1 is a flow diagram of a bearing fault sparse feature extraction method based on local feature online learning provided by an embodiment of the application; Figure 2 is a bearing vibration signal graph after normalization provided by an embodiment of the application; Figure 3 is a target feature vector signal graph provided by an embodiment of the application. DETAILED DESCRIPTION

[0018] The exemplary embodiments will be described in detail herein with reference to the attached drawings. In the following description, like reference numerals refer to like elements unless indicated otherwise. The following exemplary embodiments described herein represent the best known uses consistent with the present application. They are intended to be illustrative only and not restrictive of all the ways in which the application is construed to be in accord with the appended claims.

[0019] Figure 1 is a flowchart of a bearing fault sparse feature extraction method based on local feature online learning provided by an embodiment of the present application, referring to Figure 1 A bearing fault sparse feature extraction method based on local feature online learning includes: S11: Collecting a bearing vibration original signal of a warehouse logistics equipment; Usually obtained by an acceleration sensor arranged at a key position of the equipment, to reflect the dynamic characteristics of the bearing in the running process.

[0020] S12: Normalizing the bearing vibration original signal; In order to eliminate the influence of signal amplitude difference under different working conditions on the subsequent analysis results, the collected bearing vibration original signal is normalized to make it distributed in a unified numerical range, thereby improving the consistency of feature extraction. The bearing vibration signal after normalization is as shown in Figure 2

[0021] S13: Based on the kurtosis index, the processed bearing vibration signal is subjected to sliding window segmentation, and a preset number of signal segments with the maximum kurtosis are selected as an initial atom set; After normalization, the processed bearing vibration signal is subjected to sliding window segmentation based on the kurtosis index. By setting a fixed window length, the signal sequence is segmented piece by piece, and the kurtosis value of each segmented piece is calculated. The kurtosis value can represent the sharpness and pulse characteristics of the signal. Early bearing failure often manifests as transient impact signals, with high kurtosis characteristics. Therefore, a number of signal segments with the maximum kurtosis are selected as the initial atom set, to ensure that the selected signal segments contain rich fault feature information.

[0022] S14: Orthogonalizing the atoms in the initial atom set in turn, removing the repeated signal segments in the atoms according to the orthogonalization operation structure, and forming an online learning dictionary based on local feature learning; ​Subsequently, the atoms in the initial atomic set are sequentially subjected to an orthogonalization operation. Specifically, the current atom is subjected to a projection operation with the selected atoms, redundant components in the signal segment are removed, and the projection residual is normalized and added to the dictionary matrix as a new atom. By traversing all the initial atoms, an online learning dictionary capable of self-adaptively representing the local operation characteristics of the bearing is finally formed. The dictionary has strong adaptability and sparsity and can be dynamically updated under different working conditions.

[0023] S15: performing convolution operation on the atoms in the online learning dictionary and the normalized bearing vibration signal to obtain a sparse coefficient matrix, and performing soft threshold operation on the sparse coefficient matrix to suppress the noise components of the sparse coefficient matrix and enhance the effective features of the bearing fault signal; On the basis of the constructed online learning dictionary, convolution operation is performed on the atoms in the dictionary and the normalized bearing vibration signal to obtain a sparse coefficient matrix. The sparse coefficient matrix reflects the sparse representation of the signal under different atom bases. In order to reduce noise interference, soft threshold operation is performed on the sparse coefficient matrix to suppress low-amplitude noise components and highlight the effective feature signals related to fault impacts.

[0024] S16: calculating the kurtosis value of each sparse coefficient vector in the sparse coefficient matrix, and selecting the sparse vector with the maximum kurtosis value as a target feature vector; The kurtosis value of each sparse coefficient vector in the sparse coefficient matrix is calculated to measure the significance of the impact features in each sparse vector. The sparse vector with the maximum kurtosis value is selected as the target feature vector, ensuring that the subsequent analysis focuses on the component that best represents the bearing fault features.

[0025] The target feature vector obtained at this time is as shown in Figure 3

[0026] S17: performing envelope spectrum analysis on the target feature vector to obtain an envelope spectrum; In the envelope spectrum, the modulation frequency components caused by the fault impacts can be intuitively identified.

[0027] S18: comparing the peak frequency and the high-order harmonic frequency in the envelope spectrum with the bearing theoretical fault frequency under the same working condition to calculate an error value; S19: if the error value is within a preset range, it is judged that the bearing of the warehouse logistics equipment has a fault.

[0028] The technical scheme is aimed at the problem that the traditional vibration spectrum method is easily overwhelmed by noise in the early fault stage and the characteristic signal is not obvious. The kurtosis index is used to screen the signal segment, which can preferentially retain the waveband with impact features. Then, the online learning dictionary is used to extract sparse features, and soft threshold noise suppression is performed on the sparse coefficient matrix, effectively enhancing the observability of weak fault signals.​

[0029] The traditional simple diagnosis method relies on a preset threshold value, and lacks adaptability to new faults and complex working conditions. The local feature online learning dictionary is used to adaptively update the atom set under different operating conditions, without fixed artificial threshold value, so as to extract the sparse feature vector which can best represent the fault. The dependence on artificial experience is reduced, the adaptability to different types and new faults is improved, and the intelligent diagnosis capability is stronger.

[0030] The convolution sparse representation and soft threshold processing are adopted, which has low computational complexity in mathematics, and the online dictionary updating avoids global large-scale iterative operation.

[0031] The envelope spectrum analysis of the target sparse vector is performed, and the theoretical fault frequency is compared, so that the fault separability can be maintained in a complex noise environment, and the monitoring under multiple working conditions is suitable.

[0032] It should be noted that the bearing vibration signal after processing is divided by a sliding window based on the kurtosis index, including: The bearing vibration signal after processing is divided by a sliding window based on the set length; The mean value of each segmented bearing vibration signal is calculated; According to the bearing vibration signal and its mean value, the kurtosis value of each segment of the bearing vibration signal is calculated.

[0033] The bearing vibration signal after normalization is divided by a sliding window according to the set window length. The window length is reasonably set according to the bearing sampling frequency and the rotating speed condition, so as to ensure that each segmented signal can cover the complete vibration period characteristics. The sliding window moves in the whole vibration signal with a certain step, and several signal segments are gradually divided.

[0034] The mean value of each segmented signal is calculated, and the kurtosis value of each segmented signal is calculated by using the relationship between each segmented signal and its mean value. The kurtosis value can reflect the sharpness and pulse characteristics of the signal waveform. When the bearing appears early fault, the transient impact component is often superimposed in the vibration signal, so that the kurtosis value of the signal segment is higher.

[0035] After calculating the kurtosis value of all segmented segments by the above method, several signal segments with the highest kurtosis value can be screened out for constructing the initial atom set. These segments are rich in bearing fault impact components, and can provide high-quality basic data for the generation of online learning dictionary and sparse feature extraction.

[0036] It should be noted that the atoms in the initial atom set are sequentially subjected to an orthogonalization operation, and repeated signal segments in the atoms are removed according to the orthogonalization operation structure to form an online learning dictionary based on local feature learning, including: Constructing a dictionary matrix; Taking a current atom in the initial atom set, projecting the current atom with existing atoms in the dictionary matrix, and calculating a projection residual; Normalizing the projection residual and adding the normalized projection residual as a newly generated atom to the dictionary matrix; After the initial atom set is traversed, the dictionary matrix is used as an online learning dictionary based on local feature learning.

[0037] First, a dictionary matrix is constructed for storing atoms obtained after orthogonalization processing. The dictionary matrix is initially empty.

[0038] Secondly, a current atom in the initial atom set is sequentially taken and projected with existing atoms in the dictionary matrix to obtain a projection component of the atom in the dictionary matrix. Through the projection operation, the similarity and redundancy between the current atom and the existing atoms can be identified.

[0039] The residual between the current atom and the projection component is calculated, and the residual is normalized. The normalization processing can eliminate the influence of amplitude differences between different atoms, so that the residual signal remains consistent and comparable. The obtained residual is added to the dictionary matrix as a newly generated atom.

[0040] After the initial atom set is traversed, the atoms saved in the dictionary matrix constitute an online learning dictionary based on local feature learning. The dictionary is composed of a group of representative, mutually orthogonal or approximately orthogonal atoms, which can effectively reduce the interference of signal redundancy components in the sparse representation process.

[0041] Through the above process, the online learning dictionary can dynamically capture significant feature components in the bearing vibration signal and exclude repeated or invalid information, thereby ensuring the adaptability and sparse representation ability of the dictionary under multiple working conditions. Compared with the traditional preset dictionary method, the online learning dictionary generated by the present application can better reflect the actual running state of the equipment and is suitable for long-term online monitoring scenarios.

[0042] It should be noted that before performing the soft threshold operation on the sparse coefficient matrix, the method further includes: Normalizing the sparse coefficient vectors in the sparse coefficient matrix.

[0043] Before performing the soft threshold operation on the sparse coefficient matrix, the sparse coefficient vectors in the sparse coefficient matrix need to be normalized to make them distributed in a relatively uniform scale range. Through the normalization processing, the deviation of the subsequent soft threshold operation caused by the amplitude difference of different sparse coefficient vectors can be avoided.

[0044] It should be noted that the kurtosis value of the sparse coefficient vector in the sparse coefficient matrix is calculated respectively, including: The mean and standard deviation of the sparse coefficient vector in the sparse coefficient matrix are calculated. According to the mean and standard deviation of the sparse coefficient vector in the sparse coefficient matrix, the kurtosis value of each sparse coefficient vector in the sparse coefficient matrix is calculated.

[0045] In the embodiment, in order to effectively identify the component that can best reflect the bearing fault impact characteristics from the sparse coefficient matrix, the kurtosis value of the sparse coefficient vector therein is calculated. Specifically, first, the statistical characteristic analysis of the sparse coefficient vector is performed to obtain the mean and standard deviation thereof, so as to depict the overall level and dispersion degree of the vector. On this basis, the kurtosis index is further calculated. As a high-order statistical characteristic, the kurtosis can sensitively reflect the significance of the pulse or peak component in the signal. When the bearing is in the early failure stage, the vibration signal thereof often shows short-time impact type fluctuation, and thus the corresponding sparse vector usually presents a larger kurtosis value. In this way, the most representative candidate feature can be distinguished from the multiple sparse coefficient vectors, thereby providing a reliable basis for the subsequent target vector screening and envelope spectrum analysis.

[0046] It should be noted that the envelope spectrum analysis includes band-pass filtering, centering processing, envelope signal, envelope calculation, feature extraction and demodulation.

[0047] In order to more accurately reveal the fault modulation information contained in the sparse feature vector, the envelope spectrum analysis is performed on the target sparse vector. The envelope spectrum analysis preferably includes the steps of band-pass filtering, centering processing, envelope signal construction, envelope calculation, feature extraction and demodulation. Through the band-pass filtering, the interference signals in the irrelevant frequency band can be effectively eliminated; the centering processing can stabilize the waveform reference; and the envelope calculation and demodulation steps can clearly show the modulation components in the signal, so that the envelope spectrum feature representing the impact fault in the frequency domain is obtained.

[0048] In specific practice, the cutoff frequency band in the envelope spectrum is set to [0, Fr / 3] Hz.

[0049] In the processing of the envelope spectrum, the application preferably sets the cutoff frequency band as [0, Fr / 3] Hz, wherein Fr is the rotating frequency of the bearing. The frequency band range covers the characteristic frequency of the common bearing fault and its low-order harmonic components, avoiding high-frequency noise interference, and can highlight the energy distribution of the fault signal, making the envelope spectrum characteristics more targeted and separable.

[0050] It should be noted that the method further comprises: The theoretical fault frequency of the bearing is calculated based on the rotating speed condition of the bearing and the bearing parameters.

[0051] The fault characteristic frequency of different types of bearings can be obtained by theoretical calculation. The theoretical fault frequency of the bearing has strong correlation with the rotating speed condition and the bearing parameters, and is usually obtained based on theoretical calculation.

[0052] In specific practice, the following calculator is used: Processor: 13th Gen Intel(R) Core(TM) i7-13700HX 2.10 GHz; Machine RAM: 16GB; In the calculator configured as above, the average value of the running time of the bearing fault sparse feature extraction method based on local feature online learning is calculated for 20 times, and the average time consumption of each time is 0.085s. It is proved that the algorithm has high efficiency and good feature extraction effect, and meets the needs of online monitoring.

[0053] It can be understood that the same or similar parts in the above embodiments can be mutually referred to, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0054] It should be noted that in the description of the present application, the terms "first", "second" and the like are only used for descriptive purposes and should not be understood as indicating or implying relative importance. In addition, in the description of the present application, unless otherwise specified, the meaning of "a plurality of" is at least two.

[0055] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions or other processes, and the various preferred embodiments of the present application include additional implementations in which the functions described in the illustrated or discussed order are performed in a different order, including substantially simultaneously or in reverse order, and according to the functions involved, and this should be understood by those skilled in the art of the embodiments of the present application.

[0056] It should be understood that various parts of the present application can be implemented in hardware, software, firmware or a combination thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or their combinations, can be used to implement the hardware: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0057] Those skilled in the art can understand that all or part of the steps carried out by the above-mentioned embodiment method can be completed by a program instructing the relevant hardware, and the program can be stored in a computer readable storage medium. When the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0058] In addition, the functional units in each embodiment of the present application can be integrated into one processing module, or each unit can be physically present separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be realized in the form of hardware or in the form of a software function module. The integrated module, if realized in the form of a software function module and sold or used as an independent product, can also be stored in a computer readable storage medium.

[0059] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0060] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example" or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0061] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and cannot be understood as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for extracting sparse features of bearing faults based on online learning of local features, characterized in that, include: Collect raw vibration signals of bearings in warehousing and logistics equipment; The original bearing vibration signal is normalized. The processed bearing vibration signal is segmented by sliding window based on the kurtosis index, and the signal segments with the largest kurtosis are selected as the initial atom set. The atoms in the initial atom set are sequentially orthogonalized, and the repeated signal segments in the atoms are removed according to the orthogonalization operation structure to form an online learning dictionary based on local feature learning; The atoms in the online learning dictionary are convolved with the normalized bearing vibration signal to obtain a sparse coefficient matrix. A soft thresholding operation is then performed on the sparse coefficient matrix to suppress the noise components of the sparse coefficient matrix and enhance the effective features of the bearing fault signal. Calculate the kurtosis value for each sparse coefficient vector in the sparse coefficient matrix, and select the sparse vector with the largest kurtosis value as the target feature vector; Envelope spectrum analysis is performed on the target feature vector to obtain the envelope spectrum; The peak frequency and its higher harmonic frequencies in the envelope spectrum are compared with the theoretical failure frequency of the bearing under the same operating conditions, and the error value is calculated. If the error value is within the preset range, it is determined that the bearing of the warehousing and logistics equipment is faulty.

2. The method according to claim 1, characterized in that, Sliding window segmentation is performed on the processed bearing vibration signal based on the kurtosis index, including: The processed bearing vibration signal is segmented using a sliding window based on a set length. Calculate the mean value of the bearing vibration signal of each segment after segmentation; Based on the vibration signals of each bearing segment and their mean, the kurtosis value of the vibration signals of each bearing segment is calculated.

3. The method according to claim 1, characterized in that, The atoms in the initial atom set are sequentially orthogonalized. Based on the orthogonalization structure, repetitive signal segments are removed from the atoms to form an online learning dictionary based on local feature learning, including: Construct a dictionary matrix; Take the current atom from the initial atom set, project the current atom onto the existing atoms in the dictionary matrix, and calculate the projection residual; The projection residuals are normalized and added to the dictionary matrix as newly generated atoms; After traversing the initial set of atoms, the dictionary matrix is ​​used as an online learning dictionary based on local feature learning.

4. The method according to claim 1, characterized in that, Before performing soft thresholding on the sparse coefficient matrix, the method further includes: The sparse coefficient vector in the sparse coefficient matrix is ​​normalized.

5. The method according to claim 1, characterized in that, Calculate the kurtosis values ​​for each sparse coefficient vector in the sparse coefficient matrix, including: Calculate the mean and standard deviation of the sparse coefficient vector in the sparse coefficient matrix; Calculate the kurtosis value of each sparse coefficient vector in the sparse coefficient matrix based on the mean and standard deviation of the sparse coefficient vectors in the sparse coefficient matrix.

6. The method according to claim 1, characterized in that, The envelope spectrum analysis includes: bandpass filtering, centering, envelope signal, envelope calculation, feature extraction, and demodulation.

7. The method according to claim 1, characterized in that, The cutoff frequency in the envelope spectrum is set to [0, Fr / 3] Hz.

8. The method according to claim 1, characterized in that, The method further includes: The theoretical failure frequency of the bearing is calculated by considering the bearing's operating speed and parameters.

Images