Method for evaluating bearing capacity of oil bearing based on virtual simulation

Abstract

The invention relates to an oil bearing carrying capacity evaluation method based on virtual simulation, and the method comprises the steps: obtaining working condition data samples of an oil bearing in a plurality of typical operation states under a complex working condition of a composite load, and carrying out the marking of bearing structure partitions, so as to build a working condition partition mapping table; establishing a multi-physics field sensitive area identification model for each partition, and performing dynamic parameter adjustment on the global finite element model and the local high-precision infinitesimal sub-model by using a multi-physics field coupling calibration factor; and performing time sequence feature extraction on a multi-physical field joint simulation result after dynamic parameter adjustment to obtain a multi-dimensional parameter sequence related to the typical failure mechanism of the oil bearing.

Description

Technical Field

[0001] The present application relates to virtual simulation of workpiece working conditions, and specifically to a method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation. Background Art

[0002] At present, the evaluation of the load-bearing capacity and durability analysis of oil-containing bearings are gradually developing in the direction of high-fidelity virtual simulation and multi-physical field coupling. In particular, in the fields of aerospace, high-end manufacturing and precision machinery, higher technical requirements are put forward for the engineering reliability, failure mechanism identification and life prediction of oil-containing bearings under complex, dynamic and composite load conditions. Mainstream technologies are mostly based on global finite element analysis, supplemented by certain contact mechanics and lubrication theories, and use simulation methods to calculate the parameters of single physical fields such as structural stress, temperature, and lubrication status. However, faced with the diverse and complex working conditions such as dynamic loads, impact loads, asymmetric and superimposed loads acting simultaneously in the actual operation of bearings, as well as the highly coupled dynamic processes between multiple physical fields such as lubrication, structure, and heat, the existing methods mainly have the following technical status quo and shortcomings:

[0003] At present, international virtual simulations of complex working conditions of oil-containing bearings mostly focus on numerical analysis of single physical fields or weakly coupled multi-physical fields. Common methods include static finite element stress field simulation, classical Reynolds lubrication theory and temperature rise calculation, finite element-fluid dynamics weak coupling method, etc. These methods generally use preset load histories or typical single working conditions, such as constant loads or simplified dynamic loads, to qualitatively or quantitatively analyze the stress distribution, oil film thickness, and temperature field state of the bearing; some advanced technologies attempt to introduce fluid-solid (fluid-solid) or solid-thermal (thermo-solid) coupling into the analysis, but due to limitations such as computational cost, data synchronization interfaces, or scale differences, there are generally bottlenecks in overall resolution and real-time interactivity of multi-physical fields, making it difficult to adapt to the ever-changing complex loads and real failure processes in engineering. In addition, simulation models often use fixed parameter assumptions, lacking the ability to accurately capture the rapid changes in physical fields and local failure mechanisms under extreme working conditions, and their accuracy mainly relies on engineering experience parameters or later manual corrections.

[0004] That is, the problems existing in the prior art include:

[0005] 1. The coupling depth and accuracy of multiple physical fields are limited. Existing technologies make it difficult to achieve high-speed, accurate, and bidirectional interaction between multiple physical fields (structure, lubrication, and heat) in sub-regions or even micron-level sensitive areas in the same simulation process. They often approximate single-field superposition or weak coupling, omitting the triggering effect of key coupling mechanisms on extreme failures.

[0006] 2. Local extreme responses conflict with global model resolution. Existing finite element models are limited by their overall scale and computing power, and cannot perform targeted, ultra-fine simulations of high-risk areas. Local dynamics and micro-lubrication thermal behaviors are easily weakened or even ignored.

[0007] 3. Lack of data-driven response calibration and model adaptation mechanisms. Traditional simulations are mostly based on theoretical formulas or static settings. It is difficult to perform dynamic parameter updates and intelligent calibration based on actual historical operating conditions, experimental observations, and new failure types. The model lacks applicability and credibility, making it difficult to evolve and expand automatically.

[0008] 4. The ability to restore the failure mechanism of dynamic working conditions with composite loads is insufficient. Existing methods fail to systematically reproduce the real physical field evolution process under the alternating superposition of dynamic loads, impacts, asymmetry and other working conditions, making it difficult for failure prediction and diagnosis to cover the entire life cycle and unknown extreme events. Summary of the Invention

[0009] In order to solve the problems existing in the above-mentioned prior art, the purpose of this application is to provide a method for evaluating the load-bearing capacity of oil-containing bearings based on virtual simulation.

[0010] In order to solve the above-mentioned technical problems, the present invention provides a method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation.

[0011] The technical solution of the present invention is achieved as follows:

[0012] A method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation, comprising:

[0013] S1: Acquire working condition data samples of an oil-containing bearing under complex working conditions with composite loads and multiple typical operating states. The working condition data includes working condition labels such as dynamic load, impact load, and asymmetric load, and identify the bearing structure partitions to establish a working condition partition mapping table;

[0014] S2: for different structural partitions in the working condition partition mapping table, collecting multi-physics field time series experimental data including lubrication, temperature and mechanical stress, and performing noise reduction and standardization processing on the collected multi-physics field time series experimental data to obtain high-quality multi-physics field basic data;

[0015] S3: Based on the multi-physics basic data and partition identification, a multi-physics sensitive area identification model is established for each partition within the global finite element model framework, and the multi-physics sensitive area index and physical variable influence weight of each partition are output;

[0016] S4: For the identified multi-physics sensitive areas, establish a local high-precision multi-physics coupled microelement sub-model, including modeling of the lubrication fluid dynamic-structural thermal coupled behavior, and associate the multi-physics coupled microelement sub-model parameters with the sensitive area index in the previous step;

[0017] S5: Input the working condition data sample and the local multi-physics field coupled microelement sub-model into the global finite element model, realize the high-speed communication between global and local multi-physics field information through the multi-scale mapping method, and output the multi-scale multi-physics field interaction data;

[0018] S6: Based on multi-scale multi-physics interaction data, a machine learning model is used to predict and invert local and global physical responses driven by time-series load parameters, calibrate traditional physical modeling results, and generate multi-physics coupling calibration factors;

[0019] S7: Use multi-physics coupling calibration factors to dynamically adjust the parameters of the global finite element model and the local high-precision microelement sub-model to achieve highly adaptable modeling for sudden extreme and atypical working conditions;

[0020] S8: Extract time series features from the multi-physics field joint simulation results after dynamic parameter adjustment to obtain multi-dimensional parameter sequences related to typical failure mechanisms, such as oil film thickness distribution, fluid pressure, temperature rise, and contact stress;

[0021] S9: Determine whether the extracted multidimensional parameter sequence meets the preset durability and failure criteria. If not, automatically optimize the modeling granularity and coupling ratio of the multi-physics field sensitive area based on the feedback information to form a model adaptive iteration;

[0022] S10: Outputs high-fidelity multi-physics field-dynamic load joint simulation analysis results after adaptive optimization, which are used to predict the durability and failure mechanism of oil-containing bearings under complex dynamic conditions and to authoritatively evaluate their load-bearing capacity.

[0023] The beneficial effects of the present invention are as follows:

[0024] 1. The high-fidelity coupling and constitutive restoration capabilities of multiple physical fields are significantly improved. This invention proposes for the first time a collaborative coupling mechanism of structural partition sensitivity identification, multi-physical field principal component quantification, local high-resolution microelement model and global finite element model. Through spatial clustering and sensitive factor weight quantification methods, quantitative screening and positioning of failure-dominant factors of key physical fields such as lubrication, temperature, and mechanical stress in various functional areas of bearings are achieved. Compared with existing simulation methods based only on single fields or weak field interactions, it can dynamically capture abnormal coupling and extreme responses between physical fields, and the corresponding error in actual measurement scenarios is reduced from more than 10% to 2% to 5%, greatly enhancing the engineering credibility of the load-bearing capacity assessment results.

[0025] 2. A multi-scale, multi-resolution data interoperability mechanism optimizes simulation performance and efficiency. High-density dynamic load, oil temperature, and stress signals acquired through multi-channel acquisition and event-driven annotation are then indexed, clustered, and feature extracted for structural partitioning / sensitive areas, enabling hierarchical refinement of the simulation model from coarse to fine, and from global to local. This mechanism not only enhances the physical tracking and resolution capabilities of critical bearing damage areas, but also significantly reduces the computational resource consumption required for detailed, full-time, and global simulations. Simulation efficiency is increased by over 50% compared to traditional uniform, high-precision discretization methods, achieving a balanced balance between efficiency and accuracy.

[0026] 3. Machine learning inversion and parameter adaptive optimization greatly enhance the adaptability and generalization of the model. The present invention uses machine learning models (such as neural networks, transfer learning, etc.) to perform data-driven inversion of local-global multi-physics field time series responses, and automatically calibrates traditional physical model parameters to build a model calibration closed loop that integrates physical and data domains. By adaptively adjusting the modeling granularity and coupling parameters through feedback errors, the model migration distortion and criterion failure caused by unknown extreme conditions or atypical complex loads are significantly suppressed. Actual measurements show that under new loading conditions, the peak error of the fluid pressure extreme value output by the model simulation is reduced by 40% compared with the traditional method, greatly reducing the risk of misjudgment and missed judgment.

[0027] 4. Multidimensional traceability of failure mechanisms and rich practical evaluation criteria. This invention extracts multidimensional temporal and spatial feature criteria and classifies typical failure mechanisms. It can effectively distinguish between various typical damage-causing operating conditions, such as sudden changes in bearing oil film thickness, increased local temperature rise, and abnormal stress distribution. It outputs a multi-physics parameter database with comprehensive spatial and temporal coverage, building a durability evaluation system that better meets real-world engineering needs. Compared to previous simulation outputs that only provided simplified limit load margins, this greatly enhances the diagnostic depth and traceability of failure causes and evolution paths, providing a solid data foundation for failure prediction and maintenance recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is the process of a method for evaluating the load capacity of oil-containing bearings based on virtual simulation described in this application Figure 1 . DETAILED DESCRIPTION

[0029] In order to make the purposes, features, and advantages of the present invention more obvious and easy to understand, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described below are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0030] like Figure 1As shown, this embodiment provides a method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation, which specifically includes:

[0031] S1: Obtain working condition data samples of oil-containing bearings under complex working conditions with composite loads under multiple typical operating conditions. The working condition data includes working condition labels such as dynamic load, impact load, and asymmetric load, and identify the bearing structure partitions to establish a working condition partition mapping table.

[0032] S2: For different structural partitions in the working condition partition mapping table, multi-physics field time series experimental data including lubrication, temperature and mechanical stress are collected, and the collected multi-physics field time series experimental data are subjected to noise reduction and standardization processing to obtain high-quality multi-physics field basic data.

[0033] S3: Based on the multi-physics field basic data and partition identification, under the global finite element model framework, a multi-physics field sensitive area identification model is established for each partition, and the multi-physics field sensitive area index and physical variable influence weight of each partition are output.

[0034] S4: For the identified multi-physics sensitive areas, establish a local high-precision multi-physics coupled microelement sub-model, including modeling of the lubrication fluid dynamic-structural thermal coupling behavior, and associate the multi-physics coupled microelement sub-model parameters with the sensitive area index in the previous step.

[0035] S5: Input the working condition data samples and the local multi-physics field coupling microelement sub-model into the global finite element model, realize the high-speed communication of global and local multi-physics field information through the multi-scale mapping method, and output multi-scale multi-physics field interaction data.

[0036] S6: Based on multi-scale multi-physics interaction data, a machine learning model is used to predict and invert the local and global physical responses driven by time-series load parameters, calibrate traditional physical modeling results, and generate multi-physics coupling calibration factors.

[0037] S7: Use multi-physics coupling calibration factors to dynamically adjust the parameters of the global finite element model and the local high-precision microelement sub-model to achieve highly adaptable modeling of sudden extreme and atypical working conditions.

[0038] S8: Perform time series feature extraction on the multi-physics field joint simulation results after dynamic parameter adjustment to obtain multi-dimensional parameter sequences related to typical failure mechanisms such as oil film thickness distribution, fluid pressure, temperature rise and contact stress.

[0039] S9: Determine whether the extracted multidimensional parameter sequence meets the preset durability and failure criteria. If not, automatically optimize the modeling granularity and coupling ratio of the multi-physics field sensitive area based on the feedback information to form a model adaptive iteration.

[0040] S10: Outputs high-fidelity multi-physics field-dynamic load joint simulation analysis results after adaptive optimization, which are used to predict the durability and failure mechanism of oil-containing bearings under complex dynamic conditions and to authoritatively evaluate their load-bearing capacity.

[0041] S1: Acquiring working condition data samples of an oil-containing bearing under complex working conditions with composite loads under multiple typical operating conditions, wherein the working condition data includes working condition labels such as dynamic load, impact load, and asymmetric load, and marking the bearing structure partitions to establish a working condition partition mapping table specifically includes:

[0042] S1.1 conducts dynamic load tests on the target oil-containing bearings under typical operating conditions to obtain original operating condition data covering three types of operating conditions: dynamic load, impact load, and asymmetric load. This data set forms an operating condition-load type mapping to ensure the diversity and representativeness of the sample's operating conditions.

[0043] The input data includes the structural information of the target oil-containing bearing, the definition of typical operating conditions, and the load types (dynamic load, impact load, and asymmetric load) to be obtained under each operating condition.

[0044] A multi-channel high-frequency dynamic load test bench (parameters: loading frequency 0-500Hz, maximum load capacity 10kN), acceleration sensor (range ±100g), strain gauge (range 500με), and force sensor (accuracy 0.1N) are used to collaboratively collect data and define the driving conditions of the bearing under preset typical operating states (such as high-speed rotation, emergency stop start, and off-load operation).

[0045] Furthermore, through the programmed loading scheme of the experimental bench (parameters: working condition sequence, amplitude change, impact pulse timing), the gradual superposition and switching of dynamic loads, impact loads and asymmetric loads are achieved, and the original working condition excitation signal is collected in real time at a set synchronous sampling rate (such as 10kHz).

[0046] The synchronous acquisition module is used to perform time alignment and multi-channel synchronization of multi-source load response signals (acceleration, strain, force), and to perform sampling point interpolation and effective interval compensation for signal distortion or information loss fragments.

[0047] Furthermore, through signal preprocessing algorithms (such as bidirectional Butterworth filtering, cutoff frequency 200 Hz), dynamic baseline correction and drift correction, background interference and zero drift noise are eliminated to obtain the purified original working condition data stream.

[0048] An automatic working condition discrimination algorithm (parameters: switching interval judgment, peak threshold) is used to perform real-time segmentation and event-driven tagging on the raw data collected in the experiment, and classify and archive them into an operating state-load type mapping dataset.

[0049] Through the above-mentioned experiments and signal processing process, the original working condition data of the dynamic load, impact load and asymmetric load corresponding to the target oil-containing bearing under various typical operating conditions are systematically collected, and a structured operating state-load type mapping data set is formed to achieve full space sample coverage of different working condition combinations, thereby improving the representativeness and diversity of subsequent simulation samples.

[0050] For example, in the actual measurement scenario of a certain type of turbine high-speed oil-containing bearing, three typical states are selected: operating state A (stable speed of 18000rpm), state B (high-speed start-emergency stop), and state C (radial offset load of 1000N at the output end). A three-channel load is applied, and each channel applies a sinusoidal dynamic load with an amplitude of 500N, an impact load with a width of 4000N of 20ms, and a variable load with a periodic asymmetric distribution (maximum amplitude of 800N). The signal is collected using a TDK450 three-axis acceleration sensor and an HBMFTP-10 force sensor with a sampling frequency of 10kHz and a data synchronization error controlled at 0.05ms. The obtained working condition data is filtered with a Butterworth order of 4 and corrected for zero drift in the entire section to eliminate baseline noise. A custom working condition discrimination criterion is used. When the signal peak is greater than 3500N, it is automatically marked as an impact condition. When it is a periodic sine wave and the amplitude change of the offset load distribution is greater than 20%, it is marked as an asymmetric load. The rest are classified according to dynamic loads. The final output is a three-dimensional mapping data set of operating state (A, B, C) and load type (dynamic load, impact load, asymmetric load). The total sample volume reaches 36,000 frames, fully covering typical and extreme working conditions, laying the foundation for subsequent simulation modeling and data mining.

[0051] Based on the original working condition data samples, S1.2 uses an event-driven labeling algorithm to assign load condition labels to each time series sampling point, generate a working condition data time series label set, and lay a structured foundation for subsequent data partitioning and feature extraction.

[0052] The input conditions include the multi-channel, high-sampling-rate purified raw operating condition data stream obtained in step S1.1. The data structure contains three signals: acceleration (A(t)), strain (ε(t)), and force (F(t)). The time synchronization accuracy is better than 0.05ms, and the sampling frequency is 10kHz. The raw operating condition data has been preliminarily mapped and classified according to typical operating states (such as A, B, C) and load types (dynamic load, impact load, asymmetric load).

[0053] An event-driven annotation algorithm (with parameters including peak threshold, periodicity criteria, amplitude change rate, and switching interval determination criteria) is used to automatically identify dynamic load conditions at the sampling point level. Furthermore, a time window sliding segmentation technique (with parameters including a 100ms window length and a 10ms step) enables real-time extraction of event features within raw signal segments and clustering of multiple conditions.

[0054] Furthermore, a multi-dimensional signal discrimination index calculation algorithm is used to calculate the maximum value P for the signal data in the current time window. max =max(F(t)), average value μ=mean(F(t)), root mean square value

[0055] Time domain statistics such as standard deviation σ=std(F(t)) and periodic distribution parameters are used to extract event discrimination features for impact, dynamic load and asymmetric load.

[0056] Furthermore, by setting the peak threshold P th，impact , cycle amplitude difference threshold Δ asym and cyclical indicator C cyc , the judgment is as follows: when P max >P th，impact , the current sampling segment is marked as an impact load event; when the signal periodicity is strong and Δ asym > set value, marked as asymmetric load; the rest that meet the set amplitude and period are marked as dynamic load.

[0057] A segmented event-driven labeling result merging and error correction algorithm is adopted, and boundary smoothing criteria and logical backtracking are introduced for critical intervals or signal mutation segments to eliminate mislabeling and ensure the consistency and coherence of temporal labels.

[0058] Through the above-mentioned event-driven labeling algorithm processing method, the purified original working condition signal is converted into a time series label set of working condition data labeled according to the sampling point accuracy, realizing high-precision automatic labeling of working conditions such as dynamic loads, impact loads, and asymmetric loads, and outputting the three-dimensional time series label structure data of working condition-operating state-sampling point, providing a structured basis for subsequent structural partition mapping and multi-physics field simulation feature extraction.

[0059] For example, in the original data set obtained for the TDK450 triaxial acceleration and HBM FTP-10 force sensor (sampling frequency 10 kHz, total sample frame number 36000), the peak threshold P is used. th，impact =3500N, period amplitude criterion Δ asym = 20, 100ms sliding window and 10ms step window are used for event-driven analysis. For the signal window segment, if max(F(t))>3500N, it is directly marked as impact condition;

[0060] If the periodicity is significant and the local amplitude deviation section is greater than 20%, it is marked as an asymmetric load condition, and the rest is treated as dynamic load. Using time series label merging and 20ms boundary smoothing rules, the sudden change segment and boundary segment are identified and corrected, and the final output is a time series label set of the operating condition marked by the signal point under each operating state of the oil-containing bearing, achieving high consistency and accurate labeling without omissions for 36,000 sampling data.

[0061] The test results show that the annotation consistency rate reaches 99.5%, laying a solid foundation for subsequent structure mapping and simulation feature attribution.

[0062] S1.3 performs a partitioning algorithm on the three-dimensional CAD structural model of the oil-containing bearing, subdivides the entire bearing into functional areas (such as rolling element area, cage area, lubrication channel area, etc.), and obtains the structural partition index of each area.

[0063] S1.4 performs high-dimensional data fusion of the time series label set of the working condition data and the structural partition index, and uses a multi-label mapping strategy to generate a working condition-structure integrated mapping table for each structural partition under each load condition, thereby building an association foundation for subsequent multi-physical field partition identification.

[0064] Based on the working condition-structure integrated mapping table, S1.5 performs data consistency verification and redundancy filtering on typical structural partition samples under various operating states. Through data cleaning strategies, it obtains high-quality working condition partition mapping tables, providing support for multi-physics field basic data collection and model regionalization analysis.

[0065] S2: for different structural partitions in the working condition partition mapping table, collecting multi-physics field time series experimental data including lubrication, temperature, and mechanical stress, and performing noise reduction and standardization on the collected multi-physics field time series experimental data to obtain high-quality multi-physics field basic data, specifically including:

[0066] S2.1 performs partition identification operations on the structural partitions in the working condition partition mapping table, extracts the acquisition units of each structural partition based on the partition identification, and uses them as the target objects for physical field signal acquisition; through partition identification, it ensures that the subsequent physical field data sampling is targeted and highly representative in the working condition space, and outputs a list of structural partition acquisition units.

[0067] S2.2 is based on the list of acquisition units for structural partitions, and multi-channel sensor fusion technology is used to obtain the original experimental time series data of lubrication, temperature and mechanical stress under the partition respectively, so as to obtain a highly coupled dynamic data stream of multiple physical fields and form a multi-channel multi-physical field original data set.

[0068] S2.3 uses a time-frequency domain adaptive filtering algorithm to perform systematic noise reduction on the multi-channel and multi-physics field original data sets to eliminate environmental interference and sensor noise, improve the signal-to-noise ratio of each physical field signal, and output the noise-reduced multi-physics field purified data set.

[0069] S2.4 Based on the physical field purification data set, normalization and standardization transformations are performed on the physical quantities of each structural partition, including but not limited to Z-Score standardization and physical quantity dimension consistency processing, to eliminate the influence of different physical field measurement scales, dimensions and amplitude ranges, and obtain calibrated multi-physical field normalized data.

[0070] S2.5 introduces time series feature completion and anomaly detection technology to the normalized data of multi-physics fields, automatically corrects missing data and abnormal data points, ensures the integrity and consistency of lubrication, temperature and mechanical stress signal data of each structural partition, and generates a high-quality multi-physics field basic data set.

[0071] S3: Based on the multi-physics basic data and partition identification, under the global finite element model framework, a multi-physics sensitive area identification model is established for each partition, and the multi-physics sensitive area index and physical variable influence weight of each partition are output, including:

[0072] S3.1 structures and organizes the basic data and partition identification of multiple physical fields, adopts a data hierarchical mapping strategy, binds each structural partition with its own time series physical variables in the global finite element model, and realizes the preliminary construction of the partition-physical variable basic mapping table for subsequent sensitive area screening.

[0073] Based on the established partition-physical variable basic mapping table, S3.2 applies a local correlation analysis algorithm to perform multi-dimensional statistics on the temporal distribution characteristics of multi-physical field variables such as lubrication, temperature, and mechanical stress within the bearing structure partition, and obtains the correlation matrix of the physical variables within the partition as the basic data for subsequent sensitive measurements.

[0074] Based on the partition-physical variable basic mapping table, the input includes each structural partition label, partition time series physical variable data (lubrication, temperature, mechanical stress and other multi-physical field variables) and the structural partition identification information corresponding to each time series data.

[0075] A local correlation analysis algorithm (parameters: time series data of physical variables in each partition, analysis window length, and correlation criterion type) is used to perform pairwise correlation statistics on the multi-physical field time series variables in each structural partition, quantify the joint change relationship of each physical field variable in the partition over time, and realize the extraction of preliminary information coupling between variables.

[0076] Furthermore, through time series sliding window correlation analysis (parameters: window step size, window width), the correlation change trend of various types of physical variables in the partition in different time segments is tracked, and the deviation of the correlation estimation is corrected according to the statistical sample capacity to ensure the robustness of the variable coupling relationship under extreme or abnormal working conditions, and obtain a preliminary correlation coefficient matrix based on time segments.

[0077] Furthermore, through the multidimensional correlation matrix aggregation algorithm (parameters: variable category, partition label, time slice range), the correlation statistical data in all time series windows are spatially and temporally aggregated to form a partition-level multi-physics field correlation matrix - this correlation matrix uses variable categories as rows and columns, and each element is the correlation coefficient between corresponding variable pairs (such as Pearson correlation coefficient, mutual information value or Spearman rank correlation coefficient), which systematically reflects the degree of coupling between physical field variables in different partition spaces and dynamic working conditions.

[0078] Furthermore, through the statistical significance filtering algorithm (parameters: correlation threshold, p-value criterion, variable pair selection criteria), variable pairs with insufficient statistical sample capacity or no significant correlation are eliminated, and only regional physical variable correlation pairs with high correlation or high influence are retained, providing the original basic data for subsequent sensitivity measurement and principal component quantitative clustering.

[0079] Through the above-mentioned local correlation analysis and correlation matrix construction processing, the partition-physical variable basic mapping table is converted into a partition-level multi-physical field correlation matrix, which realizes the quantification of the in-depth, dynamic and spatial coupling relationship between physical variables and provides high-reliability basic statistical data for the multi-physical field sensitivity screening of structural partitions.

[0080] For example, in a global finite element model of a batch of oil-containing bearings, for the rolling element area numbered A01 and the lubrication channel area numbered A02, the multi-physics field normalized time series data collected has a sampling rate of 1000 Hz and a sample capacity of 180,000 items (corresponding to a typical working condition of 3 consecutive minutes);

[0081] The Pearson correlation coefficient is used as the criterion for local correlation analysis to analyze the pairwise correlation of the three physical variables of lubrication pressure, surface temperature, and mechanical stress in this partition. The corresponding formula is as follows:

[0082]

[0083] Among them, r XY is the Pearson correlation coefficient between variables X and Y, X i and Y i are the normalized physical values ​​of the i-th sampling point, and are the sampling means, and n is the number of sampling points.

[0084] With a window width of 1000 (i.e., data per second) and a sliding step of 500, the correlation coefficients of the three pairs of variables (lubrication-temperature, lubrication-stress, and temperature-stress) in each window were calculated to obtain the time correlation change series.

[0085] The statistical results of all windows were averaged and aggregated, and p < 0.05 was used as the significance standard. Weakly correlated samples with absolute values ​​of correlation coefficients lower than 0.3 were eliminated, and the multi-physics field correlation matrix of area A01 was finally obtained:

[0086]

[0087] In the A01 partition, the correlation coefficients between lubrication and temperature were 0.68, and the correlation coefficient between temperature and stress was 0.63, both exceeding the predetermined thresholds and identifying them as sensitive variable pairs. The correlation matrix for this partition served as direct input for subsequent principal component analysis and sensitivity weight calculation, significantly improving the accuracy of identifying physical coupling relationships and their engineering interpretability.

[0088] S3.3 performs quantitative sensitivity assessment on the correlation matrix and actual failure annotation data, and uses multivariate statistical methods such as principal component analysis (PCA) and factor weight normalization to calculate the sensitivity weight distribution of each physical variable on structural performance, and outputs the multi-physical field sensitivity factor matrix for each structural partition.

[0089] The input partition-level multi-physics field correlation matrix and actual failure annotation data are jointly subjected to sensitivity quantification processing to identify the influence of multi-physics field variables within each structural partition on structural performance.

[0090] The principal component analysis (PCA) method (parameters: correlation matrix, cumulative contribution rate threshold, variable category) is used to perform eigenvalue decomposition on the partition-level multi-physics field correlation matrix to obtain each principal component vector and its corresponding eigenvalue, thereby automatically screening the key physical factors that can explain most of the variance in the high-dimensional variable space.

[0091] Furthermore, using a characteristic contribution rate calculation method (parameters: principal component eigenvalue vector, total variance), we quantitatively analyze the variance contribution of each principal component, sort them in descending order, and determine the top principal components whose cumulative contribution rate meets a preset threshold (e.g., 95%). This process is used to reduce the impact of redundant variables and highlight the sensitive dimensions of the physical response.

[0092] Furthermore, the factor weight normalization algorithm (parameters: principal component loading matrix, normalization method) is applied to normalize the principal component loadings, and the absolute contribution weights of each principal component are assigned to the original physical variables, thereby quantifying the sensitivity weight of each physical variable within the structural partition.

[0093] Furthermore, based on the actual failure annotation data, regression analysis or discriminant statistics (parameters: principal component score, discriminant function, failure classification label) are used to calculate the correlation between the principal component score and the actual failure probability or failure type. Through sensitive factor residual analysis, physical variables that are highly correlated with failure are screened, and the sensitivity weights are recalibrated.

[0094] Through matrix reorganization and normalization, the sensitivity data of the physical variables in each partition, after PCA and weight normalization, are integrated to output a multi-physics sensitivity factor matrix for each structural partition. This matrix, with physical field variable categories as rows, principal component mapping indicators as columns, and matrix elements as weight coefficients, comprehensively characterizes the sensitivity response of each physical variable to the performance of the partition structure.

[0095] Through the above-mentioned principal component analysis, factor weight normalization and failure re-correction processing, the results of the prior correlation statistics are effectively converted into a multi-physical field sensitive factor matrix at the structural partition level, realizing the quantitative identification of key physical quantities of the failure mechanism, thereby laying a data foundation for the subsequent sensitive area division and local modeling process.

[0096] For example, in the rolling element partition numbered A01, the input Pearson correlation matrix is ​​as follows:

[0097]

[0098] The principal component decomposition was performed using PCA, and the eigenvalues ​​obtained were λ1 = 2.1, λ2 = 0.6, and λ3 = 0.3.

[0099] The cumulative contribution rates are as follows:

[0100]

[0101] Among them, k is the principal component number.

[0102] Select the first two principal components, and the total contribution rate is That is more than 90%.

[0103] Principal component vector (simplified example):

[0104] PC1: (0.61, 0.58, 0.53), PC2: (0.45, -0.82, 0.36).

[0105] The principal component loads of each variable were normalized by absolute value, resulting in sensitivity weights of 0.41 for lubrication, 0.39 for temperature, and 0.36 for stress. Combined with the failure sample data, lubrication and temperature were determined to be the primary sensitive fields in the A01 partition.

[0106] The output A01 partition multi-physics field sensitivity factor matrix is ​​as follows:

[0107]

[0108] Under this technical process, different working condition samples and partition scenarios can be adapted to the above analysis paradigm to achieve standardized, automated and traceable identification of regional physical variable sensitivity. The output data will be used for subsequent sensitive area clustering, key sub-model setting and global coupling strategy optimization, effectively improving the ability to restore the action mechanism of complex multi-physical field loads and the accuracy of simulation analysis.

[0109] S3.4 clusters the sensitive areas of the structural partitions in the global finite element model based on the multi-physics field sensitivity factor matrix, uses the spatial clustering algorithm and neighborhood recognition mechanism to calibrate the high-weight physical response areas, and automatically generates the sensitive area index of each structural partition as a spatial reference for the nesting of refined sub-models.

[0110] The input conditions are: partition-level multi-physics field sensitivity factor matrix, global finite element model structural partition information, and structural partition physical variable spatial distribution data.

[0111] A spatial clustering algorithm (parameters: sensitivity factor threshold, neighborhood radius, variable weight vector) is used to perform high-dimensional feature clustering on each structural partition in the global finite element model, realize automatic spatial aggregation of multi-physical field sensitive factors, and identify nodes or unit sets with similar physical response characteristics.

[0112] Furthermore, through the high-weight physical response area identification mechanism (parameters: sensitivity weight lower limit, spatial proximity constraint), weighted screening is performed on the nodes or units in the spatial clustering results according to the weight distribution in the sensitivity factor matrix to form a set of spatial candidate areas with weights significantly higher than the average level.

[0113] Furthermore, a neighborhood search and spatial connectivity detection algorithm (parameters: node adjacency matrix, connectivity threshold) is used to search for sub-regions with continuous boundaries and coordinated responses within the candidate region set, excluding scattered nodes that have no actual coupling significance physically, ensuring that sensitive areas have high physical correlation and structural coherence.

[0114] Furthermore, based on the spatial optimization function (parameters: regional volume, gradient distribution of physical variables, and model boundary conditions), the distribution boundaries of sensitive areas are optimized, and the positions and ranges of cluster partitions are adjusted so that the clustering results of sensitive areas can cover the high-sensitivity areas to the greatest extent, while avoiding redundancy and overlap, and improving the spatial utilization efficiency of subsequent sub-model nesting.

[0115] Through the numbering and indexing algorithm, the sensitive areas obtained in each structural partition are coded according to unique identifiers to generate a structural partition-sensitive area index table, thereby realizing the automatic generation and location tracing of sensitive area indexes.

[0116] Through the above-mentioned spatial clustering and neighborhood calibration algorithm processing, the numerical results of the partitioned multi-physics field sensitivity weights are converted into spatial sensitive area index data, providing a strict spatial reference for the subsequent refined sub-model nesting, and achieving high adaptability and accurate positioning of modeling objects in sensitive areas.

[0117] For example, in the A01 rolling element partition of the global finite element model of the oil-containing bearing, based on the obtained sensitivity factor weights (lubrication weight 0.41, temperature weight 0.39, stress weight 0.36), a DBSCAN-based spatial clustering algorithm was used, with the minimum number of neighboring cells set to 12 and the sensitivity weight threshold set to 0.3, to perform weighted aggregation on the three types of physical field response values ​​of each grid node;

[0118] For the high-weighted regions, further node connectivity analysis revealed two primary spatially connected subregions: the first covering the rolling element outer race-lubricant film interface, and the second located in the relative slip zone between the rolling element and the cage. After spatial boundary optimization and exclusion of isolated low-response nodes, the volume of the first sensitive region accounted for 18.7% of the A01 partition, while the second region accounted for 6.2%.

[0119] Assign a unique index to each sub-region (such as A01-R1, A01-R2) and generate a region index table of A01:[A01-R1,A01-R2];

[0120] The verification process shows that the clustering and calibration results can accurately cover the area with the maximum oil film pressure gradient and the area with concentrated temperature rise, providing a complete spatial basis for subsequent local high-precision modeling. In addition, the coupling correlation between the oil film thickness and contact stress in the sensitive sub-area is increased to 0.76 (higher than the partition average of 0.51), which greatly improves the accuracy of the model's physical field identification and dynamic response analysis.

[0121] S3.5 integrates the partition sensitive area index and its corresponding multi-physics sensitivity weight, and outputs a structural partition-sensitive area-weight mapping table, realizing the traceability of physical variable influences, location location, and quantitative response, providing accurate basic data for subsequent local high-precision multi-physics field coupled microelement sub-model selection setting and parameter initialization.

[0122] S4: For the identified multi-physics sensitive area, establish a local high-precision multi-physics coupled micro-element model, including lubrication fluid dynamic-structural thermal coupling behavior modeling, and associate the micro-element model parameters with the sensitive area index of the previous step, specifically including:

[0123] S4.1 is based on the multi-physics sensitive area index, screens the corresponding local structural geometric parameters, material parameters and working boundary conditions, and generates multi-physics sensitive area micro-element analysis objects as data input for high-precision local modeling.

[0124] The multi-physics sensitive area index, local structure geometric parameter set, material parameter set and working condition boundary condition parameters are used as data input.

[0125] A sensitive area index screening algorithm (parameters: structure partition-sensitive area-weight mapping table, high weight threshold, spatial index number) is used to achieve automatic spatial positioning of the calibrated multi-physics field sensitive areas in the global model of the target oil-containing bearing.

[0126] Furthermore, through the local structure geometric parameter extraction method (parameters: 3D CAD model, node coordinate set, sensitive area boundary), the batch export of data such as the precise geometric dimensions, spatial position, surface division contour, etc. of each node and unit in the sensitive area is achieved, and the corresponding geometric parameter subset is constructed to ensure that the local model can be modeled finely with high resolution.

[0127] Furthermore, a local material property aggregation algorithm (parameters: material identification number, physical property database, sensitive area unit list) is used to realize the classified retrieval of the elastic modulus, Poisson's ratio, density, thermal expansion coefficient, thermal conductivity and oil film physical properties of various material types involved in sensitive areas, such as the matrix, lubrication pair, and cage, to form a multi-physical field microelement material parameter table.

[0128] Furthermore, the working condition dynamic mapping algorithm (parameters: working condition partition mapping table, corresponding working condition label, stress-load-temperature time series data) is used to achieve the matching of historical working condition time series boundary conditions of the partition where the sensitive area is located, including the integration of multiple time series boundary parameters such as dynamic load, impact load size and direction, lubrication oil supply rate, heat flux density, and ambient temperature.

[0129] Furthermore, a multi-physics field infinitesimal object generation algorithm (parameters: sensitive area index, local geometric parameters, material parameters, operating boundary conditions) is adopted to encapsulate all the above local input parameters into a multi-physics field sensitive area infinitesimal analysis object for local high-precision modeling. Each infinitesimal object data has a unique spatial, physical, and operating condition attribute identifier, which facilitates subsequent modeling and adaptive initialization of the coupling interface.

[0130] Through the above-mentioned multi-algorithm chain processing, the sensitive areas identified in the global finite element model are intelligently and efficiently converted into structured local high-precision multi-physics field micro-element analysis objects, realizing the pre-parameter preparation for refined modeling of sensitive areas.

[0131] For example, in the rolling element-lubricating film interface sensitive area (A01-R1) of the partition numbered A01, the volume of the sensitive area is 1.2mm 3 For the test object, 1132 nodes and 786 units were automatically located by the spatial index screening algorithm;

[0132] By extracting local geometric parameters, the node coordinate range [x1, x2], [y1, y2], [z1, z2] and the average surface roughness R are obtained. a 0.12μm;

[0133] The local material parameters retrieved were steel (elastic modulus 209 GPa, Poisson's ratio 0.28), lubricating oil (dynamic viscosity 0.024 Pa·s, thermal conductivity 0.13 W / m·K);

[0134] Using the working condition mapping algorithm, the load sequence range is 25N~240N, the maximum pulse impact load is 320N, the oil temperature sequence range is 22℃~97℃, and the lubrication flow rate is 5.6imes10 -4 m^3 / s;

[0135] Through the multi-physics field microelement object generation algorithm, this sensitive area microelement object already includes a node-element list, a material-property parameter table, and a set of operating condition boundary timing parameters. The final output is a complete microelement object data package for S4.2 mesh refinement and S4.3 multi-physics field coupling solver initialization.

[0136] The standardized output of this object significantly improves the accuracy and efficiency of local modeling in complex loading and multi-physics field interaction scenarios.

[0137] S4.2 uses a multi-scale finite element partitioning algorithm for the micro-element analysis objects in multi-physics field sensitive areas to refine their finite element mesh structure to meet local high-precision requirements and enhance the spatial resolution of node information transmission to obtain a refined finite element mesh model.

[0138] The multi-physics field sensitive area microelement analysis object is used as input, which contains the screened local geometric parameters, material parameters and dynamic working boundary conditions, and has a unique spatial and physical identity.

[0139] A multi-scale finite element partitioning algorithm (parameters: target sensitive area node / element list, initial mesh size, refinement scale interval, accuracy classification strategy) is used to implement finite element mesh refinement on the spatial area defined by the micro-element analysis object, so as to achieve a step-by-step improvement in the spatial discreteness within the sensitive area.

[0140] Furthermore, through the regional adaptive grid generation algorithm (parameters: local physical quantity change rate threshold, maximum unit size limit, boundary layer automatic encryption factor), the grid density of key parts such as oil film interface, high stress concentration points, and temperature rise mutation areas is automatically adjusted according to the physical response gradient distribution, thereby enhancing the node distribution density in these areas and improving spatial resolution.

[0141] Furthermore, in response to the spatial coupling requirements of different physical fields, heterogeneous multi-physics field coupling grid reconstruction technology (parameters: discretization format of each physical field, coupling node pair mapping table) is adopted to establish node mapping relationships and physical quantity projection functions between the fluid, structure, and thermal physical field sub-areas, ensuring that the transmission values ​​of physical quantities between multi-physics field sub-models are consistent and the errors are controllable.

[0142] Furthermore, by refining the mesh quality optimization algorithm (parameters: element aspect ratio, acute angle threshold, boundary fitting accuracy standard), the generated local mesh structure is subjected to element reconstruction, node redistribution, and boundary contour refitting, effectively avoiding the risk of numerical instability in isolated weak elements and high deformation areas.

[0143] Furthermore, a multi-level data encapsulation scheme for the grid structure (parameters: level number, node-element correspondence table, coupling interface mark) is adopted to archive the refined finite element grid structure in the form of layered objects, which is convenient for subsequent high-precision multi-physics field coupling solution and dynamic parameter adjustment, and finally obtains a refined finite element grid model that accurately corresponds to the sensitive area index.

[0144] Through the above-mentioned multi-level, multi-algorithm coordinated refinement processing, the micro-element objects in the sensitive areas of multiple physical fields are structurally converted into local refined finite element meshes with high spatial resolution and physical quantity coupling consistency, providing standardized modeling units for subsequent high-precision coupling simulation analysis of multiple physical fields and dynamic iteration of physical field parameters, thereby achieving a significant improvement in the physical response analysis capability and numerical prediction accuracy in sensitive areas.

[0145] For example, in the sensitive area A01-R1 ​​of a typical oil-bearing bearing, the number of nodes in the initial microelement object is 1132, the number of elements is 786, and the area volume is 1.2mm 3 A multi-scale finite element partitioning algorithm was used to refine the mesh at the lubricating oil film interface and high gradient stress area based on the maximum allowable element side length of 0.05 mm, the minimum side length of 0.01 mm, and the regional physical quantity change rate threshold of 5%.

[0146] As a result, the total number of elements increased to 2684, the number of nodes increased to 3497, and the minimum element size of the local oil film interface reached 0.009mm. Through heterogeneous multi-physics field coupling mesh reconstruction, 364 key node mappings were achieved between the three sub-fields of lubrication, structure, and thermal.

[0147] Using a mesh quality optimization algorithm, the cell aspect ratio is controlled between 1 and 4, all acute angles are greater than 30°, and the boundary contour error is less than 0.5%;

[0148] Ultimately, the refined finite element mesh model effectively improved the node density and spatial resolution, and the local relative error of the numerical prediction of oil film thickness and contact stress in sensitive areas was less than 2.1%, supporting the next step of multi-physics field coupling solution to achieve higher spatiotemporal accuracy and engineering interpretability.

[0149] S4.3 uses the fluid-solid-thermal multi-physics field coupling solver to apply fluid dynamics, heat conduction and structural mechanics boundary conditions to the lubricating oil film sub-area, heat source point and load-bearing surface in the refined finite element mesh model to obtain the lubricating fluid dynamic-structural thermal multi-physics field joint reaction parameters.

[0150] The input is the microelement analysis object of the multi-physics field sensitive area and its refined finite element mesh model. The microelement object includes the extracted geometric parameters, material parameters and multi-time sequence boundary conditions, and the refined mesh contains high spatial resolution nodes and unit attributes.

[0151] The fluid-solid-heat multi-physics coupling solver (parameters: fluid sub-field coupling format, structural sub-field finite element characteristics, heat conduction solver sub-module) is used to define the physical field boundary type for each grid unit and node in the sensitive area, including fluid dynamics boundaries (inlet pressure, viscous oil film shear velocity), heat source boundary conditions (local heat flux density, adiabatic / constant temperature surface), and structural mechanics boundary conditions (load transfer surface, constrained surface, node displacement constraint), to realize the multi-physics boundary configuration of the lubricating oil film sub-area, heat source point and load-bearing surface.

[0152] Furthermore, based on the partitioned iterative algorithm of the fluid-solid / thermal three-field coupled solution (parameters: iteration step, sub-field coupling strength, convergence criterion), the basic physical equations governing lubrication flow, heat transfer, and structural deformation are combined and solved according to the time step. In the fluid dynamics sub-area, the incompressible Navier-Stokes equations are used to describe the flow behavior within the lubricating oil film:

[0153]

[0154] Where ρ is the density of the lubricating oil, u is the velocity vector, p is the pressure, μ is the viscosity, and f is the body force.

[0155] In the heat conduction sub-area, the transient heat conduction equation is used to describe the local temperature rise of the heat source point and the wall surface:

[0156]

[0157] Among them, c p is the specific heat capacity, T is the temperature field, λ is the thermal conductivity, and Q is the heat source per unit volume.

[0158] In the structural mechanics subarea, the force and displacement responses of the nodes are solved by the finite element method, using the equilibrium equation:

[0159] Ku=F ext +F int +F th

[0160] Where K is the overall stiffness matrix, u is the node displacement vector, and F ext is the external load, F int is the oil film-structure contact internal force, F th is the thermal load caused by thermal expansion.

[0161] Furthermore, through the sub-field coupling-mapping-synchronization mechanism (parameters: cascade boundary mapping table, node physical quantity transfer factor, convergence synchronization tolerance), the internal pressure distribution of the lubricating oil film is mapped to the structural nodes of the load-bearing surface at each time integration step, and the feedback of the temperature rise of the heat source point on the fluid viscosity and structural thermal expansion is synchronized to realize the dynamic transmission of data on physical properties and boundary conditions among the three fields of fluid, heat and structure.

[0162] Furthermore, a coupled field convergence judgment and error control algorithm (parameters: maximum allowable residual, coupling criterion type) is used to perform convergence judgment on the main variables of pressure, temperature, and displacement in each iteration. If the residuals of all main variables are lower than the preset threshold, the multi-physics field joint reaction parameters of the current step are output; otherwise, local re-relaxation correction is automatically performed until convergence.

[0163] Through the fluid-solid-heat multi-physics field coupling solver and the multi-step dynamic boundary transfer algorithm, the multi-field joint reaction parameters of the oil film pressure, flow velocity field, temperature distribution, node displacement and stress response of the lubricating oil film sub-area, heat source point and load-bearing surface with respect to time are obtained within the refined finite element sensitive area grid, realizing high-fidelity multi-physics field time series parameter output.

[0164] For example, for the microelement model of the sensitive area numbered A01-R1, the fluid subfield adopts the incompressible Navier-Stokes equation, and the oil film density ρ = 860 kg / m 3 , dynamic viscosity μ=0.024Pa·s; the thermal subfield uses the transient heat conduction equation, thermal conductivity λ=0.13W / m·K, initial temperature T0=40℃; the structural subfield steel stiffness modulus E=209GPa. The working condition input is the dynamic load range of 25N to 240N, the oil film inlet velocity u in =0.22m / s, the periodic internal heat generation rate of the heat source point Q = 19.4kW / m 3 Set the time step width to 0.001s, the subfield coupling strength parameter to 0.8, and the convergence tolerance to 1×10 -3In the fluid-solid thermal field iterative solution, at each iteration, the oil film pressure field of the current time step is loaded onto the structural bearing surface grid through node mapping, and the temperature response of the structural node deformation is fed back to the thermal subfield to adjust the oil film area temperature and local viscosity. Under the multi-step convergence judgment, the residuals are all less than 1×10 -4 Afterwards, the pressure profile, temperature distribution, and node deformation data for each step were stably output. Ultimately, within a 0.3-second loading cycle, 210 sets of high-frequency oil film pressure-structural stress-local temperature multi-field correlation data were acquired. The oil film pressure extremes dynamically varied from 1.9 to 3.3 MPa, the temperature in the maximum temperature rise zone rose to 91.2°C, and the maximum equivalent stress at the main contact node was 143 MPa. The high-fidelity combined reaction parameter output provided a solid data foundation for subsequent failure criterion extraction and model dynamic optimization.

[0165] Based on the above-mentioned multi-physics field joint response parameters, S4.4 executes the dynamic prediction algorithm of surface contact stress and oil film thickness, outputs high-resolution local parameters such as time-series oil film pressure distribution, temperature rise field and contact area stress response sequence, and forms a local high-precision multi-physics field response set.

[0166] S4.5 establishes correlation data mapping rules between the intermediate parameters of the local high-precision multi-physics field response set and the original global finite element model through sensitive area indexes, realizes the structural synchronous association between the local microelement sub-model parameters and the sensitive area indexes, and provides a standardized interface for subsequent multi-scale multi-physics field coupling and high-speed information intercommunication.

[0167] S5: Inputting the working condition data sample and the local multi-physics field coupled microelement sub-model into the global finite element model, realizing high-speed communication between global and local multi-physics field information through a multi-scale mapping method, and outputting multi-scale multi-physics field interaction data specifically includes:

[0168] S5.1 performs semantic label mapping on the obtained working condition data samples, takes working condition parameters such as dynamic load, impact load and asymmetric load as input, and performs data adaptation processing on each structural partition through the working condition partition mapping table to generate a partition working condition parameter set, providing multi-dimensional time-series working condition driving factors for subsequent finite element modeling.

[0169] S5.2 is based on the partitioned operating condition parameter set and adopts a multi-physics field coupling mapping algorithm. The modeling parameters, initial conditions and sensitive area indexes of the local multi-physics field coupling micro-element sub-model are used as parameter factors, integrated with the node attribute data of the global finite element model, and the partition-level multi-scale multi-physics field initial configuration is generated to achieve parameter synchronization and state initialization between partitions.

[0170] S5.3 uses the partition-level multi-scale multi-physics field initial configuration to construct a multi-physics field mapping interface for the global finite element model. Through multi-scale grid coupling technologies such as matrix mapping and node interpolation, it realizes high-speed interconnection of physical variables between the local multi-physics field coupling micro-element model and the global finite element model, completing the model coupling initialization driven by working conditions.

[0171] Based on the model coupling initialization state, S5.4 adopts a state timing synchronization mechanism to drive the global finite element model to perform multi-physics field joint simulation iteration under dynamic load timing, transfer the moment working condition parameters to each local multi-physics field coupling micro-element sub-model, and receive the high-precision local physical field response returned by the local model, realizing global-local multi-physics field dynamic data two-way feedback.

[0172] S5.5 uses a partition reconstruction algorithm to integrate the real-time return data of the global finite element model and the local multi-physics field coupled microelement sub-model, and extracts a multi-scale multi-physics field interactive data set containing multi-dimensional numerical features such as fluid pressure, oil film thickness, node temperature, contact stress, etc., laying a data foundation for subsequent machine learning inversion and durability evaluation processes.

[0173] S6: Based on multi-scale multi-physics interaction data, a machine learning model is used to predict and invert local and global physical responses driven by time series load parameters, calibrate traditional physical modeling results, and generate multi-physics coupling calibration factors, specifically including:

[0174] S6.1 performs feature engineering on multi-scale and multi-physics interaction data to extract time-series feature parameters that are highly correlated with local physical field interactions and global dynamic load responses, and generates a physical field interaction feature set.

[0175] S6.2 Based on the physical field interaction feature set, a preliminary machine learning mapping model is established for different structural partitions and sensitive areas, including a supervised learning model or a graph neural network model, to characterize the high-dimensional nonlinear interaction relationship between local and global physical responses and output local-global response prediction parameters.

[0176] S6.3 compares historical experimental data, numerical simulation results and local-global response prediction parameters, and uses transfer learning or data inversion algorithms to automatically identify systematic deviations between the physical model and actual observations, generating physical response model calibration residuals.

[0177] S6.4 uses the physical response model to calibrate the residuals. By optimizing traditional physical modeling parameters (such as material constitutive parameters, lubrication coefficient, thermal conductivity coefficient, etc.), the machine learning mapping model is iteratively trained until the error between the predicted results and the observed data meets the established convergence criteria, and the calibrated multi-physics field coupling mapping factor is obtained.

[0178] S6.5 outputs the calibrated multi-physics coupling mapping factor and the calibrated local-global physical response prediction model, which are used to dynamically adjust the subsequent global finite element model and local high-precision sub-model parameters, supporting the continuous adaptive optimization of the multi-physics field-dynamic load high-fidelity joint simulation.

[0179] The S7: using the multi-physics field coupling calibration factor to dynamically adjust the parameters of the global finite element model and the local high-precision microelement sub-model to achieve highly adaptable modeling for sudden extreme and atypical working conditions specifically includes:

[0180] S7.1 performs a categorized analysis on the multi-physics coupling calibration factors generated by the machine learning model. Based on the structural unit type and node physical property distribution in the global finite element model, a parameter mapping strategy is executed to generate a global finite element model calibration parameter set to achieve the correspondence between the multi-physics coupling calibration factors and the dynamic parameters of the global finite element model.

[0181] Based on the global finite element model calibration parameter set, S7.2 performs batch dynamic assignment and real-time adjustment of the physical property parameters of the global finite element model (including node stiffness matrix, thermal conductivity coefficient, friction contact parameters, etc.), and outputs a dynamically adjusted global finite element model parameter array to ensure the high adaptability of the global finite element model to the coordinated response of physical fields under extreme and atypical dynamic working conditions.

[0182] S7.3 performs regional positioning of the multi-physics coupling calibration factor, extracts the parameter vector of the corresponding partition sub-factor based on the multi-physics sensitive area index in the local high-precision micro-element sub-model, and generates a dynamic calibration parameter set exclusive to the micro-element sub-model to achieve hierarchical mapping of global and local calibration parameters.

[0183] S7.4 applies the dynamic calibration parameter set exclusive to the microelement sub-model to local physical fields such as lubricating fluid, structural stress, and heat conduction. Through microelement modeling algorithms such as nested numerical solution and real-time adjustment of physical field boundary conditions, dynamic parameter correction is performed on the local high-precision multi-physical field coupled microelement sub-model to obtain the adjusted microelement sub-model input parameter group.

[0184] S7.5 summarizes the adjusted global finite element model parameter array and the input parameter group of the microelement sub-model, adopts a multi-scale multi-physics field mapping algorithm to realize the synchronous communication of two-layer information, and feeds back the feedback information of dynamic parameter adjustment to the machine learning model that generates the multi-physics field coupling calibration factor in real time, realizing closed-loop optimization control of adaptive modeling parameters of complex working conditions, and outputting a highly adaptable multi-physics field model set for subsequent simulation analysis.

[0185] S8: extracting time series features from the multi-physics field joint simulation results after dynamic parameter adjustment to obtain multi-dimensional parameter sequences related to typical failure mechanisms such as oil film thickness distribution, fluid pressure, temperature rise, and contact stress, specifically includes:

[0186] S8.1 extracts multi-dimensional physical field data by region based on the multi-physics field joint simulation results after dynamic parameter adjustment. Relying on the global finite element model regional index, it takes oil film thickness, fluid pressure, temperature rise and contact stress as input variables, spatially maps them based on the partition number, and outputs partition-level multi-physics field time series raw data.

[0187] S8.2 uses a physical field adaptive preprocessing algorithm to filter and normalize the partition-level multi-physics field time series raw data to remove high-frequency noise and scale drift driven by working conditions, and achieve the generation of highly consistent, multi-physics field normalized time series data.

[0188] S8.3 takes multi-physics field normalized time series data as input, and through a dynamic feature extraction method based on a sliding window, gradually divides the oil film thickness distribution, fluid pressure changes, temperature rise curves and contact stress sequence performances in continuous time periods, and realizes the regional extraction and discrimination of sensitive characteristic parameters of typical failure modes (such as minimum oil film thickness, maximum instantaneous fluid pressure, extreme temperature rise points and local stress peaks).

[0189] S8.4 targets sensitive characteristic parameters and uses time series statistical criteria and frequency domain analysis algorithms to reprocess the extreme operating condition response patterns, extracting operating condition characteristic sequence indicators that reflect the failure mechanism, including time series structured characteristics such as periodic mutations in oil film thickness, abnormal fluctuations in fluid pressure, continuous accumulation rate of temperature rise, and time-varying distribution of contact stress.

[0190] S8.5 combines the time-series structured features with the original partition number, and based on the spatial heat map reconstruction algorithm of partition-level multi-physics field sensitive parameters, outputs multi-dimensional parameter sequences such as oil film thickness distribution, fluid pressure flow field, local temperature rise distribution, contact stress cloud map, etc. related to the global typical failure mechanism of the bearing, forming a space-time multi-physics field feature database for reference in durability criterion evaluation and the next step of model adaptive iterative optimization.

[0191] S9: Determine whether the extracted multi-dimensional parameter sequence meets the preset durability and failure criteria. If not, automatically optimize the modeling granularity and coupling ratio of the multi-physics field sensitive area according to the feedback information to form a model adaptive iteration, specifically including:

[0192] S9.1 performs criterion comparison processing on the multi-dimensional parameter sequences such as oil film thickness distribution, fluid pressure, temperature rise and contact stress in the multi-physics field joint simulation results after dynamic parameter adjustment, based on the set durability criterion and failure criterion template parameters, to obtain the criterion achievement identification results.

[0193] S9.2 is based on the identification results of the obtained criteria, and calls the characteristic deviation analysis algorithm to quantitatively calculate the technical deviation between the non-compliant multi-dimensional parameter sequence and the preset criteria, and outputs the main influencing factors and corresponding multi-physical field sensitive area change recommendations.

[0194] S9.3 Based on the suggestions for changing the multi-physics field sensitive areas, execute the sensitive area modeling granularity optimization algorithm, dynamically refine or coarsen the spatial resolution and physical variable sampling density of the original multi-physics field sensitive areas, and form optimized modeling granularity distribution data.

[0195] S9.4 takes the optimized modeling particle size distribution data as input, adopts the coupling weight adaptive adjustment algorithm, dynamically adjusts the weights of multi-physics field coupling parameters, and outputs the updated regional coupling weight distribution matrix.

[0196] S9.5 reconfigures the information interaction parameters between the multi-physics coupling micro-element model and the global finite element model based on the updated regional coupling weight distribution matrix, realizes model adaptive simulation configuration optimization, generates a new round of multi-physics joint simulation input conditions, and realizes model adaptive iterative closed loop.

[0197] S10: Outputs high-fidelity multi-physics field-dynamic load joint simulation analysis results after adaptive optimization, which are used for durability and failure mechanism prediction and authoritative load-bearing capacity evaluation of oil-containing bearings under complex dynamic conditions, specifically including:

[0198] S10.1 performs normalization and structural transformation on the multi-dimensional time series parameter sequence output by the multi-physics field joint simulation after dynamic parameter adjustment based on the feature mapping algorithm to obtain the oil film thickness distribution parameter sequence, fluid pressure parameter sequence, temperature rise parameter sequence and contact stress parameter sequence in a unified format, which serve as the basic input for subsequent durability and failure mechanism evaluation.

[0199] Based on a structured multi-dimensional time series parameter sequence, S10.2 uses a working condition-adaptive failure criterion algorithm to perform a multi-criteria aggregation evaluation of oil film thickness distribution parameters, fluid pressure parameters, temperature rise parameters, and contact stress parameters, achieving a preliminary classification prediction of the durability and failure risk of oil-containing bearings under complex working conditions.

[0200] S10.3 Based on the failure risk information predicted by the preliminary classification, high-fidelity multi-physics field-dynamic load inversion technology is used to conduct engineering traceability analysis on the local extreme response characteristics, further refine the fault failure mechanism diagnosis output, and obtain regional failure causes and evolution paths.

[0201] Based on the failure mechanism diagnosis results, S10.4 uses a model adaptive evaluation algorithm to perform secondary parameter self-optimization on the simulation modeling granularity, sensitive area coupling parameters, etc., to generate the final optimal multi-physics field modeling parameter set and a high-fidelity simulation database of global dynamic load response.

[0202] S10.5 takes the optimal multi-physics modeling parameter set and a high-fidelity simulation database as input, and uses an authoritative load-bearing capacity assessment model to provide multi-dimensional quantitative output of the load-bearing capacity boundaries, endurance life, and extreme failure modes of oil-containing bearings under various complex dynamic working conditions, forming a deliverable authoritative load-bearing capacity assessment report.

[0203] Those skilled in the art can make various other corresponding changes and deformations based on the technical solutions and concepts described above, and all of these changes and deformations should fall within the scope of protection of the claims of this application.

Claims

1. A method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation, characterized in that: include: S1. Obtaining working condition data samples of an oil-containing bearing under a typical operating state under load conditions, and identifying bearing structure partitions to establish a working condition partition mapping table; S2. For different structural partitions in the working condition partition mapping table, collect multi-physics field time series experimental data, and perform noise reduction and standardization processing to obtain multi-physics field basic data; S3. Based on the multi-physics basic data and partition identification, a multi-physics sensitive area identification model is established for each partition within the global finite element model framework, and the multi-physics sensitive area index and physical variable influence weight of each partition are output; S4. For the identified multi-physics sensitive area, establish a local multi-physics coupled micro-element sub-model, and associate the local multi-physics coupled micro-element sub-model parameters with the multi-physics sensitive area index; S5. Input the working condition data sample and the local multi-physics field coupled microelement sub-model into the global finite element model, and output the multi-scale multi-physics field interaction data; S6. Based on multi-scale multi-physics interaction data, predict and invert the local and global physical responses driven by time series load parameters to generate multi-physics coupling calibration factors; S7. Use multi-physics coupling calibration factors to dynamically adjust the parameters of the global finite element model and the local high-precision microelement sub-model; S8. Extract the timing characteristics of the multi-physics field joint simulation results after dynamic parameter adjustment to obtain the multi-dimensional parameter sequence related to the typical failure mechanism.

2. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The working condition data in S1 includes dynamic load, impact load and asymmetric load working condition labels, and the identification of the bearing structure partitions to establish the working condition partition mapping table specifically includes: Conduct dynamic load tests on the target oil-containing bearing under typical operating conditions to obtain original operating data covering three types of conditions: dynamic load, impact load, and asymmetric load, and form a load type mapping data set; Based on the original working condition data samples, the event-driven labeling algorithm is used to assign load condition labels to each time series sampling point to generate a working condition data time series label set; A partitioning algorithm is applied to the 3D CAD structural model of the oil-bearing bearing to subdivide the entire bearing into functional areas and obtain the structural partition index of each area. The time series label set of the working condition data is fused with the structural partition index in high-dimensional data, and a multi-label mapping strategy is used to generate a working condition-structure integrated mapping table for each structural partition under each load condition. Based on the working condition-structure integrated mapping table, data consistency verification and redundancy filtering are performed on typical structure partition samples under each operating state to obtain the working condition partition mapping table.

3. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The multi-physics sensitive area index and physical variable influence weight of each partition output in S3 include: The basic data and partition identification of multi-physics fields are structured and organized, and a data layer mapping strategy is adopted to bind each structural partition with its own time series physical variables in the global finite element model; Based on the established partition-physical variable basic mapping table, a local correlation analysis algorithm is applied to perform multidimensional statistics on the temporal distribution characteristics of multiple physical field variables within the bearing structure partitions to obtain the correlation matrix of the physical variables within the partitions. Perform sensitivity quantitative assessment on the correlation matrix and actual failure annotation data, calculate the sensitivity weight distribution of each physical variable on structural performance, and output the multi-physics field sensitivity factor matrix for each structural partition; Based on the multi-physics field sensitivity factor matrix, the sensitive regions of the structural partitions in the global finite element model are clustered and divided. The high-weight physical response regions are calibrated using a spatial clustering algorithm and a neighborhood recognition mechanism, and the sensitive region index of each structural partition is automatically generated. Integrate the partitioned sensitive region index with its corresponding multiphysics sensitivity weight.

4. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The step S4 of associating the microelement sub-model parameters with the sensitive area index specifically includes: Based on the multi-physics sensitive area index, the corresponding local structural geometric parameters, material parameters and working boundary conditions are screened to generate the multi-physics sensitive area micro-element analysis object; For the micro-element analysis objects in the multi-physics field sensitive area, a multi-scale finite element partitioning algorithm is used to refine the finite element mesh structure and obtain a refined finite element mesh model; Using the fluid-structure-thermal multi-physics coupling solver, fluid dynamics, heat conduction, and structural mechanics boundary conditions are applied to the lubricating oil film sub-area, heat source point, and load-bearing surface in the refined finite element mesh model to obtain the lubricating fluid dynamics-structural thermal multi-physics field combined reaction parameters; Based on the above multi-physics field joint response parameters, a dynamic prediction algorithm for surface contact stress and oil film thickness is executed to output high-resolution local parameters and form a local high-precision multi-physics field response set; The intermediate parameters of the local high-precision multi-physics field response set are associated with the original global finite element model through the sensitive area index to establish the correlation data mapping rules, so as to realize the structural synchronous association between the local microelement sub-model parameters and the sensitive area index.

5. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The S5 output multi-scale multi-physics field interaction data specifically includes: Perform semantic label mapping on the obtained working condition data samples, take the working condition parameters as input, and perform data adaptation processing on each structural partition through the working condition partition mapping table to generate a partition working condition parameter set; Based on the partitioned operating condition parameter set, a multi-physics coupling mapping algorithm is used to integrate the modeling parameters, initial conditions, and sensitive area indexes of the local multi-physics coupling micro-element sub-model as parameter factors with the node attribute data of the global finite element model to generate a partition-level multi-scale multi-physics initial configuration. Utilize the partition-level multi-scale multi-physics initial configuration to build a multi-physics mapping interface for the global finite element model. Use multi-scale grid coupling technology to complete the model coupling initialization driven by the working condition. Based on the model coupling initialization state, the state timing synchronization mechanism is adopted to drive the global finite element model to perform multi-physics field joint simulation iteration under dynamic load timing, pass the moment working condition parameters to each local multi-physics field coupling micro-element sub-model, and receive the high-precision local physical field response returned by the local model; The real-time return data of the global finite element model and the local multi-physics field coupled microelement sub-model are used to integrate the structural partition multi-physics field information through the partition reconstruction algorithm to extract the multi-scale multi-physics field interaction data set with multi-dimensional numerical features.

6. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The S6 generating the multi-physics field coupling calibration factor specifically includes: Perform feature engineering on multi-scale and multi-physics interaction data to extract time-series characteristic parameters that are highly correlated with local physics interactions and global dynamic load responses, generating a physics interaction feature set. Based on the physical field interaction feature set, a preliminary machine learning mapping model is established for different structural partitions and sensitive areas, including a supervised learning model or a graph neural network model, to output response prediction parameters; Compare historical experimental data, numerical simulation results and response prediction parameters to automatically identify systematic deviations between the physical model and actual observations and generate physical response model calibration residuals; The physical response model is used to calibrate the residuals. By optimizing the traditional physical modeling parameters, the machine learning mapping model is iteratively trained until the error between the predicted results and the observed data meets the established convergence criteria, thus obtaining the calibrated multi-physics field coupling mapping factor. Output the calibrated multi-physics field coupling mapping factor and the calibrated physical response prediction model.

7. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The S7 achieves highly adaptable modeling for sudden extreme and atypical operating conditions, specifically including: Perform a categorized analysis on the multi-physics coupling calibration factors generated by the machine learning model. Based on the structural unit types and node physical property distribution in the global finite element model, a parameter mapping strategy is implemented to generate a global finite element model calibration parameter set. Based on the global finite element model calibration parameter set, the physical property parameters of the global finite element model are dynamically assigned and adjusted in batches in real time, and a dynamically adjusted global finite element model parameter array is output; Perform regional positioning of the multi-physics coupling calibration factor, extract the parameter vector of the corresponding partition sub-factor based on the multi-physics sensitive area index in the local high-precision micro-element sub-model, and generate a dynamic calibration parameter set exclusive to the micro-element sub-model; Apply the dynamic calibration parameter set exclusive to the microelement submodel to the local physical field, and use the microelement modeling algorithm to dynamically correct the parameters of the local high-precision multi-physics field coupled microelement submodel to obtain the adjusted microelement submodel input parameter set; The adjusted global finite element model parameter array and the microelement sub-model input parameter group are summarized, and the feedback information of the dynamic parameter adjustment is fed back in real time to the machine learning model that generates the multi-physics field coupling calibration factor, outputting a highly adaptable multi-physics field model set for subsequent simulation analysis.

8. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: The step of obtaining the multidimensional parameter sequence in step S8 specifically includes: The multi-dimensional physical field data is extracted by region based on the multi-physics field joint simulation results after dynamic parameter adjustment. Relying on the global finite element model region index, the oil film thickness, fluid pressure, temperature rise and contact stress are used as input variables, and spatial mapping is performed based on the partition number to output the partition-level multi-physics field time series raw data; For the original multi-physics time series data at the partition level, remove the high-frequency noise and scale drift driven by the working conditions to generate normalized multi-physics time series data; Using multi-physics field normalized time series data as input, a dynamic feature extraction method based on a sliding window is used to gradually segment the oil film thickness distribution, fluid pressure changes, temperature rise curves, and contact stress sequence performance within continuous time periods to obtain sensitive characteristic parameters of typical failure modes. For sensitive characteristic parameters, the time series statistical criterion and frequency domain analysis algorithm are used to reprocess the extreme working condition response mode, extract the working condition characteristic sequence indicators reflecting the failure mechanism, and obtain the time series structured characteristics; The temporal structured features are combined with the original partition numbers, and a spatial heat map reconstruction algorithm based on partition-level multi-physics field sensitive parameters is used to output a multidimensional parameter sequence related to the global typical failure mechanism of the bearing, forming a spatial-temporal multi-physics field feature database.

9. The method for evaluating the load-bearing capacity of an oil-containing bearing based on virtual simulation according to claim 1, characterized in that: Also includes: S9. Determine whether the extracted multidimensional parameter sequence meets the preset durability and failure criteria. If not, automatically optimize the modeling granularity and coupling ratio of the multi-physics field sensitive area based on the feedback information to form a model adaptive iteration; S10. Output high-fidelity multi-physics field-dynamic load joint simulation analysis results after adaptive optimization, which are used for durability and failure mechanism prediction and authoritative load-bearing capacity evaluation of oil-containing bearings under complex dynamic working conditions.

Images