A parameter-free method for estimating gearbox gear meshing frequency

Through the parameterless gear meshing frequency estimation method based on vibration signals, the sliding narrowband window and Fourier transform technology are used to solve the problem of incomplete gear count parameters, and the accurate estimation and fault diagnosis of gear meshing frequency are achieved, which improves the Efficiency and accuracy.

CN114018573BActive Publication Date: 2024-11-08XIAN YINLIAN INFORMATION TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111355367.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-16
Publication Date
2024-11-08
Estimated Expiration
2041-11-16

AI Technical Summary

Technical Problem

In the prior art, the gearbox gear number parameters are incomplete, resulting in inaccurate manual empirical estimation and low efficiency of fault disassembly, and lack of effective method for estimating the meshing frequency of the gear without parameter, which affects the accuracy of fault diagnosis.

Method used

The gear meshing frequency estimation method based on the vibration signal is adopted. The rotation frequency identification is determined by the maximum frequency of the sliding narrowband window energy, combined with the fast Fourier transform and majority voting principle, the gear teeth number and the meshing frequency are estimated, and the notch is designed. Filter out the frequency of influence to ensure the unity and accuracy of estimation standards.

Benefits of technology

It realizes accurate estimation of the gear teeth number and meshing frequency when abnormal gearbox occurs, improves the accuracy and efficiency of fault diagnosis, avoids the uncertainty of manual estimation and inefficiency during disassembly.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114018573B_ABST
    Figure CN114018573B_ABST
Patent Text Reader

Abstract

The present invention discloses a parameter-free gearbox gear meshing frequency estimation method, which collects a vibration signal of the gearbox, wherein the vibration signal includes: a speed signal and an envelope signal; and obtains the rotational frequency of the gear shaft; then, a sliding narrowband window is used in the collected envelope signal to obtain the sliding narrowband window energy maximum frequency Enve_MaxFreq, and it is judged whether the sliding narrowband window energy maximum frequency Enve_MaxFreq meets the rotational frequency identification condition. The present invention is a gear tooth number estimation method based on a vibration signal, which collects the vibration signal of the gearbox in real time through a vibration acceleration sensor installed at the bearing position of each shaft of the gearbox, and estimates the gear tooth number based on the vibration signal. When an abnormality occurs in the gearbox, the gear tooth number can be estimated, and then the gear meshing frequency can be obtained. The estimation method of the present invention is accurate and also ensures the uniformity of the estimation standard.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of mechanical equipment status monitoring, and in particular relates to a parameter-free gearbox gear meshing frequency estimation method. Background Art

[0002] As a gearbox equipment widely used in the industrial field, it has the characteristics of smooth transmission, compact structure and flexibility. For the process industry, key gear transmission equipment or expensive gearbox transmission equipment, when the gear fails, the downtime or production suspension caused by the failure will cause great economic losses. Therefore, it is of great significance for enterprises to master the operating status of the gearbox in real time and predict gear failure in advance for safe and efficient production. Vibration monitoring has been widely accepted as an effective tool for rotating machinery status detection. When a gear fails (such as tooth profile error, wear, pitting and broken teeth), the gearbox vibration signal will change significantly. At this time, the gear meshing frequency is also the fault characteristic frequency. Usually, the gear meshing frequency will increase significantly when the gear fails, and some faults will also modulate the meshing frequency, and obvious side frequency components will appear next to the meshing frequency. Therefore, when diagnosing gear faults, the identification of the gear meshing frequency is particularly important.

[0003] However, for equipment manufacturers, the gearbox tooth number is difficult to obtain because there is no tooth number parameter on the gearbox nameplate and the tooth number of each gear in the gearbox is confidential for most gearbox manufacturers. This brings a lot of trouble to the accuracy of gear fault identification. Therefore, it will be very meaningful for gearbox fault diagnosis and automatic diagnosis of gear faults if a method for estimating the gear tooth number can be found.

[0004] At present, for gearboxes without parameters for the number of teeth, the number of gear teeth can only be collected by manually estimating based on experience or by disassembling the gearbox after a gearbox failure. The different manual estimation methods and experience, and the lack of a unified standard lead to large deviations in the estimation results; and the method of collecting the number of teeth parameters after disassembling the machine is very inefficient. Summary of the invention

[0005] In order to solve the problems in the prior art of incomplete gearbox tooth number parameters, inconsistent estimation standards based on manual experience, and inefficient fault disassembly collection methods, the purpose of the present invention is to provide a parameter-free gearbox gear meshing frequency estimation method, which is a gear tooth number estimation method based on vibration signals, and can estimate the gear tooth number when an abnormality occurs in the gearbox. This method ensures accurate gear tooth number estimation while also ensuring the uniformity of estimation standards.

[0006] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is a parameter-free gearbox gear meshing frequency estimation method, which collects the vibration signal of the gearbox, wherein the vibration signal includes: a speed signal and an envelope signal; and obtains the rotation frequency of the gear shaft; then, a sliding narrowband window is used in the collected envelope signal to obtain the maximum energy frequency EnveMaxFreq of the sliding narrowband window, and it is determined whether the maximum energy frequency EnveMaxFreq of the sliding narrowband window meets the rotation frequency identification condition. If so, the speed signal is searched in mode 1, and if not, the speed signal is searched in mode 2;

[0007] Search method 1, calculate the energy ratio of the search frequency band in the entire frequency band Range_energy_ratio, if the energy ratio Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component; if the energy ratio Range_energy_ratio is greater than or equal to the first threshold threshold1, then the 2 times window energy corresponding to each frequency is calculated by sliding the narrow band window of 2 times frequency in the search frequency band, take the top five frequency components of energy, and divide the top five frequency components of energy by the rotation frequency to the nearest integer, and obtain the number of gear teeth by the majority voting principle;

[0008] Search method 2, filter out the frequency components of the maximum frequency EnveMaxFreq of the sliding narrowband window energy and the 2 times maximum frequency EnveMaxFreq of the sliding narrowband window energy in the speed spectrum signal Vel_sig to obtain the filtered speed signal Filter_vel_sig, calculate the energy ratio Range_energy_ratio of the search frequency band in the entire frequency band, if the energy ratio Range_energy_ratio is less than the first threshold threshold1, if the energy ratio Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component, if the energy ratio Ra nge_energy_ratio is greater than or equal to the first threshold threshold1, then the 2 times window energy corresponding to each frequency is calculated by sliding the narrowband window with 2 times frequency in the search frequency band, and the frequency components whose energy in the narrowband window is greater than or equal to the second threshold threshold2 and whose energy difference between the 1 times narrowband window and the 2 times narrowband window is less than 3 times the third threshold threshold are added to the filtered frequency sequence Filter_freq_list, and the top five frequency components with energy in the 2 times narrowband window in the frequency sequence Filter_freq_list are taken, and the top five frequency components with energy are divided by the rotation frequency and rounded to the nearest integer, and the number of gear teeth is obtained by majority voting principle;

[0009] Finally, the gear meshing frequency is calculated by the obtained gear tooth number.

[0010] Furthermore, the vibration signal of the gear box is collected by a vibration acceleration sensor installed at the bearing position of each shaft of the gear box.

[0011] Further, when the maximum energy frequency EnveMaxFreq of the sliding narrowband window is equal to the rotation frequency or twice the rotation frequency, it is determined that the maximum energy frequency EnveMaxFreq of the sliding narrowband window satisfies the rotation frequency identification condition.

[0012] Furthermore, according to the minimum number of teeth without undercutting in gear machining, the number of spur teeth is 14 and the number of helical teeth is 12, so the search frequency band of the meshing frequency is set to: 12*rotation frequency-1KHz.

[0013] Furthermore, the search method 1 includes the following steps:

[0014] Step 11, performing fast Fourier transform (FFT) on the velocity signal collected in step 1 to obtain a velocity spectrum Vel_f_sig;

[0015] Step 12: Calculate the energy ratio of the search frequency band to the energy ratio of the entire frequency band. The calculation formula is as follows:

[0016]

[0017] Among them, index_12 is the index corresponding to 12*rotation frequency in the speed spectrum frequency series, and N is the length of the speed spectrum;

[0018] Step 13, if the energy ratio Re_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component, that is, if the search band energy is too small, it means that the gear meshing frequency is very small, basically at the same level as the spectrum bottom noise, then it can be considered that the gear meshing is good, and there is no need to search for the meshing frequency in this case; if Re_ratio is greater than the first threshold threshold1, then proceed to step 14, wherein the first threshold threshold1 is obtained through case data statistics;

[0019] Step 14: Calculate the energy in the 2 times narrowband window corresponding to each frequency in the search frequency band with a 2 times frequency sliding narrowband window, and the energy in the 2 times narrowband window of the nth frequency in the search frequency band is Energy_f n for:

[0020]

[0021] Among them, vel_Δf is the spectral resolution of the velocity spectrum, Vel_f_sig(l*f i -m*Δf) is the envelope spectrum corresponding to the frequency l*f i-m*Δf amplitude;

[0022] Energy_f n Add to the search band frequency energy sequence Search_f_e_list;

[0023] Step 15, take the top five frequency components in the search band frequency energy sequence Search_f_elist, divide the top five frequency components by the rotation frequency and round up to the nearest integer, and obtain the gear tooth number Gt_num by majority voting principle.

[0024] Furthermore, the search method 2 includes the following steps:

[0025] Step 21. Since the maximum periodic component Enve_MaxFreq in the envelope spectrum is not the rotation frequency, the frequency may be the characteristic frequency of the bearing fault or a random frequency component. If it is the characteristic frequency of the bearing fault, the frequency may appear in the speed spectrum, which will affect the gear meshing frequency search. Therefore, the frequency must be filtered out before searching, and notch filters are designed to filter out Enve_MaxFreq and 2*Enve_MaxFreq to obtain the filtered signal Filter_vel_sig;

[0026] Step 22: Perform fast Fourier transform (FFT) on the filtered signal Filter_vel_sig to obtain the velocity spectrum Filter_vel_f_sig

[0027] Step 23, calculate the energy ratio of the search frequency band to the energy of the entire frequency band Re_ratio, the calculation formula is as follows:

[0028]

[0029] Among them, index_12 is the index corresponding to 12*rotation frequency in the speed spectrum frequency sequence, and N is the length of the speed spectrum;

[0030] Step 24, if Re_ratio is less than the first threshold value threshold1, it means that there is no effective gear meshing frequency component, that is, if the search frequency band energy is too small, it means that the gear meshing frequency is very small, basically at the same level as the spectrum bottom noise, the gear meshing is good, and there is no need to search for the meshing frequency; if Re_ratio is greater than the first threshold value threshold1, then proceed to step 25; wherein the first threshold value threshold1 is obtained through case data statistics;

[0031] Step 25: Calculate the energy in the 2-fold sliding narrowband window corresponding to each frequency in the search frequency band using a 2-fold sliding narrowband window, and calculate the energy in the 2-fold sliding narrowband window corresponding to each frequency in the search frequency band. n 1 times the narrowband window energy Filter_vel_fn _1 and 2 times the narrowband window energy Filter_vel_f n _2 is:

[0032]

[0033]

[0034] Among them, vel_Δf is the spectral resolution of the velocity spectrum, Filter_vel_f_sig(f n -m*Δf) is the envelope spectrum corresponding to frequency f n -m*Δf amplitude;

[0035] Then the frequency f n The energy in the sliding narrowband window Filter_vel_f is twice n for:

[0036] Filter_vel_f n =Filter_vel_f n _1+Filter_vel_f n _2

[0037] Filter the frequencies within the search band based on the following criteria:

[0038] A. 1 times the narrowband window energy Filter_vel_f n _1 and 2 times the narrowband window energy Filter_vel_f n The frequency band energy of _2 is greater than the threshold threshold2;

[0039] B. 1 times the narrowband window energy Filter_vel_f n _1 and 2 times the narrowband window energy Filter_vel_f n The ratio of the frequency band energy of _2 is less than threshold3;

[0040] The frequency f that satisfies the above conditions A and B n Add the search band frequency energy sequence Search_f_list, frequency f n The energy in the sliding narrowband window Filter_vel_f is twice n Add to the search band frequency energy sequence Search_f_e_list;

[0041] Take the top five frequency components in the search frequency band frequency energy sequence Search_f_elist, divide the top five frequency components by the rotation frequency and round them up to the nearest integer, and use the majority voting principle to get the number of gear teeth GTnum;

[0042] Step 5: Calculate the gear meshing frequency. The calculation formula is as follows:

[0043] GMF=rotate_freq*GTnum.

[0044] Furthermore, the first threshold value threshold1 is set to 0.3.

[0045] Further, the method specifically includes the following steps:

[0046] Step 1: collect the vibration signal of the gear box in real time, the vibration signal includes: speed signal and envelope signal; and simultaneously obtain the rotation frequency rotate_freq of the gear shaft;

[0047] Step 2: Perform fast Fourier transform (FFT) on the envelope signal collected in step 1 to obtain an envelope spectrum Enve_f_sig, and use a sliding narrowband window of Wnum times in the envelope spectrum Enve_f_sig to obtain the maximum frequency EnveMaxFreq of the sliding narrowband window energy;

[0048] Step 3: Determine whether the maximum frequency Enve_MaxFreq of the sliding narrowband window energy meets the frequency rotation flag condition. When the judgment condition is met, the frequency rotation flag rotate_freq_flag is True, and when the judgment condition is not met, the frequency rotation flag rotate_freq_flag is False. The judgment conditions are as follows:

[0049] Step 4: According to the result of the frequency rotation flag rotate_freq_flag, the speed signal is processed in different ways to search for the meshing frequency. When the frequency rotation flag rotate_freq_flag is True, the meshing frequency is searched using search method 1. When rotate_freq_flag is False, the meshing frequency is searched using search method 2.

[0050] Furthermore, in step 2, the i-th frequency component f of the envelope frequency sequence f_enve_list i , f i The frequency sequence corresponding to the k-fold narrowband window frequency is [k*f i -2*Δf:k*f i +2*Δf], corresponding to frequency f i The corresponding sliding narrowband window energy calculation formula is:

[0051]

[0052] Where: Δf is the spectral resolution of the envelope spectrum, Enve_f_sig(j*f i -k*Δf) is the envelope spectrum corresponding to frequency j*fi -k*Δf amplitude; the maximum sliding narrowband window energy max(Energy_f i ) corresponds to the frequency Enve_MaxFreq.

[0053] Furthermore, in step 2, the value of Wnum is 4.

[0054] Compared with the prior art, the present invention has at least the following beneficial effects: the gear meshing frequency estimation method of the present invention collects the vibration signal of the gearbox in real time through the vibration acceleration sensor installed at the bearing position of each shaft of the gearbox, estimates the number of gear teeth based on the vibration signal, and can estimate the number of gear teeth when the gearbox is abnormal, thereby obtaining the gear meshing frequency. The estimation method of the present invention is accurate and also ensures the uniformity of the estimation standard. During the operation of the present invention, the filtered signal Filter_vel_sig is obtained by designing a notch filter to filter out Enve_MaxFreq and 2*Ehve_MaxFreq, thereby avoiding the frequency of the maximum periodic component Enve_MaxFreq in the envelope spectrum from affecting the gear meshing frequency search, thereby ensuring the accuracy and effect of the search. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the overall process of the present invention.

[0056] Figure 2 It is a schematic diagram of energy search for maximum energy frequency within a sliding narrow-band window.

[0057] Figure 3 This is a flowchart of gear tooth number search method 1.

[0058] Figure 4 This is a flowchart of gear tooth number search method 2.

[0059] Figure 5 It is a schematic diagram of meshing frequency estimation.

[0060] Figure 6 is the estimated effect of partial gear meshing frequency. DETAILED DESCRIPTION

[0061] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0062] First of all, it should be noted that the gear meshing frequency is because the gear transmission system is an elastic mechanical system. Due to the relationship between structure and motion, there is non-stationarity in motion and force. This non-stationarity leads to the meshing vibration of the gear transmission system. The frequency of this vibration is the gear meshing frequency GMF. The calculation formula is as follows:

[0063] GMF=RPM / 60*GTnum

[0064] Among them, RPM is the speed of the shaft where the gear is located, and GTnum is the number of gear teeth.

[0065] See attached Figure 1 , Figure 1 The figure is the overall flow chart of the gear meshing frequency estimation method. First, a vibration acceleration sensor is installed at the bearing position of each shaft of the gearbox to collect the vibration signal of the gearbox in real time. The vibration signal includes: speed signal and envelope signal, and the rotation frequency of the gear shaft is obtained at the same time; secondly, a sliding narrowband window is used in the envelope signal to obtain the maximum frequency of the narrowband window energy Enve_MaxFreq, and it is determined whether the maximum frequency of the narrowband window energy Enve_MaxFreq is the rotation frequency. If it is the rotation frequency, the speed signal is searched for the number of gear teeth by the gear tooth search method 1. If there is no effective meshing frequency, the search is stopped, and the gear is in good running condition; on the contrary, if there is an effective meshing frequency, the gear tooth number is searched and obtained, and the gear meshing frequency is calculated in combination with the gear shaft rotation frequency; if it is the rotation frequency, the speed signal is searched for the number of gear teeth by the gear tooth search method 2. If there is no effective meshing frequency, the search is stopped, and the gear is in good running condition; on the contrary, if there is an effective meshing frequency, the gear tooth number is searched and obtained, and the gear meshing frequency is calculated in combination with the gear shaft rotation frequency.

[0066] See attached Figure 2 , Figure 2 The diagram shows the calculation of energy in a sliding narrowband window. The energy sum in the sliding narrowband double frequency window is calculated by sliding the narrowband double frequency window on the spectrum. The center frequency of the first narrowband of the narrowband window corresponding to the maximum energy sum in the window after sliding the entire frequency band is the frequency with the maximum energy on the spectrum, that is, the main component of the spectrum.

[0067] See attached Figure 3 , Figure 3 Schematic diagram of gear tooth number search method 1. Gear tooth number search method 1 is mainly aimed at the tooth number search method with the maximum periodic component of the envelope as the rotation frequency. First, FFT is performed on the speed signal to obtain the speed spectrum. 12*rotation frequency-1KHz is used as the search frequency band, and the energy ratio of the search frequency band in the entire signal Re_ratio is calculated. If Re_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component and the gear state is running well; if Re_ratio is greater than the first threshold threshold1, the window energy of each sliding window is calculated with a 2-fold frequency sliding window, and the frequency that meets the conditions is added to the frequency sequence to be selected:

[0068] Condition 1: the energy in the first window and the energy in the second window of the sliding window are greater than the threshold threshold2; the first window is a 1x window in a 2x frequency-multiplied sliding window, and the second window is a 2x window in a 2x frequency-multiplied sliding window;

[0069] Condition 2: The ratio of the energy in the first window of the sliding window to the energy in the second window is less than the third threshold threshold3. Take the top five frequencies in the sliding window of the frequency sequence to be selected, and divide the top five frequency components by the rotation frequency and round them up to the nearest integer. The number of gear teeth GTnum is obtained by majority voting.

[0070] See attached Figure 4 , Figure 4 Schematic diagram of gear tooth number search method 2. Gear tooth number search method 1 is mainly aimed at the tooth number search method whose maximum periodic component of the envelope is not the rotation frequency. First, FFT is performed on the speed signal to obtain the speed spectrum; then, a notch filter is used on the speed spectrum to filter out the 1X and 2X frequencies of the maximum periodic component of the envelope; 12*rotation frequency-1KHz is used as the search frequency band, and the proportion of the search frequency band energy in the energy of the entire signal Re_ratio is calculated. If Re_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component and the gear state is running well; if Re_ratio is greater than the first threshold threshold1, the window energy of each sliding window is calculated using a 2x frequency sliding window, and the frequency that meets the conditions is added to the frequency sequence to be selected.

[0071] Condition 1: The energy in the first window and the energy in the second window of the sliding window are greater than the threshold threshold2;

[0072] Condition 2: The ratio of the energy in the first window of the sliding window to the energy in the second window is less than the third threshold threshold3;

[0073] Take the top five frequencies in the sliding window of the frequency sequence to be selected, divide the top five frequency components by the rotation frequency and round them up to the nearest integer, and use the majority voting principle to get the number of gear teeth GTnum.

[0074] See attached Figure 5 , Figure 5 This is a schematic diagram of the calculation of the meshing frequency GMF of the present invention, that is, the meshing frequency GMF is equal to the searched gear tooth number GTnum multiplied by the gear shaft rotation frequency.

[0075] See attached Figure 6 , Figure 6 In some specific embodiments, a partial search of the gear meshing frequency is performed and compared with the actual meshing frequency. From the comparison, it can be seen that the searched gear meshing frequency is very close to the actual meshing frequency. The gear meshing frequency estimation of the present invention has a more accurate estimation effect. Through verification of various situations, it can also be known that the gear meshing frequency estimation method of the present invention has a very good scope of application and has a good prospect for promotion and implementation.

[0076] A parameter-free gearbox gear meshing frequency estimation method includes the following steps:

[0077] Step 1: Install a vibration acceleration sensor at the bearing position of each shaft of the gearbox to collect the vibration signal of the gearbox in real time. The vibration signal includes: a speed signal and an envelope signal; and simultaneously obtain the rotation frequency rotate_freq of the gear shaft;

[0078] Step 2: Perform FFT on the envelope signal collected in step 1 to obtain the envelope spectrum Enve_f_sig, and use a sliding narrowband window of Wnum times in the envelope spectrum Enve_f_sig to obtain the maximum frequency EnveMaxFreq of the sliding narrowband window energy;

[0079] Example: The ith frequency component f of the envelope frequency sequence f_enve_list i , f i The frequency sequence corresponding to the k-fold narrowband window frequency is [k*f i -2*Δf:k*f i +2*Δf], corresponding to frequency f i The corresponding sliding narrowband window energy calculation formula is:

[0080]

[0081] Where: Δf is the spectral resolution of the envelope spectrum, Enve_f_sig(j*f i -k*Δf) is the envelope spectrum corresponding to frequency j*f i -k*Δf amplitude, Wnum is always set to 4 according to the calculation effect;

[0082] Maximum sliding narrowband window energy: max(Energy_f i )The frequency corresponding to is Enve_MaxFreq;

[0083] Step 3: Determine whether the maximum frequency Enve_MaxFreq of the sliding narrowband window energy is a rotation frequency. In actual calculations, it is found that the envelope spectrum may have a rotation frequency of 2 times greater than 1 times. Therefore, when the maximum frequency EnveMaxFreq of the sliding narrowband window energy is 2 times the rotation frequency, it is also determined to be a rotation frequency. When the judgment condition is met, the rotation frequency flag rotate_freq_flag is True, and when the judgment condition is not met, the rotation frequency flag rotate_freq_flag is False. The judgment condition is as follows: If (EnveMaxFreq == rotate_freq) | (Enve_MaxFreq == 2*rotate_freq):

[0084] rotate_freq_flag = True

[0085] else:

[0086] rotate_freq_flag = False

[0087] Step 4: According to the result of the frequency rotation flag rotate_freq_flag, the speed signal is processed in different ways to search for the meshing frequency. When the frequency rotation flag rotate_freq_flag is True, the meshing frequency is searched using search method 1. When rotate_freq_flag is False, the meshing frequency is searched using search method 2:

[0088] Search method 1 includes the following steps:

[0089] Step 11, performing FFT on the velocity signal collected in step 1 to obtain a velocity spectrum Vel_f_sig;

[0090] Step 12: According to the minimum number of teeth that do not undercut during gear processing, the number of straight teeth is 14 and the number of helical teeth is 12. Therefore, the search frequency band of the meshing frequency is set to: 12*rotation frequency-1KHz. The energy ratio of the search frequency band (12*rotation frequency-1KHz) to the energy ratio of the entire frequency band is calculated. The calculation formula is as follows:

[0091]

[0092] Among them, index_12 is the index corresponding to 12*rotation frequency in the speed spectrum frequency series, and N is the length of the speed spectrum;

[0093] Step 13, if Re_ratio is less than the first threshold value threshold1, it means that there is no effective gear meshing frequency component, that is, if the search frequency band energy is too small, it means that the gear meshing frequency is very small, basically at the same level as the spectrum bottom noise, then it can be considered that the gear meshing is good, and there is no need to search for the meshing frequency in this case; if Re_ratio is greater than the first threshold value threshold1, then proceed to step 14, wherein the first threshold value threshold1 is obtained through case data statistics, and in the preferred embodiment of the present invention, the first threshold value threshold1 is 0.3;

[0094] Step 14: Calculate the energy in the 2 times narrowband window corresponding to each frequency in the search frequency band with a 2 times frequency sliding narrowband window, and the energy in the 2 times narrowband window of the nth frequency in the search frequency band is Energy_f n for:

[0095]

[0096] Among them, vel_Δf is the spectral resolution of the velocity spectrum, Vel_f_sig(l*f i-m*Δf) is the envelope spectrum corresponding to the frequency l*f i -m*Δf amplitude;

[0097] Energy_f n Add to the search band frequency energy sequence Search_f_e_list;

[0098] Step 15, take the top five frequency components in the search band frequency energy sequence Search_f_e_list, divide the top five frequency components by the rotation frequency and round up to the nearest integer, and obtain the gear tooth number Gt_num by majority voting principle.

[0099] Search method 2 can be divided into the following steps:

[0100] Step 21. Since the maximum periodic component EnVe_MaxFreq in the envelope spectrum is not the rotation frequency, the frequency may be the characteristic frequency of the bearing fault or a random frequency component. If it is the characteristic frequency of the bearing fault, the frequency may appear in the speed spectrum, which will affect the gear meshing frequency search. Therefore, the frequency must be filtered out before searching, and a notch filter is designed to filter out Enve_MaxFreq and 2*Ehve_MaxFreq to obtain the filtered signal Filter_vel_sig;

[0101] Step 22: Perform fast Fourier transform (FFT) on the filtered signal Filter_vel_sig to obtain the velocity spectrum Filter_vel_r_sig

[0102] Step 23. According to the minimum number of teeth that do not undercut in gear processing, the number of straight teeth is 14 and the number of helical teeth is 12. Therefore, the search frequency band of the meshing frequency is set to: 12*rotation frequency-1KHz. The energy ratio of the search frequency band (12*rotation frequency-1KHz) to the energy ratio of the entire frequency band is calculated. The calculation formula is as follows:

[0103]

[0104] Among them, index_12 is the index corresponding to 12*rotation frequency in the speed spectrum frequency sequence, and N is the length of the speed spectrum;

[0105] Step 24, if Re_ratio is less than the first threshold value threshold1, it means that there is no effective gear meshing frequency component, that is, if the search band energy is too small, it means that the gear meshing frequency is very small, basically at the same level as the spectrum bottom noise, the gear meshing is good, and there is no need to search the meshing frequency; if Re_ratio is greater than the first threshold value threshold1, then proceed to step 25; wherein the first threshold value threshold1 is obtained through case data statistics, and in a preferred embodiment of the present invention, the first threshold value threshold1 is 0.3;

[0106] Step 25: Calculate the energy in the 2-fold sliding narrowband window corresponding to each frequency in the search frequency band using a 2-fold sliding narrowband window, and calculate the energy in the 2-fold sliding narrowband window corresponding to each frequency in the search frequency band. n 1 times the narrowband window energy Filter_vel_f n _1 and 2 times the narrowband window energy Filter_vel_f n _2 is:

[0107]

[0108]

[0109] Among them, vel_Δf is the spectral resolution of the velocity spectrum, Filter_vel_f_sig(f n -m*Δf) is the envelope spectrum corresponding to frequency f n -m*Δf amplitude;

[0110] Then the frequency f n The energy in the sliding narrowband window Filter_vel_f is twice n for:

[0111] Filter_vel_f n =Filter_vel_f n _1+Filter_vel_f n _2

[0112] Filter the frequencies within the search band based on the following criteria:

[0113] A. Filter_vel_f n _1 and Filter_vel_f n _2 frequency band energy is greater than or equal to the threshold threshold2, that is, (Filter_vel_f n _1≥threshold2)&(Filter_vel_f n _1≥threshold2)

[0114] B. Filter_vel_f n _1 and Filter_vel_f n The frequency band energy of Filter_vel_f _2 cannot differ by more than the third threshold value threshold3, that is, (max(Filter_vel_f n _1, Filter_vel_f n _2) / min(Filter_vel_f n _1, Filter_vel_f n _2))<third threshold value threshold3;

[0115] The frequency f that satisfies the above conditions A and B n Add the search band frequency energy sequence Search_f_list, frequency f n The energy in the sliding narrowband window Filter_vel_f is twice n Add to the search band frequency energy sequence Search_f_e_list;

[0116] Take the top five frequency components in the search frequency band frequency energy sequence Search_f_elist, divide the top five frequency components by the rotation frequency and round them up to the nearest integer, and use the majority voting principle to get the number of gear teeth GTnum;

[0117] Step 5: Calculate the gear meshing frequency. The calculation formula is as follows:

[0118] GMF=rotate_freq*GTnum.

[0119] The gear meshing frequency estimation method of the present invention collects the vibration signal of the gear box in real time through the vibration acceleration sensor installed at the bearing position of each shaft of the gear box, and the vibration signal includes: a speed signal and an envelope signal; and the rotation frequency of the gear shaft is obtained at the same time; secondly, a sliding narrowband window is used in the envelope signal to obtain the maximum frequency EnveMaxFreq of the narrowband window energy, and it is determined whether EnveMaxFreq is a rotation frequency. If it is a rotation frequency, the speed signal is processed in method 1, and if it is not a rotation frequency, the speed signal is processed in method 2.

[0120] Next, the collected speed signal is processed in two ways:

[0121] Method 1, calculate the energy ratio Range_energy_ratio of the search frequency band (12*rotation frequency-1KHz) in the entire frequency band. If Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component. If Range_energy_ratio is greater than the first threshold threshold1, then slide the narrowband window of 2 times the frequency in the search frequency band to calculate the energy in the 2 times window corresponding to each frequency, take the top five frequency components of energy, and divide the top five frequency components of energy by the rotation frequency to the nearest integer, and obtain the number of gear teeth by the majority voting principle.

[0122] Method 2, filter out the frequency components of EnveMaxFreq and 2*EnveMaxFreq in the speed spectrum signal Vel_sig to obtain the filtered speed signal Filter_vel_sig, calculate the energy ratio of the search frequency band (12*rotation frequency-1KHz) in the entire frequency band Range_energy_ratio, if Range_energy_ratio is less than the first threshold threshold1, if Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component, if Range_en ergy_ratio is greater than the first threshold threshold1, and then the 2 times window energy corresponding to each frequency is calculated by sliding the narrowband window in the search frequency band. During the search process, the frequency components whose energy in the narrowband window is greater than the threshold threshold2 and whose ratio of the energy of the 1 times narrowband window to the 2 times narrowband window is less than the third threshold threshold3 are added to the filtered frequency sequence Filter_freq_list, and the top five frequency components with the energy in the 2 times narrowband window in Filter_freq_list are taken, and the top five frequency components with the energy are divided by the rotation frequency and rounded to the nearest integer, and the number of gear teeth is obtained by the majority voting principle. Finally, the gear meshing frequency is calculated.

Claims

1. A parameter-free gearbox gear meshing frequency estimation method, characterized in that: Collect the vibration signal of the gear box, the vibration signal includes: a speed signal and an envelope signal; and obtain the rotation frequency of the gear shaft; then, use the sliding narrowband window to obtain the maximum frequency Enve_MaxFreq of the sliding narrowband window energy in the collected envelope signal, and judge whether the maximum frequency Enve_MaxFreq of the sliding narrowband window energy satisfies the rotation frequency identification condition, if so, search the speed signal in mode 1, if not, search the speed signal in mode 2; Search method 1, calculate the energy ratio of the search frequency band in the entire frequency band Range_energy_ratio, if the energy ratio Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component; if the energy ratio Range_energy_ratio is greater than or equal to the first threshold threshold1, then the 2 times window energy corresponding to each frequency is calculated by sliding the narrow band window of 2 times frequency in the search frequency band, take the top five frequency components of energy, and divide the top five frequency components of energy by the rotation frequency to the nearest integer, and obtain the number of gear teeth by the majority voting principle; Search method 2, filter out the frequency components of the maximum frequency Enve_MaxFreq of the sliding narrowband window energy and the maximum frequency Enve_MaxFreq of the sliding narrowband window energy in the speed spectrum signal Vel_sig to obtain the filtered speed signal Filter_vel_sig, calculate the energy ratio Range_energy_ratio of the search frequency band in the entire frequency band, if the energy ratio Range_energy_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component, if the energy ratio Range_energy_ratio is greater than or equal to If it is equal to the first threshold threshold1, then the 2x window energy corresponding to each frequency is calculated in the search frequency band by sliding the narrowband window at 2x frequency. During the search process, the frequency components whose energy in the narrowband window is greater than or equal to the second threshold threshold2 and whose energy difference between the 1x narrowband window and the 2x narrowband window is less than 3 times the third threshold threshold are added to the filtered frequency sequence Filter_freq_list. The top five frequency components in energy in the 2x narrowband window in the frequency sequence Filter_freq_list are taken, and the top five frequency components in energy are divided by the rotation frequency and rounded to the nearest integer, and the number of gear teeth is obtained by majority voting principle. Finally, the gear meshing frequency is calculated by the obtained gear tooth number.

2. A parameter-free gearbox gear meshing frequency estimation method according to claim 1, characterized in that: The vibration signal of the gear box is collected by a vibration acceleration sensor installed at the bearing position of each shaft of the gear box.

3. The method for estimating the gearbox gear meshing frequency without parameters according to claim 1, characterized in that: When the maximum energy frequency Enve_MaxFreq of the sliding narrowband window is equal to the rotation frequency or twice the rotation frequency, it is determined that the maximum energy frequency Enve_MaxFreq of the sliding narrowband window satisfies the rotation frequency identification condition.

4. The method for estimating the gear meshing frequency of a gearbox without parameters according to claim 1, characterized in that: According to the minimum number of teeth without undercutting in gear processing, the number of straight teeth is 14 and the number of helical teeth is 12, so the search frequency band of the meshing frequency is set to: 12*revolution frequency - 1KHz.

5. The method for estimating the gear meshing frequency of a gearbox without parameters according to claim 1, characterized in that: The search method 1 comprises the following steps: Step 11, performing fast Fourier transform on the velocity signal collected in step 1 to obtain a velocity spectrum Vel_f_sig; Step 12: Calculate the energy ratio of the search frequency band to the energy ratio of the entire frequency band. The calculation formula is as follows: in, The corresponding index of 12*revolution frequency in the speed spectrum frequency series, is the length of the velocity spectrum; Step 13, if the energy ratio Re_ratio is less than the first threshold threshold1, it means that there is no effective gear meshing frequency component, if Re_ratio is greater than the first threshold threshold1, then proceed to step 14, wherein the first threshold threshold1 is obtained through case data statistics; Step 14: Calculate the energy in the 2 times narrowband window corresponding to each frequency by sliding the narrowband window with 2 times the frequency in the search band, and calculate the energy in the 2 times narrowband window of the nth frequency in the search band. for: in, is the spectral resolution of the velocity spectrum, is the frequency corresponding to the velocity spectrum The amplitude of Will Add to the search band frequency energy sequence Search_f_e_list; Step 15, take the top five frequency components in the search band frequency energy sequence Search_f_e_list, divide the top five frequency components by the rotation frequency and round up to the nearest integer, and obtain the gear tooth number Gt_num by majority voting principle.

6. The method for estimating the gear meshing frequency of a gearbox without parameters according to claim 1, characterized in that: The search method 2 comprises the following steps: Step 21: Design a notch filter Enve_MaxFreq and 2* Enve_MaxFreq Get the filtered signal Filter_vel_sig; Step 22, perform fast Fourier transform on the filtered signal Filter_vel_sig to obtain the velocity spectrum Filter_vel_f_sig; Step 23, calculate the energy ratio of the search frequency band to the energy of the entire frequency band Re_ratio, the calculation formula is as follows: in, The index corresponding to 12*revolution frequency in the speed spectrum frequency sequence, is the length of the velocity spectrum; Step 24, if Re_ratio is less than the first threshold value threshold1, it means that there is no effective gear meshing frequency component, the gear meshing is good, and there is no need to search for meshing frequency; if Re_ratio is greater than the first threshold value threshold1, then proceed to step 25; wherein the first threshold value threshold1 is obtained through case data statistics; Step 25: Calculate the energy in the 2x sliding narrowband window corresponding to each frequency in the search frequency band using a 2x sliding narrowband window, and calculate the energy in the 2x sliding narrowband window corresponding to each frequency in the search frequency band. 1 times the energy in the narrowband window and 2 times the energy in the narrowband window For: in, is the spectral resolution of the velocity spectrum, The frequency corresponding to the filtered velocity spectrum Filter_vel_f_sig The amplitude of Then the frequency The energy in the sliding narrowband window is twice for: Filter the frequencies within the search band based on the following criteria: A. 1 times the energy in the narrowband window and twice the energy in the narrowband window The frequency band energy is greater than or equal to the threshold threshold2; B. 1 times the energy in the narrowband window and twice the energy in the narrowband window The ratio of the frequency band energy is less than threshold3; The frequency that satisfies the above conditions A and B Add the search band frequency energy sequence Search_f_list, frequency The energy in the sliding narrowband window is twice Add to the search band frequency energy sequence Search_f_e_list; Take the top five frequency components in the search frequency band frequency energy sequence Search_f_e_list, divide the top five frequency components by the rotation frequency and round them up to the nearest integer, and use the majority voting principle to get the number of gear teeth. ; Step 5: Calculate the gear meshing frequency. The calculation formula is as follows: 。 7. A parameter-free gearbox gear meshing frequency estimation method according to claim 4 or 5, characterized in that: The first threshold threshold1 is 0.

3.

8. The method for estimating the gear meshing frequency of a gearbox without parameters according to claim 1, characterized in that: The specific steps include: Step 1: Collect the vibration signal of the gearbox in real time, the vibration signal includes: speed signal and envelope signal; and obtain the rotation frequency of the gear shaft at the same time. rotate_freq ; Step 2: Perform fast Fourier transform on the envelope signal collected in step 1 to obtain the envelope spectrum Enve_f_sig , in the envelope spectrum Enve_f_sig Using sliding Wnum Times the narrowband window to obtain the maximum frequency of the sliding narrowband window energy Enve_MaxFreq; Step 3: Determine the maximum frequency of the sliding narrowband window energy Enve_MaxFreq Whether the frequency switching identification conditions are met, the frequency switching identification is performed when the judgment conditions are met rotate_freq_flag for True, When the judgment condition is not met, the frequency flag is changed rotate_ freq_flag for False, The judgment conditions are as follows: Step 4: According to the frequency conversion mark rotate_freq_flag As a result, the speed signal is processed in different ways to search for the meshing frequency. rotate_freq_flag True, use search mode 1 to search for meshing frequency. rotate_freq_flag If it is False, the meshing frequency will be searched using search method 2.

9. A parameter-free gearbox gear meshing frequency estimation method according to claim 8, characterized in that: In step 2, the frequency sequence of the envelope f_enve_list The i-th frequency component , of k The frequency sequence corresponding to the frequency in the times narrowband window is [ -2 : +2 ], corresponding to the frequency The corresponding sliding narrowband window energy calculation formula is: in: is the spectral resolution of the envelope spectrum, The envelope spectrum corresponds to the frequency The maximum sliding narrowband window energy The corresponding frequency is Enve_ MaxFreq .

10. A parameter-free gearbox gear meshing frequency estimation method according to claim 8, characterized in that: In the step 2, the Wnum The value is 4.