Bearing test bench spectrum construction method based on rotation speed-torque joint counting method

Abstract

Disclosed in the present invention is a bearing test bench spectrum construction method based on a rotation speed-torque joint counting method, comprising: step 1, acquiring and preprocessing sample load spectrum road operation data; step 2, performing rotation speed-torque joint counting on the preprocessed data; step 3, constructing a damage model for a bearing; step 4, extrapolating original data; and step 5, compiling a multi-level program load spectrum. The present invention uses a frequency extrapolation method that combines extrapolation of a Gaussian kernel function with extrapolation of an enhancement coefficient K, so as to improve the authenticity of data after load spectrum extrapolation for the bearing. On the basis of a correspondence between the rotation speed and the fatigue life of the bearing, the present invention provides a load spectrum construction method based on rotation speed-torque joint counting, which improves accuracy of fatigue life analysis. In addition, by constructing a multi-level spectrum image of rotation speed-torque joint distribution, load spectrum analysis characteristics are intuitively reflected.

Description

A test bench bearing test spectrum construction method based on a torque and rotating speed combined counting method

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] The present application claims priority to the Chinese patent application No. 202411284736.6 filed on September 13, 2024, and entitled “A test bench bearing test spectrum construction method based on a torque and rotating speed combined counting method”, the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0003] The present application belongs to the technical field of reliability and durability test of automobile bearing parts, and relates to a test bench spectrum construction method based on rotating speed and torque combined counting. BACKGROUND

[0004] On an automobile, 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 vehicle frames, steel plate springs, coil springs, shafts and rod parts that bear alternating loads. Therefore, the importance of reliability and durability analysis of vehicle part fatigue life in the process of vehicle development cannot be underestimated.

[0005] In the process of analyzing the reliability and durability of a vehicle, the load spectrum is the top input of the whole process of durability 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-examination problems. Therefore, load spectrum analysis is an important part of vehicle durability development. The conventional method of analyzing reliability and durability needs to obtain the service life through the whole cycle load spectrum. However, the amount of data obtained by test road load spectrum collection cannot cover the loading condition of the whole life cycle. Therefore, it is necessary to use test load samples to infer the extreme load and its frequency that may occur in the whole life cycle.

[0006] As a key method for obtaining load distribution, counting method is particularly important. For general non-rotating parts or shaft parts, etc., the torque fluctuation is not related to the rotation of the parts, and the rainflow counting method can be used. For bearings and other parts whose rotating speed affects the fluctuation frequency, the rotating rainflow counting method or rotating torque histogram counting method needs to be used.

[0007] However, the above two methods cannot well reflect the joint distribution law of rotating speed and torque, and cannot intuitively and efficiently reflect the influence of rotating speed and torque on bearing fatigue life. Therefore, the present application focuses on the accuracy of the joint consideration of rotating speed and torque in the process of selecting load counting method, so that the rotating speed and torque joint counting method is used for bearing load counting, and the load extrapolation is carried out through the Gaussian kernel function combined with K reinforcement coefficient extrapolation, and the test spectrum construction method for bearing reliability and durability is constructed. SUMMARY

[0008] The purpose of the present application is to provide a test spectrum construction method for bench bearing based on torque and rotating speed joint counting method. The present application adopts the frequency extrapolation method of Gaussian kernel function combined with K reinforcement coefficient extrapolation to improve the authenticity of the data after bearing load spectrum extrapolation. Since the traditional analysis of bearing stress process lacks consideration of bearing radial deformation and other factors, the present application adopts the segmented interpolation method after software simulation to calculate the bearing stress and improve the accuracy of stress analysis results. In view of the lack of analysis of the influence of rotating speed on the construction results in the current bearing test spectrum construction process, and the rotating speed directly determines the value of fatigue life curve, the present application proposes a load spectrum construction method based on torque and rotating speed joint counting according to the corresponding relationship between rotating speed and bearing fatigue life, improves the accuracy of fatigue life analysis, and reflects the load spectrum analysis characteristics intuitively through the construction of multi-level spectrum image of rotating speed and torque joint distribution.

[0009] The technical scheme adopted by the present application is:

[0010] A bearing bench test spectrum construction method based on torque and rotating speed joint counting method, comprising:

[0011] Step one, obtaining sample load spectrum road running data and preprocessing to obtain preprocessed time domain data;

[0012] Step two, rotating speed and torque joint counting of preprocessed time domain data: recording the loading frequency under each torque and rotating speed interval by partition counting method;

[0013] Step three, constructing the damage model of the bearing: performing bearing stress analysis, calculating the equivalent dynamic load according to the analysis results, considering the aISO correction coefficient of rotating speed parameter, obtaining the bearing life calculation formula with torque and rotating speed as independent variables, and constructing the damage model of the bearing combined with the joint counting results;

[0014] Step four, extrapolation of the pretreated data: according to the results of the bearing damage model, the target damage of the bearing is extrapolated, the K reinforcement coefficient is obtained, and the probability density function is obtained through the kernel density estimation. After the counting results obtained in step two are extrapolated through the Gaussian kernel function combined with the K reinforcement coefficient, the frequency matrix about the speed and torque after extrapolation is obtained;

[0015] Step five, preparation of a multi-level program load spectrum: the torque and speed are each divided into eight levels according to a certain proportion, and the frequency matrix of the torque and speed after extrapolation in step four is used to obtain the bearing 8*8 level bench test spectrum of the speed and torque by using the Palmgren-Miner rule combined with the S-N curve of the material and adopting the upward equivalent method to convert the frequency in the 128*128 level small area into the frequency on the 8*8 level large area according to the equal damage conversion principle.

[0016] Optionally, the pretreatment in step one includes outlier rejection, threshold processing, and data resampling;

[0017] The outlier rejection includes: Fourier transform is performed on the original data to obtain an original frequency spectrum, the torque data at each amplitude is filtered out according to the frequency size through Butterworth filtering, and the third wave of the amplitude data segment less than ±1 Nm is removed to complete the preliminary filtering of invalid waves;

[0018] The threshold processing refers to filling the missing data by using piecewise interpolation after the outlier rejection;

[0019] The data resampling includes: resampling by time specified frequency interpolation, extracting non-repeated time, and interpolating to obtain pretreated time domain data according to a fixed uniform sampling frequency.

[0020] Optionally, in the speed and torque joint counting process in step two, the pretreated time domain data is divided into 128*128 small area data counting according to the speed and torque target interval, the speed n(t) in the i-th speed and torque small area is integrated at time t, and each small area corresponds to time The number of torque segments in the small area is m, and the integration result is recorded as the frequency N of the interval. i The frequencies of each interval are saved as a matrix.

[0021] Optionally, in the process of constructing the bearing damage model in step three, the bearing damage D is calculated as follows:

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

[0023] Optionally, the loading frequency corresponding to the torque of each level of the outer bearing before extrapolation is obtained by step two;

[0024] The outer bearing extrapolates the life before:

[0025] 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 coefficient; P is the equivalent dynamic load; a ISO The correction coefficient is a function of the speed:

[0026] a ISO The specific calculation formula can be known according to the selected different bearings; the pollution coefficient e c of the automobile reducer bearing is selected as 0.8-0.9; the load fatigue limit C u can be obtained according to the bearing model on the supplier's website; the lubrication condition of the bearing is represented by the viscosity ratio κ, that is, the ratio of the actual kinematic viscosity and the reference kinematic viscosity:

[0027] v is the actual kinematic viscosity, and the reference kinematic viscosity v1 depends on the bearing speed and the pitch diameter D pw , and is calculated as follows:

[0028] When n < 1000 r / min, v1 = 45000*n -0.83 *D pw -0.5 ;

[0029] When n > 1000 r / min, v1 = 4500*n -0.5 *D pw -0.5 ;

[0030] P is the equivalent dynamic load, which can be calculated according to the force analysis of the bearing P = X*A a +Y*A r ;

[0031] Where A r is the radial force borne by the bearing, and A a is the axial force borne by the bearing; by setting several groups of bearing input torque T, the bearing stress segment result is obtained by using the finite element simulation software Romax, the fitting curve of the simulation result about the input torque T is obtained by interpolation, and the corresponding stress result is obtained by bringing the test torque into the fitting curve, that is, the A r and A a corresponding to each level of torque are obtained; X and Y are distribution coefficients, and the equivalent dynamic load under each level of torque is obtained by combining the torque-stress matrix.

[0032] Optionally, in step three:

[0033] The bearing life calculation formula under each torque and speed is:

[0034] f(T,n) that is, the bearing life under each torque and speed can be simplified as a function of torque and speed.

[0035] According to the final calculation of the bearing life of each level combined with the load frequency of each level, the extrapolated bearing damage matrix under 128*128 levels can be obtained:

[0036] According to the bearing damage matrix of this original data, the next step of load spectrum extrapolation can be carried out.

[0037] Optionally, in step four of the original data extrapolation process:

[0038] The overall load spectrum extrapolation adopts the method of Gaussian kernel function combined with K reinforcement coefficient extrapolation;

[0039] The load extreme value extrapolation of the direct extrapolation process of K reinforcement coefficient selects the load function method, compares and selects the distribution function with the smallest fitting goodness of each working condition under the 90% position reliability of the original data damage-frequency distribution combined with various distributions, and tests whether the initial fitting distribution model is accurate through the distribution fitting goodness test method. The Anderson-Darling test statistic determines the fitting goodness:

[0040] This test method compares the AD 2 value of the corresponding distribution cluster critical value, and at the significance level α, accepts or rejects the original hypothesis H0; F n (x) represents the probability density distribution function of experimental data, and F(x) represents the probability density distribution function of the distribution to be compared; wherein the smaller the AD value, the better the fitting effect of the distribution;

[0041] Further determine the data distribution corresponding probability density function, obtain the extrapolated extreme value;

[0042] According to the extrapolated extreme value combined with the required extrapolation target reliability, the target life total damage D 目标 is determined.

[0043] The total damage D 试验 of the current test data is obtained from the damage model in step three.

[0044] Optionally, in step four, the Gaussian kernel function combined with K reinforcement coefficient extrapolation process is:

[0045] Let x be a random variable with an unknown probability density function A sample x1,...,x n is drawn from it The distribution of the function

[0046] The probability density function is estimated from the kernel density estimation principle as

[0047] The Gaussian kernel function is chosen as

[0048] The corresponding two-dimensional kernel density estimation is

[0049] The adaptive bandwidth is chosen as

[0050] An adaptive factor λ i is introduced

[0051] where λ i is the adaptive factor is the probability density value at point u i

[0052] g is calculated as

[0053]

[0054] The final adaptive kernel density estimation is

[0055] N is the number of sample data

[0056] After combining the K reinforcement coefficient extrapolation and the Gaussian kernel function extrapolation, a set of expanded frequency data at each level of speed torque under the frequency and speed torque is obtained

[0057] Optionally, in the process of preparing a multi-stage program load spectrum in step five, the frequency matrix corresponding to the speed-torque joint count data comes from two variables of speed and torque

[0058] ​​According to the equivalent damage principle, the user's full life cycle load spectrum under the extrapolated road working condition is reconstructed as the program load spectrum as the loading input basis of the durability or reliability test; the bearing damage of 128*128 levels obtained through step four is converted into a more simple 8*8 level spectrum by using the equivalent damage conversion principle, and the proportional coefficients of the 8 level spectrum are graded according to 1, 0.95, 0.85, 0.725, 0.575, 0.425, 0.275 and 0.125; the principle is that:

[0059] 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, wherein N eq is the frequency corresponding to a group of torque and speed in the 8*8 level spectrum, L eq is the bearing fatigue life corresponding to a group of speed and torque in the 8*8 level spectrum; wherein N i is the frequency corresponding to a group of torque and speed in the 128*128 level spectrum, L i is the bearing fatigue life corresponding to a group of speed and torque in the 128*128 level spectrum; and 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.

[0060] The beneficial effects of the present application are:

[0061] (1) In the load frequency extrapolation process, compared with the traditional extrapolation method, the present application adopts the extrapolation method combining the Gaussian kernel function and the K reinforcement coefficient, the distribution characteristics of the original load are fully considered in the extrapolation process, the respective advantages of the Gaussian kernel function for the expansion of the distribution interval and the K reinforcement coefficient for the linear extrapolation of the data frequency are combined, the load spectrum after extrapolation not only meets the cycle number required by the test, but also the load boundary is reasonably expanded. The load cycles that exist in the extrapolated load spectrum but should exist are retained, and the accuracy of load extrapolation is improved.

[0062] (2) In the bearing stress analysis process, based on the traditional bearing stress analysis calculation, the segmented interpolation calculation is carried out combined with the simulation results of the finite element software, the influence of the bearing radial deformation and the deflection deformation of the transmission shaft on the bearing stress in the bearing stress analysis process is considered. The data obtained through finite element simulation improves the fitting degree of the stress analysis results and the real bearing stress, and the fitting interpolation of the simulation results expands the limited simulation results to the full torque period while saving the test time.

[0063] (3) According to the bearing life calculation formula, the speed parameter affects the reference kinematic viscosity, and then affects the viscosity ratio, so that a ISOThe coefficient size is changed, but the current most bearing test bench spectrum loading process does not consider the influence of the rotating speed, so the rotating speed torque joint counting method adopted in the application can greatly improve the accuracy and authenticity of the bearing life analysis results. BRIEF DESCRIPTION OF DRAWINGS

[0064] The accompanying drawings are included to provide a further understanding of the application, and are incorporated in and constitute a part of the specification, illustrate embodiments of the application, and together with the description serve to explain the principles of the application. In the drawings:

[0065] Fig. 1 is a flow chart of the multi-stage test bench spectrum construction method based on the rotating speed torque joint counting method of the application;

[0066] Fig. 2 is a multi-stage test bench spectrum construction schematic diagram of the application, recording and saving each Ni and the corresponding torque speed as a frequency matrix;

[0067] Fig. 3 is the final obtained 8*8 stage test spectrum about bearing rotating speed torque, i.e. the required bearing test bench test spectrum (the torque is divided into 16 stages in the figure to facilitate image analysis). DETAILED DESCRIPTION

[0068] The application will be described in detail below in combination with the drawings and specific embodiments.

[0069] First part:

[0070] Explain the complete steps of the method of the application.

[0071] In combination with Fig. 1, the multi-stage test bench spectrum construction method based on the rotating speed torque joint counting method proposed by the application, the specific embodiment mainly includes the following steps:

[0072] Step 1, obtaining sample load spectrum road running data and preprocessing: the original data directly collected by the sensor may appear phenomena such as mutation, noise and burr due to the influence of driving environment, interference signal and buildings, and needs to be preprocessed by operations such as outlier rejection.

[0073] The outlier rejection is based on the Dalai criterion, and 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 load basically does not cause fatigue damage but greatly increases the calculation amount, forming data redundancy, i.e. high frequency noise. Therefore, this kind of load needs to be removed, by doing Fourier transform on the original data, the original frequency spectrum can be obtained, by Bartlett filtering, the torque data under each amplitude can be screened out according to the frequency size, divided into primary wave, secondary wave and tertiary wave. The tertiary wave with small amplitude and high frequency (amplitude data segment less than ±1Nm) is removed, and the invalid wave can be preliminarily filtered out.

[0074] Super-threshold processing refers to filling in missing data by piecewise interpolation after removing abnormal data outside the torque speed requirement interval.

[0075] The reason for data resampling is that the sampling frequencies of each signal are inconsistent, and the data points of each signal cannot be one-to-one corresponding in time. Therefore, in order to be able to perform joint distribution statistics of each signal, all data are resampled to unify the sampling frequency. Resampling is performed by interpolation at a specified frequency of 10 Hz, extracting non-repeated time, and interpolating at a fixed uniform sampling frequency.

[0076] Step 2, joint speed and torque counting of original data: the preprocessed time domain data are counted by interval counting to count the data under each torque and speed (generally, torque and speed are divided into 128 levels), and the speed in each torque and speed interval is integrated in time:

[0077] The frequency N under each torque and speed interval is recorded i . Each Ni is recorded and saved as a frequency matrix corresponding to torque and speed. The result is shown in Figure 2.

[0078] Step 3, construction of damage model: calculate the bearing damage at each level of speed and torque during the damage process, according to the bearing damage at each level of data loading frequency N divided by the corresponding bearing life L at each level nm It can be known that:

[0079] According to step 2, the loading frequency corresponding to each level of speed and torque of the bearing before extrapolation can be obtained.

[0080] The life of the bearing before extrapolation can be known from "GB_T6391-2010 Rolling Bearing Rated Dynamic Load and Rated Life":

[0081] 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 coefficient; P is the equivalent dynamic load; a ISO The correction coefficient is a function of speed, and also related to the pollution coefficient of the bearing, the fatigue limit of the load and the lubrication condition:

[0082] a ISO The specific calculation formula of a can be known by consulting the table according to the selection of different bearings. According to the reference data: the pollution coefficient ec of the automobile reducer bearing is generally selected as 0.8-0.9; the load fatigue limit Cu can be obtained from the supplier's website according to the bearing model; the lubrication condition of the bearing is represented by the viscosity ratio κ, which is the ratio of the actual kinematic viscosity and the reference kinematic viscosity:

[0083] The actual kinematic viscosity v is related to the lubricating oil characteristics, the rotational speed, and the working temperature during the operation of the bearing. The lubricating oil characteristics are the viscosity grade of the lubricating oil, which has a special standard in ISO. The actual viscosity of the lubricating oil can be obtained by table lookup or calculation according to the viscosity grade of the lubricating oil and the working temperature.

[0084] The reference kinematic viscosity v1 depends on the bearing rotational speed and the pitch diameter D pw , and is calculated as follows:

[0085] When n < 1000 r / min, v1 = 45000 * n -0.83 *D pw -0.5 ;

[0086] When n > 1000 r / min, v1 = 4500 * n -0.5 *D pw -0.5 ;

[0087] P is the equivalent dynamic load, which can be calculated according to the force analysis of the bearing P = X * A a + Y * A r ;

[0088] where A r is the radial force borne by the bearing, and A a is the axial force borne by the bearing. By setting several groups of bearings to input the torque T, the bearing force segment results are obtained by using the finite element simulation software Romax, the simulation results are interpolated to obtain the fitting curve of the bearing force with respect to the input torque T, and then the test torque is brought into the fitting curve to find the corresponding force result, so that the A r and A a corresponding to each level of torque can be obtained; X and Y are distribution coefficients, which can be obtained by referring to the national standard GB_T6391-2010, and the equivalent dynamic load under each level of torque can be obtained by combining the torque-force matrix.

[0089] Finally, the bearing life calculation formula under each torque and speed can be simplified as:

[0090] According to the finally calculated bearing life of each level, the extrapolated bearing damage matrix under 128*128 levels can be obtained:

[0091] According to this original data bearing damage matrix, the next step of load spectrum extrapolation can be performed.

[0092] Step 4, extrapolate the original data: use the Gaussian kernel function combined with the K reinforcement coefficient extrapolation method to perform load spectrum extrapolation.

[0093] Step 4.1: K-stiffness coefficient extrapolation:

[0094] Since the K-stiffness coefficient extrapolation needs to be obtained according to the damage before and after the extrapolation, as follows:

[0095] That is, the extrapolation multiple = total damage of target life / total damage of current test data.

[0096] For the acquisition of target damage D 目标 , load extreme value extrapolation is needed:

[0097] Load extreme value extrapolation selects load function method, combines damage-frequency distribution of original data with various distributions for fitting comparison, selects the distribution function with the smallest fitting goodness of each working condition under the confidence of 90% position, and can test whether the initial fitting distribution model is accurate through distribution fitting goodness test method. Anderson-Darling test statistic determines the goodness of fitting:

[0098] This test method compares the size of AD 2 and the critical value of the corresponding distribution cluster, and accepts or rejects the original hypothesis H0, F n (x) represents the probability density distribution function of experimental data, and F(x) represents the probability density distribution function to be compared; wherein the smaller the AD value, the better the fitting effect of the distribution.

[0099] Here, the user unit mileage damage is taken 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 . Then, combined with the target life mileage

[0100] L 目标 , the total damage target D 目标 of vehicle life cycle is calculated = d 90 *L 目标 .

[0101] According to step 3, D 试验 can be calculated.

[0102] Substitute the total damage target D 目标 of vehicle life cycle with the test damage D 试验 , and the K-stiffness coefficient can be obtained.

[0103] Step 4.2: Gaussian kernel function extrapolation process:

[0104] Assume that the distribution of the corresponding value x of each level of rotational speed and torque obeys an unknown probability density function The sample x1,...,x n The distribution of the function A smoothing function K(*) is defined as the kernel function, and h is the smoothing parameter (bandwidth). The bandwidth h = b-a is the length of the interval, and 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:

[0105] According to the kernel density estimation principle, the probability density function is:

[0106] The selected Gaussian kernel function is:

[0107] The corresponding two-dimensional kernel density estimation is:

[0108] For the other factor affecting the probability density function under the kernel density estimation principle, the bandwidth, there is an optimal bandwidth selection. The larger the bandwidth coefficient, the smoother the estimated density function curve, but it may lead to underfitting, i.e. hiding some characteristics of the original sample data; the smaller the bandwidth coefficient, it may lead to overfitting of the estimated density function curve, resulting in a larger variance.

[0109] Here, the adaptive bandwidth is selected in the form of:

[0110] An adaptive factor λ i is introduced.

[0111] In the formula, λ i is the adaptive factor; is the probability density value at point u i ; g is calculated by

[0112] .

[0113] The final adaptive kernel density estimation is:

[0114] N is the number of sample data. Replace the estimated rotational speed and torque with x and y in the above formula to obtain the adaptive kernel density estimation result.

[0115] After introducing the adaptive factor, the bandwidth can change with the data, and it is smaller in data-intensive places and larger in data-sparse places. The probability density obtained by comparing the traditional fixed bandwidth is closer to the actual distribution.

[0116] Through the combination of K reinforcement coefficient extrapolation and Gaussian kernel function extrapolation, a set of extrapolated frequency data of each rotational speed torque level after the expansion of frequency and rotational speed torque can be obtained:

[0117] Step 5, preparing a multi-level program load spectrum: dividing the torque and rotational speed into eight levels according to a certain proportion, and converting the frequency in the 128*128 level small area into the frequency in the 8*8 level large area by using the Palmgren-Miner rule in combination with the damage equivalent principle and the upward equivalent method, and the 8-level proportion coefficients are graded as 1, 0.95, 0.85, 0.725, 0.575, 0.425, 0.275 and 0.125. The principle is that:

[0118] That is, the damage of each level in the 8*8 level spectrum is equal to the sum of the damages in the 128*128 level spectrum, wherein N e q is the frequency corresponding to a set of torque and rotational speed in the 8*8 level spectrum, L eq is the bearing fatigue life corresponding to a set of rotational speed and torque in the 8*8 level spectrum; wherein N i is the frequency corresponding to a set of rotational speed and torque in the 128*128 level spectrum, L i is the bearing fatigue life corresponding to a set of rotational speed and torque in the 128*128 level spectrum; and l is the number of 128 level small areas included in the interval corresponding to each level in the eight level spectrum.

[0119] The final 8*8 level test spectrum about the bearing rotational speed torque is the required bearing bench test spectrum. As shown in FIG. 3 (in the figure, the torque is divided into 16 levels to facilitate image analysis).

[0120] Second part:

[0121] Theoretical analysis of the feasibility of the present application:

[0122] (1) In the load frequency extrapolation process, compared with the traditional extrapolation method, the present application adopts the extrapolation method combining Gaussian kernel function and K reinforcement coefficient, and the distribution characteristics of the original load are fully considered in the extrapolation process, the advantages of Gaussian kernel function in expanding the distribution interval and K reinforcement coefficient in linearly extrapolating the data frequency are combined, and the extrapolated load spectrum not only meets the required cycle number, but also the load boundary is reasonably expanded. The load spectrum after extrapolation retains the load cycles that are not collected but should exist, and the accuracy of load extrapolation is improved.

[0123] (2) In the bearing stress analysis process, the application is based on the traditional bearing stress analysis calculation, combined with the segmented interpolation calculation of the finite element software simulation results, considering the influence of bearing radial deformation and deflection deformation of the transmission shaft on the bearing stress in the bearing stress analysis process. The data obtained by finite element simulation improves the fitting degree of the stress analysis results and the real bearing stress, and the fitting interpolation of the simulation results expands the limited simulation results to the full torque period while saving the test time.

[0124] (3) According to the bearing life calculation formula, the speed parameter affects the reference kinematic viscosity, and then affects the viscosity ratio, so that the a ISO coefficient changes when calculating the life, but the loading process of most bearing test benches is not considered the influence of speed, so the speed and torque joint counting method in the application can greatly improve the accuracy and authenticity of the bearing life analysis results.

[0125] In the embodiment of the application, step 1 pre-processes the speed and 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 obtain the speed and torque distribution under the expected life through the Gaussian kernel function combined with the K reinforcement coefficient extrapolation, which represents the two parameters of torque and speed that affect the reliability and durability of the reducer bearing in space, and then converts the test bench test spectrum through the Palmgren-Miner rule combined with the equal damage conversion method using the upward equivalent method. After accurately and reasonably extrapolating the test obtained load spectrum, 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.

[0126] The above only describes the embodiments of the application and is not used to limit the application. The application can have various changes and variations for those skilled in the art. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the application shall be included in the scope of the claims of the application.

Claims

1. A bearing rig test spectrum construction method based on a torque speed combined counting method, characterized in that, The application relates to a method for preparing a multi-level program load spectrum of a bearing. Step one: obtaining sample load spectrum road operation data and preprocessing the data to obtain preprocessed time domain data; Step two: jointly counting the torque and the speed of the preprocessed time domain data: the preprocessed time domain data are recorded in each torque and speed interval by means of interval counting to record the loading frequency in each torque and speed interval; Step three, build the damage model of bearing: do bearing stress analysis, according to the analysis results, calculate the equivalent dynamic load, and consider the speed parameter a ISO Correction coefficient, get the bearing life calculation formula with torque speed as independent variable, combined with the joint counting results, build the damage model of bearing; Step four: extrapolating the preprocessed data: according to the results of the bearing damage model, the K reinforcement coefficient is obtained by cooperating with the bearing extrapolation target damage, and the probability density function is obtained by means of kernel density estimation; the counting results obtained in step two are extrapolated by means of a Gaussian kernel function combined with the K reinforcement coefficient, and a frequency matrix about the torque and the speed after extrapolation is obtained; Step five: preparing a multi-level program load spectrum: the torque and the speed are each divided into eight levels according to a certain proportion; the frequency matrix of the torque and the speed after extrapolation in step four is converted into the frequency on a large interval of 8*8 levels in the small interval of 128*128 levels by means of the Palmgren-Miner rule combined with the S-N curve of the material and by means of an upward equivalent method according to the equal damage conversion principle, so that the bearing 8*8 level bench test spectrum of the torque and the speed is obtained.

2. The bearing rig test spectrum construction method based on the combined torque and rotational speed counting method according to claim 1, characterized in that, The preprocessing in step one comprises abnormal value elimination, threshold value processing and data resampling; The abnormal value elimination comprises: Fourier transforming the original data to obtain an original frequency spectrum, screening the torque data under each amplitude according to the frequency size by means of Bartlett filtering, and dividing the torque data into primary waves, secondary waves and tertiary waves; the tertiary waves of the amplitude data segments smaller than plus or minus 1Nm are removed to complete the preliminary filtering of invalid waves; The threshold value processing refers to filling the missing data by means of piecewise interpolation after the abnormal value elimination; The data resampling comprises: resampling by means of time specified frequency interpolation, extracting non-repeated time, and interpolating to obtain the preprocessed time domain data according to a fixed uniform sampling frequency.

3. The method of claim 1 or 2, wherein, In the step two torque speed joint counting process, the pretreated time domain data is first divided into 128*128 size small interval data according to the torque speed target interval, the speed n(t) in the i-th torque speed small interval is integrated on time t, and each small interval corresponds to time The number of torque segments in the small interval is m, and the integration result is recorded as the frequency N of the interval i Save the frequency of each interval as a matrix; 4. The method of claim 1 or 2, wherein, In the process of constructing the damage model of the bearing in step three, the damage D of the bearing is calculated as follows: N the frequency of data loading at each level, L nm The bearing life at each level can be known.

5. The method of claim 1 or 2, wherein, The loading frequency corresponding to each level of the torque and the speed of the bearing before extrapolation is obtained in step two; bearing outer push before life: where Cr is the basic dynamic load rating; m is the bearing life exponent, m = 3 for ball bearings and m = 10 / 3 for roller bearings; a1 is a reliability correction factor; P is the equivalent dynamic load; a ISO The correction factor is a function of rotational speed: a ISO The specific calculation formula can be known according to the bearing table selected; the pollution coefficient e of the automobile speed reducer bearing c Select 0.8-0.9; the load fatigue limit C u Can be found according to the bearing model in the supplier's website; the lubrication condition of the bearing is expressed by the viscosity ratio K, that is, the ratio of the actual kinematic viscosity and the reference kinematic viscosity: v is the actual kinematic viscosity, the reference kinematic viscosity vl depends on the bearing rotational speed and the pitch diameter D pw is calculated as follows: When n<1000r / min, v1 = 4500 * n -0.83 D pw -0.5 ; When n>1000r / min, v1 = 4500 * n -0.5 D pw -0.5 ; P is an equivalent dynamic load, which can be calculated according to the stress analysis of the bearing P = X * A a + Y * A r ; wherein A r is the radial force on the bearing, A a is the axial force on the bearing, by setting several groups of bearings input torque T, using software Romax finite element simulation calculation bearing force piece results, the simulation results interpolation to get the fitting curve of the bearing force about input torque T, and then the test torque into the fitting curve to find the corresponding stress results, that is, the corresponding A r and A a ; X and Y are distribution coefficients, combined with torque-stress matrix to get the equivalent dynamic load under each level of torque.

6. The bearing rig test spectrum construction method based on the combined torque and rotational speed counting method according to claim 5, characterized in that, In step three: The bearing life calculation formula under each torque speed is: f(T,n) indicates that the bearing life under each torque and speed can be simplified as a function of the torque and the speed; According to the final calculated bearing life of each level combined with the load frequency of each level, the extrapolated bearing damage matrix under the 128*128 level can be obtained: According to the bearing damage matrix of the original data, the next load spectrum extrapolation can be carried out.

7. The method of claim 1 or 2, wherein, In the original data extrapolation process in step four: The overall load spectrum extrapolation adopts the method of Gaussian kernel function combined with K reinforcement coefficient extrapolation; K is the load function method of load extreme value extrapolation of the direct extrapolation process, the damage-frequency distribution of the original data is fitted and compared with various distributions, the distribution function with the smallest fitting goodness of each working condition at the 90% position reliability is selected, and the fitting goodness test method is used to test whether the initial fitting distribution model is accurate, and the Anderson-Darling test statistic determines the fitting goodness: This test accepts or rejects the null hypothesis H0 at a significance level a by comparing the AD 2 value to the size of the corresponding distributional cluster critical value. n (x) represents the probability density distribution function of the experimental data, F(x) represents the probability density distribution function of the distribution to be compared; wherein the smaller the AD value, the better the fitting effect of the distribution; Then the data distribution corresponding probability density function is determined to obtain an extrapolated extreme value; The target life total damage D can be determined according to the extrapolated extreme value combined with the requirement of extrapolating the target reliability 目标 The calculation of the K reinforcement extrapolation coefficient is based on the following formula: Current test data total damage D 试验 From the damage model in step three.

8. The method of claim 1 or 2, wherein, In step four, the Gaussian kernel function combined with K reinforcement coefficient extrapolation process is: Let x be a random variable with an unknown probability density function Using the samples x1,...,x n For the function The distribution of the estimated; define a smooth function K(*) as the kernel function, h is the smooth parameter (bandwidth); bandwidth h = b-a, that is, the length of the interval, assuming that 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: The probability density function obtained according to the kernel density estimation principle is For: The Gaussian kernel function used is: The corresponding two-dimensional kernel density estimate is: The adaptive bandwidth is selected in the form of: Introducing an adaptation factor λ i ; where λ i is an adaptive factor; the probability density value at the point u i the probability density value at the point u g by The calculation result is obtained; The final adaptive kernel density estimate is: N is the number of sample data; Through the combination of K reinforcement factor extrapolation and Gaussian kernel function extrapolation, a set of extrapolated frequency data at each level of torque speed with expanded frequency and torque speed is obtained:

9. The method of claim 1 or 2, wherein, In the process of preparing the multi-level program load spectrum in step five, the frequency matrix corresponding to the speed-torque joint counting data comes from the two variables of the speed and the torque; According to the equivalent damage principle, the user's full life cycle load spectrum after extrapolation under the road working condition is reconstructed as the program load spectrum as the loading input basis of the durability or reliability test; through the 128*128 level bearing damage obtained after extrapolation in step four, the equivalent damage conversion principle is adopted to convert it into a more simple 8*8 level spectrum, and the proportion coefficients of the 8 level spectrum are graded according to 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 spectrum is equal to the sum of the damage in the included 128*128 spectrum, where N eq is the frequency corresponding to a group of torque-speed pairs in the 8*8 spectrum, L eq is the bearing fatigue life corresponding to a group of torque-speed pairs in the 8*8 spectrum; In the formula, N i,j is the frequency corresponding to a group of torque-speed in the 128*128 spectrum, L i,j is the bearing fatigue life corresponding to a group of torque-speed in the 128*128 spectrum; and l is the number of small intervals of the 128 spectrum included in the interval corresponding to a level in the eight-level spectrum.

Images