Thermal coupling simulation method and system for sliding bearing under dynamic load

Abstract

The invention discloses a thermal-mechanical coupling simulation method and system for a sliding bearing under a dynamic load, and relates to the field of multi-physics field coupling simulation, and the method comprises the steps: building a multi-physics field coupling model through multi-physics field modeling according to the properties of a three-dimensional model, a multi-physics field grid, dynamic load data, lubricating oil and a shaft material of the sliding bearing; and outputting transient multi-physical field response data, calculating the surface abrasion loss of the bearing bush through the transient multi-physical field response data, arranging the surface abrasion loss into surface abrasion distribution data of the bearing bush, and adjusting and updating the surface geometric morphology of the bearing bush by adopting a dynamic grid according to the surface abrasion distribution data of the bearing bush to obtain an updated multi-physical field grid. And synchronously updating transient multi-physical field response data, and constructing a bearing performance prediction model to obtain performance prediction and fault diagnosis results of the sliding bearing. Performance prediction and fault diagnosis of the sliding bearing are realized through the bearing performance prediction model, and the fault detection capability is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of multi-physics field coupling simulation, and in particular to a method and system for thermal-mechanical coupling simulation under dynamic load of a sliding bearing. Background Art

[0002] As a key component in mechanical transmission systems, the performance of sliding bearings directly impacts the stability and efficiency of the entire system. In recent years, the continuous advancement of industrial technology, especially the increasing demand for high-performance mechanical equipment, has placed higher demands on the precise design and optimization of sliding bearings. Traditional sliding bearing analysis methods rely primarily on experimental testing and empirical formulas. While this approach can meet engineering requirements to a certain extent, it often lacks sufficient accuracy and reliability when faced with dynamic loads under complex operating conditions.

[0003] First, the ability to simulate the wear process is limited, making it difficult to fully consider the impact of factors such as lubrication conditions, temperature changes, and material properties on the wear behavior of sliding bearings. Furthermore, there is a lack of effective methods for predicting the performance degradation of sliding bearings under long-term operating conditions. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a thermal-mechanical coupling simulation method for sliding bearings under dynamic loads to solve the problems of limited ability in simulating the wear process and lack of performance degradation prediction.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for thermal-mechanical coupling simulation of a sliding bearing under dynamic load, which comprises:

[0008] Create a 3D model of the sliding bearing, create a multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material;

[0009] Based on the three-dimensional model of the sliding bearing, the multi-physics grid, the dynamic load data, the properties of the lubricant and the material, a multi-physics coupling model is constructed using multi-physics modeling to output transient multi-physics response data.

[0010] The bearing surface wear is calculated through transient multi-physics field response data and organized into bearing surface wear distribution data;

[0011] Based on the bearing surface wear distribution data, dynamic mesh adjustment is used to update the bearing surface geometry to obtain an updated multi-physics mesh, and the transient multi-physics response data is simultaneously updated.

[0012] Construct a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings;

[0013] The performance prediction and fault diagnosis results of the sliding bearing are visualized to obtain a sliding bearing analysis report.

[0014] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing of the present invention, wherein: the creating of the three-dimensional model of the sliding bearing and dividing the multi-physics field grid refers to constructing the geometric structure of the sliding bearing using computer-aided design software to obtain the three-dimensional geometric model of the sliding bearing;

[0015] The three-dimensional geometric model is meshed in the multi-physics field simulation pre-processing software to complete the multi-physics field mesh division.

[0016] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing described in the present invention, the properties of the lubricating oil include density, specific heat capacity, thermal conductivity, viscosity, elastic modulus and thermal expansion coefficient, the properties of the material include elastic modulus and thermal expansion coefficient, and the properties of the lubricating oil and the material are organized into a material table.

[0017] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing described in the present invention, the output of transient multi-physical field response data specifically includes the following steps:

[0018] Import the three-dimensional geometric model, multi-physics mesh, dynamic load sequence, and material table of the sliding bearing into the multi-physics simulation platform, and configure the thermal calculation model, fluid calculation model, and mechanical calculation model to complete the establishment of the multi-physics coupling model;

[0019] Configure initial and boundary conditions for the multiphysics coupling model and perform multiphysics coupling simulation to generate transient multiphysics response data.

[0020] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing described in the present invention, wherein: the bearing surface wear amount is calculated through transient multi-physical field response data and organized into bearing surface wear distribution data, specifically including the following steps:

[0021] The single wear depth of the bearing surface is calculated using the Archimedean wear model based on the transient multi-physics field response data;

[0022] According to the laws of physics, the bearing surface is divided into high-pressure areas and other areas. The cumulative wear of the bearing surface within the total simulation time is calculated based on the single wear depth. The cumulative wear is compared with the experimental data to form the bearing surface wear distribution data.

[0023] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing described in the present invention, wherein: according to the wear distribution data of the bearing surface, the bearing surface geometry is updated by dynamic grid adjustment to obtain an updated multi-physics field grid, and the transient multi-physics field response data is updated synchronously, specifically comprising the following steps:

[0024] The bearing surface wear distribution data is converted into the position movement of the bearing surface grid points through dynamic grid adjustment, and the update frequency of dynamic grid adjustment is set to update the bearing surface geometry;

[0025] Based on the updated bearing surface geometry, the multiphysics mesh is regenerated, multiphysics coupling simulation is performed, and the transient multiphysics response data is updated synchronously.

[0026] As a preferred solution of the thermal-mechanical coupling simulation method under dynamic load of the sliding bearing described in the present invention, wherein: constructing a bearing performance prediction model to obtain the performance prediction and fault diagnosis results of the sliding bearing specifically includes the following steps:

[0027] Based on the transient multi-physics field response data, a state space model is established and a Kalman filter is used to generate Bayesian filter detection results.

[0028] Extracting input features and output labels from transient multi-physics field response data and Bayesian filter detection results as a training set and performing normalization processing, defining the structure of a neural network active learning framework, and performing training to obtain a neural network active learning framework, wherein the input features include gap, oil groove angle, and amplitude;

[0029] Integrate Kalman filtering and neural network active learning framework to build a bearing performance prediction model, input characteristics and transient multi-physics field response data into the bearing performance prediction model, and output the performance prediction and fault diagnosis results of the sliding bearing.

[0030] In a second aspect, the present invention provides a thermal-mechanical coupling simulation system for sliding bearings under dynamic loads, comprising:

[0031] Create a module to create a 3D model of the sliding bearing, divide the multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material;

[0032] Build a module that uses multi-physics modeling to build a multi-physics coupling model based on the three-dimensional model of the sliding bearing, dynamic load data, and the properties of the lubricant and material, and outputs transient multi-physics response data;

[0033] The wear module calculates the bearing surface wear through transient multi-physics field response data and organizes it into bearing surface wear distribution data;

[0034] The update module uses dynamic mesh adjustment to update the bearing surface geometry based on the bearing surface wear distribution data, obtains the updated multi-physics field mesh, and simultaneously updates the transient multi-physics field response data;

[0035] Prediction module, which builds a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings;

[0036] The analysis module visualizes the performance prediction and fault diagnosis results of the sliding bearing and obtains the sliding bearing analysis report.

[0037] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the thermal-mechanical coupling simulation method under dynamic load of a sliding bearing as described in the first aspect of the present invention is implemented.

[0038] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the thermal-mechanical coupling simulation method under dynamic load of a sliding bearing as described in the first aspect of the present invention is implemented.

[0039] The beneficial effects of the present invention are as follows: through transient multi-physics field response data, combined with the Archimedean wear model, the surface wear amount of the bearing is calculated and organized into wear distribution data, special consideration is given to the influence of lubrication conditions and temperature on wear, the wear behavior of the sliding bearing is quantified, and potential wear hotspots can be discovered in advance, providing a basis for long-term performance prediction; in addition, according to the bearing surface wear distribution data, dynamic grid adjustment technology is used to update the bearing surface geometry, and the transient multi-physics field response data is updated synchronously, realizing real-time reflection of the impact of wear on the geometry, maintaining the accuracy of the simulation model, and improving the timeliness and accuracy of the simulation results, and also based on the state space model and Kalman filtering technology, combined with the neural network active learning framework, realizing performance prediction and fault diagnosis of sliding bearings, and enhancing fault detection capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] 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 description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 Flowchart of the thermal-mechanical coupling simulation method for sliding bearings under dynamic loads.

[0042] Figure 2System diagram for thermal-mechanical coupling simulation of sliding bearings under dynamic loads.

[0043] Figure 3 Schematic diagram for multiphysics modeling.

[0044] Figure 4 This is the architecture diagram of the bearing performance prediction model. DETAILED DESCRIPTION

[0045] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0047] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0048] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a thermal-mechanical coupling simulation method for a sliding bearing under dynamic load, comprising the following steps:

[0049] S1. Create a 3D model of the sliding bearing, divide the multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material.

[0050] The specific steps include:

[0051] S1.1. First, construct a 3D geometric model of the sliding bearing. Use computer-aided design software (such as SolidWorks or CATIA) to create the geometry of the shaft, bearing, and lubricant film. Specifically, define the shaft geometry in the computer-aided design software, for example, a shaft diameter of 60 mm, a length of 120 mm, and a material of 40Cr steel, which has high strength and wear resistance. The bearing geometry includes an inner diameter of 60.06 mm and a thickness of 6 mm. The material is Babbitt B83, which is suitable for withstanding sliding friction. The lubricant film geometry is set to an initial thickness of 0.06 mm, filling the space between the shaft and the bearing to reflect the lubrication state.

[0052] To ensure geometric accuracy, a high-precision surface profilometer is used to measure the bearing surface roughness and obtain bearing surface roughness measurement data. This data is stored as a point cloud, a set of three-dimensional coordinate points that describes the bearing surface geometry and roughness characteristics. This data is then imported into computer-aided design software to adjust the bearing inner surface geometry and reflect actual manufacturing features, such as minor bumps or machining marks.

[0053] Preferably, computer-aided design software is used to create a three-dimensional geometric model of the sliding bearing to ensure accurate geometric dimensions, provide a realistic basis for subsequent multi-physics field simulation, and avoid calculation errors caused by geometric deviations.

[0054] S1.2. After completing the three-dimensional geometric model of the sliding bearing, load the geometric structure in the multi-physics field simulation pre-processing software (such as ANSYS) and perform meshing. The meshing steps are as follows: Use a fine tetrahedral mesh in the lubricating oil film area to capture the changes in fluid pressure and velocity, such as the pressure gradient in the center of the oil film. Use a hexahedral mesh in the shaft and bearing area to reduce the amount of calculation, such as for the calculation of the stress of the outer layer of the bearing. In the contact area between the bearing and the lubricating oil film, set a transition mesh, which can be a 5-layer transition mesh, with a unit size gradually changing from 0.01 mm to 0.2 mm, for example, from the oil film boundary to the inner wall of the bearing, to ensure numerical continuity and reduce pressure calculation errors.

[0055] After meshing is complete, physical field assignment is performed. Specifically, fluid flow and heat transfer properties are assigned to the lubricating oil film area. The oil film pressure is calculated based on the Navier-Stokes equations, and the temperature distribution is calculated based on the energy conservation equation, for example, a temperature of 20 degrees Celsius at the inlet. Elastic deformation and heat transfer properties are assigned to the shaft and bearing areas. Stress is calculated based on the elastic equations, and thermal expansion deformation is calculated based on the thermal expansion formula, for example, the deformation of the bearing when subjected to 60 MPa pressure.

[0056] Begin by setting the initial conditions for the 3D sliding bearing geometry, including the shaft rotational speed (also known as the sliding velocity), ambient temperature, lubricant inlet pressure, and external heat dissipation capacity. The external heat dissipation capacity is 11 watts per square meter per Kelvin, resulting in a heat dissipation rate of 440 watts per square meter on the bearing's outer surface. Then, save the 3D sliding bearing geometry and mesh data (i.e., the multiphysics mesh).

[0057] Preferably, fluid flow and heat transfer properties are assigned to the lubricating oil film region, and mechanical deformation and heat transfer properties are assigned to the shaft and bearing pad region, to ensure that the multi-physics coupling calculation reflects the real phenomenon.

[0058] S1.3. Refer to the historical operating data of the wind turbine gearbox, use force sensors to collect real loads, capture the force characteristics of the bearings under complex working conditions, and obtain a dynamic load sequence. The dynamic load sequence includes constant loads and vibration loads. The vibration load is composed of the superposition of three independent frequency components. Each frequency corresponds to a physical phenomenon, which is divided into low-frequency vibration, medium-frequency vibration and high-frequency vibration. For example, the frequency of low-frequency vibration is 25 Hz and the amplitude is 2.5 kN, which simulates the low-frequency vibration of gear meshing. The frequency of medium-frequency vibration is 60 Hz and the amplitude is 1.8 kN, which reflects the rotor imbalance effect. The frequency of high-frequency vibration is 110 Hz and the amplitude is 0.9 kN, which indicates the disturbance of high-speed components.

[0059] To simulate real-world operating conditions, the start times of each vibration were slightly offset—for example, low-frequency vibration was delayed by 0.01 seconds, medium-frequency vibration by 0.02 seconds, and high-frequency vibration by 0.03 seconds—to create a cumulative effect. Signal processing techniques were used in the data processing software to filter out high-frequency interference above 200 Hz, preserving the primary vibration signature.

[0060] Set the simulation time interval, generate the total load value for each time step, record the time and load value for each generated total load value, and organize the recorded time and load values ​​for each data point into a dynamic load data table containing the time and load values.

[0061] S1.4. Define the properties of lubricants and materials based on standard test data. The properties of lubricants include density, specific heat capacity, thermal conductivity, viscosity, elastic modulus, and thermal expansion coefficient. Define the properties of shaft materials (using 40Cr steel as an example), including elastic modulus and thermal expansion coefficient, and organize them into a material table.

[0062] S2. Based on the three-dimensional model of the sliding bearing, dynamic load data, and the properties of the lubricant and material, multi-physics field modeling is used to build a multi-physics field coupling model and output transient multi-physics field response data.

[0063] The specific steps include:

[0064] S2.1. Import the 3D geometric model and mesh data of the sliding bearing into the multi-physics simulation platform, load the dynamic load data table and material table, configure the thermal, fluid, and mechanical calculation models, and process the physical fields of the oil film, bearing, and shaft respectively. The first step is fluid calculation. Specifically, for the lubricating oil film area, calculate the oil film pressure distribution based on the Navier-Stokes equation, taking into account the lubricating oil viscosity, oil film thickness, and shaft rotation speed. The expression is:

[0065]

[0066] Where ρ represents density, represents the partial derivative symbol, v represents the velocity vector, t represents the time, represents the gradient operator, P represents the pressure gradient, μ represents the dynamic viscosity, represents the Laplace operator.

[0067] Next is the thermal calculation. For the oil film, bearing, and shaft, the temperature distribution is calculated based on the energy conservation equation. The transfer of frictional heat (generated by oil film shear) between the oil film, bearing, and shaft is analyzed. The expression is:

[0068]

[0069] Where ρ represents density, c p represents specific heat capacity, T represents temperature, k represents thermal conductivity, represents the friction heat source,

[0070] Finally, mechanical calculation is performed. For the bearing and shaft, deformation and stress are calculated based on the elastic mechanics equation. The expansion deformation and stress caused by the increase of oil film pressure and temperature are analyzed. The expression is:

[0071]

[0072] Where σ represents the stress tensor, f represents the body force vector, C represents the stiffness matrix, and ∈ represents the strain tensor.

[0073] For example, the stress on the inner surface of the bearing reaches 180 MPa and the deformation is 0.014 mm.

[0074] After configuring the thermal, fluid, and mechanical calculation models, we began setting up interactions, defining the coupling relationships between the physical fields. Specifically, we implemented this interaction through the multiphysics simulation platform's coupling solver (COMSOL's multiphysics interface). The coupling relationship was as follows: oil film pressure served as the external boundary condition on the bearing surface, frictional heat drove temperature changes, temperature affected lubricant viscosity by increasing temperature, enhancing molecular thermal motion and reducing lubricant viscosity, and temperature affected the material's elastic modulus by enhancing lattice vibrations and reducing the material's elastic modulus. This completed the construction of the multiphysics coupling model.

[0075] Preferably, the fluid, thermal, and mechanical calculation models are configured to process the physical fields of the oil film, bearing, and shaft respectively, laying the foundation for the multi-physics field coupling model and supporting high-precision simulation. The configured interactive relationship simulates the real working conditions of the sliding bearing, enabling the multi-physics field coupling model to accurately predict complex behaviors.

[0076] S2.2. Configure boundary conditions for the multiphysics coupling model. These boundary conditions are categorized into four types: mechanical, fluid, thermal, and kinematic, based on the physical characteristics and operating conditions of the sliding bearing. The mechanical boundary condition is that the bearing outer surface is fixed. The fluid boundary conditions are the oil film inlet and outlet pressures. The thermal boundary condition is the heat dissipation coefficient of the bearing outer surface. The kinematic boundary condition is the shaft rotation speed, for example, 4000 rpm, which drives the oil film shear.

[0077] The simulation time step must match the time interval of the dynamic load data table. The oil film pressure, temperature distribution, mechanical deformation and stress are calculated in parallel by the computer to generate transient multi-physics field response data, including oil film pressure, temperature, stress and mechanical deformation.

[0078] S3. Calculate the bearing surface wear amount through transient multi-physics field response data and organize it into bearing surface wear distribution data.

[0079] The specific steps include:

[0080] S3.1. First, prepare transient multi-physics response data, sliding velocity, and oil film thickness. Calculate the oil film thickness using the mechanical deformation of the transient multi-physics response data (e.g., 0.014 mm) and the initial thickness of the three-dimensional geometric model of the sliding bearing (e.g., 0.06 mm). Calculate the single wear depth of the bearing material using the improved wear prediction method based on the Archimedean wear model. The expression is:

[0081] Δd=k×Z×U×Δt / H;

[0082] Among them, Δd represents the single wear depth, k represents the wear coefficient, Z represents the oil film pressure, U represents the sliding velocity, Δt represents the time step, and H represents the material hardness.

[0083] The thickness of the oil film will affect the change of the wear coefficient. For example, if the oil film thickness is less than twice the average roughness, it indicates insufficient lubrication. For example, if the wear increases by 25%, the wear coefficient k will increase from the initial 1×10 -11 Adjusted to 1.25×10 -11 If the oil film thickness is greater than four times the average roughness, the lubrication is sufficient, for example, the wear is reduced by 50%, and the wear coefficient k is reduced from the initial 1×10 -11 Adjusted to 0.5×10 -11 .

[0084] Changes in temperature will also cause changes in material hardness. For example, for every 100 degrees Celsius increase in temperature, the hardness decreases by 12% and wear increases by 18%.

[0085] Preferably, the improved wear prediction method can accurately capture the dynamic changes of wear, and also integrate lubrication and temperature effects, reflecting the wear behavior of real working conditions and realistic simulations of sliding bearings.

[0086] S3.2 For high-pressure areas of the plain bearing, such as those greater than 45 MPa, use a time step faster than the time interval in the dynamic load data table to ensure that rapid wear is captured. For other areas (i.e., those less than 45 MPa), use the same time step as the time interval in the dynamic load data table. For each bearing surface grid point, calculate the cumulative wear per unit time, such as the cumulative wear per 50 seconds.

[0087] Check the uniformity of bearing surface wear. Calculate the average wear depth (e.g., 0.002 mm) for all bearing surface grid points. If the wear of a grid point exceeds twice the average, it is marked as a risk point. If no grid point exceeds the limit, it means there is no abnormality. The wear depth of all bearing surface grid points is organized into a wear depth table, and the rate per unit time (e.g., 50 seconds) is recorded. For example, the high pressure area is 8×10 per second. -8 The wear depth table and the wear rate time series are combined into the bearing surface wear distribution data.

[0088] S4. Based on the bearing surface wear distribution data, dynamic mesh adjustment is used to update the bearing surface geometry to obtain an updated multi-physics field mesh, and the transient multi-physics field response data is simultaneously updated.

[0089] The specific steps include:

[0090] Load the mesh data in the simulation software (such as COMSOL), read the bearing surface wear distribution data, and use the mesh dynamic adjustment technology to map the wear depth of the bearing surface wear distribution data to the position movement of the bearing surface grid points, reflecting the advantages of dynamic modeling. The mesh dynamic adjustment technology simulates the dynamic morphology evolution of the bearing surface under wear by moving the grid points in real time. For example, the morphology of the high-pressure area changes faster than that of the low-pressure area. At the same time, the update frequency is set to ensure the dynamic changes of the morphology and transient multi-physics field response data.

[0091] The oil film thickness is recalculated based on the updated bearing surface geometry. For example, an area with an initial oil film thickness of 0.06 mm may become 0.064 mm due to wear (e.g., 0.004 mm), a localized increase of approximately 6%. This demonstrates the power of dynamic modeling, as the dynamic adjustment of oil film thickness reflects the real-time impact of wear on lubrication conditions.

[0092] To avoid deformation of the bearing surface mesh (for example, element stretching in high-wear areas), smoothing is applied. Specifically, the position of each bearing surface mesh point is weighted by the average position of itself and its neighboring mesh points (for example, 6-10 neighboring mesh points), with a weight of 0.3 (the weight of itself is 0.7, and the average weight of neighboring mesh points is 0.3; the weight can be customized according to individual needs). Check the quality of the updated mesh and complete the smoothing of the bearing surface mesh. Smoothing updates the mesh in real time to adapt to changes in the bearing topography, ensuring that the updated mesh adapts to the dynamic working conditions and stability of the sliding bearing and meets the requirements of dynamic modeling. The mesh is then saved as the updated multiphysics mesh file.

[0093] Using the updated multi-physics mesh file, thermal, fluid, and mechanical coupling calculations are performed in the simulation platform to analyze the impact of changes in the bearing surface geometry (for example, the inner diameter increases from 60.06 mm to 60.064 mm) on the performance of the sliding bearing. Then, by repeating the previous steps to obtain transient multi-physics response data, updated transient multi-physics response data are obtained.

[0094] It is further explained that the dynamic grid adjustment technology reflects the impact of wear on the geometric morphology of the bearing surface in real time, drives the changes in the physical field, forms a dynamic modeling closed loop (wear-morphology-physical field), and updates the transient multi-physical field response data to ensure that the latest transient multi-physical field response data is used for calculations, thereby improving accuracy.

[0095] S5. Construct a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings.

[0096] The specific steps include:

[0097] S5.1. Build a state-space model based on transient multiphysics response data to describe the operating state of the sliding bearing. Specifically, first define the state vector of the state-space model, determining that the physical quantities it contains are oil film pressure, temperature, stress, and mechanical deformation, which vary over time. The state vector is the core of the state-space model, representing the operating state of the sliding bearing and encompassing the coupled relationships between flow (oil film pressure), heat (temperature), and solid (stress and mechanical deformation).

[0098] Describe the state transition process and clarify how the next state is generated. The next state is a linear transformation of the current state through the state transition matrix (4 rows and 4 columns, e.g., the oil film pressure influence coefficient on stress is 0.1), superimposed with the control input (e.g., clearance 0.05 mm, oil groove angle 10 degrees, amplitude 2 kN) through the control matrix (4 rows and 3 columns, e.g., the amplitude influence coefficient on stress is 0.05), and the process noise (e.g., Gaussian distribution with mean 0 and variance 0.01). The state transition process defines the dynamic evolution rules of the state-space model, describing how the bearing state changes over time.

[0099] Describe the observation process and clarify how observations are generated. Observations (e.g., oil film pressure 62 MPa, temperature 132°C, stress 180 MPa, and mechanical deformation 0.014 mm) are generated from the current state through the observation matrix, with observation noise superimposed. For example, an observation directly reflects the oil film pressure of 62 MPa in the transient multiphysics response data.

[0100] The state transition matrix and control matrix are fitted using the least squares method. The specific steps are: A time series (e.g., the oil film pressure series [62, 61.8, ...] MPa) is read and combined with the control input to calculate the linear relationship between the states. The state transition matrix and the control matrix are the core parameters of the state-space model, quantifying the state evolution and the influence of control inputs. The least squares method uses the transient multiphysics response data to fit the coefficients, ensuring that the state-space model is consistent with the physical behavior. This must be performed after defining the state transition and observation process, as it requires a clear matrix structure and data.

[0101] Set the observation matrix, for example, a 4-row 4-column identity matrix, which means that the observation value directly reflects the state vector. For example, the oil film pressure observation value of 62 MPa corresponds to the oil film pressure of 62 MPa in the state vector. Save the state transfer matrix, control matrix and observation matrix to complete the construction of the state space model.

[0102] S5.2. Based on the state-space model, use Kalman filtering (a linear implementation of Bayesian filtering) to dynamically monitor the state of the sliding bearing. Specifically:

[0103] Load the parameters of the state-space model (state transfer matrix, control matrix, and observation matrix), because the Kalman filter relies on the state-space model parameters to predict and update the state (for example, the oil film pressure changes from 61 MPa to 61.8 MPa). The state transfer matrix defines the state evolution, the control matrix introduces external input, and the observation matrix maps the state to the observation value. Then load the transient multi-physics field response data as the observation value. The update step of the Kalman filter requires the observation value to correct the predicted state. The representative value (center point or maximum value) is extracted because the multi-physics field grid data is too large and can still reflect the sliding bearing state after simplification.

[0104] Initialize the starting parameters of the Kalman filter, including the initial state vector, initial covariance matrix, process noise covariance, and observation noise covariance. For example, the initial covariance matrix is ​​set to a 4-row, 4-column identity matrix with a value of 0.1, indicating the uncertainty of the initial state. Use the linear algebra capabilities of the SciPy library to iteratively execute the Kalman filter, which consists of two steps: prediction and update, to dynamically monitor the oil film pressure, temperature, stress, and mechanical deformation. SciPy is an open-source Python library for scientific computing and engineering applications, providing efficient mathematical, scientific, and engineering computing tools.

[0105] Prediction involves first calculating the state at the next moment. Specifically, the current state is linearly transformed using the state transition matrix, and the control input is added to the control matrix to obtain the predicted state. The predicted covariance is then calculated by transforming the current covariance (for example, the initial value is 0.1) using the state transition matrix and adding the process noise covariance (0.01 identity matrix) to generate the predicted covariance, which represents the uncertainty of the prediction.

[0106] Next comes the update process: First, the Kalman gain is calculated. Using the prediction covariance, the observation matrix, and the observation noise covariance, a weight (the Kalman gain) is determined to balance the confidence between the prediction and observation. The predicted state is then corrected by subtracting the observed portion of the predicted state from the observed value and multiplying it by the Kalman gain. This corrects the predicted state to obtain the updated state, for example, the oil film pressure is 61.8 MPa. Finally, the covariance is updated, and the predicted covariance is adjusted using the Kalman gain to generate the updated covariance.

[0107] The Kalman filter fuses the prediction results of the state-space model and the observation values ​​of the transient multi-physics field response data to improve the monitoring accuracy. The prediction step is based on the state transfer matrix and control matrix of the state-space model to infer the state at the next moment. The update step uses the observation values ​​of the transient multi-physics field response data to correct the prediction deviation.

[0108] During the Kalman filter iteration process, the updated state is checked to see if the mechanical deformation exceeds a threshold (e.g., 0.02 mm, determined based on physical evidence). If so, the time point and state value are recorded and marked as a fault, indicating structural failure. If not, the state is marked as normal. Fault monitoring is an application goal of the Kalman filter. Mechanical deformation exceeding the threshold indicates possible structural failure of the sliding bearing, such as excessive bearing deformation leading to abnormal stress. The Kalman filter's updated state sequence and fault records are saved as Bayesian filter monitoring results.

[0109] S5.3. Based on the Bayesian filtering monitoring results, a neural network training data set is constructed, including input features and output labels, and the input features and output labels are standardized, and all features and labels are standardized to the range [0, 1].

[0110] We began developing a neural network active learning framework. We first determined the prediction target and the structure of the neural network active learning framework. The prediction target was the temperature and stress of the sliding bearing. The structure of the neural network active learning framework is an input layer with 3 neurons (corresponding to the input features: clearance, oil groove angle, and amplitude). There are two hidden layers, each with 64 neurons, using the ReLU activation function. The output layer has 2 neurons (corresponding to temperature and stress). We also introduced an attention mechanism, specifically: extracting 64-dimensional features from the hidden layer, generating query, key, and value vectors, and then calculating the force weight. The calculation content is the dot product of the query and key divided by the square root of the key dimension. The attention weight is obtained by softmax normalization, and the attention weight is multiplied by the value vector to obtain the weighted feature.

[0111] Define a loss function (mean squared error). Calculate the difference between the predicted value and the true value, square it, and average it across the training data. For example, for temperature prediction, assume the predicted temperature is 132°C and the true value is also 132°C. The difference is zero, and the loss is zero. The goal is to optimize to a temperature error of ≤1°C and a stress error of ≤1 MPa. Use the Adam optimizer during training until the optimization goals are met, completing the neural network active learning framework.

[0112] S5.4. Combining Bayesian filter monitoring results (e.g., oil film pressure of 61.8 MPa, temperature of 132°C, stress of 180 MPa, and mechanical deformation of 0.014 mm) with a neural network active learning framework (predicting temperature of 132°C and stress of 180 MPa), a bearing performance prediction model was constructed for performance prediction and fault diagnosis of sliding bearings. Specifically, the neural network active learning framework was invoked to extract and predict nonlinear features using two hidden layers and an attention mechanism, outputting predicted temperature and stress ranges.

[0113] Fault diagnosis consists of two parts. The first part, based on Bayesian filter monitoring results, detects if the mechanical deformation threshold is exceeded, marking it a fault, indicating structural failure. If the mechanical deformation threshold is not exceeded, marking it normal, indicating no fault. The mechanical deformation threshold is set according to the design specifications and material strength of the sliding bearing. The second part, based on the neural network active learning framework, detects whether the stress exceeds the stress threshold (set according to the design specifications and safety factor, for example, 205 MPa). This is marked as an abnormality, indicating a sudden stress increase. The integration of the Kalman filter and neural network active learning framework prioritizes the Bayesian filter monitoring results to check for mechanical deformation, while simultaneously running the neural network active learning framework for prediction. Example results show a temperature of 132°C, a stress of 180 MPa, a mechanical deformation of 0.014 mm, a fault flag of 0 (no fault), and an abnormality flag of 0 (no abnormality).

[0114] S6. Visualize the performance prediction and fault diagnosis results of the sliding bearing to obtain a sliding bearing analysis report.

[0115] Time series analysis can be used to observe the changing trends of various parameters over time and identify abnormal fluctuations or long-term trends. For example, if the temperature of a sliding bearing is found to be rising continuously over a period of time, it may indicate deteriorating lubrication conditions or reduced cooling efficiency.

[0116] Plotting time series of various parameters (oil film pressure, temperature, stress, mechanical deformation, wear, vibration, etc.) over time visually displays the working status of the sliding bearing and its changing patterns. This helps quickly locate the time when problems occur.

[0117] Generate a detailed sliding bearing analysis report based on the time series analysis results, including the basic operating conditions of the sliding bearing, parameter change trends, potential problem warnings, and fault diagnosis results.

[0118] This embodiment further provides a thermal-mechanical coupling simulation system for sliding bearings under dynamic loads, comprising:

[0119] Create a module to create a 3D model of the sliding bearing, divide the multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material;

[0120] Build a module that uses multi-physics modeling to build a multi-physics coupling model based on the three-dimensional model of the sliding bearing, dynamic load data, and the properties of the lubricant and material, and outputs transient multi-physics response data;

[0121] The wear module calculates the bearing surface wear through transient multi-physics field response data and organizes it into bearing surface wear distribution data;

[0122] The update module uses dynamic mesh adjustment to update the bearing surface geometry based on the bearing surface wear distribution data, obtains the updated multi-physics field mesh, and simultaneously updates the transient multi-physics field response data;

[0123] Prediction module, which builds a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings;

[0124] The analysis module visualizes the performance prediction and fault diagnosis results of the sliding bearing and obtains the sliding bearing analysis report.

[0125] This embodiment also provides a computer device, which is suitable for the case of a thermal-mechanical coupling simulation method under dynamic load of a sliding bearing, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the thermal-mechanical coupling simulation method under dynamic load of a sliding bearing proposed in the above embodiment.

[0126] The computer device may be a terminal, comprising a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner may be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse.

[0127] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the thermal-mechanical coupling simulation method for sliding bearings under dynamic loads as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0128] In summary, the present invention uses transient multi-physics field response data combined with an improved Archimedean wear model to calculate the surface wear of the bearing and organize it into wear distribution data. It especially considers the influence of lubrication conditions and temperature on wear, quantifies the wear behavior of the sliding bearing, and can discover potential wear hotspots in advance, providing a basis for long-term performance prediction; in addition, according to the bearing surface wear distribution data, dynamic grid adjustment technology is used to update the bearing surface geometry, and the transient multi-physics field response data is updated synchronously, realizing real-time reflection of the impact of wear on the geometry, maintaining the accuracy of the simulation model, and improving the timeliness and accuracy of the simulation results, and also based on the state space model and Kalman filtering technology, combined with the neural network active learning framework, realizing performance prediction and fault diagnosis of sliding bearings, and enhancing fault detection capabilities.

[0129] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A thermal-mechanical coupling simulation method for a sliding bearing under dynamic load, characterized by: include, Create a 3D model of the sliding bearing, create a multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material; Based on the three-dimensional model of the sliding bearing, the multi-physics grid, the dynamic load data, the properties of the lubricant and the material, a multi-physics coupling model is constructed using multi-physics modeling to output transient multi-physics response data. The bearing surface wear is calculated through transient multi-physics field response data and organized into bearing surface wear distribution data; Based on the bearing surface wear distribution data, dynamic mesh adjustment is used to update the bearing surface geometry to obtain an updated multi-physics mesh, and the transient multi-physics response data is simultaneously updated. Construct a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings; The performance prediction and fault diagnosis results of the sliding bearing are visualized to obtain a sliding bearing analysis report.

2. The thermal-mechanical coupling simulation method for sliding bearings under dynamic loads according to claim 1, characterized in that: The creating of the three-dimensional model of the sliding bearing and dividing the multi-physics field grid refers to constructing the geometric structure of the sliding bearing using computer-aided design software to obtain the three-dimensional geometric model of the sliding bearing; The three-dimensional geometric model is meshed in the multi-physics field simulation pre-processing software to complete the multi-physics field mesh division.

3. The thermal-mechanical coupling simulation method for sliding bearings under dynamic loads according to claim 1, wherein: The properties of the lubricant include density, specific heat capacity, thermal conductivity, viscosity, elastic modulus and thermal expansion coefficient, and the properties of the material include elastic modulus and thermal expansion coefficient, and the properties of the lubricant and material are organized into a material table.

4. The thermal-mechanical coupling simulation method for a sliding bearing under dynamic load according to claim 3, wherein: The output of transient multi-physics field response data specifically includes the following steps: Import the three-dimensional geometric model, multi-physics mesh, dynamic load sequence, and material table of the sliding bearing into the multi-physics simulation platform, and configure the thermal calculation model, fluid calculation model, and mechanical calculation model to complete the establishment of the multi-physics coupling model; Configure initial and boundary conditions for the multiphysics coupling model and perform multiphysics coupling simulation to generate transient multiphysics response data.

5. The thermal-mechanical coupling simulation method for a sliding bearing under dynamic load according to claim 4, characterized in that: The bearing surface wear amount is calculated through transient multi-physics field response data and organized into bearing surface wear distribution data. The specific steps include: The single wear depth of the bearing surface is calculated using the Archimedean wear model based on the transient multi-physics field response data; According to the laws of physics, the bearing surface is divided into high-pressure areas and other areas. The cumulative wear of the bearing surface within the total simulation time is calculated based on the single wear depth. The cumulative wear is compared with the experimental data to form the bearing surface wear distribution data.

6. The thermal-mechanical coupling simulation method for a sliding bearing under dynamic load according to claim 5, characterized in that: According to the wear distribution data of the bearing surface, dynamic mesh adjustment is used to update the bearing surface geometry to obtain the updated multi-physics field mesh, and the transient multi-physics field response data is updated synchronously. The specific steps include the following: The bearing surface wear distribution data is converted into the position movement of the bearing surface grid points through dynamic grid adjustment, and the update frequency of dynamic grid adjustment is set to update the bearing surface geometry; Based on the updated bearing surface geometry, the multiphysics mesh is regenerated, multiphysics coupling simulation is performed, and the transient multiphysics response data is updated synchronously.

7. The thermal-mechanical coupling simulation method for a sliding bearing under dynamic load according to claim 6, characterized in that: Construct a bearing performance prediction model to obtain the performance prediction and fault diagnosis results of the sliding bearing, which specifically includes the following steps: Based on the transient multi-physics field response data, a state space model is established and a Kalman filter is used to generate Bayesian filter detection results. Extract input features and output labels from transient multi-physics field response data and Bayesian filter detection results as training sets and perform standardization, define the structure of the neural network active learning framework, perform training, and obtain the neural network active learning framework; Integrate Kalman filtering and neural network active learning framework to build a bearing performance prediction model, input characteristics and transient multi-physics field response data into the bearing performance prediction model, and output the performance prediction and fault diagnosis results of the sliding bearing.

8. A thermal-mechanical coupling simulation system for a sliding bearing under dynamic load, based on the thermal-mechanical coupling simulation method for a sliding bearing under dynamic load according to any one of claims 1 to 7, characterized in that: include, Create a module to create a 3D model of the sliding bearing, divide the multiphysics mesh, collect dynamic load data, and define the properties of the lubricant and shaft material; Build a module that uses multi-physics modeling to build a multi-physics coupling model based on the three-dimensional model of the sliding bearing, dynamic load data, and the properties of the lubricant and material, and outputs transient multi-physics response data; The wear module calculates the bearing surface wear through transient multi-physics field response data and organizes it into bearing surface wear distribution data; The update module uses dynamic mesh adjustment to update the bearing surface geometry based on the bearing surface wear distribution data, obtains the updated multi-physics field mesh, and simultaneously updates the transient multi-physics field response data; Prediction module, which builds a bearing performance prediction model to obtain performance prediction and fault diagnosis results of sliding bearings; The analysis module visualizes the performance prediction and fault diagnosis results of the sliding bearing and obtains the sliding bearing analysis report.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the thermal-mechanical coupling simulation method under dynamic load of a sliding bearing according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the thermal-mechanical coupling simulation method under dynamic load of a sliding bearing according to any one of claims 1 to 7 are implemented.

Images