CN-121997138-A - Full-life rolling bearing fault diagnosis method based on mechanism data migration

CN121997138ACN 121997138 ACN121997138 ACN 121997138ACN-121997138-A

Abstract

The invention relates to the technical field of rolling bearing fault diagnosis and discloses a full-life rolling bearing fault diagnosis method based on mechanism data migration, which comprises the steps of correcting a rolling body radial displacement excitation function, time-varying contact rigidity of a rolling body and a bearing raceway and impact force of a rolling body contact defect part in a mass-spring-damping model based on Hertz contact theory to obtain a nonlinear kinetic equation of a rolling bearing; and solving a nonlinear dynamics equation of the rolling bearing by a fourth-order Dragon-Gregory tower method, generating source domain data consistent with test parameters of the rolling bearing with the whole service life, and constructing a source domain data set with a complete fault label. According to the invention, by constructing the dynamic mechanism model consistent with the full-life test bearing parameters, the generated source domain data and target domain data feature distribution difference is small, a foundation is laid for efficient knowledge migration, and the problem of mismatching of traditional source domain data and actual data is solved.

Inventors

Hu Juzhen
HUANG JIAQIU
FENG LEI
PENG XIAOYIN
JIA JUN

Assignees

巨冈精工(广东)股份有限公司

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (10)

1. The full-life rolling bearing fault diagnosis method based on mechanism data migration is characterized by comprising the following steps of: The rolling bearing is constructed into a mass-spring-damping model consisting of an outer ring, an inner ring, rolling bodies and a high-frequency resonator; Correcting a radial displacement excitation function of a rolling body, time-varying contact stiffness of the rolling body and a bearing raceway and impact force at a contact defect of the rolling body in a mass-spring-damping model based on Hertz contact theory to obtain a nonlinear kinetic equation of the rolling bearing; Solving a nonlinear dynamics equation of the rolling bearing by a fourth-order Dragon-Gregory tower method, generating source domain data consistent with test parameters of the rolling bearing with the whole service life, and constructing a source domain data set with a complete fault label; training a pre-built improved alternate migration learning model based on the source domain data set and the pre-built target domain data set; performing fault diagnosis on the full-life rolling bearing by using the trained improved alternate transfer learning model, and outputting a diagnosis result; The target domain data set covers data samples of a rolling bearing test in the whole process from normal operation to failure, wherein the data samples comprise sample data of four types of faults including normal state, inner ring fault, outer ring fault and rolling body fault, and the source domain data set comprises simulation data of four types of faults including normal state, inner ring fault, outer ring fault and rolling body fault.
2. The method for diagnosing a failure of a rolling bearing over a life span based on mechanism data migration of claim 1, wherein the nonlinear dynamical equation of the rolling bearing is: ; the first two behavior inner ring equations of motion, wherein, For the mass of the inner ring, Is the damping coefficient between the inner ring and the bearing seat, For the supporting rigidity between the inner ring and the bearing seat, 、 The displacement of the inner ring in the x and y directions is respectively, 、 In order to be able to achieve a speed, 、 In order for the acceleration to be a function of the acceleration, 、 The components of the contact force between the inner ring and the rolling bodies in the x and y directions are respectively shown, F r is the bearing load, e is the eccentricity of the inner ring, The rotational angular velocity of the inner ring; and a third and fourth behavior outer ring motion equation, wherein, For the mass of the outer ring, 、 The supporting rigidity and the damping coefficient between the outer ring and the bearing seat are respectively, 、 The horizontal displacement and the vertical displacement of the shaft and the outer ring are respectively, 、 The displacement of the center of the rolling body in the x and y directions is respectively, 、 The outer ring fault excitation forces in the x and y directions are respectively; Fifth and sixth two-behavior rolling body motion equation, wherein, For a single rolling body mass; the last behavior cage equation of motion, wherein, In order to maintain the quality of the holder, For the cage angular displacement corresponding to the jth rolling element, Represents the polar diameter of the jth rolling element, In order to maintain the angular velocity of the cage, Indicating the angular position of the jth rolling element, 、 The rolling body receives the Hertz contact force of the inner ring and the outer ring of the bearing in the radial direction.
3. The method for diagnosing a full-life rolling bearing fault based on mechanism data migration according to claim 2, wherein the rolling element radial displacement excitation function in the mass-spring-damping model is modified based on the Hertz contact theory, specifically comprising: Cage rotational speed Rotational angular velocity with inner race The relation between the two is: D is the diameter of the rolling element, D is the pitch diameter of the bearing, and beta is the contact angle; ; In the above formula, Z is the number of rolling bodies, j is a natural number greater than or equal to 1, The angular position of the first rolling element relative to the positive y-axis direction at zero time; The expression of the radial displacement excitation function s of the abrupt change of the rolling bodies entering the defect area is as follows: , ; In the above formula, R is the radius of the bearing outer ring raceway, R is the radius of the rolling body, An angle at which the rolling bodies enter the defective region; 、 the front edge and the rear edge of the defect area are respectively at the angle positions; the contact deformation radial displacement of the jth rolling body and the bearing inner ring raceway is as follows: ; Radial play of the bearing; Radial displacement of contact deformation of jth rolling element and bearing outer ring raceway The correction is as follows: 。
4. A method for diagnosing a life-span rolling bearing failure based on mechanism data migration as recited in claim 3, wherein the time-varying contact stiffness of the rolling element and the bearing raceway is corrected based on the Hertz' contact theory, specifically comprising: The bearing inner and outer ring Hertz contact force applied to the rolling bodies in the radial direction has the following formula: , ; In the formula: , , , , in the formula, 、 The rigidity coefficients of the inner race and the outer race of the bearing are respectively shown, wherein n is a load-deformation coefficient; 、 Respectively representing the relative displacement vectors of the jth rolling element and the inner ring roller path and the outer ring roller path of the bearing; the time-varying contact stiffness is calculated as follows: , ; in the formula, To modify the time-varying contact stiffness coefficient, S is a shape coefficient for the reduction coefficient; is a banding range.
5. The method for diagnosing a full-life rolling bearing fault based on mechanism data migration as claimed in claim 4, wherein the method for correcting the impact force at the contact defect of the rolling element based on the Hertz contact theory comprises the following steps: impact force generated by rolling body striking defect rear edge The calculation formula of (2) is as follows: L is the defect width, m b is the mass of the rolling element; Will be Orthogonal decomposition yields horizontal and vertical components: , : The angle between the impact force and the radial speed of the rolling body is as follows: ; in the formula, For the discriminant function of whether an impact force exists, it is defined as follows: ; the total Hertz contact force of the inner ring and the outer ring of the rolling bearing is respectively as follows: , 。
6. The method for diagnosing a failure of a rolling bearing with a life span based on mechanism data migration according to claim 1, wherein the rolling bearing is constructed as a mass-spring-damping model composed of an outer ring, an inner ring, rolling bodies and a high-frequency resonator, specifically comprising: for the rolling bearing used in the life test, the rolling bearing system was simplified into a mass-spring-damping model composed of an outer ring, an inner ring, rolling bodies and a high-frequency resonator, and the model was assumed to satisfy the following conditions: The outer ring is fixed on the rigid structure, and the inner ring maintains constant rotation speed; The rolling bodies are uniformly distributed and have no relative sliding; Irrespective of the rolling element rotation effect; The inner ring, the outer ring and the rolling bodies move and remain in the same plane.
7. The method for diagnosing the fault of the full-life rolling bearing based on the mechanism data migration according to claim 2, wherein the method for solving the nonlinear dynamics equation of the rolling bearing by a fourth-order longgrid-base method generates source domain data consistent with the test parameters of the full-life rolling bearing and constructs a source domain data set with a complete fault label, and the method specifically comprises the following steps: Solving a nonlinear dynamics equation of the rolling bearing by a fourth-order Dragon-Gregory tower method, and generating simulation data of four faults, namely a normal state, an inner ring fault, an outer ring fault and a rolling body fault; and performing fault simulation based on the generated simulation data, and constructing a source domain data set with a complete fault label.
8. The method for diagnosing a life-span rolling bearing fault based on mechanism data migration of claim 1, wherein before training the pre-built improved alternate migration learning model based on the source domain data set and the pre-built target domain data set, further comprises: the method for constructing the target domain data set comprises the following steps: collecting full-life rolling bearing test data, wherein the full-life rolling bearing test data cover the whole process from normal operation to failure of the bearing and comprise axial and radial vibration signals; Preprocessing collected full-life rolling bearing test data to construct an original data set containing four fault samples of normal state, inner ring fault, outer ring fault and rolling body fault; selecting an amplitude spectrum of a fault sample as a CS-GWO algorithm fitness function, and optimizing a mode number K and a secondary penalty factor in a VMD algorithm by using an extremely small value of an amplitude spectrum entropy as a fitness value ; Based on the optimized mode number K and the quadratic penalty factor Decomposing by using a VMD algorithm to obtain K modal components; and selecting a plurality of modal components with highest correlation with the fault sample as effective modalities, carrying out summation reconstruction, carrying out envelope spectrum analysis on the reconstructed modal components, extracting characteristic information related to the fault of the rolling bearing, and forming sample data.
9. The method for diagnosing a full-life rolling bearing fault based on mechanism data migration of claim 8, wherein the method for setting a loss function of the improved alternate migration learning model is as follows: The distance of the second order statistics of the source and target domain features is defined as the CORAL loss function L 1 : , the Frobenius norm of the mean square matrix; Wherein, the For the sample dimension number, the feature covariance matrices of the source domain dataset D S and the target domain dataset D T are C S and C T , respectively; MMD is used to measure the difference in distribution of feature sets D S and D T in the regenerated kernel hilbert space, summing MMD loss with classification loss as a loss function L 2 : , , , for regenerating the nuclear hilbert space; wherein L C is the classification loss between the real tag and the predictive tag, , Mapping function , For the balance coefficient, n S is the number of samples, y is the true label of the sample data, Labels predicted for the classifier.
10. The method for diagnosing a full-life rolling bearing fault based on mechanism data migration of claim 8, wherein the calculation formula of the amplitude spectrum entropy is: ; In the above formula, L i is the amplitude spectrum of the modal component u i , H i is the amplitude spectrum entropy of the modal component u i , and N is the length of the modal component.

Description

Full-life rolling bearing fault diagnosis method based on mechanism data migration Technical Field The invention relates to the technical field of rolling bearing fault diagnosis, in particular to a full-life rolling bearing fault diagnosis method based on mechanism data migration. Background Rolling bearings are the core component of rotating machinery, the operating conditions of which directly determine the reliability of the equipment, and about 30% of rotating machinery failures are caused by rolling bearing failure. In a full life test scene, the whole process data from normal operation to failure of the bearing contains key fault evolution information, but in practical application, two major core problems exist, namely, the fault data in the full life cycle is scarce and even missing, especially, early fault samples are insufficient, so that the training effect of a data driving model is poor, and secondly, the data collected on site is influenced by noise and working condition fluctuation, the characteristic signals are weak, and the fault information is difficult to effectively extract by the traditional diagnosis method. In the prior art, a fault diagnosis method based on transfer learning tries to solve the problem of sample deletion through knowledge transfer of source domain data, but has obvious defects that on one hand, the source domain data is dependent on laboratory simulation data, the characteristic distribution difference between the source domain data and full-life actual data is large, the transfer effect is limited, on the other hand, the loss function design is single, the complementary advantages of the multiple loss functions are not fully utilized, and the target domain data containing noise is not subjected to targeted characteristic optimization, so that the diagnosis accuracy is difficult to meet engineering requirements. Parameters of the traditional VMD (Variational mode decomposition) variational modal decomposition algorithm are set empirically, so that modal aliasing is easy to occur, and fault feature extraction accuracy is affected. Therefore, a fault diagnosis method capable of constructing a source domain model accurately matching the full-life data, optimizing the feature extraction process and improving the migration learning effect is needed in the art. Disclosure of Invention The invention aims to provide a full-life rolling bearing fault diagnosis method based on mechanism data migration, so as to solve or at least partially solve the technical problems mentioned in the background art. To achieve the purpose, the invention adopts the following technical scheme: the invention provides a full-life rolling bearing fault diagnosis method based on mechanism data migration, which comprises the following steps: The rolling bearing is constructed into a mass-spring-damping model consisting of an outer ring, an inner ring, rolling bodies and a high-frequency resonator; Correcting a radial displacement excitation function of a rolling body, time-varying contact stiffness of the rolling body and a bearing raceway and impact force at a contact defect of the rolling body in a mass-spring-damping model based on Hertz contact theory to obtain a nonlinear kinetic equation of the rolling bearing; Solving a nonlinear dynamics equation of the rolling bearing by a fourth-order Dragon-Gregory tower method, generating source domain data consistent with test parameters of the rolling bearing with the whole service life, and constructing a source domain data set with a complete fault label; training a pre-built improved alternate migration learning model based on the source domain data set and the pre-built target domain data set; performing fault diagnosis on the full-life rolling bearing by using the trained improved alternate transfer learning model, and outputting a diagnosis result; The target domain data set covers data samples of a rolling bearing test in the whole process from normal operation to failure, wherein the data samples comprise sample data of four types of faults including normal state, inner ring fault, outer ring fault and rolling body fault, and the source domain data set comprises simulation data of four types of faults including normal state, inner ring fault, outer ring fault and rolling body fault. Optionally, the nonlinear kinetic equation of the rolling bearing is: ; the first two behavior inner ring equations of motion, wherein, For the mass of the inner ring,Is the damping coefficient between the inner ring and the bearing seat,For the supporting rigidity between the inner ring and the bearing seat,、The displacement of the inner ring in the x and y directions is respectively,、In order to be able to achieve a speed,、In order for the acceleration to be a function of the acceleration,、The components of the contact force between the inner ring and the rolling bodies in the x and y directions are respectively shown, F r is the bearing load, e is the eccentricity