Rack bearing test spectrum construction method based on torque and rotating speed combined counting method

Abstract

The invention discloses a bearing bench experiment spectrum construction method based on a rotating speed and torque combined counting method. The method comprises the steps of 1, obtaining sample load spectrum road operation data and performing preprocessing; step 2, rotating speed and torque combined counting is carried out on the preprocessed data; 3, constructing a damage model of the bearing; 4, extrapolating the original data; and 5, compiling a multi-stage program load spectrum. According to the method, a frequency extrapolation method of combining a Gaussian kernel function with K strengthening coefficient extrapolation is adopted, so that the authenticity of data after bearing load spectrum extrapolation is improved; according to the corresponding relation between the rotating speed and the fatigue life of the bearing, the load spectrum construction method based on rotating speed and torque combined counting is provided, the accuracy of fatigue life analysis is improved, and meanwhile, the load spectrum analysis characteristics are visually reflected by constructing the multi-stage spectrum image of rotating speed and torque combined distribution.

Description

Technical Field

[0001] The invention belongs to the technical field of reliability and durability test of automobile bearing parts, and relates to a method for constructing a bench spectrum based on rotation speed and torque joint counting. Background Art

[0002] In automobiles, more than 90% of the fractures can be attributed to fatigue failure of parts. Fatigue failure is the most common and harmful failure mode of automobile parts. Fatigue damage often occurs in frames, leaf springs, coil springs, shafts and rods that bear alternating loads. Therefore, the importance of reliability and durability analysis of the fatigue life of vehicle parts in the process of vehicle forward development research cannot be underestimated.

[0003] In the process of analyzing vehicle reliability and durability, the load spectrum is the top-level input of the entire process of forward durability research and development. If there is a statistical deviation between the load input result and the actual user load, it will greatly affect the subsequent design analysis, resulting in over-design or under-assessment problems. Load spectrum analysis is therefore an important part of vehicle durability research and development. Conventional methods for analyzing reliability and durability require the use of full-cycle load spectra to obtain service life, but the amount of data collected by the test road load spectrum cannot cover the load conditions of the entire life cycle, so it is necessary to use test load samples to infer the extreme loads and their frequencies that may occur throughout the life cycle.

[0004] As a key method to obtain load distribution, the counting method is particularly important. For general non-rotating parts or shaft parts whose torque fluctuation is unrelated to their rotational motion, the rain flow counting method can be used. For parts such as bearings whose rotation speed affects the fluctuation frequency, the rotating rain flow counting method or the rotating torque histogram counting method is required.

[0005] However, the above two methods cannot well reflect the joint distribution law of speed and torque, and cannot intuitively and efficiently reflect the influence of speed and torque on bearing fatigue life. Therefore, in the process of selecting the load counting method, the present invention focuses on the accuracy of the joint consideration of speed and torque, so the speed and torque joint counting method is used to count the load of the bearing, and the load is extrapolated by Gaussian kernel function combined with K enhancement coefficient extrapolation, so as to construct a test spectrum construction method with great guiding significance for bearing reliability and durability. Summary of the invention

[0006] The purpose of the present invention is to provide a method for constructing a bench bearing test spectrum based on a torque-speed joint counting method. The present invention adopts a frequency extrapolation method combining a Gaussian kernel function with a K enhancement coefficient extrapolation to improve the authenticity of the data after the bearing load spectrum is extrapolated; and because the traditional analysis of the bearing force process does not take into account factors such as the radial deformation of the bearing, the present invention calculates the bearing force by piecewise interpolation after software simulation to improve the accuracy of the force analysis results. In view of the current lack of analysis of the influence of the speed on the construction results in the bearing test spectrum construction process, and the speed directly determines the value of the fatigue life curve, the present invention proposes a load spectrum construction method based on speed-torque joint counting according to the corresponding relationship between the speed and the bearing fatigue life, which improves the accuracy of the fatigue life analysis, and at the same time, by constructing a multi-level spectrum image of the joint distribution of speed and torque, the load spectrum analysis characteristics are intuitively reflected.

[0007] The technical solution adopted by the present invention is:

[0008] A method for constructing a bearing bench test spectrum based on a speed-torque joint counting method, comprising:

[0009] Step 1: Obtain sample load spectrum road operation data and preprocess it to obtain preprocessed time domain data;

[0010] Step 2: Perform speed and torque joint counting on the pre-processed time domain data: record the loading frequency in each torque and speed interval by counting the pre-processed time domain data in intervals;

[0011] Step 3: Construct a bearing damage model: Perform a bearing force analysis, calculate the equivalent dynamic load based on the analysis results, and then use the aISO correction coefficient that takes the speed parameter into consideration to obtain a bearing life calculation formula with the bearing torque and speed as independent variables. Combined with the joint counting results, construct a bearing damage model.

[0012] Step 4: Extrapolate the preprocessed data: According to the results of the bearing damage model, the K enhancement coefficient can be obtained in combination with the bearing extrapolation target damage, and then the probability density function is obtained through kernel density estimation. The counting results obtained in step 2 are extrapolated through the Gaussian kernel function combined with the K enhancement coefficient to obtain the frequency matrix of speed and torque after extrapolation;

[0013] Step 5. Prepare a multi-level program load spectrum: divide the torque and speed into eight levels according to a certain ratio. Use the Palmgren-Miner law and the S-N curve of the material to combine the frequency matrix of the torque and speed after extrapolation in step 4. Use the upward equivalent method to convert the frequency in the 128*128 level small interval into the frequency in the 8*8 level large interval according to the equal damage conversion principle, and obtain the 8*8 level bench test spectrum of the speed and torque of the bearing.

[0014] Optionally, the preprocessing in step 1 includes outlier removal, over-threshold processing and data resampling;

[0015] The removal of outliers includes: performing Fourier transform on the original data to obtain the original spectrum, filtering out the torque data under each amplitude according to the frequency through Butterworth filtering, and dividing it into main wave, secondary wave and tertiary wave; removing the tertiary wave of the amplitude data fragment less than ±1Nm, completing the preliminary filtering of invalid waves;

[0016] Super-threshold processing refers to filling missing data with piecewise interpolation after outliers are removed;

[0017] Data resampling includes: resampling according to the time specified frequency interpolation, extracting non-repeating time, and interpolating according to a fixed uniform sampling frequency to obtain pre-processed time domain data.

[0018] Optionally, in the process of speed-torque joint counting in step 2, the pre-processed time domain data is first divided into 128*128 small interval data according to the speed-torque target interval, and the speed n(t) in the i-th speed-torque small interval is integrated over time t. The time corresponding to each small interval is The number of torque segments in a small interval is m, and the integration result is recorded as the frequency Ni in this interval. The frequency of each interval is saved as a matrix;

[0019]

[0020] Optionally, in the process of constructing the bearing damage model in step 3, the bearing damage D is calculated as follows:

[0021]

[0022] N is the frequency of data loading at each level, L nm The corresponding bearing life of each level can be known.

[0023] Optionally, the loading frequency corresponding to each level of speed and torque of the front bearing can be obtained by extrapolation from step 2;

[0024] Bearing life before extrapolation:

[0025]

[0026] Where Cr is the basic rated dynamic load; m is the bearing life index, m = 3 for ball bearings and m = 10 / 3 for roller bearings; a1 is the reliability correction factor; P is the equivalent dynamic load; a ISO Correction factor as a function of the rotation speed:

[0027]

[0028] aISO The specific calculation formula can be found by looking up the table according to the selection of different bearings; the contamination coefficient e of the automobile reducer bearing is selected from 0.8 to 0.9; the load fatigue limit C u It can be found on the supplier's official website according to the bearing model; the lubrication condition of the bearing is expressed by the viscosity ratio κ, which is the ratio of the actual kinematic viscosity to the reference kinematic viscosity:

[0029]

[0030] v is the actual kinematic viscosity, and the reference kinematic viscosity v1 depends on the bearing speed and pitch diameter D pw , calculated as follows:

[0031] When n<1000r / min,

[0032] v1=45000*n -0.83 *D pw -0.5 ;

[0033] When n>1000r / min,

[0034] v1=4500*n -0.5 *D pw -0.5 ;

[0035] P is the equivalent dynamic load, which can be calculated based on the force analysis of the bearing.

[0036] P=X*A a +Y*A r ;

[0037] Among them A r A is the radial force on the bearing, a The axial force on the bearing is obtained by setting several sets of bearing input torques T and using the software Romax finite element simulation to obtain the fragmentary results of the bearing force. The simulation results are interpolated to obtain the fitting curve of the bearing force with respect to the input torque T. Then the test torque is brought into the fitting curve to find the corresponding force results, and the corresponding A of each level of torque can be obtained. r With A a ; X and Y are the distribution coefficients, and the equivalent dynamic load under each level of torque is obtained by combining the torque-force matrix.

[0038] Optional, in step 3:

[0039] The calculation formula for bearing life at each torque speed is:

[0040]

[0041] f(T, n) means that the bearing life at each torque and speed can be simplified as a function of torque and speed.

[0042] According to the final calculated bearing life at each level combined with the load frequency at each level, the pre-extrapolation bearing damage matrix at 128*128 levels can be obtained:

[0043]

[0044] The next step of load spectrum extrapolation can be carried out based on the bearing damage matrix of this original data.

[0045] Optionally, during the extrapolation of raw data in step 4:

[0046] The overall load spectrum extrapolation adopts the Gaussian kernel function combined with the K enhancement coefficient extrapolation method;

[0047] The load function method is selected for the load extreme value extrapolation in the direct extrapolation process of the K enhancement coefficient. The damage-frequency distribution of the original data is combined with various distributions for fitting and comparison. The distribution function with the minimum goodness of fit for each working condition under the appropriate 90-point position confidence is selected. The distribution fit goodness test method is used to test whether the preliminary fitting distribution model is accurate. The Anderson-Darling test statistic determines the goodness of fit:

[0048]

[0049] This test compares AD 2 and the size of the critical value of the corresponding distribution cluster, accept or reject the null hypothesis H0 at the significance level α; F n (x) represents the probability density distribution function of the experimental data, and F(x) represents the probability density distribution function of the distribution to be compared; the smaller the AD value, the better the distribution fitting effect;

[0050] Then determine the probability density function corresponding to the data distribution and obtain the inferred extreme value;

[0051] The total damage D of the target life can be determined by extrapolating the target reliability based on the inferred extreme value and the requirements. 目标 , the calculation of K strengthening extrapolation coefficient is based on the following formula:

[0052]

[0053] Total damage of current test data D 试验 It can be obtained from the damage model in step three.

[0054] Optionally, in step 4, the Gaussian kernel function combined with the K enhancement coefficient extrapolation process is:

[0055] Assume that the random variable x obeys an unknown probability density function Using the samples x1, ..., x drawn from nFunction The distribution of is estimated; a smooth function K(*) is defined as the kernel function, h is the smooth parameter (bandwidth); the bandwidth h = ba is the length of the assumed interval, and the number of samples falling into the interval [a, b] is k, then the relationship between the kernel function K(*) and the number of samples is:

[0056]

[0057] According to the kernel density estimation principle, the probability density function is for:

[0058]

[0059] The Gaussian kernel function selected is:

[0060]

[0061] The corresponding two-dimensional kernel density estimate is:

[0062]

[0063] Here, the adaptive bandwidth is selected:

[0064] Introducing an adaptive factor λ i ;

[0065]

[0066] In the formula, λ i is the adaptive factor; For point u i The probability density value at ;

[0067] g by

[0068]

[0069] Calculated;

[0070] The final adaptive kernel density estimate is:

[0071]

[0072] N is the number of sample data;

[0073] By combining the K enhancement coefficient extrapolation with the Gaussian kernel function extrapolation, a set of frequency data at each level of speed and torque after extrapolation is obtained, which is expanded in both frequency and speed and torque:

[0074]

[0075] Optionally, in the process of compiling the multi-level program load spectrum in step 5, since the data source corresponding to the frequency matrix of the speed-torque joint counting comes from the two variables of speed and torque;

[0076] According to the equivalent damage principle, the user's full life cycle load spectrum under the extrapolated road condition is reconstructed into a program loading spectrum as the loading input basis for durability or reliability tests; the 128*128-level extrapolated bearing damage obtained in step 4 is converted into a relatively simple 8*8-level spectrum using the principle of equal damage conversion. The proportional coefficients of the 8-level spectrum are graded as 1, 0.95, 0.85, 0.725, 0.575, 0.425, 0.275 and 0.125; the principle is:

[0077]

[0078] That is, the damage of each level in the 8*8 level spectrum is equal to the sum of the damage in the 128*128 level spectrum included, where N eq is the frequency corresponding to a set of torque speed in the 8*8 level spectrum, L eq is a set of speed torque corresponding to the bearing fatigue life in the 8*8 level spectrum; where N i is the frequency corresponding to a set of speed torque in the 128*128 level spectrum, L i is a set of speed-torque corresponding bearing fatigue life in the 128*128-level spectrum; l is the number of small intervals of the 128-level spectrum included in the interval corresponding to one level in the eight-level spectrum.

[0079] The beneficial effects of the present invention are:

[0080] (1) In the process of load frequency extrapolation, compared with the traditional extrapolation method, the present invention adopts an extrapolation method combining Gaussian kernel function and K enhancement coefficient. The extrapolation process fully considers the distribution characteristics of the original load, combines the expansion of the distribution interval by Gaussian kernel function and the linear extrapolation of data frequency by K enhancement coefficient, and realizes that the load spectrum after extrapolation not only meets the number of cycles required by the test, but also the load boundary is reasonably expanded. The load cycles that are not collected but should exist in the extrapolated load spectrum are retained, which improves the accuracy of load extrapolation.

[0081] (2) In the process of bearing force analysis, the present invention performs segmented interpolation calculation based on traditional bearing force analysis calculation and combined with finite element software simulation results, taking into account the influence of factors such as bearing radial deformation and transmission shaft deflection deformation on bearing force in the process of bearing force analysis. The data obtained through finite element simulation improves the fit between the force analysis results and the actual bearing force, and fitting and interpolating the simulation results not only expands the limited simulation results to the full torque cycle, but also saves test time.

[0082] (3) According to the bearing life calculation formula, the speed parameter affects the reference kinematic viscosity and then affects the viscosity ratio, making a ISO The size of the coefficient changes, but the loading process of most bearing test bench spectra currently does not take the influence of the speed into consideration. Therefore, the speed-torque joint counting method adopted in the present invention can greatly improve the accuracy and authenticity of the bearing life analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present invention but do not constitute a limitation of the present invention. In the accompanying drawings:

[0084] Figure 1 It is a flow chart of the multi-stage test bench spectrum construction method based on the speed torque joint counting method of the present invention;

[0085] Figure 2 It is a schematic diagram of constructing a multi-stage test bench spectrum of the present invention, and each Ni and the corresponding torque speed are recorded and saved as a frequency matrix;

[0086] Figure 3 The final 8*8 level test spectrum of bearing speed and torque is the required bearing bench test spectrum (the torque is divided into 16 levels in the figure to facilitate image analysis). DETAILED DESCRIPTION

[0087] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0088] Part I:

[0089] The complete steps of the method of the present invention are described.

[0090] Combination Figure 1 The multi-level test bench spectrum construction method based on the speed torque joint counting method proposed in the present invention mainly includes the following steps:

[0091] Step 1: Obtain sample load spectrum road operation data and preprocess it: The raw data collected directly by the sensor may be affected by the driving environment, interference signals and buildings, and may have mutations, noise and burrs. It is necessary to perform operations such as outlier removal to preprocess the raw data.

[0092] The removal of abnormal values ​​is based on the Darai criterion. Considering that there are a large number of small torque fluctuation signals in the torque load, a large number of small-amplitude high-frequency load cycles will be formed. Such loads basically do not cause fatigue damage but will greatly increase the amount of calculation and form data redundancy, that is, high-frequency noise. Therefore, this type of load needs to be eliminated. By performing Fourier transform on the original data, the original spectrum can be obtained. Through Butterworth filtering, the torque data under each amplitude can be screened out according to the frequency size and divided into main wave, secondary wave, and tertiary wave. By removing the tertiary wave with small amplitude and high frequency (amplitude data fragments less than ±1Nm), the initial filtering of invalid waves can be completed.

[0093] Super-threshold processing refers to filling missing data with piecewise interpolation after removing abnormal data outside the torque and speed requirement range.

[0094] The reason for data resampling is that the sampling frequencies of the signals are inconsistent, and the data points of the signals cannot correspond one to one in time. Therefore, in order to be able to perform joint distribution statistics of the signals, all data are resampled to unify the sampling frequency. Resampling is performed by interpolation at a specified frequency of 10Hz, extracting non-repeating time, and interpolating at a fixed unified sampling frequency.

[0095] Step 2: Perform speed and torque joint counting on the original data: count the data at each torque and speed by dividing the preprocessed time domain data into intervals (generally, the torque and speed are divided into 128 levels each), and integrate the speed in each speed and torque interval over time:

[0096]

[0097] Record the frequency N in each torque speed range i . Record and save each Ni and the corresponding torque speed as a frequency matrix. The result is as follows Figure 2 shown.

[0098] Step 3: Construct a damage model: Calculate the damage process of the bearing at various speeds and torques. The damage of each bearing is equal to the data loading frequency N at each level divided by the corresponding bearing life L at each level. nm It can be seen that:

[0099]

[0100] According to step 2, the loading frequency corresponding to each level of speed and torque of the front bearing can be obtained.

[0101] The life of the bearing before extrapolation can be known from GB_T 6391-2010 Dynamic load rating and rated life of rolling bearings:

[0102]

[0103] Where Cr is the basic rated dynamic load, which can be directly obtained according to the bearing model; m is the bearing life index, m = 3 for ball bearings and m = 10 / 3 for roller bearings; a1 is the reliability correction factor; P is the equivalent dynamic load; a ISO The correction factor is a function of the speed and is also related to the contamination factor, load fatigue limit and lubrication conditions of the bearing:

[0104]

[0105] a ISO The specific calculation formula can be found by looking up the table according to the different bearings selected. According to the reference data: the contamination coefficient ec of the automobile reducer bearing is generally selected as 0.8-0.9; the load fatigue limit Cu can be found on the supplier's official website according to the bearing model; the lubrication condition of the bearing is expressed by the viscosity ratio κ, that is, the ratio of the actual kinematic viscosity to the reference kinematic viscosity:

[0106]

[0107] The actual kinematic viscosity v is related to the lubricant properties, speed, and operating temperature during the operation of the bearing. The lubricant properties refer to the viscosity grade of the lubricant, which is specifically standardized in ISO. The actual viscosity of the lubricant can be obtained by looking up the table or calculating it based on the viscosity grade and operating temperature of the lubricant.

[0108] The reference kinematic viscosity v1 depends on the bearing speed and pitch diameter D pw , calculated as follows:

[0109] When n<1000r / min,

[0110] v1=45000*n -0.83 *D pw -0.5 ;

[0111] When n>1000r / min,

[0112] v1=4500*n -0.5 *D pw -0.5 ;

[0113] P is the equivalent dynamic load, which can be calculated based on the force analysis of the bearing.

[0114] P=X*A a +Y*A r ;

[0115] Among them A r A is the radial force on the bearing, aThe axial force on the bearing is obtained by setting several sets of bearing input torques T and using the software Romax finite element simulation to obtain the fragmentary results of the bearing force. The simulation results are interpolated to obtain the fitting curve of the bearing force with respect to the input torque T. Then the test torque is brought into the fitting curve to find the corresponding force results, and the corresponding A of each level of torque can be obtained. r With A a ; X and Y are the distribution coefficients. According to the national standard GB_T 6391-2010, the equivalent dynamic load under each level of torque can be obtained by combining the torque-force matrix.

[0116] Finally, the bearing life calculation formula under each torque and speed can be simplified to:

[0117]

[0118] According to the final calculated bearing life at each level combined with the load frequency at each level, the pre-extrapolation bearing damage matrix at 128*128 levels can be obtained:

[0119]

[0120] The next step of load spectrum extrapolation can be carried out based on the bearing damage matrix of this original data.

[0121] Step 4: Extrapolate the original data: Use the Gaussian kernel function combined with the K enhancement coefficient extrapolation method to extrapolate the load spectrum.

[0122] Step 4.1: Extrapolation of K enhancement factor:

[0123] The extrapolation of the K enhancement factor needs to be obtained based on the damage before and after the extrapolation, as follows:

[0124]

[0125] That is, the extrapolation multiple = total damage of target life / total damage of current test data.

[0126] For target damage D 目标 To obtain , we need to extrapolate the load extreme value:

[0127] The load function method is selected for load extreme value extrapolation. The damage-frequency distribution of the original data is combined with various distributions for fitting and comparison. The distribution function with the minimum goodness of fit for each working condition under the appropriate 90-point position confidence is selected. The distribution fit goodness test method can be used to test whether the preliminary fitting distribution model is accurate. The Anderson-Darling test statistic determines the goodness of fit:

[0128]

[0129] This test compares AD 2and the size of the critical value of the corresponding distribution cluster. Under the significance level α, accept or reject the null hypothesis H0, F n (x) represents the probability density distribution function of the experimental data, and F(x) represents the probability density distribution function of the distribution to be compared; the smaller the AD value, the better the distribution fitting effect.

[0130] Here, the user unit mileage damage is used as the strength index, which should be able to represent the original load unit mileage damage strength of 90% percentile users. The Weibull distribution probability curve is fitted through the original unit mileage damage input, and the value corresponding to P = 0.9 in the probability curve is taken as the 90% percentile user unit mileage damage target value d 90 Combined with the target life mileage L 目标 , calculate the overall damage target D of the vehicle throughout its life cycle 目标 =d 90 *L 目标 .

[0131] According to step 3, D can be calculated 试验 .

[0132] Substitute the vehicle's overall damage target D over its entire life cycle 目标 With experimental damage D 试验 Then the K strengthening coefficient can be obtained.

[0133] Step 4.2: Extrapolation process of Gaussian kernel function:

[0134] Assume that the distribution of the corresponding values ​​x of each level of speed and torque obeys an unknown probability density function Using the samples x1, ..., x drawn from n Function The distribution of is estimated. Define a smooth function K(*) as the kernel function, and h as the smooth parameter (bandwidth). The bandwidth h = ba is the length of the assumed interval. Assuming that the number of samples falling into the interval [a, b] is k, the relationship between the kernel function K(*) and the number of samples is:

[0135]

[0136] According to the kernel density estimation principle, the probability density function is for:

[0137]

[0138] The Gaussian kernel function selected is:

[0139]

[0140] The corresponding two-dimensional kernel density estimate is:

[0141]

[0142] As for the bandwidth, another factor that affects the probability density function under the kernel density estimation principle, there is an optimal bandwidth selection. The larger the bandwidth coefficient, the smoother the estimated density function curve, but it may lead to insufficient fitting, that is, covering up some characteristics of the original sample data; the smaller the bandwidth coefficient, the estimated density function curve may be overfitted, resulting in a larger variance.

[0143] Here, the adaptive bandwidth is selected:

[0144] Introducing an adaptive factor λ i ;

[0145]

[0146] In the formula, λ i is the adaptive factor; For point u i The probability density value at g is given by

[0147]

[0148] Calculated.

[0149] The final adaptive kernel density estimate is:

[0150]

[0151] N is the number of sample data. Substituting the estimated torque and speed into the above x and y, the adaptive kernel density estimation result can be obtained.

[0152] After introducing the adaptive factor, the kernel density estimation bandwidth can change with the data, and it is smaller in places with dense data and larger in places with sparse data. Compared with the probability density obtained by traditional fixed bandwidth, it is closer to the actual distribution.

[0153] By combining the K enhancement coefficient extrapolation with the Gaussian kernel function extrapolation, we can obtain a set of frequency data at each level of speed and torque after extrapolation, which is expanded in both frequency and speed and torque:

[0154]

[0155] Step 5, compile a multi-level program load spectrum: divide the torque and speed into eight levels according to a certain ratio, and use the Palmgren-Miner rule and the damage equivalence principle to convert the frequency matrix of the torque and speed after extrapolation in step 4 into the frequency of the 8*8 level large interval by the upward equivalent method. The 8-level proportional coefficient is classified into 1, 0.95, 0.85, 0.725, 0.575, 0.425, 0.275, and 0.125. The principle is:

[0156]

[0157] That is, the damage of each level in the 8*8 level spectrum is equal to the sum of the damage in the 128*128 level spectrum included, where N eq is the frequency corresponding to a set of torque speed in the 8*8 level spectrum, L eq is a set of speed torque corresponding to the bearing fatigue life in the 8*8 level spectrum; where N i is the frequency corresponding to a set of speed torque in the 128*128 level spectrum, L i is a set of speed-torque corresponding bearing fatigue life in the 128*128-level spectrum; l is the number of small intervals of the 128-level spectrum included in the interval corresponding to one level in the eight-level spectrum.

[0158] The final 8*8 level test spectrum of bearing speed and torque is the required bearing bench test spectrum. Figure 3 As shown (the torque is divided into 16 levels in the figure to facilitate image analysis).

[0159] Part II:

[0160] The feasibility of the present invention is analyzed from a theoretical perspective:

[0161] (1) In the process of load frequency extrapolation, compared with the traditional extrapolation method, the present invention adopts an extrapolation method combining Gaussian kernel function and K enhancement coefficient. The extrapolation process fully considers the distribution characteristics of the original load, combines the expansion of the distribution interval by Gaussian kernel function and the linear extrapolation of data frequency by K enhancement coefficient, and realizes that the load spectrum after extrapolation not only meets the number of cycles required by the test, but also the load boundary is reasonably expanded. The load cycles that are not collected but should exist in the extrapolated load spectrum are retained, which improves the accuracy of load extrapolation.

[0162] (2) In the process of bearing force analysis, the present invention performs segmented interpolation calculation based on traditional bearing force analysis calculation and combined with finite element software simulation results, taking into account the influence of factors such as bearing radial deformation and transmission shaft deflection deformation on bearing force in the process of bearing force analysis. The data obtained through finite element simulation improves the fit between the force analysis results and the actual bearing force, and fitting and interpolating the simulation results not only expands the limited simulation results to the full torque cycle, but also saves test time.

[0163] (3) According to the bearing life calculation formula, the speed parameter affects the reference kinematic viscosity and then affects the viscosity ratio, making a ISO The size of the coefficient changes, but the loading process of most bearing test bench spectra currently does not take the influence of the speed into consideration. Therefore, the speed-torque joint counting method adopted in the present invention can greatly improve the accuracy and authenticity of the bearing life analysis results.

[0164] In the embodiment of the present invention, step 1 pre-processes the speed torque characteristic parameter data collected by the sensor; based on this, steps 2-5 jointly count the frequency of torque and speed under each working condition according to different vehicle driving conditions, and then extrapolate the speed and torque distribution under the expected life by combining the Gaussian kernel function with the K enhancement coefficient, and express the torque and speed, two parameters that affect the reliability and durability of the reducer bearing, together in space, and then convert them into bench test spectrum by using the upward equivalent method through the Palmgren-Miner rule combined with the equal damage conversion method. In this way, after the load spectrum obtained from the test is accurately and reasonably extrapolated, the accuracy of the life analysis is greatly improved; at the same time, the three-dimensional load spectrum image is also convenient for more intuitive and efficient analysis of reliability and durability.

[0165] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A method for constructing a bearing bench test spectrum based on a speed-torque joint counting method, characterized in that: include: Step 1: Obtain sample load spectrum road operation data and preprocess it to obtain preprocessed time domain data; Step 2: Perform speed and torque joint counting on the pre-processed time domain data: record the loading frequency in each torque and speed interval by counting the pre-processed time domain data in intervals; Step 3: Construct a bearing damage model: Perform a bearing force analysis, calculate the equivalent dynamic load based on the analysis results, and then use the aISO correction coefficient that takes the speed parameter into consideration to obtain a bearing life calculation formula with the bearing torque and speed as independent variables. Combined with the joint counting results, construct a bearing damage model. Step 4: Extrapolate the preprocessed data: According to the results of the bearing damage model, the K enhancement coefficient can be obtained in combination with the bearing extrapolation target damage, and then the probability density function is obtained through kernel density estimation. The counting results obtained in step 2 are extrapolated through the Gaussian kernel function combined with the K enhancement coefficient to obtain the frequency matrix of speed and torque after extrapolation; Step 5. Prepare a multi-level program load spectrum: divide the torque and speed into eight levels according to a certain ratio. Use the Palmgren-Miner law and the S-N curve of the material to combine the frequency matrix of the torque and speed after extrapolation in step 4. Use the upward equivalent method to convert the frequency in the 128*128 level small interval into the frequency in the 8*8 level large interval according to the equal damage conversion principle, and obtain the 8*8 level bench test spectrum of the speed and torque of the bearing.

2. The method for constructing a bearing bench test spectrum based on the speed-torque joint counting method according to claim 1 is characterized in that: The preprocessing in step 1 includes outlier removal, over-threshold processing and data resampling; The removal of outliers includes: performing Fourier transform on the original data to obtain the original spectrum, filtering out the torque data under each amplitude according to the frequency through Butterworth filtering, and dividing it into main wave, secondary wave and tertiary wave; removing the tertiary wave of the amplitude data fragment less than ±1Nm, completing the preliminary filtering of invalid waves; Super-threshold processing refers to filling missing data with piecewise interpolation after outliers are removed; Data resampling includes: resampling according to the time specified frequency interpolation, extracting non-repeating time, and interpolating according to a fixed uniform sampling frequency to obtain pre-processed time domain data.

3. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In the process of speed and torque joint counting in step 2, the pre-processed time domain data is first divided into 128*128 small intervals according to the speed and torque target intervals for data counting, and the speed n(t) in the i-th speed and torque small interval is integrated over time t. The time corresponding to each small interval is The number of torque segments in a small interval is m, and the integral result is recorded as the frequency N in this interval. i , save the frequency of each interval as a matrix; 4. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In the process of constructing the bearing damage model in step 3, the calculation process of the bearing damage D is as follows: N is the frequency of data loading at each level, L nm The corresponding bearing life of each level can be known.

5. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: From step 2, the loading frequency corresponding to each speed and torque of the front bearing can be obtained; Bearing life before extrapolation: Where Cr is the basic rated dynamic load; m is the bearing life index, m = 3 for ball bearings and m = 10 / 3 for roller bearings; a1 is the reliability correction factor; P is the equivalent dynamic load; a ISO Correction factor as a function of the speed: a ISO The specific calculation formula can be found by looking up the table according to the selection of different bearings; the contamination coefficient e of the automobile reducer bearing c Select 0.8~0.9; load fatigue limit C u It can be found on the supplier's official website according to the bearing model; the lubrication condition of the bearing is expressed by the viscosity ratio κ, which is the ratio of the actual kinematic viscosity to the reference kinematic viscosity: v is the actual kinematic viscosity, and the reference kinematic viscosity v1 depends on the bearing speed and pitch diameter D pw , calculated as follows: When n<1000r / min, ν1=45000*n -0.83 *D pw -0.5 ; When n>1000r / min, v1=4500*n -0.5 *D pw -0.5 ; P is the equivalent dynamic load, which can be calculated based on the force analysis of the bearing. P=X*A a +Y*A r ; Among them A r is the radial force on the bearing, A a The axial force on the bearing is obtained by setting several sets of bearing input torques T and using the software Romax finite element simulation to obtain the fragmentary results of the bearing force. The simulation results are interpolated to obtain the fitting curve of the bearing force with respect to the input torque T. Then the test torque is brought into the fitting curve to find the corresponding force results, and the corresponding A of each level of torque can be obtained. r With A a ; X and Y are the distribution coefficients, and the equivalent dynamic load under each level of torque is obtained by combining the torque-force matrix.

6. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In step three: The calculation formula for bearing life at each torque speed is: f(T, n) means that the bearing life at each torque and speed can be simplified as a function of torque and speed; According to the final calculated bearing life at each level combined with the load frequency at each level, the pre-extrapolation bearing damage matrix at 128*128 levels can be obtained: The next step of load spectrum extrapolation can be carried out based on the bearing damage matrix of this original data.

7. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In the process of extrapolating the original data in step 4: The overall load spectrum extrapolation adopts the Gaussian kernel function combined with the K enhancement coefficient extrapolation method; The load function method is selected for the load extreme value extrapolation in the direct extrapolation process of the K enhancement coefficient. The damage-frequency distribution of the original data is combined with various distributions for fitting and comparison. The distribution function with the minimum goodness of fit for each working condition under the appropriate 90-point position confidence is selected. The distribution fit goodness test method is used to test whether the preliminary fitting distribution model is accurate. The Anderson-Darling test statistic determines the goodness of fit: This test compares AD 2 and the size of the critical value of the corresponding distribution cluster, accept or reject the null hypothesis H0 at the significance level α; F n (x) represents the probability density distribution function of the experimental data, and F(x) represents the probability density distribution function of the distribution to be compared; the smaller the AD value, the better the distribution fitting effect; Then determine the probability density function corresponding to the data distribution and obtain the inferred extreme value; The total damage D of the target life can be determined by extrapolating the target reliability based on the inferred extreme value and the requirements. 目标 , the calculation of K strengthening extrapolation coefficient is based on the following formula: Total damage of current test data D 试验 It can be obtained from the damage model in step three.

8. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In step 4, the Gaussian kernel function combined with the K enhancement coefficient extrapolation process is: Assume that the random variable x obeys an unknown probability density function Using the samples x1, ..., x drawn from n Function The distribution of is estimated; a smooth function K(*) is defined as the kernel function, h is the smooth parameter (bandwidth); the bandwidth h = ba is the length of the assumed interval, and the number of samples falling into the interval [a, b] is k, then the relationship between the kernel function K(*) and the number of samples is: According to the kernel density estimation principle, the probability density function is for: The Gaussian kernel function selected is: The corresponding two-dimensional kernel density estimate is: Here, the adaptive bandwidth is selected: Introducing an adaptive factor λ i ; In the formula, λ i is the adaptive factor; For point u i The probability density value at ; g by Calculated; The final adaptive kernel density estimate is: N is the number of sample data; By combining the K enhancement coefficient extrapolation with the Gaussian kernel function extrapolation, a set of frequency data at each level of speed and torque after extrapolation is obtained, which is expanded in both frequency and speed and torque:

9. The method for constructing a bearing bench test spectrum based on a speed-torque joint counting method according to claim 1 or 2, characterized in that: In the process of compiling the multi-level program load spectrum in step 5, since the data source corresponding to the frequency matrix of the speed-torque joint count comes from the two variables of speed and torque; According to the equivalent damage principle, the user's full life cycle load spectrum under the extrapolated road condition is reconstructed into a program loading spectrum as the loading input basis for durability or reliability tests; the 128*128-level extrapolated bearing damage obtained in step 4 is converted into a relatively simple 8*8-level spectrum using the principle of equal damage conversion. The proportional coefficients of the 8-level spectrum are graded as 1, 0.95, 0.85, 0.725, 0.575, 0.425, 0.275 and 0.125; the principle is: That is, the damage of each level in the 8*8 level spectrum is equal to the sum of the damage in the 128*128 level spectrum included, where N eq is the frequency corresponding to a set of torque speed in the 8*8 level spectrum, L eq is a set of speed torque corresponding to the bearing fatigue life in the 8*8 level spectrum; where N i is the frequency corresponding to a set of speed torque in the 128*128 level spectrum, L i is a set of speed-torque corresponding bearing fatigue life in a 128*128-level spectrum; f is the number of small intervals of the 128-level spectrum included in the interval corresponding to one level in the eight-level spectrum.

Images