A bearing residual life prediction method, device, equipment and readable storage medium

Abstract

The application provides a bearing residual life prediction method, device and equipment and a readable storage medium, relates to the technical field of transportation, and comprises the following steps: acquiring bearing operation and maintenance data, preprocessing the bearing operation and maintenance data to obtain bearing fault data; processing the bearing fault data based on a survival curve algorithm to obtain a first reliability model; analyzing the failure rate change law of each stage of the bearing in different maintenance intervals, fitting the failure rate change law by using a first function to obtain a second reliability model; processing the first reliability model and the second reliability model by using a reliability-centered maintenance method to obtain a first residual life result and a second residual life result, and calculating the average value of the first residual life result and the second residual life result to obtain a final bearing residual life prediction result. The application has the beneficial effect that the residual life of the bearing can be effectively predicted, the service mileage is directly given, and the maintenance of the maintenance unit is effectively guided.

Description

Technical Field

[0001] The present invention relates to the field of transportation technology, specifically to the field of bearing analysis technology for electric multiple units (EMUs), and more particularly to a method, device, equipment, and readable storage medium for predicting the remaining life of a bearing. Background Art

[0002] As my country's high-speed rail mileage increases, the proportion of EMUs in rail passenger transport continues to rise, and the cost of operating and maintaining these trains is also increasing. Axlebox bearings, as key components in the axlebox assembly, bear heavy loads and operate in harsh and variable conditions. Operational and maintenance data reveals that a significant number of axlebox bearing failures occur each year, leading to train slowdowns, delays, and even temporary stops.

[0003] Due to the long-distance operation of my country's high-speed railways and the harsh environment of bearings, bearings are extremely prone to failure. Therefore, it is necessary to study the bearing status management technology and the remaining life prediction technology of high-speed train axle box bearings in order to predict the time when the axle box bearing life reaches its limit, thereby guiding the maintenance of the operation and maintenance units and improving the safety of train operation. Summary of the Invention

[0004] The present invention aims to provide a method, device, apparatus, and readable storage medium for predicting the remaining life of a bearing to improve the above-mentioned problem. To achieve the above-mentioned object, the present invention adopts the following technical solutions:

[0005] In a first aspect, the present application provides a method for predicting the remaining life of a bearing, comprising:

[0006] Acquire bearing operation and maintenance data, pre-process the bearing operation and maintenance data, and obtain bearing fault data, where the operation and maintenance data includes vehicle system alarm data and data on faults to be repaired;

[0007] The bearing failure data is processed based on the survival curve algorithm to obtain the first reliability model;

[0008] Analyze the variation pattern of the failure rate of bearings at different stages within different maintenance intervals, fit the variation pattern of the failure rate using the first function, and obtain the second reliability model;

[0009] A reliability-centered maintenance method is adopted to process the first reliability model and the second reliability model respectively to obtain the first remaining life result and the second remaining life result. The average value of the first remaining life result and the second remaining life result is then calculated to obtain the final bearing remaining life prediction result.

[0010] Preferably, the bearing operation and maintenance data is preprocessed to obtain bearing fault data, which includes:

[0011] Cleaning and reviewing the acquired bearing operation and maintenance data to obtain a first processing result, wherein the cleaning includes filling in missing data, removing abnormal data, and dimensionless quantization of data;

[0012] Analyzing the first processing result to obtain a second processing result, wherein the analysis includes performing descriptive analysis, data dynamic analysis, correlation analysis, and regression analysis on the first processing result;

[0013] The second processing result is classified and stored according to keywords to obtain bearing fault data, wherein the keywords include vehicle type and maintenance level.

[0014] Preferably, the bearing fault data is processed based on the survival curve algorithm to obtain a first reliability model, which includes:

[0015] The reliability distribution function is obtained using the survival curve algorithm, where the survival curve algorithm formula is as follows:

[0016]

[0017] Where R(t) is the reliability function, is the survival distribution function, Z (i) is the variable Z i Order statistics variables, where Z i =min(X i ,Y i ), X i is a non-negative random variable, Y i is the corresponding disturbance random variable; δ (i) The parameter to determine whether it is truncated data is the truncation indicator function. If it is truncated data, then δ (i) =0; if it is the real life span, then δ (i) =1.

[0018] Based on the reliability distribution function and the functional relationship corresponding to the bearing reliability, the failure probability empirical distribution function is obtained;

[0019] The failure probability empirical distribution function is fitted using a mathematical model to obtain a first reliability model, wherein the fitting process includes selecting the distribution type of the bearing, estimating the distribution parameters, and testing the distribution type.

[0020] Preferably, the fitting process includes estimating the distribution parameters of the bearing, including:

[0021] Parameter estimation is performed using applied mathematical software. The method used for parameter estimation is the Weibull probability paper method, where the calculation formula of the Weibull probability paper is as follows:

[0022]

[0023] In the formula, F(t) is a Weibull distribution cumulative failure probability function, and a failure probability distribution function; λ is a size parameter, and α is a shape parameter, and t is a mileage / time. The above formula is transformed, and logarithmic operation is simultaneously taken on both sides of the equal sign to linearize calculation, and a linear result is obtained:

[0024]

[0025] In the formula, is a failure probability value observed at t i mileage, The definition of is a ratio of a number of products that fail before t i to a number of products that start participating in the test;

[0026] According to a least square linear regression method, the linear result is calculated to obtain estimated values of the shape parameter and the size parameter of the Weibull distribution.

[0027] Preferably, the fitting process comprises a test on a distribution type of the bearing, and the test comprises:

[0028] A theoretical distribution method is used to establish a distribution function of the bearing;

[0029] A K-S test is used to test the distribution function to obtain a maximum deviation value, and a test formula is as follows:

[0030] D n = sup |F n (t) - F0(t)|

[0031] In the formula, sup(·) is max(·), F n (t) is an empirical distribution function with a sample capacity of n, F0(t) is a theoretical distribution function, and D n is the maximum deviation value;

[0032] The maximum deviation value is compared with a preset critical value to obtain a final test result.

[0033] Preferably, the analysis of the change rule of the failure rate of each stage of the bearing in different maintenance intervals comprises:

[0034] A failure rate and a mileage in a unit time of work of the bearing are determined;

[0035] A bearing failure rate curve is obtained according to the failure rate and the mileage, and the failure rate curve comprises an early failure period, a product service life period, and a wear-out failure period;

[0036] A failure rate change rule is judged based on a slope in the failure curve.

[0037] Preferably, obtaining the first remaining life result includes:

[0038] Obtain the evolution curve of bearing failure law;

[0039] Based on the first reliability model, the bearing's working environment and pre-set safety conditions are considered, and a reliability-centered maintenance method is used to preliminarily predict the remaining life of the bearing.

[0040] Determine the bearing risk control requirements. When the bearing operation risk reaches the control requirements, re-predict the initially predicted remaining life of the bearing to obtain a first remaining life result.

[0041] In a second aspect, the present application further provides a bearing remaining life prediction device, comprising an acquisition module, a processing module, a fitting module, and a calculation module, wherein:

[0042] Acquisition module: used to acquire bearing operation and maintenance data, pre-process the bearing operation and maintenance data, and obtain bearing fault data, where the operation and maintenance data includes vehicle system alarm data and data on faults to be repaired;

[0043] Processing module: used to process bearing fault data based on the survival curve algorithm to obtain a first reliability model;

[0044] Fitting module: used to analyze the failure rate variation pattern of bearings at different stages within different maintenance intervals, and use the first function to fit the failure rate variation pattern to obtain the second reliability model;

[0045] Calculation module: used to adopt a reliability-centered maintenance method to process the first reliability model and the second reliability model respectively, obtain the first remaining life result and the second remaining life result, and calculate the average value of the first remaining life result and the second remaining life result to obtain the final bearing remaining life prediction result.

[0046] In a third aspect, the present application further provides a bearing remaining life prediction device, comprising:

[0047] Memory for storing computer programs;

[0048] A processor is used to implement the steps of the bearing remaining life prediction method when executing the computer program.

[0049] In a fourth aspect, the present application further provides a readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned method for predicting the remaining life of a bearing are implemented.

[0050] The beneficial effects of the present invention are:

[0051] The present invention pre-processes the bearing operation and maintenance data to improve the accuracy of data analysis and processing, avoids the defect of large data errors, and reduces the pressure of subsequent work.

[0052] The present invention constructs two reliability mathematical models. The two reliability mathematical models are based on different considerations, but both aim to improve the working safety and reliability of bearings, and are based on the premise of effectively predicting the remaining service life of axle box bearings and optimizing maintenance procedures. At the same time, the algorithm can intuitively give the remaining service life of the bearings in units of mileage, which can effectively guide the maintenance of maintenance units.

[0053] The present invention adopts a maintenance method centered on reliability to determine reasonable risk control requirement indicators and reliability indicators to achieve the remaining service life prediction of the bearing and the optimization of the maintenance schedule.

[0054] The present invention adopts the KS test method, which has no special requirements for sample size and can test a given deviation. It is suitable for most scenarios and makes the test results more accurate.

[0055] The bearings in the present invention are key components of bogies. Predicting their remaining service life can not only prevent accidents, but also provide an important basis for formulating policies such as repair, improvement and prevention of train accidents.

[0056] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the embodiments of the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 Schematic diagram of the flow of the method for predicting the remaining life of a bearing according to an embodiment of the present invention;

[0059] Figure 2 Schematic diagram of the structure of the bearing remaining life prediction device according to an embodiment of the present invention;

[0060] Figure 3 Schematic diagram of the structure of the bearing remaining life prediction device according to an embodiment of the present invention;

[0061] Figure 4A product failure rate curve diagram for bearing residual life prediction described in the embodiments of the present application;

[0062] Figure 5 An active risk control diagram for bearing residual life prediction described in the embodiments of the present application;

[0063] Figure 6 A maintenance schedule diagram for bearing residual life prediction described in the embodiments of the present application.

[0064] In the figure: 701, acquisition module; 7011, cleaning unit; 7012, analysis unit; 7013, storage unit; 702, processing module; 7021, first obtaining unit; 7022, second obtaining unit; 7023, fitting unit; 70231, estimation unit; 70232, calculation unit; 70233, function establishing unit; 70234, verification unit; 70235, comparison unit; 703, fitting module; 7031, determination unit; 7032, third obtaining unit; 7033, judgment unit; 704, calculation module; 7041, acquisition unit; 7042, first prediction unit; 7043, second prediction unit; 800, bearing residual life prediction device; 801, processor; 802, memory; 803, multimedia assembly; 804, I / O interface; 805, communication assembly. DETAILED DESCRIPTION

[0065] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present application.

[0066] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0067] Embodiment 1:

[0068] The embodiment provides a bearing residual life prediction method.

[0069] See also Figure 1 , the figure shows that the method includes step S100, step S200, step S300 and step S400.

[0070] S100 , obtaining bearing operation and maintenance data, pre-processing the bearing operation and maintenance data, and obtaining bearing fault data, wherein the operation and maintenance data includes vehicle-mounted system alarm data and data of faults to be repaired.

[0071] Specifically, the massive amount of axlebox bearing operation and maintenance data (mainly including vehicle system alarm data and axlebox bearing fault data discovered by maintenance personnel during inspections) has multiple problems such as large data volume, heterogeneity, multi-dimensionality, and multi-scale. It needs to be processed before subsequent analysis. If the algorithm or model is directly built on the original operation and maintenance data set, it is likely to result in inaccurate analysis results or large errors. Therefore, in order to reduce the pressure of subsequent work, it is necessary to process and analyze the bearing operation and maintenance data. The main contents are as follows:

[0072] It can be understood that step S100 includes S101, S102 and S103, wherein:

[0073] S101. Clean and review the acquired bearing operation and maintenance data to obtain a first processing result, wherein the cleaning includes filling in missing data, removing abnormal data, and dimensionless data quantization; the data review is aimed at the validity of the data and aims to improve the accuracy of the data.

[0074] S102, analyzing the first processing result to obtain a second processing result, wherein the analysis includes performing descriptive analysis, data dynamic analysis, correlation analysis, and regression analysis on the first processing result;

[0075] S103 , classifying and storing the second processing result according to keywords to obtain bearing fault data, wherein the keywords include vehicle type and maintenance level.

[0076] It should be noted that the bearing operation and maintenance data that is finally cleaned and analyzed is partitioned and stored according to keywords such as vehicle model and maintenance level to facilitate subsequent in-depth processing of the data.

[0077] In this embodiment, the reliability model needs to be constructed from two aspects. The first is to establish a corresponding mathematical model based on the evolution law of the axle box bearing reliability. After data processing of the failure data, the evolution law of the reliability performance of the axle box bearing within the current service mileage is analyzed, and at the same time, an appropriate distribution model is selected to fit the evolution law of the reliability performance of the axle box bearing. The second is to consider the impact of assembly problems caused by maintenance on the reliability performance of the axle box bearing, that is, the impact on the remaining life, based on the maintenance mileage of the axle box bearing. The main method is to establish a failure rate mathematical model by analyzing the failure rate change curve of the axle box bearing in different maintenance intervals. Then, based on the functional relationship between failure rate and reliability, a reliability mathematical model of the axle box bearing is established. In this embodiment, the first mathematical model is the first reliability model, and the second mathematical model is the second reliability model.

[0078] S200 , processing bearing fault data based on a survival curve algorithm to obtain a first reliability model.

[0079] It should be noted that due to the short service life of high-speed EMUs, the bearing failure data obtained is not full life cycle data, but data truncated at a certain point in time. Therefore, when studying the evolution of the service performance of axlebox bearings, it is necessary to apply relevant theories to process the failure data. The specific contents are as follows:

[0080] It can be understood that the step S200 includes S201, S202 and S203, wherein:

[0081] S201. A survival curve algorithm is used to obtain a reliability distribution function, wherein the survival curve algorithm formula is as follows:

[0082]

[0083] Where R(t) is the reliability function, is the survival distribution function, Z (i) is the variable Z i Order statistics variables, where Z i =min(X i ,Y i ), X i is a non-negative random variable, Y i is the corresponding disturbance random variable; δ (i) The parameter to determine whether it is truncated data is the truncation indicator function. If it is truncated data, then δ (i) =0; if it is the real life span, then δ (i) =1.

[0084] It should be noted that before using the survival curve, the truncated data needs to be processed to obtain the survival distribution probability of the bearing, also known as the product limit (PL) estimate. The calculation formula and assumptions involved are as follows:

[0085] Let X1, X2, ..., X n are non-negative random variables, independent of each other and with a common survival function S(t), Y1, Y2, ..., Y n is the corresponding interference random variable, that is:

[0086]

[0087] Where Z (1) ,Z (2) ,...,Z (n) For Z1, Z2, ..., Z n The order statistic of δ (1) ,δ (2) ,...,δ (n) For Z (1) ,Z (2) ,...,Z (n) The corresponding truncation indicator function, for simplicity, is considered as Z (1) ,Z (2) ,...,Z (n) is a constant, and

[0088]

[0089] Since the lifespan is longer than Z (i-1) There are n-i+1 individuals in total. If Z (i) is the real lifespan (δ (i) =1), it means that the life span is longer than Z (i-1) There is exactly one individual among the n-i+1 individuals in (Z (i-1) , Z (i) ] died; on the contrary, if Z (i) For truncated data (δ (i) =0), it means that the life span is longer than Z (i-1) None of the n-i+1 individuals is in (Z (i-1) , Z (i) ] died. Thus P(X>Z (i) |X>Z (i-1) A reasonable estimate of ) is:

[0090]

[0091] Where, It is an estimate of the survival probability in this interval. Assuming that there is no death, the survival probability is 1. If there is a death, the failure probability is 1 / (ni-1), and the survival probability becomes (1-1 / (ni-1)).

[0092] Arranging the above formula, the formula is:

[0093]

[0094] Where R(t) is the reliability function, is the survival distribution function, Z (i) is the variable Z i Order statistics variables, where Z i =min(X i ,Y i ), X i is a non-negative random variable, Y i is the corresponding disturbance random variable; δ (i) The parameter to determine whether it is truncated data is the truncation indicator function. If it is truncated data, then δ (i) =0; if it is the real life span, then δ (i) =1.

[0095] S202, obtaining a failure probability empirical distribution function based on the reliability distribution function and a functional relationship corresponding to the bearing reliability;

[0096] The functional relationship between reliability and failure probability is:

[0097] F(t)=1-R(t) or R(t)=1-F(t) (5)

[0098] Where F(t) is the empirical distribution function of failure probability, and R(t) is the reliability function.

[0099] S203 , fitting the failure probability empirical distribution function using a mathematical model to obtain a first reliability model, wherein the fitting process includes selecting a distribution type of the bearing, estimating distribution parameters, and testing the distribution type.

[0100] It should be noted that the failure probability evolution law of the axle box bearing can be known from the obtained failure probability empirical distribution function F(t). Then it is necessary to select a suitable mathematical model to fit the failure probability of the axle box bearing. This mainly includes the selection of distribution type, estimation of distribution parameters and verification of distribution type, so as to obtain the failure probability distribution model. The fitting process is as follows:

[0101] In this embodiment, the distribution type of the bearing is selected as follows: Common distribution forms of mechanical parts include normal distribution, exponential distribution, and Weibull distribution. In this embodiment, it is assumed that the axle box bearing conforms to the normal distribution, lognormal distribution, and Weibull distribution.

[0102] In this embodiment, the distribution parameters of the bearing are estimated as follows: the parameter estimation methods include graphical estimation method, least squares method, maximum likelihood method and moment estimation method. For normal distribution and lognormal distribution, the Gauss-Newton method is used to estimate their distribution parameters (implemented with the help of relevant functions of MATLAB software). For Weibull distribution, the Weibull probability paper method is considered to estimate its distribution parameters:

[0103] It should be noted that step S203 includes S2031 and S2032, which include:

[0104] S2031. Use applied mathematics software to perform parameter estimation. The parameter estimation method used is the Weibull probability paper method. The calculation formula of the Weibull probability paper is as follows:

[0105]

[0106] Where λ is the size parameter, α is the shape parameter, t is the mileage / time, and F(t) is the Weibull distribution function. By transforming Equation (6) and performing linear calculations on both sides of the equal sign, we can obtain the linear result:

[0107]

[0108] Where, For t i The failure probability value obtained by observing the time / mileage, The definition is at t i The ratio of the number of products that failed before the test to the number of products that started participating in the test;

[0109] Introducing the variable y i 、x i , a and b, the specific relationship expression is as follows:

[0110]

[0111] Then we can get y i =a+bx i (9)

[0112] Where a and b are introduced variables.

[0113] By introducing variables, Equations (8) and (9) retain the functional relationship of Equation (7) and simplify the calculation process.

[0114] S2032. Calculate the linear results according to the least squares linear regression method to obtain estimated values ​​of the shape parameter and size parameter of the Weibull distribution.

[0115] Through least squares linear regression, we can get

[0116]

[0117] In the formula, a is the same as a in formula (8) and formula (9). In order to obtain the shape parameter and size parameter of the Weibull distribution, it must be obtained through the transformation of the above formula;

[0118] in The observed values ​​x i and y i The mean of , according to formula (8) and (10),

[0119]

[0120]

[0121] Based on the above, we can finally get the shape parameter of the Weibull distribution and size parameters estimated value.

[0122] In this embodiment, the test of the distribution type of the bearing is as follows: Commonly used goodness of fit test methods include χ 2 The KS test method and the KS test method. Since the KS test has the advantages of accurate testing, no special requirements for sample size, and can test a given deviation, it is suitable for most scenarios. Therefore, the KS test method is used to test the hypothesized distribution type.

[0123] It should be noted that step S203 also includes S2033, S2034 and S2035, which include:

[0124] S2033. Establishing a distribution function of the bearing using a theoretical distribution method;

[0125] It should be noted that, at the significance level α, the hypothesis is established:

[0126] H: F(x)=F0(x); H1: F(x)≠F0(x)

[0127] Among them, F0(x) represents the distribution function obeyed by the assumed sample population, that is, the theoretical distribution; F(x) refers to the empirical failure probability distribution function. The above formula is the general expression of probability hypothesis testing, H is the null hypothesis, and H1 is the alternative hypothesis.

[0128] In this embodiment, the KS test is based on the fact that the cumulative distribution of the word pattern detection is very close to the true cumulative distribution. The goodness of fit is measured by finding the maximum deviation between the sub-sample and the parent population, that is, the test statistic is calculated according to formula (13), which is:

[0129] S2034. Test the distribution function based on the KS test to obtain the maximum deviation value. The test formula is as follows:

[0130] D n =sup|F n (t)-F0(t)| (13)

[0131] Where sup(·) is max(·), F n (t) is the empirical distribution function of the sample size n, F0(t) is the theoretical distribution function, D n is the maximum deviation value.

[0132] S2035. Compare the maximum deviation value with the preset critical value to obtain the final test result.

[0133] It should be noted that the maximum deviation value D n and critical value D (n,α) In comparison, if D n <D (n,α) , then accept the null hypothesis, otherwise reject it.

[0134] After the inspection step is completed, the distribution model that does not conform to the evolution law of bearing failure probability will be rejected, and the distribution model that conforms to the bearing failure law will be accepted, thereby establishing a failure probability distribution model. According to formula (5), the mathematical model of bearing reliability is established, that is, the first reliability model.

[0135] S300: Analyze the failure rate variation pattern of the bearing at each stage in different maintenance intervals, use the first function to fit the failure rate variation pattern, and obtain a second reliability model.

[0136] It should be noted that in addition to failure probability and reliability, which can be used to describe the overall reliability level of a product, the failure rate, as a characteristic quantity that predicts the ratio of the number of products that fail in the next unit time after t (t is the time the product has been working for a certain period of time) to the number of products that have not failed at that time, is also one of the commonly used quantitative characteristics of product reliability. The higher the failure rate, the lower its reliability. Its calculation formula is as follows:

[0137]

[0138] Where λ(t) is the failure rate function, t is the characteristic quantity of the ratio of the number of products that fail in the next unit time after the product has worked for a certain time to the number of products that have not failed at that time, X is a random variable, which represents a random variable in the interval (t, t+Δt], and Δt→0 + .

[0139] Assuming that the limit exists and the density function f(t) of X exists, the relationship between the failure rate λ(t) and the reliability R(t) is

[0140]

[0141] or

[0142]

[0143] Where R ′ (t) is the first-order differential of the reliability function R(t), u is a hypothetical variable used to distinguish the variable t, and f(t) is the density function.

[0144] In addition, considering that some components or the axlebox bearings themselves will be reassembled during maintenance, it will cause the related components to re-run and adapt. Therefore, in the early stage after maintenance, the failure rate will be too high. The relationship between failure rate and reliability shows that the higher the failure rate, the lower the reliability, which means that the axlebox bearings cannot maintain good working performance, thus affecting the remaining service life of the axlebox bearings. As a key component of the bogie, predicting the remaining service life of the axlebox bearings can not only prevent accidents, but also provide an important basis for formulating repair, improvement and prevention policies for trains. Therefore, it is necessary to focus on analyzing the impact of maintenance performance on the life of the axlebox bearings.

[0145] In order to study the impact of maintenance performance on the life of axlebox bearings, the failure rate change curve of axlebox bearings in different maintenance intervals is analyzed, taking the maintenance interval as the boundary. The specific method is as follows:

[0146] It can be understood that the step S300 includes S301, S302 and S303, wherein:

[0147] S301, determining the failure rate and the mileage of the bearing per unit time;

[0148] It should be noted that the bathtub curve, that is, the product failure rate curve, is first drawn based on the failure rate and mileage during working hours.

[0149] S302. Obtain a bearing failure rate curve based on the failure rate and mileage, wherein the failure rate curve includes an early failure period, a product service life period, and a wear and tear failure period;

[0150] It should be noted that if Figure 4 As shown in the figure, the change in product failure rate over time can be divided into three stages: early childhood failure, accidental failure, and wear-out failure. The accidental failure stage is the product's optimal operating period and lasts for a relatively long time, also known as the product's service life. However, in reality, within a given maintenance interval for an axlebox bearing, the failure rate curve will not completely encompass all three stages. A random combination of the three stages may occur, such as a continuous accidental failure period, or a period of early childhood failure followed by a continuous accidental failure period. Therefore, it is necessary to determine the conditions occurring within a given maintenance interval.

[0151] S303: Based on the slope of the fault curve, determine the fault rate variation pattern.

[0152] Specifically, the following is a detailed description of the change stages:

[0153] 1. Early failure stage judgment: The bathtub curve shows that when in the early failure stage, the failure rate decreases with time (mileage). Therefore, the slope of the adjacent failure rate changes can be used to determine whether the vehicle is in the early failure stage. If the slope is less than 0 for a certain continuous time (mileage) within the maintenance interval, it can be determined that the failure rate for this continuous time (mileage) is in the early failure stage.

[0154] 2. Judgment of the accidental failure period: The bathtub curve shows that when in the accidental failure period, the failure rate remains unchanged with the increase of time (mileage). Therefore, if the failure rate remains stable for a certain continuous time (mileage) within the maintenance interval, it can be determined that the failure rate of this continuous time (mileage) is in the accidental failure period.

[0155] 3. Determining the Wear-Out Failure Period: As shown by the bathtub curve, the failure rate gradually increases with time (mileage) during the wear-out failure period. The determination method is similar to that for the early failure period. If the slope is greater than 0 for a certain continuous time (mileage) within the maintenance interval, it can be determined that the failure rate for that continuous time (mileage) is in the wear-out failure stage.

[0156] Specifically, after determining the variation pattern, a suitable mathematical model is selected, including but not limited to using exponential functions, cubic functions, and logarithmic functions to fit the variation pattern of the failure rate at each stage. Since the failure rate remains constant during the accidental failure stage, according to the functional relationship between the failure rate and reliability (16), the reliability variation at this time conforms to the exponential function. Therefore, the exponential function is selected to fit the failure rate during the accidental failure period. Based on formula (2), and according to formula (15) or formula (16) between the failure rate and reliability, the reliability mathematical model of the axle box bearing in each maintenance interval can be obtained, namely the second reliability model.

[0157] S400. Using a reliability-centered maintenance method, the first reliability model and the second reliability model are processed respectively to obtain a first remaining life result and a second remaining life result, and the average of the first remaining life result and the second remaining life result is calculated to obtain a final bearing remaining life prediction result.

[0158] It can be understood that the step S400 includes S401, S402 and S403, wherein:

[0159] S401, obtaining a bearing failure law evolution curve;

[0160] S402: Based on the first reliability model, the operating environment of the bearing and the preset safety conditions are comprehensively considered, and a reliability-centered maintenance method is used to preliminarily predict the remaining life of the bearing;

[0161] S403: Determine the bearing risk control requirement. When the bearing operation risk reaches the control requirement, re-predict the initially predicted remaining life of the bearing to obtain a first remaining life result.

[0162] It should be noted that if Figure 5 As shown in the figure, a reliability-centered maintenance approach can upgrade maintenance strategies from traditional "fault prevention" to "proactive risk management." Based on the bearing failure evolution curve (the trend of the axlebox bearing's operating reliability decreasing from 1 to 0, i.e., the trend of the risk increasing from 0 to 1), a reliability-centered maintenance approach is applied to predict the remaining life of the axlebox bearing. Taking into account the axlebox bearing's operating environment and safety requirements, the axlebox bearing's risk control requirements are determined. When the axlebox bearing's operating risk reaches this control requirement, the remaining life (service mileage) of the axlebox bearing can be determined.

[0163] The maintenance process optimization based on the reliability-centered maintenance method comprehensively considers the requirements such as fault deterioration speed and redundancy, and divides the fault impact level into three levels: safety, usability and economy. Among them, safety refers to the fault level that will cause the EMU to have safety accidents such as derailment, overturning, and rear-end collision after the fault occurs. Usability refers to the fault level that will cause the EMU to stop, transfer, etc. after the fault occurs, which affects the train operation quality and passenger experience. Economy refers to the fault level that does not affect the use of the EMU after the fault occurs, but requires additional manpower and material resources for maintenance, causing unnecessary economic losses. At the same time, the fault deterioration speed is divided into fast and slow levels, and the redundancy is divided into low and high levels. In this way, the acceptable minimum reliability index for the safe operation of the EMU under different fault impact levels, different fault deterioration speeds and redundancy levels can be determined. Combined with the reliability change law curve of the axle box bearing, when the reliability of the axle box bearing drops to the reliability index, the corresponding time / mileage is the maintenance time node of the axle box bearing, that is, the maintenance process, such as Figure 6 shown.

[0164] In step S400, the average of the first remaining life result and the second remaining life result is obtained to obtain the final bearing remaining life prediction result, which includes:

[0165] It should be noted that two different bearing reliability mathematical models are obtained in the above process, namely the first reliability model and the second reliability model. Next, the two different mathematical models are calculated separately to obtain the calculation results of the remaining life and maintenance repair schedule of the axle box bearing corresponding to different models. In order to balance the difference between the two, the average value of the remaining life calculation results of the two mathematical models is considered to finally give the remaining life / remaining service mileage of the axle box bearing, and the same is true for the maintenance repair schedule.

[0166] In summary, the two axlebox bearing reliability mathematical models are based on different considerations, but both aim to improve the safety and reliability of axlebox bearings, effectively predict their remaining service life, and optimize maintenance schedules. Furthermore, the algorithm can intuitively provide the remaining service life of axlebox bearings in units of mileage, effectively guiding maintenance organizations.

[0167] Example 2:

[0168] like Figure 2 As shown, this embodiment provides a bearing remaining life prediction device, see Figure 2 The apparatus includes an acquisition module 701, a processing module 702, a fitting module 703, and a calculation module 704, wherein:

[0169] Acquisition module 701: used to acquire bearing operation and maintenance data, pre-process the bearing operation and maintenance data, and obtain bearing fault data, wherein the operation and maintenance data includes vehicle system alarm data and data of faults to be repaired;

[0170] Processing module 702: used to process bearing fault data based on a survival curve algorithm to obtain a first reliability model;

[0171] Fitting module 703: used to analyze the failure rate variation pattern of the bearing at each stage within different maintenance intervals, and fit the failure rate variation pattern using the first function to obtain a second reliability model;

[0172] Calculation module 704: used to adopt a reliability-centered maintenance method to process the first reliability model and the second reliability model respectively to obtain a first remaining life result and a second remaining life result, and to calculate the average of the first remaining life result and the second remaining life result to obtain a final bearing remaining life prediction result.

[0173] Specifically, the acquisition module 701 includes a cleaning unit 7011, an analysis unit 7012, and a storage unit 7013, wherein:

[0174] The cleaning unit 7011 is configured to clean and review the acquired bearing operation and maintenance data to obtain a first processing result, wherein the cleaning includes missing data filling processing, abnormal data elimination processing, and data non-dimensionalization processing.

[0175] The analysis unit 7012 is configured to analyze the first processing result to obtain a second processing result, wherein the analysis includes descriptive analysis, data dynamic analysis, correlation analysis, and regression analysis on the first processing result.

[0176] The storage unit 7013 is configured to store the second processing result according to a keyword to obtain bearing failure data, wherein the keyword includes a vehicle type and a maintenance level.

[0177] Specifically, the processing module 702 includes a first obtaining unit 7021, a second obtaining unit 7022, and a fitting unit 7023, wherein:

[0178] The first obtaining unit 7021 is configured to obtain a reliability distribution function by using a survival curve algorithm, wherein the survival curve algorithm is as follows:

[0179]

[0180] In the formula, R(t) is a reliability function, is a survival distribution function, Z (i) is an order statistic variable of the variable Z i , wherein Z i = min(X i , Y i ), X i is a non-negative random variable, Y i is a corresponding interference random variable; δ (i) is a parameter for judging whether it is truncated data, i.e., a truncation indicator function, if it is truncated data, then δ (i) = 0; if it is real life, then δ (i) = 1.

[0181] The second obtaining unit 7022 is configured to obtain an empirical distribution function of failure probability based on the reliability distribution function and a function relationship corresponding to the bearing reliability.

[0182] The fitting unit 7023 is configured to fit the empirical distribution function of failure probability by using a mathematical model to obtain a first reliability model, wherein the fitting process includes selection of a distribution type of the bearing, estimation of distribution parameters, and verification of the distribution type.

[0183] Specifically, the fitting unit 7023 includes an estimation unit 70231 and a calculation unit 70232, wherein:

[0184] Estimation unit 70231: used to perform parameter estimation using mathematical software. The parameter estimation method is the Weibull probability paper method, where the calculation formula of the Weibull probability paper is as follows:

[0185]

[0186] Where F(t) is the cumulative failure probability function of the Weibull distribution, λ is the size parameter, α is the shape parameter, and t is the mileage / time. By transforming the above formula and performing linear calculations on both sides of the equal sign, we can obtain the linear result:

[0187]

[0188] Where, For t i The failure probability value obtained by observing the time / mileage, The definition is at t i The ratio of the number of products that failed before the test to the number of products that started participating in the test;

[0189] Calculation unit 70232: used to calculate the linear result according to the least squares linear regression method to obtain the estimated values ​​of the shape parameter and size parameter of the Weibull distribution.

[0190] Specifically, the fitting unit 7023 further includes a function establishment unit 70233, a verification unit 70234, and a comparison unit 70235, which also includes:

[0191] Establishing function unit 70233: used for establishing the distribution function of the bearing using the theoretical distribution method;

[0192] Testing unit 70234: used to test the distribution function based on the KS test to obtain the maximum deviation value. The test formula is as follows:

[0193] D n =sup|F n (t)-F0(t)|

[0194] Where sup(·) is max(·), F n (t) is the empirical distribution function of the sample size n, F0(t) is the theoretical distribution function, D n is the maximum deviation value;

[0195] Comparison unit 70235: used to compare the maximum deviation value with a preset critical value to obtain a final test result.

[0196] Specifically, the fitting module 703 includes a determining unit 7031, a third obtaining unit 7032, and a judging unit 7033, wherein:

[0197] Determining unit 7031: used to determine the failure rate and the mileage of the bearing per unit time;

[0198] The third obtaining unit 7032 is configured to obtain a bearing failure rate curve according to the failure rate and the mileage, wherein the failure rate curve includes an early failure period, a product service life period, and a wear and tear failure period;

[0199] The judgment unit 7033 is used to judge the fault rate variation pattern based on the slope of the fault curve.

[0200] Specifically, the calculation module 704 includes an acquisition unit 7041, a first prediction unit 7042, and a second prediction unit 7043, wherein:

[0201] Acquisition unit 7041: used to obtain the bearing failure law evolution curve;

[0202] The first prediction unit 7042 is used to preliminarily predict the remaining life of the bearing using a reliability-centered maintenance method based on the first reliability model, taking into account the working environment of the bearing and preset safety conditions;

[0203] The second prediction unit 7043 is used to determine the bearing risk control requirement. When the bearing operation risk reaches the control requirement, the initially predicted remaining life of the bearing is re-predicted to obtain a first remaining life result.

[0204] It should be noted that, regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated on here.

[0205] Example 3:

[0206] Corresponding to the above method embodiment, this embodiment also provides a bearing remaining life prediction device. The bearing remaining life prediction device described below and the bearing remaining life prediction method described above can refer to each other.

[0207] Figure 3 FIG. 8 is a block diagram of a bearing remaining life prediction device 800 according to an exemplary embodiment. Figure 3 As shown, the bearing remaining life prediction device 800 includes: a processor 801 and a memory 802 . The bearing remaining life prediction device 800 also includes one or more of a multimedia component 803 , an I / O interface 804 , and a communication component 805 .

[0208] The processor 801 is configured to control overall operations of the bearing remaining useful life prediction device 800 to complete all or part of the steps of the bearing remaining useful life prediction method described above. The memory 802 is configured to store various types of data to support operations of the bearing remaining useful life prediction device 800, which can include, for example, instructions for any application or method operating on the bearing remaining useful life prediction device 800, and application-related data, such as contact data, sent and received messages, pictures, audio, video, and the like. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as a static random access memory (SRAM), an electrically erasable programmable read-only memory (EEPROM), an erasable programmable read-only memory (EPROM), a programmable read-only memory (PROM), a read-only memory (ROM), a magnetic memory, a flash memory, a magnetic disk or an optical disk. The multimedia component 803 can include a screen and an audio component. The screen can be, for example, a touch screen, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 802 or transmitted through the communication component 805. The audio component also includes at least one speaker configured to output audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, which can be a keyboard, a mouse, or a button, and the like. The buttons can be virtual buttons or physical buttons. The communication component 805 is configured to perform wired or wireless communication between the bearing remaining useful life prediction device 800 and other devices. The wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G, or 4G, or a combination of one or more of them, so the corresponding communication component 805 can include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0209] In an exemplary embodiment, the bearing remaining life prediction device 800 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned bearing remaining life prediction method.

[0210] In another exemplary embodiment, a computer-readable storage medium including program instructions is further provided. When executed by a processor, the program instructions implement the steps of the aforementioned bearing remaining life prediction method. For example, the computer-readable storage medium may be the aforementioned memory 802 including the program instructions. The program instructions may be executed by the processor 801 of the bearing remaining life prediction device 800 to implement the aforementioned bearing remaining life prediction method.

[0211] Example 4:

[0212] Corresponding to the above method embodiment, this embodiment further provides a readable storage medium. The readable storage medium described below and the bearing remaining life prediction method described above can refer to each other.

[0213] A computer program is stored on the readable storage medium, and when the computer program is executed by the processor, the steps of the bearing remaining life prediction method of the above method embodiment are implemented.

[0214] The readable storage medium may specifically be any readable storage medium that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0215] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

[0216] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for predicting the remaining life of a bearing, characterized in that: include: Acquire bearing operation and maintenance data, pre-process the bearing operation and maintenance data, and obtain bearing fault data, where the operation and maintenance data includes vehicle system alarm data and data on faults to be repaired; The bearing failure data is processed based on the survival curve algorithm to obtain the first reliability model; Analyze the variation pattern of the failure rate of bearings at different stages within different maintenance intervals, fit the variation pattern of the failure rate using the first function, and obtain the second reliability model; A reliability-centered maintenance method is used to process the first reliability model and the second reliability model respectively to obtain a first remaining life result and a second remaining life result. The first and second remaining life results are averaged to obtain a final bearing remaining life prediction result. The bearing fault data is processed based on the survival curve algorithm to obtain a first reliability model, which includes: The reliability distribution function is obtained using the survival curve algorithm, where the survival curve algorithm formula is as follows: Where R(t) is the reliability function, is the survival distribution function, Z (i) is the variable Z i Order statistics variables, where Z i =min(X i ,Y i ), X i is a non-negative random variable, Y i is the corresponding disturbance random variable; δ (i) The parameter to determine whether it is truncated data is the truncation indicator function. If it is truncated data, then δ (i) =0; if it is the real life span, then δ (i) =1, t is the characteristic quantity of the ratio of the number of products that fail in the next unit time after a certain time of operation to the number of products that have not failed at that time, n is the sample size, and i is the index number of the traversal sample; Based on the reliability distribution function and the functional relationship corresponding to the bearing reliability, the failure probability empirical distribution function is obtained; Using a mathematical model to fit the failure probability empirical distribution function to obtain a first reliability model, wherein the fitting process includes selecting a distribution type of the bearing, estimating distribution parameters, and testing the distribution type; The analysis of the failure rate variation patterns of bearings at different stages within different maintenance intervals includes: Determine the failure rate and the mileage of the bearing per unit time; The bearing failure rate curve is obtained based on the failure rate and mileage, where the failure rate curve includes the early failure period, product service life period and wear and tear failure period; Based on the slope of the fault curve, determine the change pattern of the failure rate.

2. The method for predicting the remaining life of a bearing according to claim 1, characterized in that: The bearing operation and maintenance data is preprocessed to obtain bearing fault data, including: Cleaning and reviewing the acquired bearing operation and maintenance data to obtain a first processing result, wherein the cleaning includes filling in missing data, removing abnormal data, and dimensionless quantization of data; Analyzing the first processing result to obtain a second processing result, wherein the analysis includes performing descriptive analysis, data dynamic analysis, correlation analysis, and regression analysis on the first processing result; The second processing result is classified and stored according to keywords to obtain bearing fault data, wherein the keywords include vehicle type and maintenance level.

3. The method for predicting the remaining life of a bearing according to claim 1, wherein: The fitting process includes the estimation of the distribution parameters of the bearing, including: Parameter estimation is performed using mathematical software. The parameter estimation method is the Weibull probability paper method, where the calculation formula of the Weibull probability paper is as follows: Where F(t) is the failure probability distribution function, λ is the size parameter, α is the shape parameter, and t is the mileage / time. By transforming the above formula and performing linear calculations on both sides of the equal sign, we can obtain the linear result: Where, For t i The failure probability value obtained by observing the time / mileage, The definition is at t i The ratio of the number of products that failed before the test to the number of products that started participating in the test; The linear results were calculated according to the least squares linear regression method to obtain the estimated values ​​of the shape parameters and size parameters of the Weibull distribution.

4. The method for predicting the remaining life of a bearing according to claim 1, wherein: The fitting process includes a test of the distribution type of the bearings, including: The distribution function of the bearing is established using the theoretical distribution method; The distribution function is tested based on the KS test to obtain the maximum deviation value. The test formula is as follows: D n =sup|F n (t)-F0(t)| Where sup(·) is max(·), F n (t) is the empirical distribution function of the sample size n, F0(t) is the theoretical distribution function, D n is the maximum deviation value; Compare the maximum deviation value with the preset critical value to obtain the final test result.

5. The method for predicting the remaining life of a bearing according to claim 1, wherein: The first remaining life result is obtained, which includes: Obtain the evolution curve of bearing failure law; Based on the first reliability model, the bearing's working environment and pre-set safety conditions are considered, and a reliability-centered maintenance method is used to preliminarily predict the remaining life of the bearing. Determine the bearing risk control requirements. When the bearing operation risk reaches the control requirements, re-predict the initially predicted remaining life of the bearing to obtain a first remaining life result.

6. A bearing remaining life prediction device, characterized in that: include: Acquisition module: used to acquire bearing operation and maintenance data, pre-process the bearing operation and maintenance data, and obtain bearing fault data, where the operation and maintenance data includes vehicle system alarm data and data on faults to be repaired; Processing module: used to process bearing fault data based on the survival curve algorithm to obtain a first reliability model; Fitting module: used to analyze the failure rate variation pattern of bearings at different stages within different maintenance intervals, and use the first function to fit the failure rate variation pattern to obtain the second reliability model; Calculation module: used to process the first reliability model and the second reliability model respectively using a reliability-centered maintenance method to obtain a first remaining life result and a second remaining life result, and to calculate the average of the first remaining life result and the second remaining life result to obtain a final bearing remaining life prediction result; Wherein, the processing module includes: The first obtaining unit is used to obtain the reliability distribution function using the survival curve algorithm, wherein the survival curve algorithm formula is as follows: Where R(t) is the reliability function, is the survival distribution function, Z (i) is the variable Z i Order statistics variables, where Z i =min(X i ,Y i ), X i is a non-negative random variable, Y i is the corresponding disturbance random variable; δ (i) The parameter to determine whether it is truncated data is the truncation indicator function. If it is truncated data, then δ (i) =0; if it is the real life span, then δ (i) =1; A second obtaining unit is used to obtain a failure probability empirical distribution function based on the reliability distribution function and a functional relationship corresponding to the bearing reliability; Fitting unit: used to fit the failure probability empirical distribution function using a mathematical model to obtain a first reliability model, wherein the fitting process includes the selection of the distribution type of the bearing, the estimation of the distribution parameters and the verification of the distribution type; Wherein, the fitting module includes: Determination unit: used to determine the failure rate and the mileage of the bearing per unit time; The third obtaining unit is used to obtain a bearing failure rate curve according to the failure rate and mileage, wherein the failure rate curve includes an early failure period, a product service life period, and a wear and tear failure period; Judgment unit: used to judge the fault rate change pattern based on the slope of the fault curve.

7. A bearing remaining life prediction device, characterized in that: include: memory for storing computer programs; A processor is configured to implement the steps of the method for predicting the remaining life of a bearing as claimed in any one of claims 1 to 5 when executing the computer program.

8. A readable storage medium, characterized in that: The readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for predicting the remaining life of a bearing according to any one of claims 1 to 5.

Images