Micro-pitting fatigue life prediction method fusing physical information and surface integrity parameters

Abstract

This invention relates to a fretting fatigue life prediction method that integrates physical information and surface integrity parameters, belonging to the field of fretting fatigue life prediction methods for metallic materials. First, a basic fretting fatigue dataset is acquired, and a fretting contact finite element model of fretting fatigue is constructed. Simulation analysis is performed to obtain the corresponding stress-strain curves, determine the maximum normal stress on the plane containing the maximum principal strain range, and obtain the corrected fretting damage parameters. Second, the basic fretting fatigue dataset is combined to construct a training dataset. Finally, based on a composite loss function that integrates physical constraints, a fretting fatigue life prediction model integrating physical information is constructed, trained, and then used for fretting fatigue life prediction. This invention can effectively characterize the influence of surface condition on fretting fatigue behavior; under small sample conditions, it maintains high prediction accuracy and generalization ability, is applicable to different materials and different fretting conditions, and has engineering application value.

Description

Technical Field

[0001] This invention belongs to the field of fretting fatigue life prediction methods for metallic materials, and relates to a fretting fatigue life prediction method that integrates physical information and surface integrity parameters. Background Technology

[0002] During the service life of aero-engines, the turbine disk tenon joint structure is subjected to the coupled effects of centrifugal force, thermal load, aerodynamic load, and vibration load. In this complex and variable service environment, the contact area between the turbine disk tenon and groove is prone to fretting, leading to fretting fatigue damage, premature crack initiation, and ultimately, fatigue failure of the turbine disk structure. With the continuous improvement of the performance indicators of new-generation aero-engines, key load-bearing components such as turbine disks operate under high temperature, high-frequency vibration, and multi-load coupling. Under conditions of temperatures exceeding 700°C and high-frequency vibrations above 3000Hz, the damage accumulation rate in the fretting contact area significantly increases. Therefore, research on fretting fatigue life prediction methods for turbine disk tenon joint structures under complex service conditions is of great significance for improving the accuracy of life assessment, service reliability, and safety design level of key load-bearing components in aero-engines.

[0003] Studies have shown that the surface integrity parameters of the contact surface in tenon joint structures have a significant impact on fretting fatigue life. The morphology and roughness of the contact surface determine the actual contact state and local stress distribution, thus affecting oxidation behavior and crack propagation paths. Increasing surface hardness helps reduce fretting wear and delay crack initiation. Different residual stress states on the surface may promote or inhibit crack initiation and propagation. Therefore, improving the accuracy of fretting fatigue life prediction for turbine disk tenon joint structures when considering surface integrity parameters has become a crucial issue that urgently needs to be addressed.

[0004] However, existing fretting fatigue life prediction methods mostly rely on classical fatigue damage models, indirectly considering the influence of working conditions through empirical coefficients or correction factors, making it difficult to describe the damage evolution under high temperature, high frequency vibration, and multi-parameter coupling conditions. Furthermore, with the continuous expansion of data scale, the accuracy and applicability of classical fatigue life prediction methods have certain limitations. With the development of machine learning technology, data-driven methods have shown advantages in modeling complex nonlinear relationships, capable of establishing complex mapping relationships between input and output variables from large amounts of data in high-dimensional space. However, data-driven models mainly rely on training with large amounts of data to obtain prediction results, focusing more on the correlation between data samples, and are highly dependent on data scale and quality, resulting in relatively weak physical interpretability. Therefore, there is an urgent need to construct a fretting fatigue life prediction method that integrates surface integrity parameters and physical damage mechanisms to improve the physical interpretability and prediction accuracy of the model.

[0005] Currently, there are some patented technologies for machine learning methods to predict the fretting fatigue life of turbine blade tenon joints, but there are still certain limitations.

[0006] Chinese invention patent application CN112052615B discloses a "method for predicting fretting fatigue performance based on artificial neural networks." This patent obtains fretting fatigue life through numerical simulation and uses it as a constraint to optimize the artificial neural network, thus achieving prediction of fretting fatigue life. However, the numerical model used in this method does not consider the influence of surface integrity and its evolution on fretting fatigue life, resulting in limited accuracy of the fretting fatigue life obtained under complex working conditions, which to some extent affects the prediction accuracy and generalization ability of the neural network model. In addition, the neural network model constructed by this method lacks an effective physical constraint mechanism and is a typical data-driven model. The correspondence between its prediction results and the fretting fatigue damage mechanism is still unclear, and its physical interpretability needs further improvement.

[0007] Chinese invention patent CN117993304B discloses a "physical information-driven machine learning prediction method for notched fatigue life of metallic materials." This technology calculates fatigue damage parameters analytically and considers the influence of geometric discontinuities on local stress concentration, thus achieving prediction of notched fatigue life. However, in fretting contact interfaces with complex contact states, relying solely on analytical methods is insufficient to accurately characterize the actual contact state, thereby limiting the application of this method in the field of fretting fatigue life prediction to some extent.

[0008] Chinese invention patent CN114297887B discloses "A method for constructing a high-temperature fretting fatigue life prediction model considering surface hardness." This patent, based on nonlinear cumulative damage theory, combines finite element simulation with a surface hardness-related damage accumulation rate factor and a temperature-related correction term to predict high-temperature fretting fatigue life under dissimilar material contact conditions. However, this method only considers the evolution of hardness with temperature and does not incorporate surface integrity parameters such as surface morphology, roughness, and residual stress. Therefore, its predictive ability for the coupled effects of multiple factors under complex service conditions remains very limited. Furthermore, this method is essentially still a life prediction model based on finite element simulation and a continuous damage model, making it difficult to accurately establish the complex nonlinear relationship between surface integrity parameters, fretting condition parameters, and fretting fatigue life when dealing with large datasets.

[0009] In summary, there is an urgent need to propose a fretting fatigue life prediction method that can comprehensively consider fretting condition parameters and surface integrity parameters, introduce physical damage parameters, and integrate physical constraints during model training, so as to achieve high-precision, strong generalization and high physical interpretability prediction of fretting fatigue life under complex service conditions. Summary of the Invention

[0010] To overcome the limitations of existing fretting fatigue life prediction methods, which struggle to simultaneously consider the coupled effects of high-temperature, high-frequency vibration conditions and surface integrity parameters, and suffer from insufficient physical interpretability, this invention proposes a fretting fatigue life prediction method that integrates physical information and surface integrity parameters. This invention comprehensively incorporates fretting condition parameters, surface integrity parameters, and physical damage parameters, and integrates physical constraint mechanisms during model training to achieve high-precision prediction of fretting fatigue life under complex service conditions. This provides technical support for the safety assessment and life design of tenon joint structures in aero-engines.

[0011] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A fretting fatigue life prediction method integrating physical information and surface integrity parameters, comprising the following steps: Step S1: Obtain the basic dataset for fretting fatigue. Specifically: Fretting fatigue tests were conducted on the target material under high temperature and high frequency vibration conditions. Fretting condition parameters, surface integrity parameters, and corresponding fretting fatigue life data were collected to obtain a basic dataset of fretting fatigue. The fretting condition parameters included the fretting frequency. f , micro displacement amplitude d Normal load F n Tangential load F t and ambient temperature T The surface integrity parameters include surface morphology and residual stress. s res Surface hardness HV and surface roughness Ra .

[0012] Furthermore, the fretting fatigue test is conducted in a high-temperature environment, with a test temperature range of 600–800℃, a fretting frequency range of 500–3000Hz, and a fretting displacement amplitude range of 50–200μm.

[0013] Step S2: Construct a finite element model of fretting fatigue in fretting contact. Specifically: By setting the fretting pad and fatigue specimen as a contact pair, the coordinate parameters of the two-dimensional morphology of the original surface after specimen processing are extracted to establish a fretting contact finite element model with a realistic surface morphology. Simultaneously, residual stress is introduced in the form of a prestress field, and hardness is introduced through the Arcard wear model. This results in a fretting contact finite element model that simultaneously considers surface integrity parameters such as surface morphology, residual stress, and hardness.

[0014] Furthermore, the surface hardness is introduced into the finite element model of fretting contact using the Archard wear model, and the local wear depth in the contact area is calculated according to formula (1): (1) in, Dh This represents the increment of local wear depth within one load cycle; k represents the wear coefficient. Δs H represents the cumulative relative slip distance of the contact area within one load cycle; H represents the surface hardness of the material. This represents the local contact pressure in the contact area. After calculating the local wear depth at each location in the contact area according to formula (1), the geometry of the contact surface is updated, and the finite element method is re-solved based on the updated contact surface to obtain the local stress-strain response of the contact area considering the influence of surface morphology, residual stress and surface hardness.

[0015] Step S3: Based on classical contact mechanics theory, the finite element model of fretting contact constructed in step S2 is simulated and analyzed. Under a given fretting condition, the stress-strain response of the fretting contact area within one load cycle is extracted to obtain the corresponding stress-strain curve. The maximum normal stress on the plane where the maximum principal strain range is located is determined. Based on this, the corrected fretting damage parameter is calculated using formula (2). (2) in, This represents the fretting damage parameter that takes into account corrections for surface integrity parameters such as surface morphology, residual stress, and hardness. s max This represents the maximum normal stress on the plane containing the maximum principal strain range; This indicates the maximum principal strain range on the plane containing the maximum principal strain range. Step S4: Based on the fretting fatigue baseline dataset obtained in step S1, and combined with the corrected fretting damage parameters calculated in S3. We constructed a training dataset for training the micro-motion fatigue life prediction model. Step S5: Construct a composite loss function that incorporates physical constraints. The composite loss function Includes data-driven loss term Composite loss function with physical constraint loss term As shown in formulas (3) to (5): (3) (4) (5) in, NThis represents the number of samples used in training the micro-motion fatigue life prediction model. i Indicates the ID of any training sample; j Representation and training samples i The number of another training sample that constitutes the physical constraint relationship; This indicates the first value predicted by the fretting fatigue life prediction model. i The micro-motion fatigue life of each training sample; This indicates the first value predicted by the fretting fatigue life prediction model. j The micro-motion fatigue life of each training sample; Indicates the first i The experimentally measured fretting fatigue life of a training sample; Ω represents the weighting coefficient of the physical constraint loss term, used to adjust the influence of the physical constraint loss term in the composite loss function; Ω is the set of training sample pairs that satisfy the relationship between the magnitudes of the micro-motion damage parameters. K This represents the number of training sample pairs in the training sample pair set Ω.

[0016] Step S6: Based on the composite loss function established in step S5 as shown in formula (3), construct a micro-motion fatigue life prediction model that integrates physical information.

[0017] The fretting fatigue life prediction model includes an input layer, a feature extraction layer, a feature fusion layer, and an output layer. The input layer is used to input fretting condition parameters, surface integrity parameters, and the corrected fretting damage parameters obtained in step S3. The feature extraction layer is used to extract features from the fretting condition parameters, surface integrity parameters, and fretting damage parameters respectively. The feature fusion layer is used to fuse the extracted feature results. The output layer outputs the corresponding fretting fatigue life prediction result. Specifically: For any training sample pair in the training sample pair set Ω, i , j When the training sample is determined based on the corrected micromotion damage parameters obtained in step S3... i The damage parameter is smaller than that of another training sample. j The damage parameters, and the training samples predicted by the fretting fatigue life prediction model. i Fretting fatigue life Larger than another training sample j Fretting fatigue life This is considered to follow the evolution law of fretting fatigue damage; conversely, when the training sample is judged based on the corrected fretting damage parameters obtained in step S3... i The damage parameter is greater than that of another training sample. j The damage parameters, the training samples predicted by the fretting fatigue life prediction model iFretting fatigue life On the contrary, it is greater than another training sample. j Fretting fatigue life This is considered a violation of the evolution law of fretting fatigue damage, and the fretting fatigue life prediction model is penalized by a physical constraint loss term, thereby guiding the fretting fatigue life prediction model to learn a life prediction mapping relationship that conforms to the evolution law of fretting fatigue damage.

[0018] Step S7: Training the fretting fatigue life prediction model and predicting the fretting fatigue life.

[0019] The fretting fatigue life prediction model obtained in step S6 is trained using the training dataset from step S4, by minimizing the composite loss function. The parameters of the fretting fatigue life prediction model are iteratively optimized; when the preset termination condition is reached, the iteration is stopped and the trained fretting fatigue life prediction model is obtained; the fretting working condition parameters and surface integrity parameters of the component to be evaluated are input into the trained fretting fatigue life prediction model, and the corresponding fretting fatigue life prediction results are output.

[0020] Furthermore, the preset termination condition includes at least one of the following: The number of iterations has reached the preset maximum number of iterations; The rate of change of the composite loss function over several consecutive iterations is less than a preset threshold; the preset threshold range is 1×10. -3 ~1×10 -5 ; The prediction error on the validation dataset no longer decreases in several consecutive iterations.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention improves the physical interpretability of fretting fatigue life prediction by introducing fretting damage parameters as inputs and constraints to the fretting fatigue life prediction model. Explicitly incorporating surface integrity parameters such as residual stress and roughness into the fretting fatigue life prediction model effectively characterizes the influence of surface condition on fretting fatigue behavior. Even under small sample conditions, this invention maintains high prediction accuracy and generalization ability. It is applicable to different materials and different fretting conditions, and has good engineering application value. Attached Figure Description

[0022] Figure 1 A flowchart of the fretting fatigue life prediction method for a certain metallic material provided by the present invention; Figure 2 This is a diagram of the physical information-neural network architecture built in step S6 of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the specific embodiments of this invention will be further described in detail below with reference to the accompanying drawings, taking the prediction of fretting fatigue life of nickel-based superalloys as an example. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0024] This embodiment provides a method for predicting fretting fatigue life by integrating physical information and surface integrity parameters. See [link to relevant documentation]. Figure 1 The specific steps are as follows: Step S1: Obtain the basic dataset for fretting fatigue. Specifically: Fretting fatigue tests were conducted on the target material under high temperature and high frequency vibration conditions. Fretting condition parameters, surface integrity parameters, and corresponding fretting fatigue life data were collected to obtain a basic dataset of fretting fatigue. The fretting condition parameters included the fretting frequency. f , micro displacement amplitude d Normal load F n Tangential load F t and ambient temperature T The surface integrity parameters include surface morphology and residual stress. s res Surface hardness HV and surface roughness Ra .

[0025] In this embodiment, the target material was selected as a certain type of nickel-based superalloy, and a fretting fatigue test was carried out using a flat plate sample.

[0026] In the embodiment, the fretting fatigue test was conducted in a high-temperature environment with a test temperature range of 600, 650, 700, 750, and 800°C, a fretting frequency range of 500, 1000, 1500, 2000, and 3000 Hz, and a fretting displacement amplitude range of 50, 80, 100, 150, and 200 μm.

[0027] In this embodiment, surface residual stress was measured by X-ray diffraction, surface roughness was measured by a three-dimensional surface topography instrument, and surface hardness was measured by a micro Vickers hardness tester.

[0028] Step S2: Construct a finite element model of fretting fatigue in fretting contact. Specifically: By setting the fretting pad and fatigue specimen as a contact pair, the coordinate parameters of the two-dimensional morphology of the original surface after specimen processing are extracted to establish a fretting contact finite element model with a realistic surface morphology. Simultaneously, residual stress is introduced in the form of a prestress field, and hardness is introduced through the Arcard wear model. This results in a fretting contact finite element model that simultaneously considers surface integrity parameters such as surface morphology, residual stress, and hardness.

[0029] In this embodiment, the surface hardness is introduced into the micro-motion contact finite element model through the Archard wear model, and the local wear depth in the contact area is calculated according to formula (1): (1) in, Dh This represents the increment of local wear depth within one load cycle; k represents the wear coefficient. Δs H represents the cumulative relative slip distance of the contact area within one load cycle; H represents the surface hardness of the material. This represents the local contact pressure in the contact area. After calculating the local wear depth at each location in the contact area according to formula (1), the geometry of the contact surface is updated, and the finite element method is re-solved based on the updated contact surface to obtain the local stress-strain response of the contact area considering the influence of surface morphology, residual stress and surface hardness.

[0030] In one specific embodiment, the two-dimensional surface topography coordinate parameters are obtained from the measured data of the three-dimensional surface topography instrument, and the key contours are extracted by data processing software for finite element modeling.

[0031] In one specific embodiment, the finite element model sets the micro-movement pad and the fatigue specimen to face-to-face contact, and the contact area is divided with locally refined mesh to improve the accuracy of stress-strain calculation in the critical area.

[0032] Step S3: Based on classical contact mechanics theory, the finite element model of fretting contact constructed in step S2 is simulated and analyzed. Under a given fretting condition, the stress-strain response of the fretting contact area within one load cycle is extracted to obtain the corresponding stress-strain curve. The maximum normal stress on the plane where the maximum principal strain range is located is determined. Based on this, the corrected fretting damage parameter is calculated using formula (2). (2) in, This represents the fretting damage parameter that takes into account corrections for surface integrity parameters such as surface morphology, residual stress, and hardness. s max This represents the maximum normal stress on the plane containing the maximum principal strain range; This indicates the maximum principal strain range on the plane containing the maximum principal strain range. Step S4: Based on the basic dataset of fretting fatigue obtained in step S1, and combined with the corrected fretting damage parameters calculated in S3, construct a training dataset for training the fretting fatigue life prediction model. Step S5: Construct a composite loss function that incorporates physical constraints. The composite loss function Includes data-driven loss term Composite loss function with physical constraint loss term As shown in formulas (3) to (5): (3) (4) (5) in, N This represents the number of samples used in training the micro-motion fatigue life prediction model. i Indicates the ID of any training sample; j Representation and training samples i The number of another training sample that constitutes the physical constraint relationship; This indicates the first value predicted by the fretting fatigue life prediction model. i The micro-motion fatigue life of each training sample; This indicates the first value predicted by the fretting fatigue life prediction model. j The micro-motion fatigue life of each training sample; Indicates the first i The experimentally measured fretting fatigue life of a training sample; Ω represents the weighting coefficient of the physical constraint loss term, used to adjust the influence of the physical constraint loss term in the composite loss function; Ω is the set of training sample pairs that satisfy the relationship between the magnitudes of the micro-motion damage parameters. K This represents the number of training sample pairs in the training sample pair set Ω.

[0033] Step S6: Based on the composite loss function established in step S5 as shown in formula (3), construct a fretting fatigue life prediction model that integrates physical information. The fretting fatigue life prediction model includes an input layer, a feature extraction layer, a feature fusion layer, and an output layer. The input layer of the fretting fatigue life prediction model is used to input fretting condition parameters, surface integrity parameters, and the corrected fretting damage parameters obtained in step S3; the feature extraction layer is used to extract features from the fretting condition parameters, surface integrity parameters, and fretting damage parameters respectively; the feature fusion layer is used to fuse the above feature extraction results; and the output layer is used to output the corresponding fretting fatigue life prediction results. Specifically: For any training sample pair in the training sample pair set Ω, i ,j When the training sample is determined based on the corrected micromotion damage parameters obtained in step S3... i The damage parameter is smaller than that of another training sample. j The damage parameters, and the training samples predicted by the fretting fatigue life prediction model. i Fretting fatigue life Larger than another training sample j Fretting fatigue life This is considered to follow the evolution law of fretting fatigue damage, and the physical constraint loss calculated according to formula (5) is 0; conversely, when the training sample is judged according to the corrected fretting damage parameters obtained in step S3... i The damage parameter is greater than that of another training sample. j The damage parameters, the training samples predicted by the fretting fatigue life prediction model i Fretting fatigue life On the contrary, it is greater than another training sample. j Fretting fatigue life If it is considered to violate the law of fretting fatigue damage evolution, the physical constraint loss is calculated according to formula (5) and the fretting fatigue life prediction model is penalized, thereby guiding the fretting fatigue life prediction model to learn the life prediction mapping relationship that conforms to the law of fretting fatigue damage evolution.

[0034] In one specific embodiment, the fretting fatigue life prediction model adopts a multi-layer neural network structure, the overall architecture of which is as follows: Figure 2 As shown.

[0035] Step S7: Training the fretting fatigue life prediction model and predicting the fretting fatigue life.

[0036] The fretting fatigue life prediction model obtained in step S6 is trained using the training dataset from step S4, by minimizing the composite loss function. The parameters of the fretting fatigue life prediction model are iteratively optimized. When a preset termination condition is reached, the iteration stops, and the trained fretting fatigue life prediction model is obtained. The fretting condition parameters, surface integrity parameters, and corrected fretting damage parameters of the component to be evaluated are input into the trained fretting fatigue life prediction model, and the corresponding fretting fatigue life prediction results are output. In this embodiment, the preset termination condition is that the rate of change of the composite loss function in several consecutive iterations is less than 1×10⁻⁶. -3 .

[0037] The above embodiments are merely illustrative of the implementation methods of the present invention, but should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the protection scope of the present invention.

Claims

1. A method of predicting the fretting fatigue life by fusing physical information and surface integrity parameters, characterized in that, The fretting fatigue life prediction method includes the following steps: Step S1: Obtain the basic dataset of fretting fatigue; specifically: Fretting fatigue tests were conducted on the target material under high temperature and high frequency vibration conditions. Fretting condition parameters, surface integrity parameters and corresponding fretting fatigue life data were collected to obtain a basic dataset of fretting fatigue. Step S2: Construct a finite element model of fretting fatigue in fretting contact; specifically: The fretting pad and fatigue specimen are set as a contact pair. The coordinate parameters of the two-dimensional morphology of the original surface of the specimen after processing are extracted to establish a fretting contact finite element model with a real surface morphology. At the same time, residual stress is introduced in the form of a prestress field and hardness is introduced through the Arcard wear model. Finally, a fretting contact finite element model that considers surface integrity parameters such as surface morphology, residual stress and hardness is established. Step S3, simulation analysis is performed on the fretting contact finite element model constructed in step S2, the stress-strain response of the fretting contact area in a load cycle is extracted under a given fretting working condition, the corresponding stress-strain curve is obtained, the maximum normal stress on the plane where the maximum principal strain range is located is determined, and then the corrected fretting damage parameter is calculated Step S4, based on the micro-oscillation fatigue basic data set obtained in step S1, and combined with the corrected micro-oscillation damage parameters calculated in S3 , a training data set for training the micro-oscillation fatigue life prediction model is constructed; Step S5, constructing a composite loss function with physical constraints , the composite loss function containing a data-driven loss term and a physical constraint loss term ; Step S6: Based on step S5, establish a composite loss function and construct a fretting fatigue life prediction model that integrates physical information; The fretting fatigue life prediction model includes an input layer, a feature extraction layer, a feature fusion layer, and an output layer. The input layer is used to input fretting condition parameters, surface integrity parameters, and the corrected fretting damage parameters obtained in step S3. The feature extraction layer is used to extract features from the fretting condition parameters, surface integrity parameters, and fretting damage parameters, respectively. The feature extraction results are fused through the feature fusion layer, and the fretting fatigue life prediction results are output through the output layer. Step S7: Training the fretting fatigue life prediction model and predicting the fretting fatigue life. The micro-fatigue life prediction model obtained in step S6 is trained by using the training data set in step S4, and the composite loss function is minimized The micro-fatigue life prediction model parameters are iteratively optimized; when the preset termination condition is reached, the iteration is stopped, and the trained micro-fatigue life prediction model is obtained; the fretting working condition parameters and the surface integrity parameters of the component to be evaluated are input into the trained micro-fatigue life prediction model, and the corresponding micro-fatigue life prediction result is output.

2. The method of predicting the fretting fatigue life by fusing physical information and surface integrity parameters according to claim 1, characterized in that, In step S1: The fretting fatigue test was conducted in a high-temperature environment, with a test temperature range of 600–800℃, a fretting frequency range of 500–3000Hz, and a fretting displacement amplitude range of 50–200μm.

3. The method of predicting the fretting fatigue life by fusing the physical information and the surface integrity parameter according to claim 2, characterized in that, In step S1: The micro-motion working condition parameters include micro-motion frequency f , micro-motion displacement amplitude δ , normal load F n , tangential load F t and ambient temperature T , and the surface integrity parameters include surface topography, residual stress σ res , surface hardness HV and surface roughness Ra .

4. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 3, characterized in that, In step S2: Surface hardness is introduced into the micro-motion contact finite element model through the Archard wear model, and the local wear depth in the contact area is calculated according to formula (1): (1) in, Δh This represents the increment of local wear depth within one load cycle; k represents the wear coefficient. Δs H represents the cumulative relative slip distance of the contact area within one load cycle; H represents the surface hardness of the material. This indicates the local contact pressure in the contact area; After calculating the local wear depth at each location in the contact area according to formula (1), the geometry of the contact surface is updated, and the finite element method is re-solved based on the updated contact surface to obtain the local stress-strain response of the contact area considering the influence of surface morphology, residual stress and surface hardness.

5. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 4, characterized in that, In step S3, the corrected fretting damage parameters are calculated using formula (2). (2) in, σ max This represents the maximum normal stress on the plane containing the maximum principal strain range; This indicates the maximum principal strain range on the plane containing the maximum principal strain range.

6. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 5, characterized in that, In step S5, the composite loss function that incorporates physical constraints is shown in equations (3) to (5): (3) (4) (5) in, N This represents the number of samples used in training the micro-motion fatigue life prediction model. i Indicates the ID of any training sample; j Representation and training samples i The number of another training sample that constitutes the physical constraint relationship; This indicates the first value predicted by the fretting fatigue life prediction model. i The micro-motion fatigue life of each training sample; This indicates the first value predicted by the fretting fatigue life prediction model. j The micro-motion fatigue life of each training sample; Indicates the first i The experimentally measured fretting fatigue life of a training sample; Ω represents the weighting coefficient of the physical constraint loss term, used to adjust the influence of the physical constraint loss term in the composite loss function; Ω is the set of sample pairs that satisfy the relationship between the magnitudes of the fretting damage parameters. K This represents the number of sample pairs in the sample pair set Ω.

7. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 6, characterized in that, In step S6: For any sample pair in the set of sample pairs Ω ( i , j When judging the sample based on the corrected fretting damage parameters obtained in step S3... i The damage parameter is smaller than that of another sample. j The damage parameters, and the sample predicted by the fretting fatigue life prediction model. i Fretting fatigue life Larger than another sample j Fretting fatigue life This is considered to follow the evolution law of fretting fatigue damage; conversely, when judging the sample based on the corrected fretting damage parameters obtained in step S3... i The damage parameter of one sample is greater than that of another sample. j The damage parameters, the sample predicted by the fretting fatigue life prediction model i Fretting fatigue life On the contrary, it was larger than another sample. j Fretting fatigue life This is considered a violation of the evolution law of fretting fatigue damage, and the fretting fatigue life prediction model is penalized by a physical constraint loss term, thereby guiding the fretting fatigue life prediction model to learn a life prediction mapping relationship that conforms to the evolution law of fretting fatigue damage.

8. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 7, characterized in that, In step S7, the preset termination condition includes at least one of the following: The number of iterations has reached the preset maximum number of iterations; The rate of change of the composite loss function in several consecutive iterations is less than a preset threshold; The prediction error on the validation dataset no longer decreases in several consecutive iterations.

9. The fretting fatigue life prediction method based on the fusion of physical information and surface integrity parameters according to claim 8, characterized in that, The preset threshold range is 1 x 10 -3 ~ 1 x 10 -5 .

Images