Multiphysics Simulation Method for Motor Cooling Devices

Abstract

This invention relates to the field of motor cooling technology and discloses a multiphysics simulation method for motor cooling devices. The method includes a multiphysics coupling layer, an optimization layer, and a simulation verification model. The multiphysics coupling layer and the optimization layer are connected to form an end-to-end numerical calculation model. In the first simulation stage, first experimental data of motor operation, actual cooling strategy information, adjustment commands, and verification feedback information are acquired, input into the corresponding models to obtain prediction results, and the model parameters are adjusted based on these results. Future experimental data can also be obtained to further optimize the parameters, and a second simulation stage is included. This method can accurately simulate the multiphysics of motor cooling devices, optimize cooling strategies, improve simulation accuracy and model adaptability, save R&D costs, and improve motor performance and reliability.

Description

Technical Field

[0001] This invention relates to the field of motor cooling technology, specifically to a multiphysics simulation method for motor cooling devices. Background Technology

[0002] In modern industry and technology, electric motors, as core power components, are widely used in many industries such as automobiles, aerospace, and new energy. As electric motors develop towards higher power density, higher speed, and smaller size, the heat they generate during operation increases dramatically. If this heat cannot be dissipated effectively and promptly, the motor temperature will continue to rise, leading to a series of serious problems.

[0003] Excessive temperature can degrade the insulation performance of motor windings, shorten the motor's lifespan, and may even cause short circuits and damage the motor. Furthermore, uneven temperature distribution can lead to inconsistent thermal expansion of motor components, generating additional stress and affecting the motor's structural stability and operational accuracy. In addition, poor motor cooling reduces efficiency and increases energy consumption, which is extremely detrimental to modern industry's pursuit of high efficiency and energy conservation.

[0004] Traditional methods for designing and optimizing motor cooling devices rely primarily on experience and repeated testing. This approach not only consumes significant human, material, and time resources, but also fails to provide a comprehensive and in-depth understanding of the cooling device's performance under various operating conditions due to the complex and variable operating environment of actual motors, which involves the interaction of multiple physical fields such as temperature, flow, and stress fields.

[0005] In the early stages, researchers primarily focused on single physical fields of motors, such as studying only the effect of the temperature field on motor cooling, neglecting the coupling effects of other factors such as flow fields and stress fields. With technological advancements, although the concept of multiphysics has been gradually introduced, the lack of effective simulation methods and tools makes it difficult to accurately simulate the complex interrelationships between multiple physics fields. For example, when simulating the flow of cooling media, it is impossible to accurately consider its impact on the temperature field distribution, or the effect of temperature changes on the stress field.

[0006] Existing simulation methods often suffer from insufficient accuracy. On the one hand, the simplification of the model makes it impossible to accurately reflect the actual physical process. For example, the simplification of the complex internal structure of the motor results in a large deviation between the simulation results and the actual situation. On the other hand, in terms of data processing and model training, there is a lack of effective strategies to make full use of experimental data, resulting in weak generalization ability of the model, making it difficult to adapt to different motor operating conditions and cooling device designs.

[0007] Furthermore, how to formulate a reasonable cooling strategy based on actual operating conditions is a pressing issue in the design of motor cooling devices. Currently, most cooling strategies are based on fixed rules or simple empirical formulas, which cannot be dynamically adjusted according to the real-time operating status of the motor and multiphysics information, making it difficult to optimize the cooling effect. In summary, existing motor cooling technologies have many shortcomings in multiphysics simulation and cooling strategy optimization, and there is an urgent need for a more advanced, efficient, and accurate multiphysics simulation method for motor cooling devices to meet practical engineering needs. Summary of the Invention

[0008] The purpose of this invention is to provide a multiphysics simulation method for motor cooling devices to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a multiphysics simulation method for a motor cooling device, the method comprising:

[0010] The simulation method includes a multiphysics coupling layer, an optimization layer, and a simulation verification model. The multiphysics coupling layer and the optimization layer are connected to form an end-to-end numerical computation model, enabling the optimization layer to directly generate cooling strategy information based on the output of the multiphysics coupling layer. The method includes a first simulation phase.

[0011] The first simulation stage includes:

[0012] The system acquires multiple first experimental data during motor operation, first real cooling strategy information corresponding to the first experimental data, and first adjustment instructions and first verification feedback information for the first real cooling strategy information. The first experimental data includes first operating parameters of the motor and first measured physical field information for the cooling device. The first measured physical field information includes current and historical measurement data of temperature field, flow field and stress field.

[0013] The first input sample, including the first experimental data, is input into the multiphysics coupling layer to obtain the first implicit field representation output by the multiphysics coupling layer;

[0014] The first intermediate sample, including the first implicit field representation, is input into the optimization layer to obtain the first predicted cooling strategy information output by the optimization layer.

[0015] The first intermediate sample is input into the simulation verification model to obtain the first prediction adjustment instruction and the first prediction verification feedback information output by the simulation verification model.

[0016] The parameters of the simulation verification model are adjusted based on the first adjustment instruction, the first prediction adjustment instruction, the first verification feedback information, and the first prediction verification feedback information.

[0017] Preferably, the parameters of the multiphysics coupling layer and the optimization layer are adjusted based on the first prediction adjustment instruction, the first prediction verification feedback information, the first prediction cooling strategy information, and the first actual cooling strategy information.

[0018] The first adjustment instruction and the first prediction adjustment instruction are used to indicate whether manual correction strategy and correction timing are needed during the simulation process, respectively. The first verification feedback information and the first prediction verification feedback information include manually labeled and cooling performance evaluation results generated by the simulation model, respectively.

[0019] The first verification feedback information includes at least one of the following:

[0020] Information on temperature distribution uniformity, flow field stability, cooling efficiency, peak structural stress, number of thermal cycles, and material deformation threshold.

[0021] Preferably, the first simulation stage further includes:

[0022] Obtain first future experimental data for the operating environment of the cooling device.

[0023] The output of the optimization layer also includes first future predicted physics field information for the operating environment of the cooling device.

[0024] Furthermore, based on the first prediction adjustment instruction, the first prediction verification feedback information, the first prediction cooling strategy information, and the first actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted as follows:

[0025] Based on the first future experimental data, the first future predicted physics information, the first prediction adjustment instruction, the first prediction verification feedback information, the first prediction cooling strategy information, and the first real cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

[0026] Preferably, the first future experimental data includes measured data of the future temperature field, measured data of the future flow field, and / or a future implicit field representation corresponding to the measured data.

[0027] Preferably, it also includes a second simulation stage following the first simulation stage.

[0028] The second simulation stage includes:

[0029] Acquire multiple second experimental data during motor operation, second real cooling strategy information corresponding to the second experimental data, and second adjustment instructions and second verification feedback information for the second real cooling strategy information, wherein the second experimental data includes the second operating parameters of the motor and the second measured physical field information for the cooling device;

[0030] The second input sample, including the second experimental data, is input into the multiphysics coupling layer to obtain the second implicit field representation output by the multiphysics coupling layer;

[0031] The second intermediate sample, including the second implicit field representation, is input into the optimization layer to obtain the second predicted cooling strategy information output by the optimization layer;

[0032] The second intermediate sample is input into the simulation verification model to obtain the second prediction adjustment instruction and the second prediction verification feedback information output by the simulation verification model.

[0033] The parameters of the simulation verification model are adjusted based on the second adjustment instruction, the second prediction adjustment instruction, the second verification feedback information, and the second prediction verification feedback information; and

[0034] Based on the second prediction adjustment instruction, the second prediction verification feedback information, the second prediction cooling strategy information, and the second actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

[0035] Preferably, the adjustment instruction includes adjustment decision information, which indicates whether the strategy needs to be manually corrected and the specific time point for correction. The predictive adjustment instruction includes predictive adjustment decision information, which indicates whether correction will be triggered during the simulation process and the predicted trigger time.

[0036] Preferably, the adjustment instruction includes an adjustment identifier, which is used to characterize whether the first real cooling strategy information has been adjusted by external intervention, and the predictive adjustment instruction includes a predictive adjustment identifier, which is used to characterize whether the predicted cooling strategy information needs to be adjusted by external intervention.

[0037] Preferably, inputting a first input sample including the first experimental data into the multiphysics coupling layer to obtain the first implicit field representation output by the multiphysics coupling layer includes:

[0038] Based on the first verification feedback information, the multiple first experimental data are classified according to working conditions.

[0039] Based on the classification results, multiple first input sample data are obtained by sampling from the multiple first experimental data in proportion to each category; and

[0040] The first input sample, including the first input sample data, is input into the multiphysics coupling layer to obtain the first implicit field representation.

[0041] Preferably, the experimental data is divided into multiple working condition categories based on the verification feedback information, and the difference in the number of samples corresponding to each category in the multiple first input sample data does not exceed a preset threshold.

[0042] Preferably, the method further includes:

[0043] Before the first simulation stage, the multiphysics coupling layer and optimization layer are pre-computed offline so that the simulation model can generate the first predicted cooling strategy information based on the first input sample.

[0044] The first simulation stage further includes:

[0045] An initial simulation is performed using the simulation model obtained through offline pre-computation. During the simulation, the first experimental data and the corresponding first real cooling strategy information, as well as the first adjustment instruction and the first verification feedback information for the first real cooling strategy information, are acquired.

[0046] Preferably, the offline pre-computation includes:

[0047] Acquire third experimental data and third real cooling strategy information corresponding to the third experimental data, wherein the third experimental data includes historical operating parameters of the test motor and historical physical field information for the cooling device;

[0048] The third input sample, including the third experimental data, is input into the multiphysics coupling layer to obtain the third implicit field representation output by the multiphysics coupling layer;

[0049] The third intermediate sample, including the third implicit field representation, is input into the optimization layer to obtain the third predicted cooling strategy information output by the optimization layer; and

[0050] Based on the third predicted cooling strategy information and the third actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

[0051] Preferably, the third experimental data further includes third verification feedback information regarding the third real cooling strategy information.

[0052] The adjustment of the parameters of the multiphysics coupling layer and the optimization layer based on the third predicted cooling strategy information and the third actual cooling strategy information includes:

[0053] Based on the third verification feedback information, the third predicted cooling strategy information, and the third actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

[0054] Preferably, the present invention further includes a multiphysics simulation device for a motor cooling device, the device comprising:

[0055] The experimental data acquisition unit is configured to acquire experimental data, corresponding real cooling strategy information, adjustment instructions, and verification feedback information during motor operation.

[0056] The coupling layer processing unit is configured to input experimental data into the multiphysics coupling layer and output implicit field representations;

[0057] The optimization layer processing unit is configured to input the implicit field representation into the optimization layer to generate predictive cooling strategy information;

[0058] The verification model processing unit is configured to input implicit field representations into the simulation verification model to output prediction adjustment instructions and prediction verification feedback information;

[0059] The parameter adjustment unit is configured to adjust the simulation verification model parameters based on adjustment instructions, predicted adjustment instructions, verification feedback information, and predicted verification feedback information, and to adjust the coupling layer and optimization layer parameters based on predicted adjustment instructions, predicted verification feedback information, predicted cooling strategy information, and actual cooling strategy information.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] To improve simulation accuracy, this method comprehensively processes multi-physics information such as temperature, flow, and stress fields during motor operation through a multi-physics coupling layer. It considers not only current measurement data but also historical measurement data, enabling a more comprehensive and accurate reflection of the complex interactions between multiple physics fields. Compared to traditional simulation methods, it avoids accuracy losses caused by ignoring certain physical field factors or simplifying the model. For example, in simulating motor cooling, it can accurately calculate the impact of cooling medium flow on temperature distribution and stress changes caused by temperature variations, making the simulation results closer to reality and providing reliable data support for the design and optimization of motor cooling devices.

[0062] In terms of optimizing the cooling strategy, the optimization layer directly generates cooling strategy information based on the output of the multiphysics coupling layer. Furthermore, this process combines data from multiple sources, including actual cooling strategy information, predictive adjustment commands, and verification feedback, to adjust parameters. This allows the cooling strategy to be dynamically optimized based on the motor's real-time operating status and multiphysics information. For example, when changes in motor load cause a temperature increase, the optimized cooling strategy can promptly adjust the flow rate and velocity of the cooling medium to ensure the motor always operates within a suitable temperature range, effectively improving the motor's cooling efficiency and operational stability.

[0063] In terms of model training and adaptability, a large amount of experimental data is acquired, including motor operating parameters under different working conditions and measured physical field information of the cooling device. Combined with corresponding adjustment commands and verification feedback, the multiphysics coupling layer, optimization layer, and simulation verification model are continuously trained and their parameters adjusted. This approach improves the model's generalization ability, enabling it to adapt to different motor operating conditions and cooling device designs. Whether under high-load, low-speed motor conditions or with cooling devices of different structures and materials, the model can accurately perform simulations and strategy optimization.

[0064] In terms of cost savings, traditional motor cooling device designs rely on numerous experiments, requiring significant manpower, resources, and time. The simulation method of this invention, however, performs simulations and optimizations in a virtual environment, greatly reducing the number of actual tests. For example, in the development of new motor cooling devices, multiple design schemes can be evaluated and screened first using simulation methods, and then a few optimized schemes can be verified through actual testing, effectively reducing development costs and shortening the product development cycle.

[0065] In terms of improving motor performance and reliability, precise simulation and optimized cooling strategies result in more uniform temperature distribution, a more stable flow field, and effective control of structural stress. This helps improve the motor's insulation performance, reduce the number of thermal cycles, and lower the risk of material deformation, thereby significantly improving motor performance and reliability, extending motor lifespan, and providing strong support for the stable development of related industries. Attached Figure Description

[0066] Figure 1 This is a schematic diagram illustrating the working principle of the multiphysics simulation method for the motor cooling device described in this invention.

[0067] Figure 2 Flowchart for First Future Experiment Data Construction;

[0068] Figure 3 This is the flowchart for the second simulation stage;

[0069] Figure 4The flowchart shows the offline pre-calculation and initialization process for the first simulation stage. Detailed Implementation

[0070] 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 some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0071] Please see Figures 1-4 This invention provides a multiphysics simulation method for a motor cooling device, with the following specific implementation steps:

[0072] During motor operation, multiple sets of initial experimental data are collected. These data include the motor's initial operating parameters, such as motor speed and load, as well as initial measured physical field information for the cooling device, covering current and historical measurement data of the temperature field, flow field, and stress field. Simultaneously, initial real-world cooling strategy information corresponding to the initial experimental data is acquired, such as the cooling medium flow rate setting and cooling channel switching control, along with initial adjustment commands and initial verification feedback information based on the initial real-world cooling strategy information.

[0073] The first input sample, containing the first experimental data, is fed into the multiphysics coupling layer. The multiphysics coupling layer analyzes and processes the input data, considering the interactions between the temperature field, flow field, and stress field, and outputs a first implicit field representation. This implicit field representation is a comprehensive description of the current state of the motor cooling device, containing the correlation information between the various physical fields.

[0074] The first intermediate sample, containing the first implicit field representation, is input into the optimization layer. Based on the first implicit field representation, the optimization layer uses specific algorithms and strategies, considering the motor's operating requirements and the cooling device's performance limitations, to output first predicted cooling strategy information. This predicted cooling strategy information is an optimized cooling scheme provided by the optimization layer for the current motor operating state and cooling device state.

[0075] The first intermediate sample is input into the simulation verification model. Based on the input first intermediate sample, the simulation verification model simulates the operation of the motor cooling device under the predicted cooling strategy, and outputs the first prediction adjustment command and the first prediction verification feedback information. The first prediction adjustment command is used to indicate whether the cooling strategy needs to be manually corrected during the simulation process and when to make the correction; the first prediction verification feedback information includes the evaluation results of cooling performance, such as the evaluation of temperature distribution uniformity, flow field stability, etc.

[0076] Based on the first adjustment instruction, the first prediction adjustment instruction, the first verification feedback information, and the first prediction verification feedback information, the parameters of the simulation verification model are adjusted. By continuously adjusting the parameters of the simulation verification model, it can more accurately simulate the actual operation of the motor cooling device, thereby improving the accuracy and reliability of the simulation.

[0077] The present invention will be further described below with reference to Examples 1 to 6: Example 1

[0078] In practical applications, in addition to the basic steps mentioned above, the first simulation stage needs to consider more factors. After acquiring the first experimental data, the first real cooling strategy information, the first adjustment command, and the first verification feedback information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted based on the first prediction adjustment command, the first prediction verification feedback information, the first prediction cooling strategy information, and the first real cooling strategy information.

[0079] The first adjustment instruction and the first predictive adjustment instruction are used to indicate whether manual correction of the strategy is needed during the simulation process and when correction should be made. For example, when the motor load suddenly changes significantly, the first adjustment instruction may indicate that manual intervention is needed to adjust the cooling strategy to ensure that the motor can operate normally under the new load conditions. The first predictive adjustment instruction, on the other hand, is based on the simulation verification model's prediction of whether a correction strategy is needed and the predicted timing of the correction. If the simulation verification model predicts that the motor temperature will exceed the safety threshold at some point in the future, it will output the first predictive adjustment instruction, prompting that the cooling strategy needs to be adjusted in advance.

[0080] The first verification feedback information and the first prediction verification feedback information include manually labeled and simulation model-generated cooling performance evaluation results, respectively. The first verification feedback information includes at least one of the following: temperature distribution uniformity information, flow field stability information, cooling efficiency, peak structural stress, number of thermal cycles, and material deformation threshold. Taking temperature distribution uniformity information as an example, during motor operation, an unreasonable design of the cooling device may lead to significant temperature differences in different parts of the motor, affecting its performance and lifespan. Analyzing the temperature distribution uniformity information allows for the evaluation of the effectiveness of the cooling strategy. If the first verification feedback information indicates uneven temperature distribution, the cooling strategy needs to be adjusted, such as changing the flow path of the cooling medium or increasing its flow rate. Simultaneously, the parameters of the multiphysics coupling layer and optimization layer are adjusted based on this feedback information, enabling the model to better adapt to actual conditions and improve prediction accuracy.

[0081] When acquiring initial experimental data on the operating environment of the cooling system, all possible variables must be fully considered. Measured future temperature field data may be affected by changes in ambient temperature; if the motor is used outdoors, the ambient temperature will vary with the seasons and time. Measured future flow field data may be affected by wear or blockage of internal components of the cooling system, leading to changes in the flow state of the cooling medium. The corresponding implicit future field representation reflects the combined impact of these changes on the motor cooling system.

[0082] In addition to the first predicted cooling strategy information, the output of the optimization layer also includes the first future predicted physics field information for the operating environment of the cooling device. Based on the first future experimental data, the first future predicted physics field information, the first predicted adjustment command, the first predicted verification feedback information, the first predicted cooling strategy information, and the first actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted. This allows the model to not only predict and adjust based on current experimental data, but also consider possible future situations, optimize the cooling strategy in advance, and improve the performance and reliability of the motor cooling device. Example 2

[0083] In the first simulation phase, data acquisition and processing are crucial. Obtaining initial experimental data specific to the operating environment of the cooling device is essential for improving the accuracy and predictability of the simulation.

[0084] The first set of future experimental data includes measured data of the future temperature field, measured data of the future flow field, and / or the corresponding implicit field representation of the future. Measured data of the future temperature field can be acquired by placing temperature sensors around the cooling device. These sensors can monitor changes in ambient temperature and temperature trends in key components inside the motor in real time. For example, high-precision temperature sensors can be installed near the motor windings and on the surface of the core to record temperature data over a future period. Analysis of this data reveals the temperature variation patterns of the motor under different operating conditions, providing a basis for subsequent simulations and strategy adjustments.

[0085] Acquiring future flow field measurement data will require specialized measuring equipment, such as particle image velocimeters (PIV) or hot-wire anemometers. These devices can measure parameters such as the velocity, flow rate, and flow direction of the cooling medium within the cooling channel. By monitoring the future flow field measurement data, abnormalities occurring during the cooling medium flow process can be detected in a timely manner, such as excessively low flow velocity or uneven flow.

[0086] Assume the flow rate of the cooling medium is Flow velocity is The cross-sectional area of ​​the cooling channel is They satisfy the formula ,in This indicates the volume of cooling medium passing through a cross-section of the cooling channel per unit time, expressed in cubic meters per second. ); This indicates the flow rate of the cooling medium, measured in meters per second (m / s). ); This indicates the cross-sectional area of ​​the cooling channel, in square meters (m²). If the flow rate in a certain area of ​​the cooling channel is found to be significantly lower than that in other areas, it may be due to blockage or excessive resistance in that area. In this case, the cooling device needs to be inspected and maintained, and this factor should be considered in the simulation model, and the relevant parameters should be adjusted.

[0087] The implicit field representation corresponding to future measured data is a comprehensive abstract representation of information such as future temperature and flow fields. It integrates the interrelationships between multiple physical quantities, and can more comprehensively reflect the changes in the operating environment of the cooling device. When acquiring the implicit field representation, some advanced data processing methods can be used, such as principal component analysis (PCA) in machine learning algorithms or autoencoders in deep learning. These methods can extract key features from a large amount of future measured data to generate a concise and effective implicit field representation.

[0088] To measure how well the future implicit field representation retains the features of the original data, an information retention rate is introduced. The calculation formula is: ,in Indicates the number of principal components extracted. Indicates the number of features in the original data. Indicates the first The eigenvalues ​​corresponding to each principal component Indicates the first The feature values ​​corresponding to each original feature. The range of values ​​is within arrive Between, the closer The higher the degree to which the implicit field representation of the future preserves the features of the original data.

[0089] After acquiring these first future experimental data, they are integrated into the simulation process. When generating the first predicted cooling strategy information, the optimization layer fully considers the trends and changes reflected in these future experimental data. If measured data from the future temperature field indicate that the ambient temperature will rise in the near future, the optimization layer may increase the flow rate of the cooling medium in advance to ensure the motor can still operate normally under high-temperature conditions. Simultaneously, based on the first future experimental data, the first future predicted physics field information, the first predicted adjustment command, the first predicted verification feedback information, the first predicted cooling strategy information, and the first actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted, enabling the model to better adapt to future changes and improving the accuracy and reliability of the simulation. Example 3

[0090] After completing the first simulation phase, the present invention also includes a second simulation phase. In the second simulation phase, multiple second experimental data during motor operation, second real cooling strategy information corresponding to the second experimental data, and second adjustment instructions and second verification feedback information based on the second real cooling strategy information are first acquired.

[0091] The second set of experimental data includes the motor's second operating parameters and the second measured physical field information for the cooling device. Similar to the first simulation phase, the motor's second operating parameters cover information such as speed and load during this phase. These parameters will change depending on the actual usage of the motor; for example, the motor may need to operate at different speeds and loads under different working conditions. The second measured physical field information for the cooling device also includes relevant data on the temperature field, flow field, and stress field. By monitoring these data in real time, the operating status of the cooling device in the second phase can be understood.

[0092] A second input sample, including the second experimental data, is input into the multiphysics coupling layer. The multiphysics coupling layer analyzes and processes the second input sample, considering the interactions between the temperature field, flow field, and stress field, and outputs a second implicit field representation. This process is the same as the working principle of the multiphysics coupling layer in the first simulation stage, but due to the different input data, the output second implicit field representation will reflect the specific state of the current motor and cooling device.

[0093] A second intermediate sample, including a second implicit field representation, is input into the optimization layer. Based on the second implicit field representation, and considering the motor's operating requirements and the performance limitations of the cooling device, the optimization layer outputs second predicted cooling strategy information. For example, if the second implicit field representation shows an upward trend in temperature at a certain part of the motor, the optimization layer may adjust the cooling strategy, increasing the cooling medium flow rate at that part or changing the flow path of the cooling medium.

[0094] The second intermediate sample is input into the simulation verification model. Based on the input second intermediate sample, the simulation verification model simulates the operation of the motor cooling device under the predicted cooling strategy, and outputs a second prediction adjustment command and second prediction verification feedback information. The second prediction adjustment command is used to indicate whether manual correction of the cooling strategy is needed during the simulation process and when to make the correction; the second prediction verification feedback information includes the evaluation results of cooling efficiency, such as the evaluation of temperature distribution uniformity, flow field stability, etc.

[0095] The parameters of the simulation verification model are adjusted based on the second adjustment command, the second prediction adjustment command, the second verification feedback information, and the second prediction verification feedback information. By continuously adjusting the parameters of the simulation verification model, it can more accurately simulate the actual operation of the motor cooling device. For example, if the second verification feedback information shows a large deviation between the simulation results and the actual situation, it is necessary to adjust the parameters of the simulation verification model, optimize the model's algorithm and structure, and improve the simulation accuracy.

[0096] Based on the second prediction adjustment command, the second prediction verification feedback information, the second prediction cooling strategy information, and the second actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted. This allows the multiphysics coupling layer and the optimization layer to better adapt to the operating state of the motor and cooling device in the second stage, improves the accuracy of prediction, and provides more reliable support for subsequent cooling strategy optimization. Example 4

[0097] Throughout the simulation, adjustment commands and predictive adjustment commands play crucial roles. Adjustment commands include adjustment decision information, indicating whether manual strategy correction is needed and the specific timing of such correction; predictive adjustment commands include predictive adjustment decision information, indicating whether correction will be triggered during the simulation and the predicted trigger time.

[0098] In practical applications, when the operating conditions of a motor change significantly, such as a sudden increase in motor load, the motor may generate more heat, and the original cooling strategy may not be able to meet the heat dissipation requirements. In this case, adjustment decision information may indicate the need for manual modification of the cooling strategy, specifying the exact timing of the modification, such as immediately increasing the flow rate of the cooling medium. After the cooling strategy is manually modified, corresponding first-real cooling strategy information is generated, which will serve as the basis for subsequent simulations and model adjustments.

[0099] Predictive adjustment decision information is generated by the simulation verification model based on simulation results. The simulation verification model predicts the future operation of the motor cooling system. If it predicts that the motor temperature will exceed the safety threshold at some point in the future, or that the flow field will show an unstable trend, it will output predictive adjustment decision information. For example, if the simulation verification model predicts that the motor winding temperature will reach the warning value within the next 5 minutes, the predictive adjustment decision information will instruct the cooling strategy to be corrected 5 minutes in advance, such as increasing the speed of the cooling fan.

[0100] By comparing the adjustment decision information and the predicted adjustment decision information, the accuracy of the simulation verification model can be evaluated. If the two indications are basically consistent, it indicates that the simulation verification model can predict the operation of the motor cooling device well; if there are significant differences, the simulation verification model needs to be adjusted and optimized. Simultaneously, these adjustment instructions and predicted adjustment instructions also affect the parameter adjustments of the multiphysics coupling layer and the optimization layer. Based on the adjustment decision information and the predicted adjustment decision information, combined with the verification feedback information and the predicted verification feedback information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted, enabling the model to more accurately predict the cooling strategy and improve the performance and reliability of the motor cooling device.

[0101] The adjustment instructions also include adjustment identifiers, and the predicted adjustment instructions include predicted adjustment identifiers. The adjustment identifier indicates whether the initial true cooling strategy information has undergone external intervention, while the predicted adjustment identifier indicates whether the predicted cooling strategy information requires external intervention. When abnormal conditions occur during motor operation, such as excessively high temperature or unstable flow field, operators may manually adjust the cooling strategy. In this case, the adjustment identifier records this external intervention. During subsequent simulations, analyzing the adjustment identifiers reveals when external intervention is needed and its impact on cooling performance. The predicted adjustment identifier helps determine whether the predicted cooling strategy information requires further manual intervention. If the predicted adjustment identifier indicates that the predicted cooling strategy information requires external intervention, preparations can be made in advance to adjust the cooling strategy at the appropriate time, ensuring the normal operation of the motor cooling device. Example 5

[0102] During the input of the first experimental data into the multiphysics coupling layer, some data processing operations are required. Based on the first verification feedback information, the operating conditions of multiple first experimental data are classified. The first verification feedback information contains rich information about the operating status of the motor cooling device, such as temperature distribution uniformity and flow field stability. By analyzing this information, the operating conditions of the motor can be divided into different categories. For example, if the motor has a uniform temperature distribution, stable flow field, and high cooling efficiency within a certain period of time, this operating condition can be classified as a normal operating condition; if the motor temperature is locally too high, the flow field is somewhat turbulent, and the cooling efficiency decreases, it can be classified as an abnormal operating condition.

[0103] Based on the classification results, multiple first-stage experimental datasets were sampled proportionally by category to obtain multiple first-stage input sample data. This was done to ensure that each operating condition category had reasonable representativeness in the input samples, preventing any one category from dominating the dataset and affecting the model's training and prediction performance. For example, if the amount of data for normal operating conditions is much larger than that for abnormal operating conditions, the model might be more inclined to learn the features of data under normal operating conditions during training, while its predictive ability for abnormal operating conditions would be weaker. By sampling proportionally by category, the model can better learn the features of different operating conditions, improving its generalization ability.

[0104] During the sampling process, it is essential to ensure that the difference in the number of samples corresponding to each category among multiple first input sample data does not exceed a preset threshold. The preset threshold needs to be adjusted based on the actual situation. Generally, the preset threshold should not be too large, otherwise the balance of data across categories cannot be guaranteed; nor should it be too small, otherwise insufficient data may affect the model's training effect. The first input samples, including the first input sample data, are input into the multiphysics coupling layer to obtain the first implicit field representation. After the above data processing steps, the first input samples input to the multiphysics coupling layer can more comprehensively reflect the different operating conditions of the motor cooling device, making the first implicit field representation output by the multiphysics coupling layer more accurate and reliable. This provides a good foundation for the subsequent optimization layer to generate accurate first predicted cooling strategy information, and also helps improve the accuracy of the simulation verification model and the performance of the entire simulation system. Example 6

[0105] Before the first simulation phase, offline pre-computation of the multiphysics coupling layer and optimization layer is required so that the simulation model can generate the first predicted cooling strategy information based on the first input sample.

[0106] The offline pre-calculation first requires acquiring the third experimental data and the corresponding third real-world cooling strategy information. The third experimental data includes the historical operating parameters of the test motor and historical physical field information for the cooling device. The historical operating parameters of the test motor record information such as speed and load under different operating times and conditions, reflecting various operating states of the motor. The historical physical field information for the cooling device includes past measurement data of the temperature field, flow field, and stress field. By analyzing this historical data, the performance of the cooling device under different operating conditions can be understood.

[0107] The third input sample, including the third experimental data, is input into the multiphysics coupling layer. The multiphysics coupling layer analyzes and processes the input data, considering the interaction between the temperature field, flow field, and stress field, and outputs a third implicit field representation. This process is the same as the principle by which the multiphysics coupling layer processes the first input sample in the first simulation stage; both obtain a comprehensive implicit field representation through the analysis of the input data.

[0108] Next, the third intermediate sample, including the third implicit field representation, is input into the optimization layer. Based on the third implicit field representation, the optimization layer uses specific algorithms and strategies, considering the operating requirements of the motor and the performance limitations of the cooling device, to output the third predicted cooling strategy information.

[0109] If the third experimental data also includes third verification feedback information regarding the third true cooling strategy, then adjusting the parameters of the multiphysics coupling layer and optimization layer needs to be based on the third verification feedback information, the third predicted cooling strategy information, and the third true cooling strategy information. The third verification feedback information includes evaluation results of cooling performance, such as temperature distribution uniformity and flow field stability. For example, if the third verification feedback information shows that the temperature distribution of the cooling device is uneven under a certain operating condition, and the third predicted cooling strategy information fails to effectively solve this problem, then the parameters of the multiphysics coupling layer and optimization layer need to be adjusted. By adjusting the parameters, the model can better learn the patterns in historical data, improve the model's prediction accuracy of cooling strategies under different operating conditions, and provide a more reliable model foundation for the subsequent first simulation stage. In the first simulation stage, the simulation model obtained through offline pre-calculation is used to perform the initial simulation. During the simulation, the first experimental data and the corresponding first true cooling strategy information, as well as the first adjustment instructions and first verification feedback information regarding the first true cooling strategy information, are acquired, thereby continuously optimizing the simulation model and improving the accuracy and reliability of the multiphysics simulation of the motor cooling device.

[0110] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0111] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multiphysics simulation method for a motor cooling device, characterized in that, The simulation method includes a multiphysics coupling layer, an optimization layer, and a simulation verification model. The multiphysics coupling layer and the optimization layer are connected to form an end-to-end numerical calculation model, enabling the optimization layer to directly generate cooling strategy information based on the output of the multiphysics coupling layer. The method includes a first simulation phase. The first simulation stage includes: The system acquires multiple first experimental data during motor operation, first real cooling strategy information corresponding to the first experimental data, and first adjustment instructions and first verification feedback information for the first real cooling strategy information. The first experimental data includes first operating parameters of the motor and first measured physical field information for the cooling device. The first measured physical field information includes current and historical measurement data of temperature field, flow field and stress field. The first input sample, including the first experimental data, is input into the multiphysics coupling layer to obtain the first implicit field representation output by the multiphysics coupling layer; The first intermediate sample, including the first implicit field representation, is input into the optimization layer to obtain the first predicted cooling strategy information output by the optimization layer. The first intermediate sample is input into the simulation verification model to obtain the first prediction adjustment instruction and the first prediction verification feedback information output by the simulation verification model. The parameters of the simulation verification model are adjusted based on the first adjustment instruction, the first prediction adjustment instruction, the first verification feedback information, and the first prediction verification feedback information. The first simulation phase also includes: Obtain first future experimental data for the operating environment of the cooling device. The output of the optimization layer also includes first future predicted physics field information for the operating environment of the cooling device. Furthermore, based on the first prediction adjustment instruction, the first prediction verification feedback information, the first prediction cooling strategy information, and the first actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted as follows: Based on the first future experimental data, the first future predicted physics information, the first prediction adjustment command, the first prediction verification feedback information, the first prediction cooling strategy information, and the first real cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted. The first future experimental data includes measured data of the future temperature field, measured data of the future flow field, and / or a future implicit field representation corresponding to the measured data.

2. The multiphysics simulation method for motor cooling devices according to claim 1, characterized in that, Based on the first prediction adjustment instruction, the first prediction verification feedback information, the first prediction cooling strategy information, and the first actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted. The first adjustment instruction and the first prediction adjustment instruction are used to indicate whether manual correction strategy and correction timing are needed during the simulation process, respectively. The first verification feedback information and the first prediction verification feedback information include manually labeled and cooling performance evaluation results generated by the simulation verification model, respectively. The first verification feedback information includes at least one of the following: Information on temperature distribution uniformity, flow field stability, cooling efficiency, peak structural stress, number of thermal cycles, and material deformation threshold.

3. The multiphysics simulation method for motor cooling devices according to any one of claims 1-2, characterized in that, It also includes a second simulation phase following the first simulation phase. The second simulation stage includes: Acquire multiple second experimental data during motor operation, second real cooling strategy information corresponding to the second experimental data, and second adjustment instructions and second verification feedback information for the second real cooling strategy information, wherein the second experimental data includes the second operating parameters of the motor and the second measured physical field information for the cooling device; The second input sample, including the second experimental data, is input into the multiphysics coupling layer to obtain the second implicit field representation output by the multiphysics coupling layer; The second intermediate sample, including the second implicit field representation, is input into the optimization layer to obtain the second predicted cooling strategy information output by the optimization layer; The second intermediate sample is input into the simulation verification model to obtain the second prediction adjustment instruction and the second prediction verification feedback information output by the simulation verification model. The parameters of the simulation verification model are adjusted based on the second adjustment instruction, the second prediction adjustment instruction, the second verification feedback information, and the second prediction verification feedback information; and Based on the second prediction adjustment instruction, the second prediction verification feedback information, the second prediction cooling strategy information, and the second actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

4. The multiphysics simulation method for motor cooling devices according to any one of claims 1-2, characterized in that, The adjustment instruction includes adjustment decision information, which indicates whether the strategy needs to be manually corrected and the specific time point for correction. The predictive adjustment instruction includes predictive adjustment decision information, which indicates whether correction will be triggered during the simulation process and the predicted trigger time.

5. The multiphysics simulation method for a motor cooling device according to any one of claims 1-2, characterized in that, The adjustment instruction includes an adjustment identifier, which is used to indicate whether the first real cooling strategy information has been adjusted by external intervention. The predictive adjustment instruction includes a predictive adjustment identifier, which is used to indicate whether the predicted cooling strategy information needs to be adjusted by external intervention.

6. The multiphysics simulation method for a motor cooling device according to any one of claims 1-2, characterized in that, Inputting a first input sample, including the first experimental data, into the multiphysics coupling layer to obtain the first implicit field representation output by the multiphysics coupling layer includes: Based on the first verification feedback information, the multiple first experimental data are classified according to working conditions. Based on the classification results, multiple first input sample data are obtained by sampling from the multiple first experimental data in proportion to each category; and The first input sample, including the first input sample data, is input into the multiphysics coupling layer to obtain the first implicit field representation.

7. The multiphysics simulation method for motor cooling devices according to claim 6, characterized in that, The experimental data is divided into multiple working condition categories based on the verification feedback information, and the difference in the number of samples corresponding to each category in the multiple first input sample data does not exceed a preset threshold.

8. The multiphysics simulation method for motor cooling devices according to any one of claims 1-2, characterized in that, Also includes: Before the first simulation stage, the multiphysics coupling layer and optimization layer are pre-computed offline so that the simulation verification model can generate the first predicted cooling strategy information based on the first input sample. The first simulation stage further includes: The initial simulation is performed using the simulation verification model obtained through offline pre-computation. During the simulation, the first experimental data and the corresponding first real cooling strategy information, as well as the first adjustment instruction and the first verification feedback information for the first real cooling strategy information, are acquired.

9. The multiphysics simulation method for motor cooling devices according to claim 8, characterized in that, The offline pre-computation includes: Acquire third experimental data and third real cooling strategy information corresponding to the third experimental data, wherein the third experimental data includes historical operating parameters of the test motor and historical physical field information for the cooling device; The third input sample, including the third experimental data, is input into the multiphysics coupling layer to obtain the third implicit field representation output by the multiphysics coupling layer; The third intermediate sample, including the third implicit field representation, is input into the optimization layer to obtain the third predicted cooling strategy information output by the optimization layer; and Based on the third predicted cooling strategy information and the third actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

10. The multiphysics simulation method for motor cooling device according to claim 9, characterized in that, The third experimental data also includes third verification feedback information regarding the third real cooling strategy information. The adjustment of the parameters of the multiphysics coupling layer and the optimization layer based on the third predicted cooling strategy information and the third actual cooling strategy information includes: Based on the third verification feedback information, the third predicted cooling strategy information, and the third actual cooling strategy information, the parameters of the multiphysics coupling layer and the optimization layer are adjusted.

11. A multiphysics simulation device for a motor cooling system, used to implement the multiphysics simulation method for a motor cooling system as described in any one of claims 1-10, characterized in that, include: The experimental data acquisition unit is configured to acquire experimental data, corresponding real cooling strategy information, adjustment instructions, and verification feedback information during motor operation. The coupling layer processing unit is configured to input experimental data into the multiphysics coupling layer and output implicit field representations; The optimization layer processing unit is configured to input the implicit field representation into the optimization layer to generate predictive cooling strategy information; The verification model processing unit is configured to input implicit field representations into the simulation verification model to output prediction adjustment instructions and prediction verification feedback information; The parameter adjustment unit is configured to adjust the simulation verification model parameters based on adjustment instructions, predicted adjustment instructions, verification feedback information, and predicted verification feedback information, and to adjust the coupling layer and optimization layer parameters based on predicted adjustment instructions, predicted verification feedback information, predicted cooling strategy information, and actual cooling strategy information.

Images