A calculation method for bearing rotation frequency, a device, and a fault determination method

Abstract

An embodiment of the present invention provides a method for calculating the rotational frequency of a bearing, a device, and a method for fault determination. The method for calculating the rotational frequency of the bearing includes: obtaining a vibration signal of the bearing; filtering the vibration signal to obtain a first signal; obtaining a resonance frequency band of the vibration signal, and determining an envelope signal according to the resonance frequency band; fusing the first signal and the envelope signal to obtain a fused signal; and substituting the fused signal into a multi-component collaborative cost function to obtain the rotational frequency of the bearing. This method accurately determines the rotational frequency of the bearing without a tachometer, overcomes the deficiencies of the prior art, and realizes the accurate calculation of the rotational frequency of the bearing.

Description

Technical Field

[0001] The present invention belongs to the field of signal processing, and particularly relates to a method for calculating the rotational frequency of a bearing, a device, and a method for fault determination. Background Art

[0002] Rolling bearings are one of the most widely used components in rotating machinery. The complex working environment and long-term alternating loads can easily cause faults in the bearings. If maintenance measures are not taken in time, it is very easy to trigger major safety accidents. Real-time monitoring of the operating state of the bearings is a necessary link to ensure the normal operation of high-end mechanical equipment, and it is also an important means to improve the operating safety, stability, and reliability of mechanical equipment. Although the fault diagnosis technology of rolling bearings has made great progress and development in theoretical innovation and practical application, most of the current rolling bearing monitoring and diagnosis technologies and methods focus on the constant speed working conditions. In actual industrial applications, variable speed working conditions are inevitable. Even rolling bearings operating at a constant speed will have speed fluctuations due to reasons such as load changes.

[0003] Under variable speed working conditions, the impacts caused by rolling bearing faults will no longer appear in an equal-period form. This change makes the fault characteristic frequency lose its original physical meaning, resulting in the failure of the classic rolling vibration fault diagnosis algorithms represented by the resonance demodulation technology, which are premised on a stable speed. In recent years, order analysis has played an important role in the field of rolling bearing fault diagnosis under variable speed working conditions. By sampling the signal with a fixed angular increment instead of a fixed time increment, the time-varying vibration signal is transformed into the angular domain and fault diagnosis is carried out in the angular domain through the fault order. Implementing order analysis in the angular domain requires the support of a tachometer in terms of hardware and requires calculation of differences in terms of calculation. However, a tachometer is not always available on the equipment to be analyzed, and the accuracy of order analysis in the angular domain is limited by the interpolation calculation. Therefore, it is particularly important to develop a method for diagnosing rolling bearing faults at variable speeds without a tachometer. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is a method for calculating the rotational frequency of a bearing, a device, and a method for fault determination. The method for calculating the rotational frequency of the bearing comprehensively considers the amplitudes and continuities of the time-frequency ridge line distributions of the rotational frequency and the fault characteristic frequencies of each bearing, and realizes the accurate calculation of the rotational frequency of the bearing. Compared with other rotational speed estimation technologies, it can accurately determine the rotational frequency of the bearing without a tachometer, overcomes the deficiencies of the prior art, and has a good noise reduction effect.

[0005] To achieve the above object, an embodiment of the present invention provides a method for calculating the rotational frequency of a bearing, including: obtaining the vibration signal of the bearing; filtering the vibration signal to obtain a first signal; obtaining the resonance frequency band of the vibration signal, and determining the envelope signal according to the resonance frequency band; fusing the first signal and the envelope signal to obtain a fused signal; substituting the fused signal into a multi-component collaborative cost function to obtain the rotational frequency of the bearing.

[0006] Optionally, the vibration signal of the bearing is obtained by a vibration acceleration sensor.

[0007] Optionally, the obtaining the resonance frequency band of the vibration signal and determining the envelope signal according to the resonance frequency band includes obtaining the fast Hoyer spectrogram of the vibration signal, determining the resonance frequency band according to the Hoyer index of the spectrogram; performing band-pass filtering on the resonance frequency band to obtain a filtered signal, and determining the envelope signal of the filtered signal.

[0008] Optionally, the Hoyer index of the Hoyer spectrogram is the Hoyer index, which is used to select the optimal filtering frequency band.

[0009]

[0010]

[0011] Wherein, l2 / l1norm is the norm, N is the signal length, n is 1, 2... N, and x n is the vibration signal of the nth data point collected.

[0012] Optionally, the performing band-pass filtering on the resonance frequency band to obtain a filtered signal and determining the envelope signal of the filtered signal includes: the envelope signal is |z(t)|.

[0013]

[0014]

[0015] Wherein, x(t) is the real part value of the filtered signal. is the imaginary part value of the filtered signal, and j is the imaginary unit.

[0016] Optionally, the first signal and the envelope signal are added with a fixed ratio threshold to obtain the fused signal; the time-frequency distribution of the fused signal includes a rotational frequency ridge component and a bearing fault feature ridge component.

[0017] Optionally, the multi-component collaborative cost function CF k includes:

[0018]

[0019] Among them, TF(t,f) is the amplitude corresponding to the frequency f at time t in the fusion signal, k is 2, 3... m, m is the number of columns of the matrix TF(t,f), f r (c) is the rotational frequency corresponding to the minimum value of the cost function, f r (i) is all candidate frequencies for searching f r (c), α is the energy ratio of the signal after band-pass filtering and the signal after low-pass filtering, e k is the weight coefficient corresponding to the amplitude of the ridge matrix element, TF(t k ,f k (i)) is the amplitude corresponding to the frequency f k at time t k (i), TF(t k ,f qk (i)) is the amplitude corresponding to each theoretical fault order ridge of the frequency f k at time t qk (i).

[0020] Optionally, the weight coefficient e k of the frequency band FB k is:

[0021] FB k =[f k-1 (c)-f w ,f k-1 (c)+f w ,

[0022]

[0023] Among them, f k (c) is the frequency corresponding to the minimum value of the cost function CF k , f k-1 (c) is the frequency corresponding to the minimum value of the cost function among the previous frequency points of f k (c), f w is half of the range of the frequency band FB k .

[0024] Optionally, the multi-component collaborative cost function is minimized to determine the frequency points within the frequency band threshold range.

[0025] Correspondingly, an embodiment of the present invention further provides a method for determining bearing faults, which is characterized by including the frequency points determined according to any one of the above-mentioned bearing rotation frequency calculation methods, and connecting not less than two of the frequency points to form a rotation frequency ridge line, where the rotation frequency ridge line is used to determine the change law of the bearing rotation frequency; determining the theoretical fault feature ridge line of the bearing according to the multi-component collaborative cost function, including: theoretical fault feature ridge line = rotation frequency ridge line × theoretical value of fault feature order; the rotation frequency ridge line and the theoretical fault feature ridge line are obtained through VNCMD to get the actual extraction result, and the fault feature ridge line with the smallest variance is determined as the actual fault feature ridge line; determining the actual value of the fault feature order according to the actual fault feature ridge line; comparing the theoretical value of the fault feature order and the actual value of the fault feature order to determine the fault type of the bearing.

[0026] Optionally, the fault types of the bearing at least include outer ring fault, inner ring fault, rolling element fault and cage fault.

[0027] Optionally, the theoretical value of the fault feature order is:

[0028]

[0029] where k O is the theoretical value of the fault feature order of the outer ring of the bearing, k I is the theoretical value of the fault feature order of the inner ring, k B is the theoretical value of the fault feature order of the cage, k C is the theoretical value of the fault feature order of the rolling element, z is the number of bearing balls, f r is the rotation frequency of the bearing, d is the diameter of the bearing ball, D is the pitch diameter of the bearing, is the contact angle, f O is the outer ring fault frequency, f I is the inner ring fault frequency, f B is the rolling element fault frequency, f C is the cage fault frequency.

[0030] Correspondingly, an embodiment of the present invention further provides a calculation device for bearing rotation frequency, which is characterized by including: an information acquisition module for acquiring the vibration signal of the bearing; a processing module for processing the vibration signal to obtain the rotation frequency of the bearing, including: filtering the vibration signal to obtain a first signal; and determining the envelope signal through the resonance frequency band of the vibration signal; fusing the first signal and the envelope signal to obtain a fused signal; and substituting the fused signal into the multi-component collaborative cost function to obtain the rotation frequency of the bearing.

[0031] Optionally, determining the envelope signal according to the resonance frequency band of the vibration signal includes obtaining the fast Hoyer spectrogram of the vibration signal, determining the resonance frequency band according to the Hoyer index of the spectrogram; performing band-pass filtering on the resonance frequency band to obtain a filtered signal, and determining the envelope signal of the filtered signal.

[0032] Optionally, the multi-component collaborative cost function CF k includes:

[0033]

[0034] where TF(t,f) is the amplitude corresponding to the frequency f at time t in the fusion signal, k is 2, 3... m, m is the number of columns of the matrix TF(t,f), f r (c) is the rotational frequency corresponding to the minimum value of the cost function, f r (i) is all candidate frequencies for searching f r (c), α is the energy ratio of the signal after band-pass filtering and the signal after low-pass filtering, e k is the weight coefficient corresponding to the amplitude of the ridge matrix element, TF(t k ,f k (i)) is the amplitude corresponding to the frequency f k at time t k (i), TF(t k ,f qk (i)) is the amplitude corresponding to each theoretical fault order ridge of the frequency f k at time t qk (i).

[0035] Optionally, the weight coefficient e k of the frequency band FB k is:

[0036] FB k =[f k-1 (c)-f w ,f k-1 (c)+f w ,

[0037]

[0038] where f k (c) is the frequency corresponding to the minimum value of the cost function CF k , f k-1 (c) is the frequency corresponding to the minimum value of the cost function among the previous frequency points of f k (c), f w is half of the range of the frequency band FB k .

[0039] Optionally, minimize the multi-component collaborative cost function to determine the frequency points within the frequency band threshold range.

[0040] Through the above technical solution, the present invention fuses the signals after low-pass filtering and band-pass filtering to highlight the rotation frequency information and fault feature information in the original signal. Based on the local cost function of multi-component collaboration, comprehensively consider the continuity and amplitude of the ridges of the rotation frequency and fault feature frequency in the time-frequency distribution, and obtain the rotation frequency of the bearing from the time-frequency distribution of the fused signal. Compared with other rotational speed estimation techniques, it can accurately determine the bearing rotation frequency without a tachometer, overcomes the deficiencies of the prior art, and has a good noise reduction effect.

[0041] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent specific implementation part. Brief Description of the Drawings

[0042] The drawings are used to provide a further understanding of the embodiments of the present invention, and constitute a part of the specification. Together with the following specific implementation, they are used to explain the embodiments of the present invention, but do not constitute a limitation to the embodiments of the present invention. In the drawings:

[0043] Figure 1 is a schematic flowchart of the calculation method of the bearing rotation frequency of the present invention;

[0044] Figure 2 is the time-domain waveform of the vibration signal of the bearing in an embodiment of the present invention;

[0045] Figure 3 is the envelope spectrum diagram of the vibration signal of the bearing in an embodiment of the present invention;

[0046] Figure 4 is the fast Hoyer spectrum diagram corresponding to the vibration signal of the bearing in an embodiment of the present invention;

[0047] Figure 5 is the time-frequency distribution of the fused signal in an embodiment of the present invention;

[0048] Figure 6 is the result comparison of the actual rotation frequency calculated according to the key phase signal and the multi-component collaborative estimated rotation frequency of the present invention. Detailed Description of the Invention

[0049] The following details the specific implementation of the embodiments of the present invention in conjunction with the drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiments of the present invention, and is not used to limit the embodiments of the present invention.

[0050] Figure 1 is a schematic flowchart of the calculation method of the bearing rotation frequency of the present invention.

[0051] Step S101 is to obtain the vibration signal of the bearing. Rolling bearings are one of the most widely used components in rotating machinery. The complex working environment and long-term alternating loads can easily cause faults in the bearings. If maintenance measures are not taken in time, it is very easy to trigger major safety accidents. Real-time monitoring of the bearing operating state is a necessary link to ensure the normal operation of high-end mechanical equipment, and it is also an important means to improve the operating safety, stability, and reliability of mechanical equipment. In actual industrial applications, variable-speed operating conditions are inevitable. Even rolling bearings operating at a constant speed will have speed fluctuations due to reasons such as load changes. According to a preferred implementation manner, the present invention obtains the vibration signal of the bearing through a vibration acceleration sensor. The vibration velocity sensor is mainly installed on the bearing covers of various rotating machinery devices (such as steam turbines, compressors, fans, and pumps, etc.). It is an electromagnetic sensor that outputs voltage by the moving coil cutting magnetic lines of force. Therefore, it has the characteristics of not requiring a power supply during operation and being easy to install.

[0052] Step S102 is to filter the vibration signal to obtain a first signal. Preferably, a low-pass filter is used to filter the signal. A low-pass filter is an electronic filtering device that allows signals below the cut-off frequency to pass through, but signals above the cut-off frequency cannot pass through. For different filters, the strength of the signal at each frequency is different. In this application, an elliptic filter is preferably used.

[0053] Step S103 is to obtain the resonance frequency band of the vibration signal. The fast Hoyer spectrogram is used to determine the resonance frequency band where the bearing fault impact is located. According to a preferred implementation manner, the fast Hoyer spectrogram of the vibration signal is obtained (this fast Hoyer spectrogram can be obtained in the art). The resonance frequency band is determined according to the Hoyer index of this spectrogram. The fast Hoyer spectrogram replaces the kurtosis index in the fast kurtosis diagram with the Hoyer index. The Hoyer index of the Hoyer spectrogram is Hoyerindex, which is used to select the optimal filtering frequency band.

[0054]

[0055]

[0056] Among them, l2 / l1norm is the norm, N is the signal length, n is 1, 2…N, and x n is the vibration signal of the nth data point collected. The Hoyer index can be used to indicate the richness of the bearing fault feature information contained in different filtering frequency bands, so as to facilitate the selection of the optimal filtering frequency band.

[0057] Step S104 is to determine the envelope signal according to the resonance frequency band. First, perform band-pass filtering on the resonance frequency band to obtain a filtered signal, and then determine the envelope signal of the filtered signal. The band-pass filtering is a filtering method that filters out high- and low-frequency signals and retains intermediate-frequency signals during the process of processing the signal. The envelope signal is |z(t)|,

[0058]

[0059]

[0060] where x(t) is the real part value of the filtered signal, is the imaginary part value of the filtered signal, and j is the imaginary unit. Among them Analytical signal is the result after Hilbert transform. After obtaining the analytical form that can be formed is: Taking the modulus of the analytical signal can obtain the envelope signal: The envelope signal in the signal refers to a high-frequency amplitude-modulated signal whose amplitude changes according to the low-frequency modulation signal. If the peak points of the high-frequency amplitude-modulated signal are connected, a curve corresponding to the low-frequency modulation signal can be obtained, and this curve is the envelope line.

[0061] Step S105 is to fuse the first signal and the envelope signal to obtain a fused signal. According to a preferred implementation manner, the first signal and the envelope signal are added with a fixed ratio threshold to obtain the fused signal; the time-frequency distribution of the fused signal includes a rotation frequency ridge component and a bearing fault feature ridge component. The signal fusion method is to add the low-pass filtered signal and the band-pass filtered signal with a fixed ratio threshold. The fixed ratio threshold can be defined artificially, and it is necessary to ensure that the time-frequency distribution of the fused signal contains a rotation frequency ridge component and a bearing fault feature ridge component.

[0062] Step S106 is to substitute the fused signal into the multi-component collaborative cost function to obtain the rotation frequency of the bearing. The multi-component collaborative cost function CF k includes:

[0063]

[0064] where TF(t,f) is the amplitude corresponding to the frequency f at time t in the fused signal, k is 2, 3…m, m is the number of columns of the matrix TF(t,f), f r (c) is the rotation frequency corresponding to when the cost function reaches the minimum value, f r (i) is to search for f rAll candidate frequencies at (c), α is the energy ratio of the signal after band-pass filtering and the signal after low-pass filtering, and is a proportionality coefficient used to ensure the continuity of the rotational frequency in the multi-component collaborative cost function and the dominant position of the amplitude size of the rotational frequency. For the signal x(t), its energy calculation formula is: e k is the weight coefficient corresponding to the amplitude of the ridge matrix element, TF(t k ,f k (i)) is the amplitude corresponding to the frequency f k at time t k (i), and TF(t k ,f qk (i)) is the amplitude corresponding to each theoretical fault order ridge of the frequency f k at time t qk (i). The weight coefficient e k of the frequency band FB k is:

[0065] FB k =[f k-1 (c)-f w ,f k-1 (c)+f w ,

[0066]

[0067] The weight coefficient e k of the frequency band FB k also changes with time. The frequency band FB k is preferably 5Hz. Among them, f k (c) is the frequency corresponding to the minimum value of the cost function CF k , f k-1 (c) is the frequency corresponding to the minimum value of the cost function among the previous frequency points of f k (c), and f w is half of the range of the frequency band FB k . This method limits the ridge extraction within a small frequency band range and selects the frequency that makes the value in this local frequency region reach the local maximum as the candidate.

[0068] Minimize the multi-component collaborative cost function to determine the frequency points within the frequency band threshold range.

[0069] By dynamically minimizing the multi-component collaborative cost function, other data points on the desired ridge line are gradually extracted, and the line connecting these points is the variation curve of the selected target frequency. According to a specific implementation manner, the multi-component collaborative cost function is minimized to determine the optimal ridge line constituent frequency points within a certain frequency band at a certain moment. Taking the rotational frequency as an example, the operation is cycled. First, the initial rotational frequency is determined, that is, at time 0. Starting from this point, for example, within the range of plus or minus 5 Hz, the point that can minimize the value of CF k is searched for, and then the search for the next moment is carried out until the end of the signal time. The rotational frequency ridge line is formed by connecting no less than two of the said frequency points, and the plus or minus 5 Hz therein is the said frequency band threshold range.

[0070] The embodiment of the present invention also provides a method for judging bearing faults, including the frequency points determined according to any one of the above-mentioned bearing rotational frequency calculation methods, and connecting no less than two of the said frequency points to form a rotational frequency ridge line, which is used to determine the variation law of the bearing rotational frequency; determining the theoretical fault characteristic ridge line of the bearing according to the multi-component collaborative cost function, including: theoretical fault characteristic ridge line = rotational frequency ridge line × theoretical value of the fault characteristic order; the rotational frequency ridge line and the theoretical fault characteristic ridge line are subjected to VNCMD to obtain the actual extraction result, and the fault characteristic ridge line with the smallest variance is determined as the actual fault characteristic ridge line. The VNCMD is a new non-linear frequency modulation modal decomposition method, which is an extended method of variational mode decomposition (VMD) and is mainly applied in matlab; determining the actual value of the fault characteristic order according to the actual fault characteristic ridge line; comparing the theoretical value of the fault characteristic order and the actual value of the fault characteristic order to judge the fault type of the bearing. The theoretical value of the fault characteristic order is:

[0071]

[0072] where k O is the theoretical value of the fault characteristic order of the outer ring of the bearing, k I is the theoretical value of the fault characteristic order of the inner ring, k B is the theoretical value of the fault characteristic order of the cage, k C is the theoretical value of the fault characteristic order of the rolling element, z is the number of bearing balls, f r is the rotational frequency of the bearing, d is the diameter of the bearing ball, D is the pitch diameter of the bearing, is the contact angle, f O is the outer ring fault frequency, f I is the inner ring fault frequency, f B is the rolling element fault frequency, f CThe cage fault frequency. The contact angle, i.e., the bearing contact angle, is one of the characteristic parameters of the bearing, representing the angle between the rolling element load vector and the bearing radial plane at the midpoint of the rolling element and raceway contact area. The fault characteristic order is a theoretical value. Each type of fault has its corresponding value, which exists regardless of whether the fault occurs and is only related to the bearing model.

[0073] By comparing the theoretical value and the actual value of the fault characteristic order, the fault type of the bearing is determined. For example, when the rotating shaft has a single fault, if the theoretical inner ring fault order is 3.5 and the experimentally obtained fault order is 3.45 or 3.55, it can be considered an inner ring fault. The criterion for determining the fault type is whether the absolute value of the difference between the experimental fault order and the theoretical fault order is within the fault A threshold range. If it is within the fault A threshold range, the rotating shaft can be determined to be of fault type A. The fault types of the bearing at least include outer ring fault, inner ring fault, rolling element fault, and cage fault.

[0074] The rotating frequency and the ridge line of the bearing fault characteristic frequency are extracted by using the calculation method of the bearing rotating frequency, and the extraction results are used as the input parameters of the VNCMD method to decompose the signal by VNCMD, so as to extract the rotating frequency and the bearing vibration components. By analyzing the fault characteristic order to determine the rolling bearing fault type, it effectively avoids the inaccurate ridge line extraction caused by the single-peak detection method only focusing on the amplitude of the time-frequency distribution and not considering the continuity characteristics of the ridge line, and finally the situation where the bearing fault type cannot be judged, improving the accuracy of VNCMD analysis.

[0075] The specific implementation of an embodiment of a group of experimental signals with noise in the present invention includes:

[0076] Step 1: Figure 2 It is the time-domain waveform of the vibration signal of the bearing in an implementation manner of the present invention. As Figure 2 shown, the vibration signal of the variable-speed rolling bearing is obtained from the test bench through an acceleration sensor. The sampling frequency is 20480 Hz, the signal length is 6 s, and its envelope spectrum is as Figure 3 shown.

[0077] Step 2: Filter the signal by using a low-pass filter, and obtain the vibration signal of the bearing through a vibration acceleration sensor.

[0078] Step 3: Use the fast Hoyer spectrogram to determine the resonance frequency band where the bearing fault impact is located, perform band-pass filtering on the signal, and obtain the envelope of the filtered signal;

[0079] Step 3.1: Filter the signal by using the fast Hoyer spectrogram;

[0080] The Hoyer index is The normalized form of the norm, and the calculation method of the Hoyer index value for each frequency band is as shown in Equations 1 and 2

[0081]

[0082]

[0083] Step 3.2: Obtain the envelope signal of the filtered signal, and the process is as shown in Equation 3:

[0084]

[0085] Where: x(t) is the original signal, the analytic signal is the result after Hilbert transform, and after obtaining , the constructed analytic form can be:

[0086]

[0087] Taking the modulus of the analytic signal can obtain the envelope signal:

[0088]

[0089] The fast Hoyer spectrogram of the signal is as Figure 4 shown. The filtered frequency band with the largest Hoyer index is the resonance frequency band where the fault characteristic frequency is located. Its filtering center frequency is 8320 Hz, and the filtering bandwidth is 1280 Hz. After filtering, taking the envelope of it, the time-frequency distribution of the fused signal is as Figure 5 shown, Figure 5 There are multiple ridge lines related to the rotation frequency in it, but their true meanings cannot be determined. Only by determining the rotation frequency ridge line can the meanings represented by other ridge lines be determined.

[0090] Step 5: Calculate the characteristic order of each fault type according to the bearing signal, and the specific calculation method is as shown in Equation 6:

[0091]

[0092] Where z represents the number of bearing balls, f r represents the rotation frequency of the bearing, d is the diameter of the bearing ball, D is the pitch diameter of the bearing, is the contact angle, f O , f I , f B and f C respectively represent the outer race fault frequency, inner race fault frequency, rolling element fault frequency and cage fault frequency. The ratios of the four fault characteristic frequencies to the rotation frequency are all constants, and this constant is called the fault order. Therefore, the fault characteristic orders of the four common faults are respectively:

[0093]

[0094] In this experiment, the fault characteristic orders of the outer ring, inner ring, cage, and rolling elements of the bearing are 3.572, 5.428, 0.3969, and 2.32 respectively.

[0095] Step 6: Construct the multi-component collaborative cost function:

[0096]

[0097] where m is the number of columns of the matrix TF(t,f), e k is the weight corresponding to the amplitude of the ridge matrix element, f r is the rotation frequency, f r (i) are all candidate frequencies when searching for f r (c), f r (c) is the rotation frequency corresponding to the minimum value of the cost function, α is the proportionality coefficient, which is used to ensure the continuity of the rotation frequency in Equation 8 and the dominant position of the amplitude size, and can be determined by the energy ratio of the signal after band-pass filtering and the signal after low-pass filtering. For the signal x(t), its energy is calculated as shown in Equation 11. This method limits the ridge extraction within a small frequency band range, and selects the frequency that makes the value of TF(t,f) reach the local maximum in this local frequency region as the candidate. The frequency band FB k The weight coefficient e k also changes with time:

[0098] FB k = [f k-1 (c) - f w , f k-1 (c) + f w Equation 9

[0099]

[0100]

[0101] where f w represents half of the analyzed frequency band range. Given a starting point (t1,f1) and an appropriate f w value, by dynamically minimizing the local cost function in Equation 8, other data points on the expected ridge are gradually extracted, and the line connecting these points is the change curve of the selected target frequency. The finally obtained change law of the rotation frequency is as Figure 6 shown, Figure 6 In it, the solid line is the change law of the rotation frequency with time obtained by the proposed algorithm, for reference. The dashed line is the change law of the rotation frequency calculated by the key phase signal. By comparison, it can be found that the two are quite close. Therefore, it can be considered that the rotation frequency information extracted by the proposed method is accurate and effective. Minimize CFk The function can gradually extract the data points of the ridge line in the time-frequency distribution diagram to obtain a relatively rough rotation frequency ridge line and four bearing fault characteristic ridge lines. However, in fact, the bearing only contains one fault, and the other three ridge lines should not exist or be disordered. Taking these five ridge lines as the input of VNCMD can further extract fine ridge lines. By taking the ratio of the four bearing fault ridge lines to the rotation frequency ridge line respectively, four groups of fault characteristic orders can be obtained. The fault characteristic order is a fixed value and does not change with the change of the rotational speed. Therefore, by finding the fault order with the smallest fluctuation and the most stability and seeing which theoretical value of the fault its order value is close to, the fault type of the bearing can be judged. In this experiment, the theoretical values of the fault characteristic orders of the outer ring, inner ring, cage and rolling elements of the bearing are 3.572, 5.428, 0.3969 and 2.32 respectively. The actual value of the fault characteristic order of the inner ring of the bearing is 5.2039. By comparing the theoretical value of the fault characteristic order and the actual value of the fault characteristic order respectively, it is determined that the fault type of the bearing is: inner ring fault of the bearing.

[0102] The calculation method, device and fault determination method of the bearing rotation frequency proposed by the present invention first fuse the signals after low-pass filtering and band-pass filtering to highlight the rotation frequency information and fault characteristic information in the original signal. Based on the local cost function of multi-component collaboration, comprehensively considering the continuity and amplitude size of the ridge lines of the rotation frequency and fault characteristic frequencies in the time-frequency distribution, the rotation frequency ridge line is extracted from the time-frequency distribution of the fused signal. Compared with other rotational speed estimation technologies, it can accurately determine the bearing rotation frequency without a tachometer, overcomes the deficiencies of the prior art, improves the accuracy of the rotational speed estimation of rotating machinery, and has a good noise reduction effect, providing an important reference basis for the accurate diagnosis of rolling bearing faults under variable rotational speed conditions.

[0103] For the specific implementation details and effects of a bearing rotation frequency calculation device provided by an embodiment of the present invention, reference may be made to the foregoing embodiments, and details will not be described herein again.

[0104] The optional implementation manners of the embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation manners. Within the technical concept scope of the embodiments of the present invention, various simple variants can be made to the technical solutions of the embodiments of the present invention, and these simple variants all belong to the protection scope of the embodiments of the present invention.

[0105] In addition, it should be noted that, in the various specific technical features described in the above specific implementation manners, they can be combined in any suitable manner without conflict. To avoid unnecessary repetition, the embodiments of the present invention will not separately describe various possible combination manners.

[0106] Those skilled in the art can understand that all or part of the steps in implementing the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program is stored in a storage medium, including several instructions for causing a single-chip microcomputer, a chip or a processor to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0107] In addition, any combination can be made among the various different implementation manners of the embodiments of the present invention, as long as it does not violate the idea of the embodiments of the present invention, and it should also be regarded as the content disclosed by the embodiments of the present invention.

Claims

1. A method for determining bearing faults, characterized in that, Obtain the vibration signal of the bearing; Filter the vibration signal to obtain a first signal; Obtain the resonance frequency band of the vibration signal and determine the envelope signal according to this resonance frequency band; Fuse the first signal and the envelope signal to obtain a fused signal; Substitute the fused signal into the multi-component collaborative cost function to obtain the rotation frequency of the bearing; Minimize the multi-component collaborative cost function to determine the frequency points within the frequency band threshold; Connect no less than two of the frequency points to form a rotation frequency ridge line, and the rotation frequency ridge line is used to determine the variation law of the rotation frequency of the bearing; Determine the theoretical fault feature ridge line of the bearing according to the multi-component collaborative cost function, including: theoretical fault feature ridge line = rotational frequency ridge line The theoretical value of the fault feature order; The rotation frequency ridge line and the theoretical fault feature ridge line pass through VNCMD to obtain the actual extraction result, and determine the fault feature ridge line with the smallest variance as the actual fault feature ridge line; Determine the actual value of the fault feature order according to the actual fault feature ridge line; Compare the theoretical value of the fault feature order and the actual value of the fault feature order to determine the fault type of the bearing; The multi-component collaborative cost function includes: Among them, is the amplitude corresponding to the f frequency at time t in the fusion signal, k is 2, 3... m, where m is the number of columns of the matrix ​ is the rotational frequency corresponding to when the multi-component collaborative cost function reaches the minimum value, for searching all candidate frequencies at that time, is the energy ratio of the band-pass filtered signal to the low-pass filtered signal, is the weight coefficient corresponding to the amplitude of the ridge line matrix element, At the moment the amplitude corresponding to the frequency At the moment the amplitudes corresponding to the ridge lines of each theoretical fault order of the frequency.

2. The method according to claim 1, characterized in that, Obtain the vibration signal of the bearing through a vibration acceleration sensor.

3. The method according to claim 1, characterized in that The obtaining of the resonance frequency band of the vibration signal and determining the envelope signal according to this resonance frequency band includes Obtain the fast Hoyer spectrogram of the vibration signal and determine the resonance frequency band according to the Hoyer index of this spectrogram; Perform band-pass filtering on the resonance frequency band to obtain a filtered signal, and determine the envelope signal of this filtered signal.

4. The method according to claim 3, characterized in that, The Hoyer index of the Hoyer spectrogram is , which is used to select the optimal filtering frequency band. ， ， Among them, is the norm, N is the signal length, n is 1, 2... N, is the vibration signal of the nth data point collected.

5. The method according to claim 3, characterized in that, The performing of band-pass filtering on the resonance frequency band to obtain a filtered signal and determining the envelope signal of this filtered signal includes: The envelope signal is , ， ， Among them, is the real part value of the filtered signal, is the imaginary part value of the filtered signal, and j is the imaginary unit.

6. The method according to claim 1, characterized in that, Sum the first signal and the envelope signal at a fixed ratio threshold to obtain the fused signal; The time-frequency distribution of the fused signal includes a rotation frequency ridge line component and a bearing fault feature ridge line component.

7. The method according to claim 1, characterized in that, Frequency band weight coefficient is as follows: ， ， Among them, is the frequency corresponding to the minimum value of the multi-component collaborative cost function reached, is the frequency corresponding to the minimum value of the multi-component collaborative cost function among the previous frequency points is half of the range of the frequency band .

8. The method according to claim 1, characterized in that, The fault type of the bearing includes at least one of outer ring fault, inner ring fault, rolling element fault and cage fault.

9. The method according to claim 1, characterized in that, The theoretical value of the fault feature order is: Among them, is the theoretical value of the fault characteristic order of the outer ring of the bearing, is the theoretical value of the fault characteristic order of the inner ring. is the theoretical order value of the fault feature of the cage, is the theoretical order value of the fault feature of the rolling element, z is the number of bearing balls, d is the diameter of the bearing ball, and D is the pitch diameter of the bearing. is the contact angle.

Images