An automatic matching system for a bearing assembly line

Abstract

This invention relates to the field of bearing assembly technology, and more particularly to an automatic matching system for bearing assembly lines. This system categorizes component operating conditions into three types based on historical data: constant speed, variable load, and frequent start-stop. It also acquires the thermophysical characteristics of bearing assemblies. Combining the thermodynamic characteristics of the operating conditions with a pre-built rule base, it determines clearance compensation requirements and generates an initial matching scheme. Subsequently, it uses historical temperature rise data to mine clearance decay and thermal offset patterns to optimize the scheme. Simultaneously, it establishes a thermal deformation prediction model integrating heat sources, heat dissipation, and structural dimensions, outputting the thermal clearance prediction results. Finally, it compares real-time matching parameters with the prediction results, continuously refining the matching scheme and thresholds through data accumulation and feedback mechanisms, constructing an adaptive closed-loop system. This invention achieves a leap from static, experience-based matching to dynamic, precise matching, effectively solving the problem of clearance runaway caused by thermal expansion, and significantly improving bearing assembly accuracy, operational reliability, and the level of production line intelligence.

Description

Technical Field

[0001] This invention relates to the field of bearing assembly technology, and in particular to an automatic selection system for bearing assembly lines. Background Technology

[0002] With increasingly stringent requirements for bearing performance, precision, and reliability in high-end equipment manufacturing, the control of bearing thermal behavior under complex operating conditions has become a crucial factor affecting its service life and operational stability. During bearing operation, frictional heat generation and external loads cause varying degrees of thermal expansion in components, leading to changes in the initial assembly clearance. This can result in problems such as thermal seizure due to insufficient clearance or increased vibration and noise due to excessive clearance. Traditional bearing assembly and selection methods primarily rely on static dimensional tolerance matching and empirical clearance settings, which struggle to fully consider the impact of dynamic temperature field changes on thermal deformation under actual operating conditions. Especially under typical conditions such as constant high speed, variable load, or frequent start-stop cycles, they lack the ability to accurately predict and adaptively compensate for temperature rise characteristics and thermal expansion behavior.

[0003] In recent years, digital twin technology has provided a new technical path for realizing real-time interaction and collaborative optimization between physical manufacturing systems and virtual models. Although some studies have attempted to apply thermo-mechanical coupling models to bearing thermal deformation analysis, most methods remain at the offline simulation stage, failing to deeply integrate with the actual assembly line selection process and lacking a closed-loop optimization mechanism based on historical data and real-time feedback. Furthermore, existing selection strategies typically ignore the influence of implicit factors such as material microstructure and residual stress distribution on thermal conductivity and thermal expansion behavior, resulting in insufficient accuracy in thermophysical feature identification and making it difficult to achieve personalized, high-precision assembly matching.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide an automatic matching system for bearing assembly lines, which aims to solve the technical problem that existing bearing assembly and matching methods rely on static size matching and lack accurate prediction and closed-loop compensation capabilities for temperature rise characteristics and thermal expansion behavior under dynamic operating conditions, making it difficult to solve the technical problem of jamming or vibration noise caused by uncontrolled thermal clearance.

[0006] To achieve the above objectives, the present invention provides an automatic selection system for a bearing assembly line, the system comprising: The working condition classification module is used to obtain historical processing and inspection data of the parts to be selected in the bearing assembly line. Based on the historical processing and inspection data, the working condition adaptation type of the assembled parts is divided into three types: constant speed working condition, variable load working condition, and frequent start-stop working condition. The feature extraction module is used to obtain the thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled, so as to obtain the current thermophysical characteristics of the bearing components. The requirement determination module is used to determine the assembly clearance compensation requirements for different operating conditions based on the thermodynamic characteristics of the three operating conditions and in conjunction with a pre-established thermal expansion matching rule database. The scheme generation module is used to obtain a thermal expansion compensation scheme that conforms to the current operating conditions based on the current thermophysical characteristics and the thermal expansion selection rule database. The regularity optimization module is used to acquire historical operating temperature rise data under different operating conditions, obtain the change law of bearing clearance decay and thermal offset cycle under the influence of different operating parameters, and optimize the corresponding thermal expansion compensation selection scheme. The model prediction module is used to establish a thermal deformation prediction model for bearings based on the intensity of the heat source, heat dissipation conditions, and structural dimensions, and to obtain the thermal clearance prediction results of the thermal deformation prediction model under different inputs. The closed-loop feedback module is used to obtain real-time matching parameters based on the optimized thermal expansion compensation matching scheme, and compare the real-time matching parameters with the thermal clearance prediction results. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

[0007] Optionally, the step of acquiring historical processing and inspection data of the components to be assembled on the bearing assembly line, and classifying the working condition adaptation type of the assembled components into three types based on the historical processing and inspection data: constant speed type, variable load type, and frequent start-stop type, including: Obtain historical processing and testing data of the components to be selected in the bearing assembly line, including material microstructure distribution data; The similarity between the material's microstructure distribution data and a pre-established standard material library is calculated to obtain the material's thermal type and write it into the working condition classification database. Based on the thermal type of the material, the proportion of residual stress and hardness distribution in the material is determined to obtain the physicochemical measurement results of the material. The non-uniformity of thermal conductivity was quantitatively analyzed based on the physicochemical measurement results of the material, and the quantitative analysis results were obtained. The application scenario type value of the bearing is obtained. Combined with the material thermal type, material physicochemical measurement results and quantitative analysis results, the working condition adaptation type of the assembled parts is divided into three types: constant speed type, variable load type and frequent start-stop type.

[0008] Optionally, obtaining the thermal expansion coefficients and thermal conductivity data of the inner ring, outer ring, and rolling elements of the bearing to be assembled, to obtain the current thermophysical characteristics of the bearing components, includes: The thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled are obtained. The thermal expansion coefficient and thermal conductivity data are processed by an adaptive Kalman filter and a wavelet denoiser to obtain thermophysical processing data. Based on the thermophysical processing data, the size expansion values ​​and temperature field values ​​of the past preset batches are extracted from the thermophysical database. The size expansion values ​​and temperature field values ​​are plotted using a time series diagram to obtain the expansion change curve and the temperature rise distribution curve. For the expansion change curve and the temperature rise distribution curve, the rate of change of the statistical thermal environment is calculated using derivatives, and the thermal critical point is marked to obtain the physical value of the thermal critical point. The thermal characteristics of the physical values ​​of the thermal critical point are classified by Mahalanobis distance hierarchical clustering, and the current thermophysical characteristics of the bearing components are determined by combining them with a pre-established thermophysical feature database.

[0009] Optionally, determining the assembly clearance compensation requirements for different operating conditions based on the thermodynamic characteristics of the three operating conditions and in conjunction with a pre-established thermal expansion matching rule database includes: Based on the thermodynamic characteristics of the three working conditions, the thermal deformation of the three working conditions is obtained from the pre-established thermal expansion matching rule database, and the thermal expansion difference level of the three working conditions is obtained by variance analysis. The selection clearance adjustment frequency value, the coordinate value of the fit tolerance zone, the temperature sensor type value, and the thermal parameter acquisition accuracy value are obtained through the thermal expansion selection rule database. The selection requirement group is obtained by grouping according to the parameter dimension. The optional requirements are classified using a density peak clustering tool, and the assembly clearance compensation requirements for different operating conditions are selected by combining tolerance zone planning, optional time interval division, and temperature sensor.

[0010] Optionally, after obtaining a thermal expansion compensation matching scheme that conforms to the current operating condition characteristics based on the current thermophysical characteristics and the thermal expansion matching rule database, the method further includes: Obtain the clearance anomaly threshold, thermal offset threshold, optional data storage format, and optional data upload cycle in the thermal expansion compensation optional scheme, and upload the data to the optional log table according to the optional data upload cycle.

[0011] Optionally, the step of acquiring historical operating temperature rise data under different operating conditions, obtaining the bearing clearance decay and thermal offset periodic variation law under the influence of different operating parameters, and optimizing the corresponding thermal expansion compensation selection scheme includes: Acquire historical operating temperature rise data for different operating conditions, and extract the bearing clearance attenuation distribution and thermal offset change rate characteristics from the historical operating temperature rise data; By analyzing the correlation rules between historical operating temperature rise data and environmental thermal field data, the clearance decay and thermal offset periodic variation law of the bearing under the influence of different operating parameters are obtained, and the corresponding thermal expansion compensation selection scheme is optimized.

[0012] Optionally, the step of obtaining real-time matching parameters based on the optimized thermal expansion compensation matching scheme, comparing the real-time matching parameters with the thermal clearance prediction results, and continuously improving the thermal expansion compensation matching scheme and thermal expansion threshold through data accumulation and feedback to form an adaptive automatic thermal expansion matching closed loop includes: The real-time selection parameters are obtained based on the optimized thermal expansion compensation selection scheme, and the real-time selection parameters are compared with the thermal clearance prediction results to obtain the parameter comparison results. Based on the parameter comparison results, combined with the selection clearance adjustment frequency, fit tolerance zone area, temperature sensor type and thermal parameter acquisition accuracy parameters in the thermal expansion compensation matching scheme, as well as the thermal deformation prediction model, the automatic matching device of the bearing assembly line is deployed and debugged, and wireless data is uploaded according to the data upload cycle. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

[0013] Optionally, the step of establishing a thermal deformation prediction model for the bearing based on the intensity of the heat source, heat dissipation conditions, and structural dimensions, and obtaining the thermal clearance prediction results of the thermal deformation prediction model under different inputs, includes: Construct a finite element digital twin that includes bearing geometric parameters, material physical properties, and lubricating oil rheological characteristics; Based on fluid dynamics lubrication theory and heat conduction equations, the frictional heat generation and heat transfer path of bearings under different speeds and loads are simulated to generate a three-dimensional temperature field distribution cloud map inside the bearing. Based on the three-dimensional temperature field distribution cloud map, calculate the non-uniform thermal deformation of the inner ring, outer ring and rolling element under thermal equilibrium state. The non-uniform thermal deformation amount is vector-superimposed with the initial assembly clearance to obtain the thermal clearance prediction results of the thermal deformation prediction model under different inputs.

[0014] Optionally, after obtaining real-time matching parameters based on the optimized thermal expansion compensation matching scheme and comparing the real-time matching parameters with the thermal clearance prediction result, the method further includes: Set the allowable deviation range for hot clearance. If the comparison result shows that the difference exceeds the allowable deviation range for hot clearance, the error correction mechanism is triggered. In a digital twin, the fit tolerance of selected components is iteratively optimized using a genetic algorithm or a particle swarm optimization algorithm until the calculated hot clearance prediction result falls within the allowable deviation range of the hot clearance. The final fit tolerance parameters obtained through iterative optimization are used as the corrected real-time fit parameters and sent to the assembly execution unit.

[0015] Optionally, the step of continuously improving the thermal expansion compensation selection scheme and thermal expansion threshold through data accumulation and feedback includes: Collect measured temperature data and vibration spectrum data of the assembled bearing during actual trial operation; The measured temperature data and vibration spectrum data are input into the digital twin, and residual analysis is performed between them and the simulation results of the thermal deformation prediction model. If the residual analysis error is greater than the preset accuracy, the measured temperature data and vibration spectrum data are used to reverse the thermal boundary conditions and convective heat transfer coefficient of the thermal deformation prediction model. Based on the revised thermal deformation prediction model, the assembly clearance compensation requirements in the thermal expansion matching rule database are updated.

[0016] In the automatic matching system for a bearing assembly line of this invention, a differentiated clearance compensation mechanism is constructed for constant speed, variable load, and start-stop conditions by comprehensively considering historical processing and inspection data, material thermophysical properties, and actual operating condition classification. This significantly improves the adaptability of the assembly scheme to the actual operating environment, effectively avoids problems of excessively small or large clearance due to thermal expansion, and extends the service life of the bearing. A thermal deformation prediction model based on heat source intensity, heat dissipation conditions, and structural dimensions, combined with digital twin simulation technology, can accurately predict the trend of thermal clearance changes under different operating conditions. By comparing real-time matching parameters with prediction results and introducing a feedback learning mechanism, a perception, decision-making, optimization, and iteration process is formed. An adaptive closed-loop system continuously improves the intelligence and generalization ability of the selection strategy. By utilizing historical temperature rise data to mine the periodic patterns of clearance decay and thermal offset, the system optimizes the compensation scheme and continuously corrects the thermal expansion threshold and selection rule database through data accumulation. This enables the system to learn and evolve autonomously with time and changing operating conditions, adapting to the flexible assembly requirements of multiple varieties and variable batches. This method achieves full-process integration from operating condition identification, thermal characteristic analysis, scheme matching to automatic selection execution, reducing reliance on human experience, supporting the intelligent and automated operation of assembly lines, improving assembly efficiency and consistency, and is suitable for large-scale, high-quality production of high-precision bearing components in high-end equipment manufacturing. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of the first embodiment of the automatic selection system for the bearing assembly line of the present invention; Figure 2 This is a flowchart illustrating the specific steps involved in obtaining the current thermophysical characteristics of bearing components in the automatic selection system of the bearing assembly line of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0020] In one embodiment, such as Figure 1 As shown, an automatic selection system for a bearing assembly line is provided, the system comprising: The working condition classification module 10 is used to obtain historical processing and inspection data of the parts to be selected in the bearing assembly line. Based on the historical processing and inspection data, the working condition adaptation type of the assembled parts is divided into three types: constant speed working condition, variable load working condition, and frequent start-stop working condition. In this context, the components to be selected for the bearing assembly line can be bearing assemblies in the assembly process that have not yet been finalized, including inner rings, outer rings, and rolling elements. These are used as the objects of selection, and their geometric and material properties directly affect the final assembly clearance. For example, the components to be selected for the bearing assembly line may include the bearing inner ring, bearing outer ring, and rolling elements. Historical processing and inspection data can be a collection of data recording the dimensions, geometric tolerances, and process parameters of bearing components during manufacturing and inspection. This data can be used to provide input for classifying operating conditions, reflecting the manufacturing consistency and potential thermal behavior of the components. In a specific embodiment, historical processing and inspection data can be obtained from CMM measurements, online inspection equipment, or MES systems. Constant speed operating conditions refer to the type of operating condition where the bearing operates at a stable speed for a long time. This type corresponds to the thermal behavior characteristics of a stable heat source and a temperature rise that tends to be balanced, requiring the setting of steady-state thermal expansion compensation. Variable load operating conditions refer to the type of operating condition where the bearing is subjected to periodic or randomly changing loads. This type can lead to fluctuations in heat source intensity, requiring consideration of transient thermal response and cumulative thermal deformation effects. Frequent start-stop conditions refer to the type of operating conditions in which the bearing undergoes multiple start-stop cycles. This can lead to unsteady thermal cycling, which can easily cause thermal fatigue and clearance drift, requiring special compensation strategies.

[0021] The thermal characteristic sensing module 20 is used to acquire the thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled, so as to obtain the current thermophysical characteristics of the bearing components. The bearing inner ring can be a ring-shaped component that mates with and rotates with the shaft. Its thermal expansion directly affects the fit between the inner ring and the shaft, as well as the internal clearance. Furthermore, the thermophysical parameters of the bearing inner ring can be obtained from the material grade and heat treatment records. The bearing outer ring can be a ring-shaped component fixed in the bearing housing. Its thermal expansion affects the fit between the outer ring and the housing bore, as well as the overall clearance distribution. In an exemplary embodiment, the thermophysical parameters of the bearing outer ring can be obtained from the material grade and heat treatment records. The rolling element can be a spherical or roller-shaped element that rolls and transmits load between the inner and outer ring raceways. Its thermal expansion affects the contact stress distribution and dynamic changes in clearance. For example, the thermophysical parameters of the rolling element can be obtained from the material grade and manufacturing batch information. The coefficient of thermal expansion (CTE) can be the linear dimensional change rate of a material per unit temperature when the temperature changes. It can be used to determine the amount of expansion of a component under temperature rise and is a key input for calculating thermal deformation. Furthermore, the CTE can be obtained from material handbooks, experimental measurements, or process databases. Thermal conductivity data can be a set of physical parameters describing a material's ability to conduct heat, including thermal conductivity, and can be used to influence the rate of temperature rise and temperature field distribution, thereby affecting the spatial non-uniformity of thermal deformation. In a specific embodiment, thermal conductivity data can be retrieved from material handbooks, experimental measurements, or process databases. Current thermophysical characteristics can be a comprehensive description of the coefficient of thermal expansion and thermal conductivity of the components to be assembled, and can be used as input to the scheme matching module to reflect individual differences in thermal behavior. Furthermore, current thermophysical characteristics can be generated by the thermal characteristic sensing module by integrating the thermophysical parameters of various components.

[0022] The parameter determination module 30 is used to determine the assembly clearance compensation requirements for different operating conditions based on the thermodynamic characteristics of the three operating conditions and in conjunction with a pre-established thermal expansion matching rule database. Thermodynamic properties can refer to the heat generation, conduction, and deformation patterns exhibited by the bearing system under different operating conditions. These properties can serve as the basis for the parameter determination module, connecting the operating condition type with compensation requirements. In a specific embodiment, thermodynamic properties may include steady-state thermal equilibrium characteristics, transient thermal shock characteristics, and cyclic thermal fatigue characteristics. Assembly clearance compensation requirements can be the initial clearance amount that needs to be reserved to offset the effects of thermal expansion under specific operating conditions. This can guide the scheme matching module in generating specific selection parameters, ensuring that the hot clearance is within a reasonable range. Furthermore, assembly clearance compensation requirements can be mapped from a rule base by the parameter determination module based on the operating condition type and thermodynamic properties. The thermal expansion selection rule database can be a knowledge base storing the mapping relationship between thermophysical characteristics and assembly clearance compensation requirements under different operating conditions. This can provide structured rule support for parameter determination and scheme matching, realizing the conversion from thermal characteristics to assembly parameters. In an exemplary embodiment, the thermal expansion selection rule database can be constructed from historical assembly data, simulation results, or expert experience, and supports dynamic updates. For example, the thermal expansion matching rule database may include a constant speed rule sub-library, a variable load rule sub-library, a start-stop condition rule sub-library, etc.

[0023] The scheme matching module 40 is used to obtain a thermal expansion compensation scheme that conforms to the current operating conditions based on the current thermophysical characteristics and the thermal expansion selection rule database. The thermal expansion compensation matching scheme can be an executable assembly instruction containing specific component combinations and initial clearance settings. This instruction can be directly used for assembly execution to ensure that the clearance meets design requirements during hot operation. Furthermore, the thermal expansion compensation matching scheme can be generated by a scheme matching module based on current thermophysical characteristics and a rule base. Based on the current thermophysical characteristics and the thermal expansion matching rule database, a thermal expansion compensation matching scheme that conforms to the current operating conditions is obtained. This can be achieved by matching the current thermophysical characteristics with conditions in the rule base, outputting the optimal component combination and initial clearance. In an exemplary embodiment, this operation can be implemented by selecting the most similar historical case based on nearest neighbor search and solving feasible solutions based on constraint satisfaction problems, thereby achieving the technical effect of ensuring that the initial clearance setting has operating condition adaptability.

[0024] The data optimization module 50 is used to acquire historical operating temperature rise data under different working conditions, obtain the bearing clearance decay and thermal offset cycle change law under the influence of different operating parameters, and optimize the corresponding thermal expansion compensation selection scheme. Historical operating temperature rise data can be historical data recording the temperature changes of the bearing over time during actual operation, providing a data foundation for exploring the periodic patterns of clearance decay and thermal offset. In a specific embodiment, historical operating temperature rise data can be obtained from online temperature sensors, remote monitoring systems, or after-sales maintenance records. Clearance decay refers to the gradual reduction of bearing clearance due to factors such as thermal cycling and wear during long-term operation, and can be used as a key indicator for optimizing compensation schemes, reflecting long-term thermal stability. Furthermore, clearance decay can include thermo-induced plastic deformation decay, fretting wear decay, and material creep decay. The periodic variation pattern of thermal offset can be the periodic drift pattern of clearance caused by thermal deformation of the bearing under periodic operating conditions, which can be used to predict future thermal clearance change trends and optimize the foresight of compensation strategies. In an exemplary embodiment, the periodic variation pattern of thermal offset can be extracted from historical temperature rise data through time series analysis or Fourier transform.

[0025] The model prediction module 60 is used to establish a thermal deformation prediction model for bearing bonding heat source intensity, heat dissipation conditions and structural dimensions, and to obtain the thermal clearance prediction results of the thermal deformation prediction model under different inputs. The thermal deformation prediction model can be a mathematical or simulation model describing the thermal deformation behavior of a bearing under specific heat source intensity, heat dissipation conditions, and structural dimensions. It can be used to predict the trend of hot clearance changes under different operating conditions, providing a forward-looking basis for selecting a suitable configuration. In an exemplary embodiment, the thermal deformation prediction model can be constructed based on the heat conduction equation and thermoelastic theory, combined with actual structural parameters for parametric modeling. Heat source intensity can be the amount of heat input per unit time generated by friction during bearing operation, which can be used as a key input to the thermal deformation prediction model, determining the temperature rise. Furthermore, the heat source intensity can be calculated from operating parameters such as rotational speed, load, and lubrication status. Heat dissipation conditions can be the bearing system's ability to dissipate heat to the environment, affected by cooling methods and ambient temperature, and can be used to influence the steady-state temperature level, serving as an important boundary condition for the thermal deformation prediction model. Exemplary heat dissipation conditions can include natural convection cooling, forced air cooling, and oil cooling. Structural dimensions can be the geometrical parameters of various bearing components, including diameter, width, and wall thickness, which can be used to determine the heat conduction path and thermal deformation stiffness, affecting the spatial distribution of thermal deformation. In one specific embodiment, structural dimensions can be obtained from CAD models or inspection data. The hot clearance prediction result can be the clearance value output by the thermal deformation prediction model when the bearing reaches thermal equilibrium under specific operating conditions. This value can be used as a comparison benchmark for the closed-loop optimization module to evaluate the rationality of the selected configuration. Furthermore, the hot clearance prediction result can be output by the model prediction module after running the thermal deformation prediction model.

[0026] The closed-loop optimization module 70 is used to obtain real-time matching parameters based on the optimized thermal expansion compensation matching scheme, and compare the real-time matching parameters with the thermal clearance prediction results. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

[0027] The real-time configuration parameters can be the actual component combinations and initial clearance settings used in the current assembly task. These parameters reflect the actual configuration results and are used for deviation analysis compared to the predicted results. Furthermore, the real-time configuration parameters can be output by the scheme matching module and confirmed by the assembly execution system. The thermal expansion threshold can be a critical value used to determine whether thermal expansion exceeds the allowable range. It can be used to trigger compensation scheme adjustments or alarm mechanisms to ensure assembly quality. In an exemplary embodiment, the thermal expansion threshold can be continuously corrected by the closed-loop optimization module based on historical data and feedback.

[0028] Taking the assembly of high-end CNC machine tool spindle bearings as an example, the automatic matching system for the bearing assembly line in this embodiment can be as follows: On the assembly line, the system first acquires the historical machining data of a certain batch of spindle bearings, and the working condition classification module identifies that it is suitable for constant high-speed working conditions; the thermal characteristic sensing module reads the thermal expansion coefficient and thermal conductivity of the inner ring GCr15 material; the parameter determination module retrieves the compensation requirements under high-speed steady-state working conditions from the thermal expansion matching rule database; the scheme matching module generates a specific matching scheme based on the current thermophysical characteristics; at the same time, the model prediction module predicts that the hot clearance is 8 micrometers based on the structural dimensions of the bearing, the expected heat source intensity and oil cooling heat dissipation conditions; after the assembly is executed, the closed-loop optimization module compares the actual matching parameters with the prediction results, finds that the deviation is less than the threshold, and confirms that the scheme is effective; if the subsequent operation data shows that the clearance has a slow decay trend, the data optimization module will extract the pattern and update the rule base, so that the compensation amount of subsequent similar bearings is appropriately increased, thereby achieving adaptive evolution.

[0029] In one embodiment, historical processing and inspection data of the components to be assembled on the bearing assembly line are acquired. Based on this data, the operating condition adaptation types of the assembled components are categorized into three types: constant speed operating condition, variable load operating condition, and frequent start-stop operating condition. Obtain historical processing and testing data of the components to be selected in the bearing assembly line. The historical processing and testing data includes material microstructure distribution data. Among them, the material microstructure distribution data can be detection data describing the internal grain structure, phase composition, and distribution characteristics of bearing component materials. It can be used to reflect the microscopic root causes of the material's thermal expansion and thermal conductivity, providing a basis for identifying hidden thermal behavior differences. In this embodiment, the material microstructure distribution data can be obtained through material characterization methods such as metallographic microscopy, electron backscatter diffraction, or X-ray diffraction.

[0030] The thermal type of the material is obtained by calculating the similarity between the material's microstructure distribution data and a pre-established standard material library, and then written into the working condition classification database. The standard material library can be a benchmark database storing the correspondence between the microstructure and thermophysical properties of known materials under different heat treatment states. It can serve as a reference for similarity calculations, supporting the inference of material thermal types from microstructure. For example, the standard material library can be constructed and continuously calibrated using historical experimental data, material handbooks, or simulation results. Similarity calculations based on the material's microstructure distribution data and the pre-established standard material library can involve distance measurement or pattern matching between the microstructure feature vector of the material under test and samples in the standard library. Further, this operation is achieved by using Euclidean distance to calculate the similarity of microstructure features and using a convolutional neural network to extract microstructure image features followed by cosine similarity matching, thereby enabling the transformation from implicit microstructure to identifiable thermal types. Obtaining the material's thermal type and writing it into the operating condition classification database can be achieved by using the standard category with the highest matching degree in the similarity calculation results as the material's thermal type and storing it in the operating condition classification database. In a specific embodiment, this operation can be achieved by directly writing the label of the highest matching category and writing Top-K candidate types and their confidence levels, thereby establishing a persistent association between the material's microstructure properties and subsequent operating condition classifications. Material thermal type can be a category label characterizing the thermophysical behavior tendency of a material, derived from microstructure similarity classification. It can be used to transform implicit microstructural features into structured inputs that can participate in operating condition classification. Furthermore, material thermal type can be assigned after similarity matching between material microstructure distribution data and a standard material library. For example, material thermal type can include, but is not limited to, high thermal conductivity and low expansion type, medium thermal conductivity and medium expansion type, and low thermal conductivity and high expansion type. The operating condition classification database can be a data set storing the mapping relationship between material thermal types and their associated operating condition adaptation types. It can be used to support subsequent operating condition classification decisions, realizing the association from intrinsic material properties to operating environment types. In this embodiment, the operating condition classification database is dynamically written by the system during the classification process, recording the corresponding records of material thermal types and final operating condition labels.

[0031] Based on the thermal type of the material, the distribution of residual stress and hardness in the material is determined to obtain the physical and chemical measurement results of the material. Residual stress distribution can be described as the spatial distribution of non-uniform stress fields remaining within a material due to processing or heat treatment. It can influence local thermal expansion behavior and heat conduction paths, and is a key factor inducing thermal deformation non-uniformity. In an exemplary embodiment, residual stress distribution can be obtained through non-destructive testing methods such as X-ray stress analysis, neutron diffraction, or ultrasonic measurement. Hardness distribution ratio can be a statistical description of the area or volume ratio of different hardness ranges on a material cross-section. It can be used to indirectly reflect the uniformity of microstructure and the consistency of heat treatment, and correlate spatial variations in thermal conductivity and expansion. Furthermore, the hardness distribution ratio can be calculated by combining microhardness indentation array testing with image analysis. Determining the residual stress distribution and hardness distribution ratio in a material, based on its thermal type, can be achieved by calling the corresponding testing procedure to obtain the spatial distribution of internal stress and hardness for the identified material thermal type. Furthermore, this operation can be achieved through destructive testing of representative samples from the same batch and non-destructive sampling of in-process products, thereby revealing the coupling effect of internal material non-uniformity on thermal behavior. Obtaining the material's physicochemical measurement results can be achieved by integrating the residual stress distribution and hardness distribution ratio data to form a unified description of the material's internal state. Furthermore, this operation can be achieved by constructing multi-dimensional feature vectors and generating weighted comprehensive scores, thus providing an input basis for thermal conductivity non-uniformity analysis. The material physicochemical measurement results can be a quantitative description of the internal non-uniformity of the material formed by combining the proportion of residual stress distribution and hardness distribution. This can be used to reveal the coupling effect of the material's implicit structure on thermal behavior, improving the fidelity of thermophysical modeling. In this embodiment, the material physicochemical measurement results are generated by integrating the measured proportions of residual stress distribution and hardness distribution.

[0032] The non-uniformity of thermal conductivity was quantitatively analyzed based on the physicochemical measurement results of the materials, and the quantitative analysis results were obtained. Thermal conductivity inhomogeneity can refer to the phenomenon where the thermal conductivity of a material varies at different locations in space. This can lead to distortion of the temperature field distribution, resulting in asymmetric thermal deformation and affecting clearance stability. In an exemplary embodiment, thermal conductivity inhomogeneity can be indirectly inferred from material physicochemical measurements or estimated through infrared thermal imaging inversion. Quantitative analysis of thermal conductivity inhomogeneity based on material physicochemical measurements can be performed using empirical formulas, physical models, or data-driven methods to map the physicochemical measurements into thermal conductivity heterogeneity indices. Furthermore, this operation can be achieved by deriving local thermal conductivity based on the empirical relationship between hardness and thermal conductivity, and estimating the thermal conductivity distribution through finite element inversion of the temperature field, thereby upgrading the homogeneous material assumption to a refined model considering spatial heterogeneity. The quantitative analysis results can be output as numerical evaluation indices of thermal conductivity inhomogeneity. In one embodiment, this operation can be achieved by outputting a dispersion index in the form of standard deviation, and outputting the direction and magnitude of the maximum gradient, thereby providing quantitative parameters that can participate in operating condition classification decisions. The quantitative analysis results can be quantitative indicators output after numerically evaluating the non-uniformity of thermal conductivity, which can be used to provide calculable thermophysical heterogeneity parameters for operating condition classification. In this embodiment, the quantitative analysis results are calculated using statistical or physical models based on the material's physicochemical measurement results. For example, the quantitative analysis results may include, but are not limited to, variance coefficient indicators, gradient magnitude indicators, and anisotropy ratio indicators.

[0033] The application scenario type value of the bearing is obtained. Combined with the thermal type of the material, the physical and chemical measurement results of the material, and the quantitative analysis results, the working condition adaptation type of the assembled parts is divided into three types: constant speed type, variable load type, and frequent start-stop type.

[0034] The application scenario type value can be a structured code describing the bearing's expected service environment, such as machine tool spindle, wind turbine gearbox, or rail transportation. This value, combined with the intrinsic properties of the material, can jointly determine the appropriate operating condition, enhancing the engineering relevance of the classification. Furthermore, the application scenario type value can be extracted from product order information, BOM configuration, or design specifications. In an exemplary embodiment, the application scenario type value may include, but is not limited to, precision machine tool scenario values, heavy machinery scenario values, and high-speed transportation scenario values. Obtaining the bearing's application scenario type value can be achieved by reading the code for the bearing's intended application area from the product configuration system or order information. Furthermore, this operation can be implemented by parsing the application field in the BOM and matching the product model with a preset scenario mapping table, thereby introducing prior knowledge of the external service environment and enhancing the engineering applicability of the operating condition classification. By combining material thermal type, material physicochemical measurement results, and quantitative analysis results, the operating condition adaptation types of assembled components are classified into constant speed operating conditions, variable load operating conditions, and frequent start-stop operating conditions. This can be achieved by fusing the intrinsic thermo-mechanical-structural characteristics of the material with the application scenario type value, and determining the most suitable operating condition type through rules or models. Furthermore, this operation can be achieved through end-to-end operating condition prediction using multi-attribute decision tree classification and multi-input neural networks, thereby realizing accurate operating condition classification rooted in the deep coupling of material properties and service environment.

[0035] Taking the selection of high-speed spindle bearings as an example, the automatic selection system of the bearing assembly line in this embodiment can acquire historical processing data of a batch of M50 steel bearings, including the martensitic lath orientation distribution measured by EBSD; by performing similarity calculation with high-temperature bearing steel samples in the standard material library, it identifies the bearing as a high thermal conductivity, low expansion thermal type and writes it into the working condition classification database; then, it measures the surface residual compressive stress depth distribution and core hardness gradient of the batch of materials to obtain the material physicochemical measurement results; based on this, it quantitatively analyzes that the thermal conductivity decreases radially, and the non-uniformity index is 0.18; at the same time, it obtains the application scenario type value as aerospace high-speed spindle; finally, the system integrates the above multi-dimensional features to determine that the bearing is suitable for constant speed working conditions, rather than the variable load type that may be misjudged by the traditional method based solely on speed parameters, thereby ensuring that the subsequent clearance compensation scheme accurately matches its true thermal behavior.

[0036] In one embodiment, the thermal expansion coefficients and thermal conductivity data of the inner ring, outer ring, and rolling elements of the bearing to be assembled are obtained to determine the current thermophysical characteristics of the bearing components, including: The thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled are obtained. The thermal expansion coefficient and thermal conductivity data are processed by an adaptive Kalman filter and a wavelet denoiser to obtain thermophysical data. The adaptive Kalman filter can be a recursive state estimation algorithm that dynamically adjusts the covariance matrix of process noise and observation noise. It can suppress random measurement noise in thermal expansion coefficient and thermal conductivity data while adapting to time-varying statistical characteristics. In this embodiment, the adaptive Kalman filter can update the noise covariance parameter online based on real-time residual feedback to achieve optimal estimation of non-stationary signals. The wavelet denoiser is a processing unit that uses the multi-scale decomposition characteristics of wavelet transform to separate and reconstruct signals from noise. It can be used to remove high-frequency noise while preserving abrupt changes (such as phase transition points and stress release points) in thermal property data. In an exemplary embodiment, the wavelet denoiser can perform wavelet decomposition on the original signal, apply thresholding to the detail coefficients, and then reconstruct the signal. The thermal physics processing data can be thermal expansion coefficient and thermal conductivity data processed jointly by the adaptive Kalman filter and the wavelet denoiser. This data can be used as high-quality input for subsequent historical data association and curve construction, improving the signal-to-noise ratio and feature fidelity. In one specific embodiment, thermophysical processing data can be obtained by first smoothing the trend using an adaptive Kalman filter and then preserving abrupt changes using a wavelet denoiser. Processing thermal expansion coefficient and thermal conductivity data using an adaptive Kalman filter and wavelet denoiser to obtain thermophysical processing data can be achieved by first using an adaptive Kalman filter to perform state estimation and smoothing on the original thermophysical property data, and then using a wavelet denoiser to threshold the detail coefficients to preserve abrupt changes. Furthermore, this operation can be implemented through serial processing of Kalman filtering followed by wavelet denoising or parallel processing of joint optimization of Kalman and wavelet filters, thereby improving the signal-to-noise ratio and dynamic feature fidelity of the original thermophysical property data, providing reliable input for subsequent analysis.

[0037] Based on the thermophysical processing data, the dimensional expansion values ​​and temperature field values ​​of the past preset batches are extracted from the thermophysical database. The dimensional expansion values ​​and temperature field values ​​are plotted using time series diagrams to obtain expansion change curves and temperature rise distribution curves. The thermophysical database can be a structured collection storing historical batches of bearing components' dimensional expansion and temperature field values ​​under different temperature conditions. It can provide historical evolution references for constructing current thermophysical characteristics and support dynamic thermal behavior modeling. In this embodiment, the thermophysical database can be accumulated and structured through test bench testing, digital twin simulation, or service monitoring. The preset batch can be a predefined set of bearing production batches with the same materials and process conditions within the thermophysical database. This can be used to limit the scope of historical data retrieval and ensure process consistency between dimensional expansion and temperature field data. The dimensional expansion value can be the incremental value of the actual size of the bearing component at a specific temperature or time point relative to a reference size. It can be used to construct expansion change curves and quantify the dynamic process of thermal expansion. In a specific embodiment, the dimensional expansion value can be collected in a temperature-controlled experiment using a high-precision displacement sensor or optical measurement equipment.

[0038] Temperature field values ​​can be a set of temperature measurements at key locations of the bearing system at a specific moment, which can be used to construct temperature rise distribution curves and characterize the spatial non-uniformity of heat conduction. In an exemplary embodiment, temperature field values ​​can be acquired through a distributed temperature sensor array or an infrared thermal imaging system. A time series diagram can be a two-dimensional visualization chart with time as the horizontal axis and physical quantities as the vertical axis, which can be used to convert discrete dimensional expansion values ​​and temperature field values ​​into continuous evolution curves. Furthermore, the time series diagram can be plotted after interpolation or spline fitting of data points sorted by timestamps. The expansion change curve can be a continuous function curve describing the change in the dimensions of bearing components with time or temperature, which can be used to reflect the dynamic expansion behavior of materials under thermal loads and reveal nonlinear and time-varying characteristics. In a specific embodiment, the expansion change curve can be visualized based on the dimensional expansion values ​​within a preset batch using a time series diagram. The temperature rise distribution curve can be a continuous function curve describing the evolution of the internal temperature field of the bearing system with time, which can be used to characterize heat conduction paths and heat accumulation effects, and identify local hotspot areas. In this embodiment, the temperature rise distribution curve can be visualized based on the temperature field values ​​within a preset batch using a time series diagram.

[0039] Based on thermophysical processing data, dimensional expansion and temperature field values ​​from past preset batches are extracted from the thermophysical database. This can be achieved by searching the database for historical experimental or simulation data of the same preset batch, using the material grade and process number of the current component as keys. Furthermore, this operation can be implemented through precise batch number matching or nearest-neighbor retrieval based on material and process similarity, thereby establishing a dynamic thermal behavior correlation between the current individual and historical groups, supporting the construction of time-series curves. Dimensional expansion and temperature field values ​​are plotted using time-series graphs to obtain expansion change curves and temperature rise distribution curves. This can be achieved by plotting the time-ordered dimensional expansion and temperature field values ​​as continuous curves. Further, this operation can be achieved by using linear interpolation to create a line graph or using cubic spline fitting to create a smooth curve, thereby transforming static parameters into a dynamic evolution process and revealing the nonlinear and time-varying characteristics of the thermal response.

[0040] For the expansion change curve and the temperature rise distribution curve, the rate of change of the statistical thermal environment is calculated using derivatives, and the thermal critical point is marked to obtain the physical value of the thermal critical point. The thermal environment change rate can be the first derivative of the expansion change curve and the temperature rise distribution curve, representing the rate of change of thermal response per unit time. It can be used to identify acceleration, deceleration, or abrupt changes in thermal behavior, supporting the marking of thermal critical points. In one specific embodiment, the thermal environment change rate can be calculated by numerical differentiation or Savitzky-Golay filtering of the curve. The thermal critical point can be a key time node or temperature node where the thermal environment change rate undergoes a significant inflection or abrupt change. It can be used to identify the boundary conditions for the transition of bearing thermal behavior modes, such as temperature rise inflection points or abrupt changes in expansion rate. In this embodiment, the thermal critical point can be obtained by differentiating the expansion change curve and the temperature rise distribution curve and detecting extreme values ​​or step changes. The physical value of the thermal critical point can be a multidimensional feature vector composed of the size expansion value, temperature value, and their derivative values ​​corresponding to the thermal critical point. It can be used as input for Mahalanobis distance hierarchical clustering to characterize the comprehensive state of key thermal behavior nodes. In an exemplary embodiment, the physical value of the thermal critical point can be obtained by extracting the values ​​of all relevant physical quantities at that moment after marking the thermal critical point.

[0041] The rate of change of the thermal environment is calculated by using derivatives to analyze the expansion curve and the temperature rise distribution curve. This can be achieved by numerically differentiating the two curves separately and calculating their first derivatives as the rate of change of the thermal environment. Furthermore, this operation can be implemented by using the central difference method to calculate the derivative or by using a Savitzky-Golay filter for simultaneous smoothing and differentiation, thereby quantifying the dynamic rate of change of the thermal response and providing a mathematical basis for critical point detection. Marking the thermal critical point and obtaining its physical value can be achieved by detecting extreme points, inflection points, or abrupt changes on the rate of change of the thermal environment curve and combining the physical quantities at the corresponding moments to obtain the thermal critical point's physical value. Further, this operation can be implemented by marking inflection points based on the zero-crossing points of the second derivative or by detecting abrupt changes in variance based on a sliding window, thereby capturing key thermal behavior nodes of the bearing under complex operating conditions and forming a high-dimensional feature representation.

[0042] By using Mahalanobis distance hierarchical clustering to classify the thermal characteristics of the thermal critical point physical values, and combining this with a pre-established thermophysical feature database, the current thermophysical characteristics of bearing components are determined.

[0043] Mahalanobis distance hierarchical clustering can be a distance metric method that considers the covariance structure between variables. It is used to perform hierarchical clustering of multidimensional thermal critical point physical values. It can overcome the distortion problem of Euclidean distance under heterogeneous and correlated variables, and achieve refined classification of implicit thermal characteristics. In one embodiment, Mahalanobis distance hierarchical clustering can calculate Mahalanobis distance based on the sample covariance matrix and construct a clustering tree using an agglomerative hierarchical clustering algorithm. The thermal physical feature database can be a knowledge base that stores the clustered thermal critical point physical values ​​and their corresponding thermal behavior pattern labels. It can be used to provide a classification benchmark for the current thermal physical feature determination of components and support high-fidelity thermal behavior recognition. In this embodiment, the thermal physical feature database can be composed of historical thermal critical point data that has been labeled and stored after Mahalanobis distance hierarchical clustering.

[0044] Classifying the thermal characteristics of thermal critical point physical values ​​using Mahalanobis distance hierarchical clustering involves calculating the Mahalanobis distance between each thermal critical point physical value, constructing a cluster tree using a hierarchical clustering algorithm, and classifying thermal behavior categories. Furthermore, this operation can be further refined by using the sample covariance matrix to calculate the Mahalanobis distance or by using a regularized covariance matrix to prevent singular realizations, thereby achieving a refined classification of the material's implicit thermal characteristics and overcoming distance distortion under multidimensional heterogeneous data. Combining a pre-established thermophysical feature database, the current thermophysical characteristics of bearing components are determined. This can be achieved by matching the current clustering results with the category labels in the thermophysical feature database, assigning a corresponding thermal behavior pattern description to the current component. Further, this operation can be implemented through nearest neighbor category matching or confidence-based multi-label fusion judgment, thereby outputting high-fidelity current thermophysical features and providing accurate input for differentiated clearance compensation.

[0045] Taking the assembly of aero-engine main shaft bearings as an example, the automatic selection system of the bearing assembly line in this embodiment can be as follows: When assembling a new type of aero-engine bearing, the system first obtains the original thermal expansion coefficient data of its inner ring GCr15 material; an adaptive Kalman filter dynamically suppresses sensor drift noise, and a wavelet denoiser retains the small abrupt changes caused by the release of residual stress from heat treatment; then, it retrieves 10 sets of historical temperature control experimental data of the same batch of heat treatment from the thermophysical database, and plots the radial expansion change curve of the inner ring and the temperature rise distribution curve of the raceway; by differentiating the curves, it is found that there is an inflection point in the rate of temperature rise near 120 degrees Celsius, which is marked as the thermal critical point; the expansion amount, temperature and derivative value of this point are extracted to form the physical value of the thermal critical point; through Mahalanobis distance hierarchical clustering, this value is classified into the transient thermal shock category; finally, combined with the thermophysical feature database, the system determines that the component has the latent characteristic of high thermal conductivity but low thermal stability, thereby increasing the initial clearance compensation in subsequent selection to avoid the risk of thermal seizure under high-speed start-stop conditions.

[0046] In one embodiment, based on the thermodynamic characteristics of three operating conditions and in conjunction with a pre-established database of thermal expansion matching rules, the assembly clearance compensation requirements for different operating conditions are determined, including: Based on the thermodynamic characteristics of the three working conditions, the thermal deformation of the three working conditions was obtained from the pre-established thermal expansion matching rule database, and the thermal expansion difference level of the three working conditions was obtained by variance analysis. The thermal deformation amounts for the three operating conditions can be typical deformation amounts of the bearing system due to thermal expansion under constant speed, variable load, and frequent start-stop conditions, and can be used as basic data for quantifying the differences in thermal response under different operating conditions. Furthermore, the thermal deformation amounts for the three operating conditions can be obtained by retrieving pre-stored simulation or measured thermal deformation data from the thermal expansion matching rule database according to the operating condition type. In an exemplary embodiment, there is a statistical correlation between the thermal deformation amounts for the three operating conditions and the thermal expansion difference level. The thermal expansion difference level can be the classification result of the significant differences in thermal expansion behavior of the three types of operating conditions obtained through variance analysis, and can be used to provide a statistical basis for subsequent differential compensation. In a specific embodiment, the thermal expansion difference level can be obtained by performing a one-way variance analysis on the three sets of thermal deformation amounts, determining the significance of the differences based on the F-value and p-value, and classifying them. Based on the thermodynamic characteristics of the three operating conditions, the thermal deformation amounts for the three operating conditions are obtained from the pre-established thermal expansion matching rule database, which can be done by using the operating condition type as an index to query the corresponding typical thermal deformation amount data in the thermal expansion matching rule database. Furthermore, this operation can be achieved by retrieving structured databases using SQL queries or reading from memory cache via key-value pair mapping, thereby obtaining the basic thermal response data for difference analysis. Analysis of variance (ANOVA) is used to obtain the thermal expansion difference levels for the three operating conditions. This can be achieved by performing a one-way ANOVA on three groups of thermal deformation samples and classifying the significance levels of the differences based on the statistical test results. Alternatively, this operation can be achieved by using the ANOVAF test combined with post-hoc multiple comparisons or by employing the non-parametric Kruskal-Wallis test, thereby scientifically identifying the essential differences in thermal responses under different operating conditions and providing a basis for differentiated processing.

[0047] The selection clearance adjustment frequency value, the coordinate value of the fit tolerance zone area, the temperature sensor type value, and the thermal parameter acquisition accuracy value are obtained by acquiring the selection requirement group according to the parameter dimension through the thermal expansion selection rule database. The optional clearance adjustment frequency value can be the recommended update frequency of clearance compensation parameters for a specific operating condition, which can be used to guide the division of the selection time interval. In one specific embodiment, the optional clearance adjustment frequency value is derived from the thermal expansion selection rule database and is based on historical clearance drift rate statistics. The fit tolerance zone coordinate value can be a set of numerical values ​​describing the boundary coordinates of the bearing fit dimension tolerance zone in multi-dimensional geometric space, which can be used as input for fit tolerance zone planning. Further, the fit tolerance zone coordinate value is stored in the thermal expansion selection rule database and associated with the operating condition type and the thermophysical properties of the material. The temperature sensor type value can be a code or label identifying the type of temperature sensor suitable for a specific operating condition, which can be used to support the temperature sensor selection operation. In an exemplary embodiment, the temperature sensor type value is pre-stored in the thermal expansion selection rule database and configured according to the thermal response characteristics of the operating condition and the installation conditions. The thermal parameter acquisition accuracy value can be a requirement index for the required temperature measurement accuracy under a specific operating condition, which can be used to constrain the selection of temperature sensor. Further, the thermal parameter acquisition accuracy value is set by the thermal expansion selection rule database according to the thermal deformation sensitivity.

[0048] Parameter dimension grouping can be a structured organization method that divides relevant parameters into multiple dimensional subsets according to attribute categories, which can facilitate subsequent clustering and rule mapping. In a specific embodiment, parameter dimension grouping is classified according to the semantic attributes of the parameters, such as time dimension, geometric dimension, sensing dimension, and accuracy dimension. The selection requirement group can be a set of multi-dimensional feature vectors composed of selection clearance adjustment frequency values, fit tolerance zone area coordinate values, temperature sensor type values, and thermal parameter acquisition accuracy values, which can be used as input for density peak clustering tools. Furthermore, the selection requirement group integrates multi-source parameters to form structured data units through parameter dimension grouping. Obtaining selection clearance adjustment frequency values, fit tolerance zone area coordinate values, temperature sensor type values, and thermal parameter acquisition accuracy values ​​from the thermal expansion selection rule database can be used to batch read multi-dimensional selection parameters associated with the current operating condition type. Furthermore, this operation can be achieved by pulling multiple parameters at once by field name list or dynamically loading the required fields through parameter templates, thereby constructing a structured selection requirement input and breaking through the limitation of single-size matching. Grouping by parameter dimensions to obtain the selection requirement group can be achieved by reorganizing the acquired multidimensional parameters into a structured feature vector according to preset dimension categories. Furthermore, this operation can be implemented by concatenating the parameters into a vector in a fixed dimensional order or by preserving the dimensional semantic labels in dictionary form, thereby improving the organization of high-dimensional data and the efficiency of subsequent clustering.

[0049] The optional requirements are classified by density peak clustering tool, and the assembly clearance compensation requirements for different working conditions are selected by combining tolerance zone planning, optional time interval division and temperature sensor.

[0050] The density peak clustering tool can be an unsupervised clustering algorithm based on the local density of data points and the minimum distance to higher density points, which can be used to automatically classify high-dimensional fitting requirement groups. In a specific embodiment, the density peak clustering tool selects points with both high density and large relative distance as cluster centers by calculating the local density and relative distance of each data point. The tolerance zone planning can be a process of defining and optimizing the feasible region of the bearing fit dimension tolerance zone in coordinate space based on the thermal expansion difference level under operating conditions and the fitting requirement group, which can be used to determine suitable geometric matching intervals for different operating conditions. Furthermore, the tolerance zone planning combines the thermal deformation prediction results with the assembly clearance target interval to infer the allowable range of the initial fit dimension. The fitting time interval division can be based on the fitting clearance adjustment frequency value and the dynamic characteristics of the operating conditions, dividing the assembly task into execution cycles with different time granularities, which can be used to coordinate the assembly rhythm and the speed of thermal behavior evolution. In an exemplary embodiment, the fitting time interval division sets the optimal fitting update cycle based on the historical temperature rise response speed and clearance decay rate. Temperature sensor selection can be based on the accuracy of thermal parameter acquisition and the type of operating conditions, determining the most suitable sensor configuration from various temperature sensing solutions to ensure the accuracy of thermophysical characteristic sensing. Furthermore, temperature sensor selection is achieved by matching sensor type values ​​with the thermal response frequency of the operating conditions, installation space, and environmental interference levels.

[0051] Classifying the optional requirements using density peak clustering tools involves calculating local density and relative distance for all optional requirements groups, automatically identifying cluster centers, and assigning cluster labels. Furthermore, this operation can be implemented using the original DPeak algorithm or an improved adaptive cutoff distance DPeak variant, enabling unsupervised discovery of the inherent patterns in the optional strategies and avoiding human intervention. By combining tolerance zone planning, optional time interval division, and temperature sensor selection of assembly clearance compensation requirements for different operating conditions, executable differentiated compensation requirements can be generated by integrating clustering results with three types of planning / division / selection operations. Further, this operation can be achieved by executing the three types of operations in parallel and then fusing the output, or by sequentially constraining according to priority to generate the final requirements, thereby enabling the compensation strategy to incorporate dynamic thermal behavior, measurement configuration, and process rhythm, improving engineering feasibility.

[0052] For example, in the scenario of a flexible assembly line for aero-engine main shaft bearings, the automatic matching system of the bearing assembly line in this embodiment can identify that a batch of bearings is suitable for variable load conditions, extract its thermal deformation amount of 12 micrometers from the thermal expansion matching rule database, and simultaneously obtain the deformation amounts of 8 micrometers and 15 micrometers under constant speed and start-stop conditions, respectively; after variance analysis confirms that the differences among the three are highly significant, they are divided into three difference levels; subsequently, the system reads the matching clearance adjustment frequency value (every 2 hours) and the coordinate value of the fit tolerance zone area (inner ring diameter) corresponding to this condition. The system sets the following parameters: lower tolerance limit -3μm, upper limit +1μm; temperature sensor type (thin-film thermocouple); and thermal parameter acquisition accuracy (±0.5℃). These parameters are grouped by dimension to form a selection requirement group. The density peak clustering tool clusters these parameters together with the historical requirement group and automatically assigns them to the high dynamic response cluster. Finally, based on the characteristics of this cluster, the system performs tolerance zone planning (tightening the upper limit to 0μm), selection time interval division (set to 90 minutes), and temperature sensor selection (specifying high-temperature resistant thin-film thermocouples), and outputs customized assembly clearance compensation requirements.

[0053] In one embodiment, after obtaining a thermal expansion compensation matching scheme that conforms to the current operating condition characteristics based on the current thermophysical characteristics and the thermal expansion matching rule database, the method further includes: Obtain the clearance anomaly threshold, thermal offset threshold, optional data storage format, and optional data upload cycle in the thermal expansion compensation optional scheme, and upload the data to the optional log table according to the optional data upload cycle.

[0054] The clearance anomaly threshold can be a critical deviation value used to determine whether the actual clearance deviates from the allowable range during hot operation. It can provide a quantitative basis for anomaly diagnosis and adaptive adjustment, preventing jamming or vibration caused by uncontrolled thermal expansion. In an exemplary embodiment, the clearance anomaly threshold can be generated by the scheme matching module based on the current operating condition characteristics and rule base. The thermal offset threshold can be a critical value used to measure whether the geometric center offset caused by thermal deformation exceeds the allowable limit. It can be used to monitor the alignment stability of the bearing assembly under hot conditions, avoiding additional stress or wear caused by excessive thermal offset. For example, the thermal offset threshold is embedded in the thermal expansion compensation selection scheme and set according to the operating condition type and structural stiffness characteristics. The selection data storage format can be a structured data specification that defines the thermal expansion compensation selection scheme and related parameters stored internally in the system or in logs. It can be used to ensure the consistency and resolvability of selection data under different batches and operating conditions, supporting cross-product data fusion analysis. In this embodiment, the selection data storage format can be used as metadata attributes of the thermal expansion compensation selection scheme, predefined by the system or dynamically bound with the scheme. The optional data upload cycle can be defined as the time interval or trigger condition for submitting optional result data to the optional log table. It can be used to control the data archiving frequency, balance real-time performance and system load, and support time-series traceability management. In one specific embodiment, the optional data upload cycle can be one of the configuration parameters of the thermal expansion compensation optional scheme, set by process strategy or system strategy. For example, the optional data upload cycle can include, but is not limited to, fixed time interval upload, event-driven upload, and batch cumulative upload.

[0055] Obtaining the clearance anomaly threshold, thermal offset threshold, optional data storage format, and optional data upload cycle in the thermal expansion compensation optional scheme can be achieved by extracting these four types of meta-parameters from the generated thermal expansion compensation optional scheme data structure. Furthermore, this operation can be implemented by directly indexing the scheme object attributes through field names and uniformly reading configuration parameters through a predefined metadata interface, thus providing structured input for subsequent data archiving, anomaly monitoring, and strategy feedback. Uploading data to the optional log table according to the optional data upload cycle can be achieved by writing the current optional scheme into the optional log table according to a specified storage format based on a set upload cycle. In an exemplary embodiment, this operation can be implemented by periodically triggering uploads through a scheduled task, checking the cycle condition after each optional configuration, and deciding whether to upload immediately, enabling time-series archiving and traceable management of assembly data. The optional log table can be a structured data table used to persistently store each optional scheme and its execution parameters, enabling traceable archiving of assembly process data and providing a data source for closed-loop optimization, offline review, and predictive maintenance. In this embodiment, the optional log table can be deployed in a database system to receive structured optional data from the closed-loop optimization process.

[0056] Taking the assembly of high-precision bearings in small batches of multiple varieties as an example, the automatic matching system for the bearing assembly line in this embodiment can extract the clearance anomaly threshold of ±5 micrometers, the thermal offset threshold of 10 micrometers, the storage format of JSON, and the upload cycle of each batch after the thermal expansion compensation matching of a certain type of turbine shaft bearing is completed. When all 20 sets of bearings in the batch are assembled, the system automatically packages all matching records in JSON format and writes them to the matching log table. The subsequent data analysis module can identify the trend of thermal offset approaching the threshold in a certain sub-batch based on the log, triggering the rule base to fine-tune the compensation amount of similar material combinations, thereby improving the robustness of subsequent assembly.

[0057] In one embodiment, historical operating temperature rise data under different operating conditions are acquired to obtain the bearing clearance decay and thermal offset periodic variation patterns under the influence of different operating parameters, and the corresponding thermal expansion compensation selection scheme is optimized, including: Acquire historical operating temperature rise data for different operating conditions, and extract the bearing clearance attenuation distribution and thermal offset change rate characteristics from the historical operating temperature rise data.

[0058] The clearance attenuation distribution can be a statistical distribution describing the reduction of bearing clearance over time or operating cycles during service. It can reflect the cumulative impact of thermo-plastic deformation, wear, and other mechanisms on clearance under different operating conditions, providing a quantitative basis for compensation schemes. In an exemplary embodiment, the clearance attenuation distribution can be obtained by time-series modeling or probability density estimation of clearance measurements from historical operating temperature rise data. The thermal offset rate of change feature can be a set of dynamic parameters characterizing the clearance drift speed caused by bearing thermal deformation. It can be used to characterize the speed and stability of thermal response and to identify sensitivity to transient thermal shock. For example, the thermal offset rate of change feature can be extracted by calculating the derivative or difference sequence of clearance change per unit time in historical temperature rise data. Extracting the bearing clearance attenuation distribution and thermal offset rate of change feature from historical operating temperature rise data can involve signal processing and feature engineering of the historical temperature rise data to separate the statistical distribution of clearance attenuation and the rate of change of thermal offset. Furthermore, this operation can be achieved by using kernel density estimation to fit the clearance attenuation distribution, using sliding window difference to calculate the thermal offset rate, or by using wavelet transform to decompose the trend and fluctuation components and extract features respectively. This can quantify the dynamic impact of thermal expansion on assembly clearance and provide structured input for subsequent regularity analysis.

[0059] By analyzing the correlation rules between historical operating temperature rise data and environmental thermal field data, the law of bearing clearance decay and thermal offset periodic variation under the influence of different operating parameters is obtained, and the corresponding thermal expansion compensation selection scheme is optimized.

[0060] The environmental thermal field data can be a dataset of temperature field, cooling medium state, and heat exchange boundary conditions in the external environment of the bearing. This data can be used as external constraints for thermal behavior analysis, improving the accuracy of thermal deformation prediction and pattern discovery. In this embodiment, the environmental thermal field data can originate from sensors at the assembly site, equipment cooling system monitoring, or environmental monitoring records. For example, environmental thermal field data may include, but is not limited to, cooling oil temperature data, ambient air temperature data, and radiation intensity data from nearby heat sources. The association rules can be rule models describing the statistical dependency between historical operating temperature rise data and environmental thermal field data. These rules can be used to reveal the nonlinear mapping relationship between operating parameters and thermal deformation response, supporting pattern identification. In a specific embodiment, the association rules can be learned from multi-source data using association rule mining algorithms (such as Apriori, FP-Growth) or causal inference methods.

[0061] Analyzing the correlation rules between historical operating temperature rise data and environmental thermal field data can be achieved by aligning the two types of data and inputting them into a correlation analysis model to uncover the potential dependencies between operating parameters, environmental conditions, and thermal deformation response. Furthermore, this operation can be achieved by applying multivariate time series correlation rule mining and constructing graph neural networks to model the causal relationships between variables, thereby revealing the mechanism by which thermal behavior is influenced by both internal and external factors and improving the comprehensiveness of pattern identification. Obtaining the periodic variation patterns of bearing clearance decay and thermal offset under the influence of different operating parameters can be achieved by combining correlation rules and feature extraction results to summarize the periodic or trend patterns of thermal behavior under specific combinations of speed, load, and start-stop frequency. Further, this operation can be achieved by using cluster analysis to identify typical operating condition-response pattern pairs and using regression trees to divide the dominant patterns under different parameter intervals, thereby forming reusable knowledge of thermal evolution patterns to support the forward-looking optimization of compensation strategies.

[0062] Taking the long-term service data analysis of wind turbine main shaft bearings as an example, the automatic selection system of the bearing assembly line in this embodiment can acquire three years of historical operating temperature rise data of a certain type of wind turbine bearing under variable load and frequent start-stop conditions, and simultaneously collect environmental thermal field data such as the internal ambient temperature of the tower and the lubricating oil temperature. Through the feature extraction module, it is identified that the clearance exhibits a step-like decay distribution, and the thermal offset shows an accelerated characteristic during low-temperature start-up in winter. Association rule analysis reveals that when the ambient temperature is below -10 degrees Celsius and the start-stop interval is less than 2 hours, the thermal offset rate increases significantly. Based on this pattern, the system automatically adjusts the initial clearance compensation amount under such conditions and adds a low-temperature high-frequency start-stop condition sub-rule to the selection rule library. The thermal clearance stability of similar bearings assembled subsequently is significantly improved in actual operation, verifying the effectiveness of data-driven optimization.

[0063] In one embodiment, the step of obtaining real-time matching parameters based on the optimized thermal expansion compensation matching scheme, comparing the real-time matching parameters with the thermal clearance prediction results, and continuously improving the thermal expansion compensation matching scheme and thermal expansion threshold through data accumulation and feedback to form an adaptive automatic thermal expansion matching closed loop includes: The real-time selection parameters are obtained based on the optimized thermal expansion compensation selection scheme, and the real-time selection parameters are compared with the thermal clearance prediction results to obtain the parameter comparison results. The parameter comparison result can be an output of deviation or consistency assessment between the real-time selected parameters and the hot clearance prediction result. This can be used as a feedback signal to determine the effectiveness of the current selection scheme under hot conditions and trigger subsequent debugging or optimization actions. In this embodiment, the parameter comparison result can generate quantitative or qualitative evaluation indicators by calculating the differences between the two in terms of value, trend, or range. For example, the parameter comparison result may include, but is not limited to, absolute clearance deviation, relative error rate, and hot-state range compliance. Obtaining the parameter comparison result can be achieved by numerically or logically comparing the real-time selected parameters and the hot clearance prediction result, and outputting a deviation assessment. Furthermore, the parameter comparison result can be obtained by calculating the Euclidean distance as a deviation metric and determining whether the predicted clearance falls within the hot-state range allowed by the selection scheme, thereby generating an error signal that can be used for feedback adjustment to support closed-loop optimization decisions.

[0064] Based on the parameter comparison results, combined with the selection clearance adjustment frequency, fit tolerance zone area, temperature sensor type and thermal parameter acquisition accuracy parameters in the thermal expansion compensation matching scheme, as well as the thermal deformation prediction model, the automatic matching device of the bearing assembly line is deployed and debugged. The data is wirelessly uploaded according to the data upload cycle. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

[0065] The optional clearance adjustment frequency can be the time interval or triggering condition for the system to recalculate or update the assembly clearance compensation amount. It can be used to determine the response rhythm of the compensation strategy and balance real-time performance with computational overhead. In an exemplary embodiment, the optional clearance adjustment frequency may include, but is not limited to, fixed-cycle adjustment, event-triggered adjustment, and condition-change driven adjustment. The fit tolerance zone can be the allowable dimensional fit range between the bearing inner ring and shaft, or between the outer ring and housing bore. It can be used to constrain the geometric feasibility boundary of component assembly during automatic fitting, ensuring assembly feasibility. The temperature sensor type can be the category of sensing device used to collect the bearing operating temperature. It can affect the spatial resolution, response speed, and measurement stability of thermal field perception, thereby affecting the quality of thermal parameter input. The thermal parameter acquisition accuracy parameters can be a set of technical indicators describing the measurement accuracy and repeatability of the thermal data acquisition system for temperature, heat flow, and other heat-related data. It can be used to determine the reliability of the input data of the thermal deformation prediction model and affect the confidence of the prediction results. An automated fitting device can be a physical assembly equipment that executes a thermal expansion compensation fitting scheme. It includes sorting, pairing, and assembly execution units, and can be used to translate digital fitting instructions into actual component combinations and assembly actions, achieving closed-loop implementation. In this embodiment, the automated fitting device can configure its robotic arm path, fixture specifications, and detection thresholds according to parameters in the fitting scheme, and perform dynamic calibration in conjunction with a thermal deformation prediction model. Furthermore, the automated fitting device can collaborate with the thermal deformation prediction model: the model provides expected thermal behavior, and the device adjusts initial assembly parameters accordingly; the automated fitting device can also collaborate with a wireless upload mechanism: the device's operating status and results are transmitted back through this mechanism, supporting feedback learning. For example, the automated fitting device may include, but is not limited to, a robotic sorting and assembly unit, a track-type automated pairing table, and a vision-guided precision pressing unit.

[0066] Based on parameter comparison results, and combined with the selection clearance adjustment frequency, tolerance zone, temperature sensor type, and thermal parameter acquisition accuracy parameters in the thermal expansion compensation selection scheme, as well as the thermal deformation prediction model, the deployment and debugging of the automatic selection device for the bearing assembly line can be achieved by comprehensively configuring multi-dimensional parameters and model outputs, and dynamically setting or calibrating the hardware parameters, control logic, or detection thresholds of the automatic selection device. Furthermore, this operation can be achieved by adjusting the preload of the robotic arm online to compensate for thermal expansion deviations and recalibrating the thermal parameter input filter according to the temperature sensor type, thereby ensuring that the physical execution unit and the digital model remain synchronized between thermal behavior prediction and actual assembly, improving selection consistency. The data upload cycle can be the time interval at which the system uploads operating data to the edge or cloud platform, which can be used to control the data return frequency, affecting the timeliness of feedback learning and network load. The wireless upload mechanism can be a data transmission method that transmits assembly line operating data to a remote system via a wireless communication protocol, which can be used to achieve data connectivity between the physical assembly line and the digital twin platform, supporting remote monitoring and centralized optimization. In this embodiment, the wireless upload mechanism can establish a secure and reliable data channel based on industrial IoT protocols (such as MQTT, OPCUA over WiFi / 5G). Wireless uploads are performed according to a data upload cycle, which can be achieved by packaging and uploading selected parameters, comparison results, and operating status to the data platform at preset time intervals via a wireless communication module. Furthermore, this operation can be implemented by periodically publishing JSON format data packets using the MQTT protocol and triggering immediate uploads when abnormal deviations are detected, thereby enabling continuous feedback of operational data and providing a foundation for long-term learning and rule base updates.

[0067] Taking a flexible assembly line for wind turbine main shaft bearings as an example, the automatic matching system of the bearing assembly line in this embodiment can identify a batch of wind turbine bearings as operating under frequent start-stop conditions. The system generates a thermal expansion compensation matching scheme and outputs real-time matching parameters. The model predicts that its hot clearance should be 12 micrometers, but after actual assembly, the deviation is found to be 4 micrometers, forming a parameter comparison result. The system then combines the daily matching clearance adjustment frequency, H7 / g6 fit tolerance zone, PT100 temperature sensor type, and ±0.5℃ thermal parameter acquisition accuracy set in the scheme to fine-tighten the preload force of the rolling element sorting fixture and the inner ring pressing depth of the automatic matching device. At the same time, according to the data upload cycle of 2 hours, the comparison result and adjustment log are uploaded to the edge server through the 5G wireless upload mechanism. The accumulated data triggers the rule base update, lowering the thermal expansion threshold of the same operating condition by 3 micrometers, so that subsequent assembly can more accurately adapt to the actual thermal behavior.

[0068] In one embodiment, a thermal deformation prediction model is established based on the bearing's combined heat source intensity, heat dissipation conditions, and structural dimensions. The thermal deformation prediction model is then used to obtain thermal clearance prediction results under different inputs, including: Construct a finite element digital twin that includes bearing geometric parameters, material physical properties, and lubricating oil rheological characteristics.

[0069] The finite element digital twin can be a high-fidelity finite element simulation model that integrates bearing geometric parameters, material physical properties, and lubricating oil rheological characteristics. It can serve as a carrier for thermal deformation prediction models, enabling refined digital mapping of the thermo-mechanical coupling behavior of bearings under complex operating conditions. In this embodiment, the finite element digital twin can be imported from a CAD geometric model, assigning material constitutive relations and lubrication boundary conditions to construct a multi-physics coupled finite element mesh. The bearing geometric parameters can be structural data describing the dimensions and fit relationships of components such as the inner ring, outer ring, and rolling elements. These parameters form the geometric basis of the finite element digital twin, influencing the heat conduction path and deformation stiffness. Material physical properties can include intrinsic material parameters related to thermo-mechanical response, such as the coefficient of thermal expansion, thermal conductivity, and elastic modulus. These properties can determine the deformation characteristics of components under temperature fields, improving the physical realism of the digital twin. For example, material physical properties can be obtained from material handbooks, experimental tests, or process databases. Lubricating oil rheological properties describe the viscosity changes of lubricating oil under different shear rates and temperatures. These properties can influence the intensity of frictional heat generation and the thickness of the lubricating film, thereby altering the heat source distribution and heat dissipation efficiency. In a specific embodiment, the lubricating oil rheological properties can be obtained through rheometer testing or from the supplier's technical parameter sheet.

[0070] Constructing a finite element digital twin that includes bearing geometric parameters, material physical properties, and lubricating oil rheological characteristics can integrate geometric models, material data, and lubrication boundary conditions to establish a multi-physics coupled finite element model. Furthermore, this operation can be achieved by manually modeling and assigning parameters in commercial CAE software, or by automatically importing a fully parameterized model into a PLM system via API. This overcomes the limitations of traditional models that neglect lubrication and material details, thus improving the fidelity of the digital twin.

[0071] Based on fluid dynamics lubrication theory and heat conduction equations, the frictional heat generation and heat transfer path of bearings under different speeds and loads are simulated to generate a three-dimensional temperature field distribution cloud map inside the bearing.

[0072] The hydrodynamic lubrication theory serves as a theoretical framework describing the formation and load-bearing mechanism of the lubricating oil film between relatively moving surfaces. It can be used to calculate the frictional shear stress in the contact area, serving as a key input for frictional heat generation. In one specific embodiment, the hydrodynamic lubrication theory can solve for the oil film pressure and velocity field using the Reynolds equation. The heat conduction equation can be a partial differential equation describing the heat transfer process in a solid medium, used to simulate the establishment and evolution of the internal temperature field of the bearing. For example, the heat conduction equation can be discretized in the finite element method to solve for Fourier's law of heat conduction. Frictional heat generation can be the heat generated by the relative sliding between the rolling elements and the raceway and the shearing of the lubricating oil. It can be used as the main heat source term in the thermal deformation prediction model, driving the temperature field upwards. In an exemplary embodiment, frictional heat generation can be calculated using the hydrodynamic lubrication theory to obtain shear power and convert it into a volumetric heat source. The heat transfer path can be the path from the heat source (contact area) through the rolling elements, inner and outer rings to the bearing housing or the environment. It can be used to determine the spatial distribution of the temperature field, affecting the non-uniformity of thermal deformation. Furthermore, the heat transfer path can include, but is not limited to, solid conduction paths, convective heat transfer paths, and radiative heat dissipation paths. A three-dimensional temperature field distribution cloud map can be a visualization of the internal temperature distribution of a bearing, generated through simulations of heat conduction and frictional heat generation. It can reflect the non-uniform distribution of heat within the bearing structure, providing accurate input for thermal deformation calculations. In this embodiment, the three-dimensional temperature field distribution cloud map can be generated by a finite element solver outputting nodal temperature data after steady-state or transient thermal analysis, and rendered in the form of a color gradient.

[0073] Based on fluid dynamics lubrication theory and heat conduction equations, this method simulates the frictional heat generation and heat transfer paths of bearings under different speeds and loads. This can be achieved by simultaneously solving the Reynolds equations and the heat conduction equations to calculate dynamic heat sources and temperature evolution. Furthermore, this operation can be implemented using a sequential coupling method to calculate lubrication first and then heat, or a fully coupled method to simultaneously solve the fluid-thermal-solid fields. This allows for precise characterization of the heat source generation mechanism and heat diffusion process, supporting high-precision temperature field prediction. A three-dimensional temperature field distribution cloud map of the bearing's interior can be generated, visualizing the finite element thermal analysis results as spatial node temperatures. Further, this operation can be achieved by outputting the data in VTK format for post-processing software rendering or by directly embedding it into a digital twin platform for real-time display. This provides a clear view of temperature non-uniformity and offers high-resolution input for subsequent thermal deformation calculations.

[0074] Based on the three-dimensional temperature field distribution cloud map, calculate the non-uniform thermal deformation of the inner ring, outer ring, and rolling elements under thermal equilibrium.

[0075] The thermal equilibrium state can be considered a steady-state thermal condition where the bearing's heat generation and dissipation reach a dynamic balance during continuous operation. This state can serve as a benchmark for calculating non-uniform thermal deformation, ensuring that the predicted results correspond to typical service conditions. Non-uniform thermal deformation refers to the spatial position and shape changes of various bearing components due to the uneven three-dimensional temperature field. This can be used to more realistically describe actual thermal expansion behavior and avoid prediction errors caused by the uniform expansion assumption. In this embodiment, the non-uniform thermal deformation can be calculated by applying the three-dimensional temperature field as a thermal load to the structural mechanics model and solving the thermoelastic equation to obtain the displacement field. Based on the three-dimensional temperature field distribution cloud map, the non-uniform thermal deformation of the inner ring, outer ring, and rolling elements in the thermal equilibrium state can be calculated by applying the temperature field as a thermal load to the structural mechanics model and solving the thermoelastic displacement field. Furthermore, this operation can be achieved by using a linear thermal expansion constitutive model for small deformation analysis and a nonlinear material model to handle large temperature difference conditions, thereby overcoming the bias of the uniform expansion assumption and realistically reflecting the thermal warpage and local expansion of components.

[0076] By vector superimposing the non-uniform thermal deformation amount with the initial assembly clearance, the thermal deformation prediction model is obtained as the thermal clearance prediction result under different inputs.

[0077] The initial assembly clearance can be the internal clearance of the bearing when it is assembled at room temperature under no-load conditions. It can be used as a reference value for vector superposition and, together with the amount of thermal deformation, determines the hot clearance. In an exemplary embodiment, the initial assembly clearance can be output by the scheme matching module or obtained by actual assembly measurement. Vector superposition can be a mathematical operation that geometrically synthesizes the displacement components in each direction caused by non-uniform thermal deformation with the initial clearance. It can be used to accurately reflect the influence of thermal deformation on the actual clearance, rather than a simple algebraic addition. For example, vector superposition can be based on displacement field projection superposition based on coordinate transformation or clearance recalculation based on contact geometry reconstruction. Vector superposition of non-uniform thermal deformation with the initial assembly clearance can be used to recalculate the minimum clearance between the rolling element and the raceway based on the deformed geometric configuration. Furthermore, this operation can be achieved by calculating the minimum distance through a contact detection algorithm and estimating the effective clearance through analytical geometric projection, thereby achieving accurate prediction of the hot clearance, rather than empirical estimation. Obtaining the hot clearance prediction results of the thermal deformation prediction model under different inputs can be achieved by running the above process on different speed and load combinations and outputting the corresponding hot clearance values. Furthermore, this operation can be achieved by generating predictive surfaces through batch parametric scanning and quickly responding by calling proxy models in real time as needed, thus providing a highly reliable predictive benchmark for differentiated compensation and closed-loop optimization.

[0078] Taking the prediction of hot clearance of main bearings for aero-engines as an example, the automatic matching system for the bearing assembly line in this embodiment can be designed for a specific type of aero-engine bearing. The system constructs a finite element digital twin that includes the precision raceway geometry, the material properties of M50 steel, and the rheological characteristics of high-temperature synthetic lubricating oil. Under simulated heavy-load conditions at 30,000 rpm, the system calculates the high-shear heat source in the raceway contact area based on hydrodynamic lubrication theory, and solves the heat conduction equation by combining forced oil-gas cooling boundary conditions, generating a three-dimensional temperature field cloud map showing that the inner ring temperature is significantly higher than that of the outer ring. Based on this, it is calculated that the radial expansion of the inner ring is greater than that of the outer ring, resulting in clearance contraction. By vector superposition of the initial clearance of 12 micrometers and the non-uniform deformation, the predicted hot clearance is 4.5 micrometers, which is within the critical safety range. This result is used by the closed-loop optimization module to verify whether the current matching scheme needs to increase the initial clearance, thereby avoiding the risk of thermal seizure during high-altitude operation.

[0079] In one embodiment, after obtaining real-time matching parameters based on the optimized thermal expansion compensation matching scheme and comparing the real-time matching parameters with the hot clearance prediction results, the method further includes: Set the allowable deviation range for hot clearance.

[0080] The allowable deviation range of hot clearance can be the upper and lower limits of acceptable hot clearance under specific operating conditions, used to determine whether the predicted clearance meets operational stability requirements. Furthermore, the allowable deviation range of hot clearance can be used as a trigger threshold and optimization target boundary for an error correction mechanism, ensuring that the hot clearance remains within a safe and stable range. In an exemplary embodiment, the allowable deviation range of hot clearance can be set based on bearing type, operating condition severity, and historical failure data, and dynamically adjusted by a closed-loop optimization module.

[0081] If the comparison results show that the difference exceeds the allowable deviation range of the hot clearance, the error correction mechanism will be triggered.

[0082] The error correction mechanism can be an automatic tolerance re-optimization process initiated when the predicted thermal clearance of the real-time selected parameters exceeds the allowable deviation range. Furthermore, the error correction mechanism can be used to compensate for deficiencies in the initial selection scheme by generating corrected parameters that meet the thermal performance target through virtual iteration. In a specific embodiment, the error correction mechanism can be activated by the closed-loop optimization module after detecting an excessive deviation, and the optimization algorithm in the digital twin can be invoked to perform optimization. This operation can be an automatic activation of the optimization process when the predicted thermal clearance of the real-time selected parameters exceeds a set range. Further, this operation can be achieved by immediately interrupting the current assembly process and initiating correction, or by marking the current scheme as to be optimized and queuing it for processing, thereby enabling autonomous intervention in cases of exceeding tolerances and preventing unqualified assembly schemes from being executed.

[0083] In a digital twin, the fit tolerances of selected components are iteratively optimized using genetic algorithms or particle swarm optimization algorithms.

[0084] The digital twin can be a virtual simulation environment that integrates a thermal deformation prediction model with the mapping relationship of assembly parameters, supporting rapid evaluation and optimization of fit tolerances. Furthermore, the digital twin can be used as a computational platform for error correction mechanisms, completing multiple rounds of parameter iteration and verification without interfering with the physical assembly line. In an exemplary embodiment, the digital twin can be constructed by extending the thermal deformation prediction model based on the model prediction module and embedding a tolerance-clearance mapping function and an optimization solver. The fit tolerances of the selected components can be the allowable manufacturing tolerance combinations between the bearing inner ring, outer ring, and rolling elements in terms of dimensional matching, directly affecting the initial clearance and hot clearance. Furthermore, the fit tolerances of the selected components can be used as operational variables for iterative optimization, adjusting their values ​​to ensure that the hot clearance prediction results fall within the allowable deviation range. In one embodiment, the fit tolerances of the selected components can be provided by historical processing and inspection data, serving as the search boundary for the optimization algorithm. The genetic algorithm can be a metaheuristic optimization algorithm simulating the biological evolution process, searching for the optimal solution in the solution space through selection, crossover, and mutation operations. Furthermore, genetic algorithms can be used to efficiently search for parameter combinations that meet thermal clearance requirements in high-dimensional, nonlinear, and multi-constrained fit tolerance design spaces. Particle swarm optimization (PSO) is a swarm intelligence-based optimization algorithm that finds the optimal solution by simulating the velocity and position updates of particles in the solution space. Furthermore, PSO can be applied to continuous variable optimization problems, quickly converging to feasible tolerance combinations within the allowable deviation range of the thermal clearance. This operation can use fit tolerance as the decision variable and the conformity between the predicted thermal clearance and the allowable deviation range as the objective function, running an optimization algorithm. Furthermore, this operation can be achieved by using a genetic algorithm for global exploration or a particle swarm optimization algorithm for local fast convergence, thus efficiently searching for tolerance combinations that meet thermal performance targets in complex nonlinear spaces, overcoming the limitations of human experience.

[0085] This continues until the calculated hot clearance prediction falls within the allowable deviation range of the hot clearance.

[0086] This operation can be a continuous iterative optimization until the predicted hot clearance value obtained from a certain calculation meets the preset boundary conditions. Furthermore, this operation can be prevented from infinite looping by setting a maximum number of iterations or by introducing an early stopping mechanism to terminate the operation early after the accuracy requirement is met, thereby ensuring the operational stability of the final selected scheme under hot conditions.

[0087] The final fit tolerance parameters obtained through iterative optimization are used as the corrected real-time fitting parameters, and the corrected real-time fitting parameters are sent to the assembly execution unit.

[0088] The corrected real-time fitting parameters can be the final combination of tolerance parameters that meets the allowable deviation range of thermal clearance, output after iterative optimization by the digital twin. Furthermore, the corrected real-time fitting parameters can be used to replace the initial fitting parameters as the basis for high-precision assembly, ensuring that thermal performance meets standards. In a specific embodiment, the corrected real-time fitting parameters can be generated by an error correction mechanism optimized within the digital twin and verified for validity. This operation can involve extracting the optimal tolerance combination after convergence of the optimization algorithm and encapsulating it into an executable fitting instruction. Further, this operation can be achieved by directly replacing the original real-time fitting parameters or generating incremental correction instructions and superimposing them onto the original parameters, thereby generating high-precision assembly parameters that meet thermal performance standards. The assembly execution unit can be an automated device or control system that receives the fitting parameters and completes the physical assembly operation. Further, the assembly execution unit can be used to transform the virtual optimization results into actual assembly actions, achieving a precise mapping from the digital twin to the physical world. In an exemplary embodiment, the assembly execution unit can receive the corrected real-time fitting parameters through an industrial communication protocol and drive a robotic arm or sorting mechanism to perform matching. This operation can be performed by transmitting optimized parameters to the assembly equipment control system via an industrial communication interface. Furthermore, this operation can be implemented in real-time via the OPCUA protocol or asynchronously pushed through the MES system task queue, thereby enabling the precise implementation of virtual optimization results into physical assembly.

[0089] Taking the high-precision assembly of main shaft bearings for high-end power equipment as an example, the automatic selection system of the bearing assembly line in this embodiment can be as follows: After the system initially generates a selection scheme for a certain high-temperature alloy bearing, the model prediction module calculates that its hot clearance under the takeoff-cruise cycle is 12 micrometers, while the allowable deviation range of the hot clearance is set to 6 to 10 micrometers, which is judged to be out of tolerance; the error correction mechanism is then triggered, and the particle swarm algorithm is used to iteratively optimize the inner and outer ring diameter tolerances in the digital twin; after 18 simulation iterations, the algorithm finds a set of fit tolerance combinations, which makes the predicted hot clearance converge to 9.3 micrometers, falling within the allowable range; the corrected real-time selection parameters are sent to the six-axis assembly robot, which selects the inner and outer rings with the corresponding tolerance levels to complete the press-fit, ensuring that the engine maintains a stable clearance under extreme thermal cycles and avoiding the risk of thermal seizure.

[0090] In one embodiment, the thermal expansion compensation selection scheme and thermal expansion threshold are continuously improved through data accumulation and feedback, including: Collect measured temperature data and vibration spectrum data of the bearing after assembly during actual trial operation.

[0091] The assembled bearing can be a bearing assembly that has completed component matching and is in the trial operation verification stage. It can be used as a physical data source to provide thermal and vibration response data under real working conditions. In an exemplary embodiment, the assembled bearing may include, but is not limited to, bench test bearings, integrated machine testing bearings, and pre-shipment verification bearings. The actual trial operation process can be a short-term operation test phase conducted under simulated or real working conditions after the bearing is assembled. It can be used to expose thermal expansion and dynamic response characteristics, providing an effective data window for model calibration. For example, the actual trial operation process can adopt constant temperature bench test operation, variable load simulation test operation, and start-stop cycle test operation. The measured temperature data can be a record of the temperature change of key parts of the bearing over time, collected by temperature sensors during the trial operation. It can be used to reflect the real heat source intensity and heat dissipation efficiency, and to calibrate thermal boundary conditions. Furthermore, the measured temperature data can be collected in real time through embedded thermocouples, infrared thermometry, or wireless temperature tags. The vibration spectrum data can be the frequency domain vibration characteristics collected by accelerometers and obtained by Fourier transform during the trial operation. It can be used to indirectly reflect the thermal clearance state and assist in verifying the accuracy of the thermal deformation model. In one specific embodiment, vibration spectrum data is obtained by acquiring time-domain signals through vibration sensors and then performing spectrum analysis. The measured temperature and vibration spectrum data of the assembled bearing during actual trial operation can be acquired synchronously via a sensor network, and then preprocessed to form structured data. Furthermore, this operation can be achieved through wired multi-point temperature acquisition combined with a vibration analyzer and the deployment of wireless sensor nodes to achieve interference-free monitoring. This establishes a data channel between the physical bearing and its digital twin, providing realistic feedback for model calibration.

[0092] The measured temperature data and vibration spectrum data are input into the digital twin, and residual analysis is performed between them and the simulation results of the thermal deformation prediction model.

[0093] The digital twin can be a virtual simulation system that integrates a thermal deformation prediction model, real-time sensor data, and a mapping relationship between physical bearing behavior. It can serve as a unified platform for comparing measured data and simulation results, supporting residual analysis and model correction. In this embodiment, the digital twin is built based on the thermal deformation prediction model and achieves state synchronization by accessing measured temperature and vibration spectrum data through an interface. Furthermore, the digital twin can include, but is not limited to, geometric twin layers, thermal behavior twin layers, and dynamic response twin layers. The simulation results of the thermal deformation prediction model can be numerical results of the thermal clearance, temperature field, or deformation field output by the model under given input conditions. These can be used as a benchmark for comparison with measured data to evaluate model fidelity. In an exemplary embodiment, the simulation results of the thermal deformation prediction model are integrated with the digital twin as its core computing engine; its output participates in residual analysis. Residual analysis can be a statistical or numerical method for quantitatively evaluating the difference between measured data and simulation results. It can be used to determine whether the thermal deformation prediction model meets preset accuracy requirements and trigger a correction mechanism. For example, residual analysis is achieved by calculating the mean square error, maximum deviation, or correlation coefficient of temperature or equivalent clearance-related indicators. A preset accuracy can be a threshold standard for determining whether the residual is acceptable, and can be used as a decision-making basis for whether to initiate model backpropagation. In a specific embodiment, the preset accuracy may include, but is not limited to, temperature residual tolerance, vibration feature matching tolerance, and hot clearance deviation tolerance. Inputting measured temperature data and vibration spectrum data into the digital twin can be achieved by mapping the measured data to the corresponding state variables or observation points of the digital twin through a data interface. Furthermore, this operation can be achieved through real-time synchronization via the OPCUA protocol or batch import of CSV files for offline comparison, thereby achieving state alignment between the physical world and the virtual model and supporting consistency assessment. Performing residual analysis with the simulation results of the thermal deformation prediction model can be achieved by calculating the differences between the measured data and the simulation output in key indicators and assessing whether they exceed the allowable range. Furthermore, this operation can be achieved by judging based on temperature peak deviation and evaluating hot clearance error based on vibration dominant frequency offset, thereby quantifying model prediction deviation and identifying insufficient thermal boundary modeling or parameter mismatch problems.

[0094] If the residual analysis error is greater than the preset accuracy, the thermal boundary conditions and convective heat transfer coefficient of the thermal deformation prediction model are corrected in reverse using measured temperature data and vibration spectrum data.

[0095] Thermal boundary conditions, which describe the mode and intensity of heat exchange between the bearing system and the external environment, are physical constraints that determine how heat is conducted from the bearing to the surrounding structure or medium, and are a key input for predicting thermal deformation. In an exemplary embodiment, the thermal boundary conditions and the convective heat transfer coefficient together constitute a specific parameterized expression of the heat dissipation conditions, which is affected by the actual installation environment. The convective heat transfer coefficient can be a physical parameter characterizing the convective heat transfer efficiency between the fluid (such as air or lubricating oil) and the bearing surface, and can be used to directly affect the local heat dissipation rate, thereby affecting the temperature gradient and thermal deformation distribution. Furthermore, the convective heat transfer coefficient, as a core component of the thermal boundary conditions, together with the heat source intensity, determines the steady-state temperature field. Inverse correction of the thermal boundary conditions and convective heat transfer coefficient of the thermal deformation prediction model using measured temperature data and vibration spectrum data can be achieved by using measured data as observations and adjusting the thermal boundary conditions and convective heat transfer coefficient through optimization algorithms to make the simulation results approximate the measured values. Furthermore, this operation can be achieved by using the least squares method to fit the optimal heat transfer coefficient and using a Bayesian inversion framework to fuse prior knowledge with measured data, thereby dynamically calibrating key implicit parameters that affect heat dissipation efficiency and improving model fidelity.

[0096] Based on the revised thermal deformation prediction model, update the assembly clearance compensation requirements in the thermal expansion matching rule database.

[0097] The modified thermal deformation prediction model can be an updated thermal boundary condition and convective heat transfer coefficient model calibrated based on measured data feedback. This model can improve the accuracy of predicting the thermal behavior under real operating conditions, supporting more reliable fitting decisions. In one specific embodiment, the modified thermal deformation prediction model outputs updated assembly clearance compensation requirements to the thermal expansion fitting rule database. The assembly clearance compensation requirements in the thermal expansion fitting rule database can be clearance compensation entries stored in the rule base, corresponding to specific operating conditions and thermophysical characteristics, and can be used to directly guide the initial clearance setting of subsequent assembly tasks. For example, the assembly clearance compensation requirements in the thermal expansion fitting rule database are updated driven by the modified thermal deformation prediction model, reflecting the impact of model evolution on the rules. Updating the assembly clearance compensation requirements in the thermal expansion fitting rule database based on the modified thermal deformation prediction model can be achieved by running the modified model to generate a thermal clearance prediction under new operating conditions, and adjusting the compensation amount of the corresponding entry in the rule base accordingly. Furthermore, this operation can be achieved by automatically overwriting the original rule entries, generating new rules with version tags, and setting priorities, thereby enabling the selected rules to continuously reflect the real thermal-mechanical coupling response and achieve the evolution of personalized adaptation capabilities.

[0098] Taking the factory verification of aero-engine main shaft bearings as an example, the automatic matching system of the bearing assembly line in this embodiment can be used to simulate flight cycle testing after a high-precision angular contact ball bearing is assembled and placed on the test bench. The system simultaneously collects the measured temperature data at the inner raceway and the vibration spectrum of the housing. After receiving this data, the digital twin compares it with the steady-state temperature of 85 degrees Celsius and the corresponding clearance output by the original thermal deformation prediction model. It finds that the measured temperature reaches 92 degrees Celsius and abnormal high-frequency components appear in the vibration spectrum. Residual analysis shows that the error exceeds the preset accuracy by 5%. The system then initiates reverse correction, adjusting the thermal boundary conditions of the contact surface between the bearing housing and the outer race and the convective heat transfer coefficient of the lubricating oil, so that the simulated temperature converges to 91.5 degrees Celsius. Based on the corrected model, the system automatically updates the assembly clearance compensation requirement for this type of high-speed variable load condition from +6 micrometers to +9 micrometers and writes it into the thermal expansion matching rule database. Subsequent assembly of bearings of the same model will use the new compensation value, effectively avoiding the risk of thermal seizure.

[0099] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. An automatic selection system for a bearing assembly line, characterized in that, The system includes: The working condition classification module is used to obtain historical processing and inspection data of the parts to be selected in the bearing assembly line. Based on the historical processing and inspection data, the working condition adaptation type of the assembled parts is divided into three types: constant speed working condition, variable load working condition, and frequent start-stop working condition. The feature extraction module is used to obtain the thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled, so as to obtain the current thermophysical characteristics of the bearing components. The requirement determination module is used to determine the assembly clearance compensation requirements for different operating conditions based on the thermodynamic characteristics of the three operating conditions and in conjunction with a pre-established thermal expansion matching rule database. The scheme generation module is used to obtain a thermal expansion compensation scheme that conforms to the current operating conditions based on the current thermophysical characteristics and the thermal expansion selection rule database. The regularity optimization module is used to acquire historical operating temperature rise data under different operating conditions, obtain the change law of bearing clearance decay and thermal offset cycle under the influence of different operating parameters, and optimize the corresponding thermal expansion compensation selection scheme. The model prediction module is used to establish a thermal deformation prediction model for bearings based on the intensity of the heat source, heat dissipation conditions, and structural dimensions, and to obtain the thermal clearance prediction results of the thermal deformation prediction model under different inputs. The closed-loop feedback module is used to obtain real-time matching parameters based on the optimized thermal expansion compensation matching scheme, and compare the real-time matching parameters with the thermal clearance prediction results. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

2. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The process involves acquiring historical processing and inspection data of the components to be assembled on the bearing assembly line. Based on this data, the operating condition adaptation types of the assembled components are categorized into three types: constant speed operating condition, variable load operating condition, and frequent start-stop operating condition. Obtain historical processing and testing data of the components to be selected in the bearing assembly line, including material microstructure distribution data; The similarity between the material's microstructure distribution data and a pre-established standard material library is calculated to obtain the material's thermal type and write it into the working condition classification database. Based on the thermal type of the material, the proportion of residual stress and hardness distribution in the material is determined to obtain the physicochemical measurement results of the material. The non-uniformity of thermal conductivity was quantitatively analyzed based on the physicochemical measurement results of the material, and the quantitative analysis results were obtained. The application scenario type value of the bearing is obtained. Combined with the material thermal type, material physicochemical measurement results and quantitative analysis results, the working condition adaptation type of the assembled parts is divided into three types: constant speed type, variable load type and frequent start-stop type.

3. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The process of acquiring the thermal expansion coefficients and thermal conductivity data of the inner ring, outer ring, and rolling elements of the bearing to be assembled, to obtain the current thermophysical characteristics of the bearing components, includes: The thermal expansion coefficient and thermal conductivity data of the inner ring, outer ring and rolling elements of the bearing to be assembled are obtained. The thermal expansion coefficient and thermal conductivity data are processed by an adaptive Kalman filter and a wavelet denoiser to obtain thermophysical processing data. Based on the thermophysical processing data, the size expansion values ​​and temperature field values ​​of the past preset batches are extracted from the thermophysical database. The size expansion values ​​and temperature field values ​​are plotted using a time series diagram to obtain the expansion change curve and the temperature rise distribution curve. For the expansion change curve and the temperature rise distribution curve, the rate of change of the statistical thermal environment is calculated using derivatives, and the thermal critical point is marked to obtain the physical value of the thermal critical point. The thermal characteristics of the physical values ​​of the thermal critical point are classified by Mahalanobis distance hierarchical clustering, and the current thermophysical characteristics of the bearing components are determined by combining them with a pre-established thermophysical feature database.

4. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, Based on the thermodynamic characteristics of the three operating conditions and in conjunction with a pre-established database of thermal expansion matching rules, the assembly clearance compensation requirements for different operating conditions are determined, including: Based on the thermodynamic characteristics of the three working conditions, the thermal deformation of the three working conditions is obtained from the pre-established thermal expansion matching rule database, and the thermal expansion difference level of the three working conditions is obtained by variance analysis. The selection clearance adjustment frequency value, the coordinate value of the fit tolerance zone, the temperature sensor type value, and the thermal parameter acquisition accuracy value are obtained through the thermal expansion selection rule database. The selection requirement group is obtained by grouping according to the parameter dimension. The optional requirements are classified using a density peak clustering tool, and the assembly clearance compensation requirements for different operating conditions are selected by combining tolerance zone planning, optional time interval division, and temperature sensor.

5. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, After obtaining a thermal expansion compensation matching scheme that conforms to the current operating condition characteristics based on the current thermophysical characteristics and the thermal expansion matching rule database, the process further includes: Obtain the clearance anomaly threshold, thermal offset threshold, optional data storage format, and optional data upload cycle in the thermal expansion compensation optional scheme, and upload the data to the optional log table according to the optional data upload cycle.

6. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The process involves acquiring historical operating temperature rise data under different working conditions to obtain the bearing clearance decay and thermal offset periodic variation patterns under the influence of different operating parameters, and optimizing the corresponding thermal expansion compensation selection scheme, including: Acquire historical operating temperature rise data for different operating conditions, and extract the bearing clearance attenuation distribution and thermal offset change rate characteristics from the historical operating temperature rise data; By analyzing the correlation rules between historical operating temperature rise data and environmental thermal field data, the clearance decay and thermal offset periodic variation law of the bearing under the influence of different operating parameters are obtained, and the corresponding thermal expansion compensation selection scheme is optimized.

7. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The process of obtaining real-time matching parameters based on the optimized thermal expansion compensation matching scheme, comparing the real-time matching parameters with the thermal clearance prediction results, and continuously improving the thermal expansion compensation matching scheme and thermal expansion threshold through data accumulation and feedback to form an adaptive automatic thermal expansion matching closed loop includes: The real-time selection parameters are obtained based on the optimized thermal expansion compensation selection scheme, and the real-time selection parameters are compared with the thermal clearance prediction results to obtain the parameter comparison results. Based on the parameter comparison results, combined with the selection clearance adjustment frequency, fit tolerance zone area, temperature sensor type and thermal parameter acquisition accuracy parameters in the thermal expansion compensation matching scheme, as well as the thermal deformation prediction model, the automatic matching device of the bearing assembly line is deployed and debugged, and wireless data is uploaded according to the data upload cycle. Through data accumulation and feedback, the thermal expansion compensation matching scheme and thermal expansion threshold are continuously improved to form an adaptive thermal expansion automatic matching closed loop.

8. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The establishment of a thermal deformation prediction model for the bearing, considering the intensity of the heat source, heat dissipation conditions, and structural dimensions, and obtaining the thermal clearance prediction results of the thermal deformation prediction model under different inputs, includes: Construct a finite element digital twin that includes bearing geometric parameters, material physical properties, and lubricating oil rheological characteristics; Based on fluid dynamics lubrication theory and heat conduction equations, the frictional heat generation and heat transfer path of bearings under different speeds and loads are simulated to generate a three-dimensional temperature field distribution cloud map inside the bearing. Based on the three-dimensional temperature field distribution cloud map, calculate the non-uniform thermal deformation of the inner ring, outer ring and rolling element under thermal equilibrium state. The non-uniform thermal deformation amount is vector-superimposed with the initial assembly clearance to obtain the thermal clearance prediction results of the thermal deformation prediction model under different inputs.

9. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, After obtaining real-time selection parameters based on the optimized thermal expansion compensation selection scheme and comparing the real-time selection parameters with the thermal clearance prediction results, the method further includes: Set the allowable deviation range for hot clearance. If the comparison result shows that the difference exceeds the allowable deviation range for hot clearance, the error correction mechanism is triggered. In a digital twin, the fit tolerance of selected components is iteratively optimized using a genetic algorithm or a particle swarm optimization algorithm until the calculated hot clearance prediction result falls within the allowable deviation range of the hot clearance. The final fit tolerance parameters obtained through iterative optimization are used as the corrected real-time fit parameters and sent to the assembly execution unit.

10. The automatic selection system for bearing assembly lines as described in claim 1, characterized in that, The process of continuously improving the thermal expansion compensation selection scheme and thermal expansion threshold through data accumulation and feedback includes: Collect measured temperature data and vibration spectrum data of the assembled bearing during actual trial operation; The measured temperature data and vibration spectrum data are input into the digital twin, and residual analysis is performed between them and the simulation results of the thermal deformation prediction model. If the residual analysis error is greater than the preset accuracy, the measured temperature data and vibration spectrum data are used to reverse the thermal boundary conditions and convective heat transfer coefficient of the thermal deformation prediction model. Based on the revised thermal deformation prediction model, the assembly clearance compensation requirements in the thermal expansion matching rule database are updated.

Images