A motor temperature prediction method, device, equipment and medium

Abstract

The application relates to the technical field of thermal management and control, and discloses a motor temperature prediction method, device, equipment and medium. Current working condition parameters of a target motor are acquired; the current working condition parameters and a benchmark thermal state coefficient are substituted into a thermal state evolution equation to perform thermal state deduction, and an initial thermal state coefficient is obtained; in the case that the initial thermal state coefficient does not satisfy a preset physical residual condition, the initial thermal state coefficient is subjected to online mechanism evolution adjustment based on an optimization target of minimizing a physical residual, and a corrected thermal state coefficient is obtained; and a current temperature field is obtained through spatial mapping according to the corrected thermal state coefficient and a thermal mode base function. The beneficial effect is that the thermal state coefficient output by the thermal state evolution equation can be adaptively adjusted according to the actual error accumulation of the target motor, the working condition generalization capability of the temperature prediction method is improved, and the reliability and accuracy of the temperature prediction result are improved.

Description

Technical Field

[0001] This application relates to the field of thermal management and control technology, and in particular to a method, device, equipment and medium for predicting motor temperature. Background Technology

[0002] As electric vehicles evolve towards higher voltage, higher speed, and higher power density, the thermal load on their drive motors increases dramatically during operation, posing an increasingly severe challenge to temperature management. In industrial applications, contact sensors or purely data-driven virtual sensing technologies are commonly used to monitor drive motor temperature changes. However, in reality, blind spots exist within drive motors where sensors cannot be installed, inherently limiting contact sensor measurement methods and making it difficult to monitor the true internal thermal state of the drive motor. Furthermore, purely data-driven virtual sensing technologies heavily rely on massive amounts of training data, resulting in poor generalization ability when the drive motor's operating state differs from the training data, leading to significant errors in temperature monitoring results. Summary of the Invention

[0003] This application provides a method, apparatus, device, and medium for predicting motor temperature. Based on the actual error accumulation of the target motor, the thermal state coefficient output by the thermal state evolution equation is adaptively adjusted, which improves the generalization ability of the temperature prediction method and enhances the reliability and accuracy of the temperature prediction results.

[0004] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide a method for predicting motor temperature, the method comprising: Obtain the current operating parameters of the target motor at the current moment; Substituting the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation, thermal state deduction is performed to obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient of the target motor over time. If the initial thermal state coefficient does not meet the preset physical residual condition, the initial thermal state coefficient is adjusted online by mechanism evolution based on the optimization objective of minimizing the physical residual to obtain the corrected thermal state coefficient. The current temperature field of the target motor is obtained by spatial mapping based on the modified thermal state coefficient and thermal mode basis function; wherein the thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

[0005] The motor temperature prediction method proposed in this application pre-extracts historical full-domain temperature data and historical operating parameters of the target motor through offline spatial dimensionality reduction to obtain thermal pattern basis functions and constructs the thermal state evolution equation of the target motor. Substituting the current operating parameters of the target motor into the thermal state evolution equation, the method deduces the change in the thermal state of the target motor from the previous moment to the current moment, obtaining initial thermal state coefficients. Conditional judgment is performed on the initial thermal state coefficients; if the initial thermal state coefficients do not meet the preset physical residual conditions, they are adjusted to corrected thermal state coefficients. Spatial mapping is then performed based on the corrected thermal state coefficients and the thermal pattern basis functions to reconstruct the current temperature field of the target motor. Compared with related technologies, this application pre-extracts thermal pattern basis functions using historical full-domain temperature data and historical operating parameters of the target motor to represent the temperature field of the target motor in the full-domain space. In the actual prediction process, the thermal state coefficients of the target motor are predicted through the thermal state evolution equation, and the temperature field of the target motor is reconstructed based on the thermal state coefficients and the thermal pattern basis functions. This enables the acquisition of temperature data in areas where temperature sensors cannot be installed, filling the temperature measurement blind spots of the target motor and improving the comprehensiveness of temperature measurement. Meanwhile, this application also judges the initial thermal state coefficient based on the preset physical residual conditions, and performs online mechanism evolution adjustment of the initial thermal state coefficient when the preset physical residual conditions are not met, so that the prediction result of the thermal state coefficient can be adaptively adjusted according to the actual working state of the target motor, thereby improving the working condition generalization ability of the thermal state evolution equation and enhancing the reliability and accuracy of the temperature prediction result.

[0006] Optionally, the thermal state evolution equation can be trained in the following manner: The historical global temperature data is projected onto the low-dimensional space of the thermal model basis function to obtain the historical thermal state coefficients corresponding to the historical time of the historical global temperature data. The historical operating condition parameters and the historical thermal state coefficients are substituted into the thermal state evolution equation to perform data fitting on the training coefficients in the thermal state evolution equation, thereby completing the training process of the thermal state evolution equation.

[0007] Optionally, the online mechanistic evolution adjustment of the initial thermal state coefficient based on the optimization objective of minimizing the physical residual to obtain the corrected thermal state coefficient includes: The local nonlinear characteristics of the target motor at key nodes and the thermodynamic physical residuals of the thermal state evolution equation at the current moment are obtained; wherein, the key nodes are grid nodes in the target motor that are sensitive to temperature changes and physical parameters, and the local nonlinear characteristics are the physical parameters at the key nodes; With minimizing the thermodynamic physical residual as the optimization objective, the initial thermal state coefficient is iteratively optimized to obtain the corrected thermal state coefficient.

[0008] Optionally, the thermodynamic physical residuals can be calculated in the following manner: Substitute the initial thermal state coefficient, the current operating condition parameters, and the local nonlinear characteristics into the thermal state evolution equation to calculate the baseline evolution rate of the initial thermal state coefficient; The initial thermal state coefficient is calculated based on the reference thermal state coefficient to determine its temporal variation, and the temporal variation rate of the initial thermal state coefficient is calculated. The deviation between the baseline evolution rate and the time-series change rate is taken as the thermodynamic residual.

[0009] Optionally, the key nodes can be obtained in the following way: Based on the historical global temperature data, temperature coupling analysis is performed on the historical physical parameters to obtain the physical temperature characteristic parameters corresponding to each of the historical physical parameters. Based on the physical temperature characteristic parameters, feature model extraction and modal energy calculation are performed to obtain nonlinear modal vectors and their corresponding nonlinear modal energies. The energy ratio of the nonlinear modal vectors is then filtered based on the nonlinear modal energies to construct nonlinear spatial basis functions. Nonlinear residuals are calculated for all grid nodes in the target motor based on the nonlinear spatial basis function to obtain the nonlinear characteristic residuals of all grid nodes. Then, the key nodes are obtained by filtering all grid nodes based on the nonlinear characteristic residuals.

[0010] Optionally, the method further includes: Predict future operating parameters based on the current operating parameters; Based on the modified thermal state coefficient and the future operating parameters, the thermal state evolution equation is used to perform thermal state deduction and spatial mapping on the target motor to obtain the temperature evolution trajectory of the target motor at the subsequent time. Based on the temperature evolution trajectory and the tolerance limit curve of the target motor, a safety boundary search is performed on the target motor to obtain the remaining tolerance time of the target motor.

[0011] Optionally, the target motor is divided into multiple heat-sensitive zones; the method further includes: For any one of the plurality of heat-sensitive zones, a safety control setting is performed based on the material properties of the heat-sensitive zone to obtain the temperature safety threshold of the heat-sensitive zone. Based on the remaining tolerance time of any heat-sensitive partition and the temperature safety threshold, the corresponding safety management strategy is invoked to perform safety management on any heat-sensitive partition.

[0012] Secondly, embodiments of this application provide a motor temperature prediction device, the device comprising: The current operating condition acquisition module is used to acquire the current operating condition parameters of the target motor at the current moment; The thermal state coefficient prediction module is used to substitute the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation to perform thermal state deduction and obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient over time. The physical residual correction module is used to adjust the initial thermal state coefficient online based on the optimization objective of minimizing the physical residual when the initial thermal state coefficient does not meet the preset physical residual condition, so as to obtain the corrected thermal state coefficient. The temperature field reconstruction module is used to perform spatial mapping based on the corrected thermal state coefficient and the thermal mode basis function to obtain the current temperature field of the target motor; wherein, the thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

[0013] Thirdly, embodiments of this application provide a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in any of the above embodiments.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the method described in any one of the above embodiments.

[0015] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which are used to cause a computer to perform the method described in any of the above embodiments. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating the steps of the motor temperature prediction method provided in this application embodiment; Figure 2 This is a flowchart illustrating the steps of training the thermal state evolution equation in an embodiment of this application; Figure 3 This is a flowchart illustrating the steps of online mechanism evolution adjustment in the embodiments of this application; Figure 4 This is a flowchart illustrating the steps involved in calculating the dynamic residuals in an embodiment of this application. Figure 5 This is a flowchart illustrating the steps involved in obtaining key nodes in the embodiments of this application. Figure 6 This is a flowchart illustrating the steps involved in obtaining the remaining tolerance time in an embodiment of this application. Figure 7 This is a flowchart of the dynamic security boundary prediction and adaptive adjustment in the embodiments of this application; Figure 8 This is a flowchart illustrating the steps involved in implementing fine-grained security management of heat-sensitive zones in an embodiment of this application. Figure 9 This is a flowchart of the zoned refined thermal management scheme in the embodiments of this application; Figure 10 A block diagram of the motor temperature prediction device provided in the embodiments of this application; Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] As electric vehicles evolve towards higher voltage, higher speed, and higher power density, the thermal load on their drive motors increases dramatically during operation, posing an increasingly severe challenge to temperature management. In industrial applications, contact sensors or purely data-driven virtual sensing technologies are commonly used to monitor drive motor temperature changes. However, in reality, blind spots exist within drive motors where sensors cannot be installed, inherently limiting contact sensor measurement methods and making it difficult to monitor the true internal thermal state of the drive motor. Furthermore, purely data-driven virtual sensing technologies heavily rely on massive amounts of training data, resulting in poor generalization ability when the drive motor's operating state differs from the training data, leading to significant errors in temperature monitoring results.

[0020] To address the aforementioned issues, this application provides a method, apparatus, device, and medium for predicting motor temperature. The method involves obtaining the current operating parameters of the target motor; substituting the current operating parameters and the reference thermal state coefficient from the previous moment into the thermal state evolution equation to perform thermal state deduction, thereby obtaining the initial thermal state coefficient; if the initial thermal state coefficient does not meet the preset physical residual condition, adjusting the initial thermal state coefficient online based on the optimization objective of minimizing the physical residual, thereby obtaining the corrected thermal state coefficient; and performing spatial mapping based on the corrected thermal state coefficient and the thermal mode basis function to obtain the current temperature field.

[0021] The motor temperature prediction method provided in this application first performs offline spatial dimensionality reduction extraction on the historical global temperature data and historical operating parameters of the target motor to obtain thermal model basis functions, and constructs the thermal state evolution equation of the target motor; substitutes the current operating parameters of the target motor into the thermal state evolution equation to deduce the change in the thermal state of the target motor from the previous moment to the current moment, and obtains the initial thermal state coefficients; performs condition judgment on the initial thermal state coefficients, and adjusts the initial thermal state coefficients to corrected thermal state coefficients if the initial thermal state coefficients do not meet the preset physical residual conditions; and performs spatial mapping based on the corrected thermal state coefficients and the thermal model basis functions to reconstruct the current temperature field of the target motor.

[0022] Compared with related technologies, this application pre-extracts thermal pattern basis functions by dimensionality reduction using historical global temperature data and historical operating parameters of the target motor to represent the temperature field of the target motor in the global space. In the actual prediction process, the thermal state coefficient of the target motor is predicted through the thermal state evolution equation, and the temperature field of the target motor is reconstructed based on the thermal state coefficient and thermal pattern basis functions. This enables the acquisition of temperature data in areas where temperature sensors cannot be installed, filling the temperature measurement blind spot of the target motor and improving the comprehensiveness of temperature measurement.

[0023] Meanwhile, this application also judges the initial thermal state coefficient based on the preset physical residual conditions, and performs online mechanism evolution adjustment of the initial thermal state coefficient when the preset physical residual conditions are not met, so that the prediction result of the thermal state coefficient can be adaptively adjusted according to the actual working state of the target motor, thereby improving the working condition generalization ability of the thermal state evolution equation and enhancing the reliability and accuracy of the temperature prediction result.

[0024] According to an embodiment of this application, a method for predicting motor temperature is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0025] Reference Figure 1 As shown, this embodiment provides a method for predicting motor temperature, the method including: S100. Obtain the current operating parameters of the target motor at the current moment.

[0026] S200. Substitute the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation to perform thermal state deduction and obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient of the target motor over time.

[0027] S300. If the initial thermal state coefficient does not meet the preset physical residual condition, the initial thermal state coefficient is adjusted online by mechanism evolution based on the optimization objective of minimizing the physical residual to obtain the corrected thermal state coefficient.

[0028] S400. Spatial mapping is performed based on the modified thermal state coefficient and thermal mode basis function to obtain the current temperature field of the target motor; wherein, the thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

[0029] The target motor has a corresponding global space, the size of which is the same as the external dimensions of the target motor. The global space can be divided into multiple grids, each with its own grid nodes.

[0030] The thermal state evolution equation can be obtained by constructing an initial reduced-order equation based on the heat transfer equation in the offline stage, and then training the initial reduced-order equation with parameters based on historical global temperature data and historical operating parameters. This equation is used to extrapolate the thermal state of any grid node in the target motor and output the thermal state coefficients at that grid node. The thermal state coefficients represent the contribution of each thermal model basis function to the thermal state of the target motor at a given time. The input conditions for the thermal state evolution equation can include the operating parameters of the target motor and the thermal state coefficients from the previous time step. After obtaining the updated thermal state coefficients, the temperature field of the target motor is constructed based on the updated thermal state coefficients and the thermal model basis functions, serving as the result of the temperature distribution prediction.

[0031] In some embodiments, the process of obtaining the initial reduced-order equations may include: projecting multiple heat transfer equations onto a low-dimensional space composed of thermal mode basis functions to obtain mechanistic projection equations. Utilizing the orthogonality between thermal mode basis functions, the mechanistic projection equations are reduced in order and terms are eliminated to obtain thermal state evolution equations concerning thermal state coefficients. It is understood that the heat transfer equations can be partial differential equations describing heat transfer modes such as heat conduction and convection, including but not limited to transient heat conduction equations, steady-state heat conduction equations, and thermal radiation equations. The reduction-order elimination process may be used to eliminate spatial derivative terms in the mechanistic projection equations.

[0032] The thermal state evolution equation can be expressed as: in, The input operating parameters represent the input parameters during the operation of the target motor. The term is a nonlinear term, representing the physical parameters of the target motor at the corresponding time, that is, the parameters that change due to temperature during the operation of the target motor; and The system thermal resistance and thermal capacity characteristic parameter matrix of the target motor; , and In the offline phase, all coefficients can be used as training factors, obtained through iterative training using historical full-domain temperature data and historical operating condition parameters. In the online phase... It can be replaced with the actual physical parameters of the target motor.

[0033] The thermal mode basis functions can be extracted offline using spatial dimensionality reduction based on the historical global temperature data and historical operating parameters of the target motor. Each thermal mode basis function represents a thermal mode in the target motor, and different thermal modes correspond to different operating parameters. For example, the thermal mode basis functions can be expressed as: in, , , and These are the basis functions for the thermal mode; This represents the total number of basis functions for the hot mode.

[0034] In some embodiments, the extraction of thermal mode basis functions may include: acquiring historical global temperature data and historical operating parameters of the target motor at multiple historical moments; constructing a basic snapshot space based on the historical global temperature data and historical operating parameters to describe the temperature distribution of the target motor under different historical operating parameters at multiple historical moments; using multiphysics numerical computation tools to simulate the global temperature field of the target motor based on the basic snapshot space to obtain the historical simulated temperature field of the target motor at the corresponding historical moment, which serves as the basic snapshot matrix; correcting the basic snapshot matrix for the time dimension based on the historical global temperature data and historical operating parameters to obtain a hybrid snapshot set; and performing spatial feature decomposition on the hybrid snapshot set to obtain thermal mode basis functions to construct a low-dimensional space for describing the temperature distribution in the target motor. In some embodiments, the specific methods of spatial feature decomposition may include, but are not limited to, orthogonal mode decomposition, manifold learning, and principal component analysis, wherein orthogonal mode decomposition may include methods such as Proper Orthogonal Decomposition (POD).

[0035] The specific process of time dimension correction may include: In the time dimension, synchronizing the basic snapshot matrix, historical global temperature data, and historical operating parameters along the same time axis; based on the sensor spatial coordinates corresponding to the historical global temperature data and historical operating parameters, using the grid nodes in the target motor's global space corresponding to the sensor spatial coordinates as spatial feature points, and extracting simulated data of the spatial feature point positions from the basic snapshot matrix as feature vectors for the spatial feature points; for any spatial feature point, adjusting the feature vectors based on the historical global temperature data and historical operating parameters. The vectors are used to calculate residuals to obtain simulated residual data. A correction operator is constructed for any spatial feature point based on the simulated residual data, and a feature correction matrix is ​​constructed based on the correction operators for all spatial feature points. Considering the connectivity between spatial feature points and other grid nodes in the global space, and based on spatial topological correlation, the feature correction matrix is ​​interpolated and extrapolated in the global space of the target motor to obtain the global correction matrix. The base snapshot matrix is ​​corrected based on the global correction matrix to obtain the corrected snapshot matrix of the target motor. The base snapshot matrix and the corrected snapshot matrix are combined to construct a hybrid snapshot set of the target motor at the corresponding historical moment. The correction operator can be an additive correction operator, a multiplicative correction operator, or a combination of both. The combination operation of the base snapshot matrix and the corrected snapshot matrix can be linear superposition or nonlinear fusion operation.

[0036] Understandably, historical full-range temperature data can be the actual temperature data of the target motor across the entire spatial domain at a historical moment. This data can be obtained using temperature sensors and can include temperature distribution data of the target motor under various operating conditions and during transitions between different conditions. These various operating conditions can include the normal operating state of the target motor, as well as extreme operating conditions, including but not limited to stalling and continuous launch in vehicles equipped with the target motor. Similarly, historical operating condition parameters can be the actual operating condition data of the target motor across the entire spatial domain at a historical moment. This data can be obtained using sensors specific to the type of operating condition parameter, such as operating current and operating voltage. Both historical full-range temperature data and historical operating condition parameters can be actual data obtained from real-vehicle operation tests using actual vehicles equipped with the target motor, or sensor data obtained from testing the target motor on a motor test bench. This sensor data can include NTC (Negative Temperature Coefficient) sample values ​​and current-voltage sequences, etc.

[0037] It should be noted that since the historical global temperature data and historical operating parameters are all acquired in the global space of the target motor, covering all grid nodes in the global space, the thermal model basis functions obtained by spatial feature extraction based on the historical global temperature data and historical operating parameters have the ability to characterize the thermal models in the global space. After reconstructing the temperature field based on the thermal state coefficient and thermal model basis functions, it is possible to predict the temperature distribution of any grid node in the global space, thereby filling the temperature measurement blind spot of the target motor.

[0038] Specifically, the actual operating phase of the target motor is defined as the online phase. During the online phase, the target motor's current operating parameters are acquired through its motor controller, serving as the data basis for temperature distribution prediction. These current operating parameters can be motor signal characteristics exhibited by the target motor during operation, including but not limited to current vector, speed, and bus voltage.

[0039] Furthermore, the current operating parameters and the thermal state coefficient of the target motor at the previous moment are substituted into the thermal state evolution equation to perform thermal state deduction on the target motor, so as to predict and evaluate the contribution of each thermal mode under the current operating parameters, and output the initial thermal state coefficient of the corresponding thermal mode basis function based on the contribution of each thermal mode.

[0040] In some embodiments, the process of acquiring all physical parameters of the target motor at each time step requires a long time and a lot of computing resources, making it difficult to ensure the real-time performance of the thermal state coefficient iteration, thus affecting the timeliness of motor temperature prediction. To improve the real-time performance of the thermal state coefficient iteration, in some embodiments, the physical parameters of some grid nodes in the global space can be used as local nonlinear features to interpolate and map the global space of the target motor, thereby quickly updating the global nonlinear features of the target motor. Specifically, this includes: extracting the mapping sub-matrices corresponding to multiple key nodes in the thermal mode basis functions; performing a dimensionality reduction space mapping between the initial thermal state coefficients and the mapping sub-matrices to directly obtain the node temperature data of the target motor at the key nodes; substituting the node temperature data into the material thermal property formula for calculation to obtain the node physical parameters of multiple key nodes; and linearly mapping the node physical parameters to the global space through the interpolation mapping matrix to update all physical parameters of the target motor, obtaining the global nonlinear features of the target motor at the current time step. Here, the key nodes can be grid nodes in the target motor that are sensitive to temperature changes and physical parameters.

[0041] It should be noted that the construction of the interpolation mapping matrix may include: selecting the nonlinear spatial basis functions corresponding to each key node from multiple nonlinear spatial basis functions as node basis functions to form a node basis function matrix; wherein, the nonlinear spatial basis functions can represent the mode of change of physical parameters in the target motor with temperature; performing inverse matrix calculation on the node basis function matrix to obtain the node basis function inverse matrix, and obtaining the interpolation mapping matrix based on the node basis function matrix and the node basis function inverse matrix.

[0042] The acquisition of multiple nonlinear spatial basis functions can include: constructing a snapshot matrix of nonlinear physical terms for all key nodes based on the physical parameters at each key node; extracting feature models from the snapshot matrix of nonlinear physical terms to obtain multiple nonlinear mode vectors; calculating the modal energy of each nonlinear mode vector; and performing energy filtering on the multiple nonlinear mode vectors to obtain multiple nonlinear spatial basis functions.

[0043] For example, the interpolation mapping matrix can be represented as: in, Let be the inverse matrix of the node basis functions. The global physical parameters obtained based on the interpolation mapping matrix can be expressed as: in, These are global physical parameters; These are the node's physical parameters.

[0044] To address the aforementioned issues, this embodiment sets a preset physical residual condition for the initial thermal state coefficient and judges the initial thermal state coefficient based on the preset physical residual condition. For example, the preset physical residual condition may be that the thermodynamic physical residual corresponding to the initial thermal state coefficient does not exceed a preset residual threshold. The thermodynamic physical residual can be calculated based on the initial thermal state coefficient, current operating parameters, and the physical parameters of the target motor, and can represent the error between the initial thermal state coefficient and the actual thermal state of the target motor.

[0045] When the initial thermal state coefficients satisfy the preset physical residual conditions, it means that the initial thermal state coefficients output by the thermal state evolution equation closely approximate the actual thermal state of the target motor, accurately characterizing the current temperature field of the target motor. In this case, the initial thermal state coefficients can be directly used as corrected thermal state coefficients to reconstruct the current temperature field. When the initial thermal state coefficients do not satisfy the preset physical residual conditions, it means that there is a significant difference between the initial thermal state coefficients output by the thermal state evolution equation and the actual thermal state of the target motor. In this case, the initial thermal state coefficients can be adjusted based on the aforementioned differences to minimize the thermodynamic physical residual between the initial thermal state coefficients and the actual thermal state, obtaining corrected thermal state coefficients. These corrected thermal state coefficients then represent the true thermal state of the target motor at the current moment, used to reconstruct the current temperature field.

[0046] In some embodiments, the thermal state evolution equation can output the same thermal state coefficient for similar operating conditions or operating condition transitions. However, in reality, the target motor is subject to multiple influences such as long-term operation, material aging, and environmental factors, causing changes in the target motor's thermal characteristics. These changes in thermal characteristics result in different temperature distributions of the target motor under the same operating conditions, causing the thermal state evolution equation to accumulate errors during the prediction process. This leads to prediction results deviating from reality, thus limiting the accuracy of temperature monitoring of the target motor. To address the above problem, preset long-term coefficient fine-tuning conditions can be set. These preset long-term coefficient fine-tuning conditions may include at least one of the following: Time / Mileage Threshold: When the cumulative running time or cumulative mileage of the target motor exceeds the preset aging cycle threshold, the parameters will be automatically adjusted adaptively.

[0047] Steady-state error: When the target motor is in steady-state operation, the actual temperature feedback of the target motor is obtained, such as the critical node temperature derived from the real-time electrical parameters of the motor, or the critical node temperature collected by physical sensors; the deviation between the actual temperature feedback and the predicted temperature at the same location in the current temperature field is calculated as the predicted temperature deviation; when the average value of the predicted temperature deviation in multiple consecutive steady-state evaluation cycles exceeds the preset aging cycle threshold, it is considered that the thermal state evolution equation has generated a long-term steady-state deviation, and parameter adaptive fine-tuning is performed.

[0048] To overcome model distortion caused by long-term aging of the target motor, a long-period parameter adaptive fine-tuning mechanism can be set for the thermal state evolution equation to adaptively fine-tune the parameters. Specifically, this involves: under the condition of satisfying preset long-term coefficient fine-tuning, performing sensitivity analysis on the system thermal resistance and thermal capacity characteristic parameter matrix in the thermal state evolution equation to identify the physical parameters to be updated that are highly sensitive to temperature deviations. With minimizing the predicted temperature deviation as the optimization objective, a system identification algorithm is used to perform online fitting and solving for the physical parameters to be updated. The solution results are then used to replace the system thermal resistance and thermal capacity characteristic parameter matrix in the thermal state evolution equation, completing the long-term physical mechanism update of the thermal state evolution equation.

[0049] Furthermore, spatial mapping is performed based on the modified thermal state coefficients and thermal mode basis functions. The thermal mode basis functions are then linearly combined using the modified thermal state coefficients to obtain the current temperature field of the target motor at the current moment. It can be understood that for subsequent moments, the modified thermal state coefficients can represent the historical state of the target motor, used to deduce the thermal state coefficients of the target motor at subsequent moments and reconstruct the temperature field.

[0050] For example, the current temperature field of the target motor can be expressed as: in, express The coordinates of the target motor at time t are Temperature field on the grid nodes; This is the set of basis functions for the thermal mode; for The vector form of the thermal state coefficients corresponding to the basis functions of each thermal mode at time t.

[0051] The motor temperature prediction method provided in this embodiment first performs offline spatial dimensionality reduction extraction on the historical global temperature data and historical operating parameters of the target motor to obtain thermal model basis functions, and constructs the thermal state evolution equation of the target motor; substitutes the current operating parameters of the target motor into the thermal state evolution equation to deduce the change in the thermal state of the target motor from the previous moment to the current moment, and obtains the initial thermal state coefficient; performs condition judgment on the initial thermal state coefficient, and adjusts the initial thermal state coefficient to a corrected thermal state coefficient if the initial thermal state coefficient does not meet the preset physical residual condition; and performs spatial mapping based on the corrected thermal state coefficient and the thermal model basis functions to reconstruct the current temperature field of the target motor.

[0052] Compared with related technologies, this application pre-extracts thermal pattern basis functions by dimensionality reduction using historical global temperature data and historical operating parameters of the target motor to represent the temperature field of the target motor in the global space. In the actual prediction process, the thermal state coefficient of the target motor is predicted through the thermal state evolution equation, and the temperature field of the target motor is reconstructed based on the thermal state coefficient and thermal pattern basis functions. This enables the acquisition of temperature data in areas where temperature sensors cannot be installed, filling the temperature measurement blind spot of the target motor and improving the comprehensiveness of temperature measurement.

[0053] Meanwhile, this application also judges the initial thermal state coefficient based on the preset physical residual conditions, and performs online mechanism evolution adjustment of the initial thermal state coefficient when the preset physical residual conditions are not met, so that the prediction result of the thermal state coefficient can be adaptively adjusted according to the actual working state of the target motor, thereby improving the working condition generalization ability of the thermal state evolution equation and enhancing the reliability and accuracy of the temperature prediction result.

[0054] Reference Figure 2 As shown, in one embodiment of this application, the thermal state evolution equation is trained in the following manner: S210. Project the historical global temperature data onto the low-dimensional space of the thermal model basis functions to obtain the historical thermal state coefficients corresponding to the historical time of the historical global temperature data.

[0055] S220. Substitute the historical operating condition parameters and historical thermal state coefficients into the thermal state evolution equation to perform data fitting on the training coefficients in the thermal state evolution equation, thus completing the training process of the thermal state evolution equation.

[0056] Specifically, for any given historical moment, the historical global temperature data at that moment is projected onto a low-dimensional space constructed by the thermal model basis functions to obtain the historical thermal state coefficients at that moment. By combining the thermal model basis functions based on these historical thermal state coefficients, the reconstructed historical temperature field for the corresponding historical moment can be obtained. It is understood that the method for obtaining the reconstructed historical temperature field can be the same as the method for obtaining the current temperature field.

[0057] In some embodiments, a training optimization objective function can be set when training the thermal state evolution equation. This objective function is used to iteratively optimize the historical thermal state coefficients at each historical moment to minimize the physical residual between the historical thermal state coefficients and the actual state of the target motor at the corresponding historical moment. The thermal state evolution equation is then trained based on the iteratively optimized historical thermal state coefficients, thereby improving the reliability and accuracy of the thermal state evolution equation. For example, the training optimization objective function can include multiple constraints, including but not limited to data constraints and physical constraints. Data constraints can be constraints set on the error between actual temperature data and predicted temperature data, used to minimize the error between historical global temperature data and actual temperature data at spatial feature points, ensuring that the historical thermal state coefficients closely match the actual temperature data. Physical constraints can be constraints set on the difference between the temperature distribution in the target motor and the heat transfer equation, used to minimize the residual of the heat transfer equation under the historical reconstructed temperature field, ensuring that the historical thermal state coefficients follow the fundamental laws of thermodynamics.

[0058] Furthermore, before training is complete, the thermal state evolution equation can contain multiple training coefficients, corresponding to input variables such as input operating condition parameters and thermal state coefficients from the previous time step. Historical operating condition parameters and historical thermal state coefficients are input into the thermal state evolution equation containing unknown training coefficients. The training coefficients are then fitted using this historical data to determine their specific values, resulting in the completed thermal state evolution equation. In some embodiments, data fitting can employ global least squares or further introduce regularization terms.

[0059] Understandably, historical global temperature data and historical operating condition parameters include temperature distribution data of the target motor under various operating conditions and during transitions between different operating conditions. The thermal state evolution equation trained based on this data has pre-learned the preset state mode of the target motor under the corresponding operating conditions, obtaining the thermal state coefficients of the reconstructed temperature field under this preset state mode. Therefore, in the actual prediction process, the thermal state evolution equation can output the same thermal state coefficients for similar operating conditions or operating condition transitions.

[0060] Reference Figure 3 As shown, in one embodiment of this application, based on the optimization objective of minimizing physical residuals, the initial thermal state coefficient is adjusted online through mechanistic evolution to obtain a corrected thermal state coefficient, including: S310. Obtain the local nonlinear characteristics of the target motor at the key nodes, and the thermodynamic physical residuals of the thermal state evolution equation at the current moment; where the key nodes are the grid nodes in the target motor that are sensitive to temperature changes and physical parameters, and the local nonlinear characteristics are the physical parameters at the key nodes.

[0061] S320. With minimizing the thermodynamic physical residual as the optimization objective, the initial thermal state coefficient is iteratively optimized to obtain the corrected thermal state coefficient.

[0062] Specifically, the target motor contains multiple key nodes, which correspond to active physical characteristics, exhibiting both temperature change sensitivity and physical parameter sensitivity. Temperature change sensitivity indicates that temperature data at key nodes is easily affected by operating parameters, leading to drastic changes. Physical parameter sensitivity indicates that the physical parameters of the target motor at key nodes are easily affected by temperature data, leading to drastic changes. For example, key nodes can be distributed in areas such as the core region at the winding ends and the cooling water inlet of the target motor. The physical nodes used to update global physical parameters can be key nodes. The physical parameters at multiple key nodes in the current temperature field are acquired as local nonlinear characteristics of the target motor. These local nonlinear characteristics can be parameters that change with temperature, including but not limited to resistivity, thermal conductivity, and coefficient of thermal expansion.

[0063] Furthermore, the thermodynamic physical residual of the thermal state evolution equation at the current moment is obtained. Minimizing this thermodynamic physical residual is used as the optimization objective. Based on this residual, an objective function is constructed for the target motor, resulting in the optimization objective function. For example, the optimization objective function can be expressed as: in, For thermodynamic physical residuals, The initial thermal state coefficient; This is a regularization term used to improve the smoothness of the current preset temperature distribution; and These are weighting coefficients used to adjust the relative relationships between various objectives.

[0064] Numerical optimization algorithms are used to iteratively optimize the objective function to gradually approach its minimum value. In some embodiments, gradient optimization can be used to iteratively update the initial thermal state coefficients. Specifically, this can include: in the initial time step of the iterative update, substituting the initial thermal state coefficients into the objective function for calculation to obtain the current iterative value and current iterative gradient of the objective function; numerically optimizing the initial thermal state coefficients based on the current iterative gradient to obtain iterative thermal state coefficients; in any subsequent time step, substituting the iterative thermal state coefficients into the objective function for numerical and gradient calculations to obtain the current iterative value and current iterative gradient of the objective function at that time step; if neither the current iterative value nor the current iterative gradient satisfies the preset convergence condition, numerically optimizing the iterative thermal state coefficients of the previous time step based on the current iterative gradient to obtain the iterative thermal state coefficients of that time step for iterative updates in subsequent time steps; if either the current iterative value or the current iterative gradient satisfies the preset convergence condition, the iteration stops, and the iterative thermal state coefficients at this point are output as corrected thermal state coefficients.

[0065] It is understood that in this embodiment, the initial thermal state coefficient is iteratively updated based on the thermodynamic physical residual to obtain a corrected thermal state coefficient that is closer to the actual thermal state of the target motor. This enables the thermal state evolution equation to be adaptively adjusted according to the actual operating state of the target motor, thereby improving the generalization ability of the thermal state evolution equation and enhancing the reliability and accuracy of the temperature prediction results.

[0066] Reference Figure 4 As shown, in one embodiment of this application, dynamic residual calculation is performed based on the initial thermal state coefficient, current operating parameters, and local nonlinear characteristics to obtain the thermodynamic physical residual of the thermal state evolution equation at the current moment, including: S322. Substitute the initial thermal state coefficient, current operating parameters, and local nonlinear characteristics into the thermal state evolution equation to calculate the baseline evolution rate of the initial thermal state coefficient.

[0067] S324. Calculate the time-series change of the initial thermal state coefficient based on the reference thermal state coefficient, and calculate the time-series change rate of the initial thermal state coefficient.

[0068] S326. The deviation between the baseline evolution rate and the time-series rate of change is taken as the thermodynamic residual.

[0069] Specifically, in the online phase, the nonlinear terms are replaced with the physical parameters of the target motor during actual operation to obtain the thermal state evolution equation for the online phase. For example, the thermal state evolution equation for the online phase can be expressed as: in, The physical parameters of the target motor during actual operation can be substituted with local nonlinear characteristics. The initial thermal state coefficient, current operating parameters, and local nonlinear characteristics are then substituted into the thermal state evolution equation for mechanism deduction, yielding the baseline evolution rate of the initial thermal state coefficient. It can be understood that the baseline evolution rate represents the theoretical thermal state change trend influenced by the actual thermophysical state of the target motor at the current moment and the current operating parameters, under the premise of strictly adhering to the inherent thermodynamic theorems of the target motor.

[0070] Furthermore, based on the reference thermal state coefficient of the target motor at the previous moment, the time-series variation of the initial thermal state coefficient is calculated to determine the trend of the initial thermal state coefficient changing with time, thus obtaining the time-series variation rate. Residual calculations are performed on the time-series variation rate and the reference evolution rate to determine the trend error between them, obtaining the thermodynamic physical residual. For example, the thermodynamic physical residual can be expressed as: in, The initial thermal state coefficient; The rate of change over time; These are the current operating parameters. It is understandable that... The baseline evolution rate represents the initial thermal state coefficient. Under the condition that the thermal state evolution equation is accurate and conforms to the physical laws, the value of the thermodynamic physical residual should always be equal to 0.

[0071] Reference Figure 5 As shown, in one embodiment of this application, the key nodes are obtained in the following manner: S312. Perform temperature coupling analysis on historical physical parameters based on historical global temperature data to obtain the physical temperature characteristic parameters corresponding to each historical physical parameter.

[0072] S314. Based on the physical temperature characteristic parameters, feature model extraction and modal energy calculation are performed to obtain nonlinear modal vectors and their corresponding nonlinear modal energies. The energy ratio of the nonlinear modal vectors is then selected based on the nonlinear modal energies to construct nonlinear spatial basis functions.

[0073] S316. Perform nonlinear residual calculation on all grid nodes of the target motor according to the nonlinear spatial basis function to obtain the nonlinear characteristic residual of all grid nodes, and perform node screening on all grid nodes according to the nonlinear characteristic residual to obtain the key nodes.

[0074] Specifically, in the offline phase, historical physical parameters at each historical moment are acquired. Temperature coupling analysis is performed on these historical physical parameters based on historical global temperature data from the mixed snapshot set to determine the thermal impact of historical global temperature data on the historical physical parameters at each historical moment, thus obtaining the corresponding physical temperature characteristic parameters for each historical physical parameter. The parameter types of the historical physical parameters can be the same as those of the current physical parameters, including but not limited to resistivity, thermal conductivity, and coefficient of thermal expansion. Correspondingly, the physical temperature characteristic parameters can include resistivity field distribution varying with temperature, or local convective heat transfer coefficient field distribution, etc.

[0075] In some embodiments, the temperature coupling analysis can be performed using finite element analysis software. In this case, the temperature coupling analysis is typically performed separately for multiple mesh nodes of the target motor in the global space, and the analysis results can be output according to the mesh node numbering order. Based on the above analysis results, the spatial index order of the mesh nodes is determined according to their spatial positions in the global space. The analysis results are then rearranged based on the spatial index order to obtain a snapshot matrix of nonlinear physical terms containing physical temperature characteristic parameters. It can be understood that each data point in the snapshot matrix of nonlinear physical terms represents the nonlinear source intensity of the mesh node in the global space at a specific historical moment, reflecting the degree to which the physical parameters are affected by temperature.

[0076] Furthermore, feature model extraction is performed on the snapshot matrix of nonlinear physical terms to extract mixed spatial features from historical global temperature data and historical physical parameters, resulting in multiple nonlinear mode vectors. Modal energy is then calculated for each nonlinear mode vector to obtain its corresponding nonlinear mode energy. Based on the energy proportion of the nonlinear mode vectors, nonlinear mode vectors that are significantly affected by temperature are selected, and nonlinear spatial basis functions are constructed using these nonlinear mode vectors.

[0077] For example, the specific method for feature model extraction and modal energy calculation can be Singular Value Decomposition (SVD). Singular value decomposition is performed on the snapshot matrix of the nonlinear physics term to obtain multiple singular values ​​and their corresponding singular vectors. The singular values ​​can serve as a measure of nonlinear modal energy, and the singular vectors can serve as nonlinear modal vectors. The singular values ​​are sorted in descending order, and the sum of squares of the multiple singular values ​​with higher values ​​is calculated. If the ratio of the sum of the squares of multiple singular values ​​to the total sum of all singular values ​​is greater than a preset singular value proportion, the number of singular values ​​from which the sum of squares is calculated is used as the upper limit for modal truncation. Nonlinear spatial basis functions are then constructed based on the singular vectors corresponding to these singular values. The preset singular value proportion can be 99%.

[0078] Further, nonlinear residuals are calculated for all grid nodes in the target motor based on the nonlinear spatial basis functions to obtain the nonlinear characteristic residuals of all grid nodes. All grid nodes are then sorted in descending order based on these nonlinear characteristic residuals, and a node selection process is performed to identify the grid nodes with the largest nonlinear characteristic residuals as the key nodes of the target motor. In some embodiments, the search method for key nodes can be a greedy algorithm, a clustering algorithm, a genetic algorithm, or a gradient-based optimization algorithm, etc. For example, a key node can be represented as: in, This represents the total number of critical nodes. It is understandable that during the online phase, processes such as temperature distribution prediction and mechanism evolution adjustment can be performed based on critical nodes. For example, initial thermal state coefficients can be output for critical nodes using thermal state evolution equations to predict the temperature field data of the target motor at the critical nodes, thereby reducing the computational load during the online phase and meeting the computing power limitations of automotive chips.

[0079] Reference Figure 6 As shown, in one embodiment of this application, the method further includes: S510. Predict future operating parameters based on current operating parameters.

[0080] S520. Based on the corrected thermal state coefficient and future operating parameters, the thermal state evolution equation is used to perform thermal state deduction and spatial mapping on the target motor to obtain the temperature evolution trajectory of the target motor in subsequent moments.

[0081] S530. Based on the temperature evolution trajectory and the tolerance limit curve of the target motor, a safety boundary search is performed on the target motor to obtain the remaining tolerance time of the target motor.

[0082] Specifically, within a future time window following the current moment, future operating conditions are predicted based on the current operating parameters at the current moment to obtain the future operating parameters of the target motor within the future time window. In some embodiments, the future operating condition prediction can be performed by keeping the current operating load constant, so that the future operating parameters are the same as the current operating parameters. In other embodiments, future operating parameters can also be predicted based on the current operating parameters using a linear fitting method. For example, the system state simulation process can be represented as follows: in, Parameters for future operating conditions; This is the correlation function between operating parameters and thermal state coefficients, which can be obtained from the thermal state evolution equation. The length of the future time window can be 10 seconds, 60 seconds, etc.

[0083] Furthermore, using the corrected thermal state coefficient as initial conditions, the system state is simulated based on future operating parameters using the thermal state evolution equation to predict the temperature changes of the target motor at key nodes, thus obtaining the temperature evolution trajectory of the target motor within a future time window. It can be understood that the process of simulating the system state and reconstructing the temperature field of the target motor using the thermal state evolution equation based on the corrected thermal state coefficient and future operating parameters to obtain the temperature evolution trajectory of the target motor at subsequent times is the same as the process of reconstructing the current temperature field of the target motor using the thermal state evolution equation and current operating parameters.

[0084] Furthermore, the characteristics of each component and material within the target motor are queried to obtain the tolerance limits and physical properties of each component and material, determining the highest tolerable temperature that the target motor can withstand without failure. Based on this highest tolerable temperature, the tolerance limit curve of the target motor is obtained. A limit comparison is performed between the temperature evolution trajectory and the tolerance limit curve, and a safety boundary search is conducted on the target motor based on the comparison results to obtain the remaining tolerance time. The tolerance limit curve can be a constant or a dynamic curve. A constant curve indicates that the highest tolerable temperature of the target motor does not change with time, while a dynamic curve indicates that the highest tolerable temperature of the target motor changes with time. The remaining tolerance time can be the length of time the target motor can continue to operate before the temperature distribution exceeds the highest tolerable temperature.

[0085] Understandably, if no temperature data in the temperature evolution trajectory exceeds the tolerance limit curve, it indicates that the temperature distribution within the future time window will not pose a safety risk to the target motor, and the remaining tolerance time of the target motor can cover the future time window. Conversely, if temperature data in the temperature evolution trajectory exceeds the tolerance limit curve, it indicates that the temperature distribution within the future time window may pose a safety risk to the target motor. The remaining tolerance time of the target motor is obtained based on the time points corresponding to the temperature data exceeding the tolerance limit curve in the temperature evolution trajectory.

[0086] For example, the remaining tolerance time can be obtained by cross-identifying the temperature evolution trajectory and the tolerance limit curve. If there is no overlap between the temperature evolution trajectory and the tolerance limit curve, the target motor can operate safely within a future time window. If there is an overlap between the temperature evolution trajectory and the tolerance limit curve, the time node corresponding to the intersection point is determined. In this case, the time interval from the current moment to that time node can be used as the remaining tolerance time of the target motor.

[0087] Reference Figure 7 As shown, this application designs a dual safety mechanism for the target motor, including a short-term prediction mechanism and a long-term optimization mechanism. The short-term prediction mechanism can be as described in this embodiment, which predicts the remaining tolerance time of the target motor within a future time window. After obtaining the remaining tolerance time, the remaining tolerance time is fed back to the motor controller of the target motor to prompt the motor controller to reduce the output power of the target motor or to plan a derating route for the target motor in advance.

[0088] The long-term optimization mechanism may include: for a vehicle equipped with a target motor, constructing a long-term window during the vehicle's steady-state operation phase, and calculating the error integral between the predicted results of the thermal state evolution equation and the reference observations within the long-term window to obtain the long-term error integral. The steady-state operation phase of the vehicle can be a phase where current and temperature changes are relatively gradual, and the long-term error integral can be the result of calculations for any component of the target motor, such as the average temperature of the stator resistance.

[0089] If the long-term error integral does not exceed a preset long-term threshold, the vehicle and target motor are deemed to be operating safely. If the long-term error integral exceeds the preset long-term threshold, an online identification algorithm is triggered for the target motor to construct an inverse heat conduction problem. Deviation calculations are performed based on the target motor's current operating parameters, current temperature field, and reference temperature field to obtain motor loss parameters. The thermal state evolution equation is then optimized and updated based on these parameters. The reference temperature field can be obtained from the average temperature of the target motor during steady-state operation. Specific methods for deviation calculation may include, but are not limited to, least squares method and Kalman filter parameter estimation.

[0090] Reference Figure 8 As shown in one embodiment of this application, the target motor is divided into multiple heat-sensitive zones; the method further includes: S540. For any heat-sensitive zone among multiple heat-sensitive zones, safety control settings are made according to the material characteristics of any heat-sensitive zone to obtain the temperature safety threshold of any heat-sensitive zone.

[0091] S550. Based on the remaining tolerance time and temperature safety threshold of any heat-sensitive zone, invoke the corresponding safety management strategy to perform safety management on any heat-sensitive zone.

[0092] The heat-sensitive zones can be divided based on the physical properties of the components and materials in the target motor. These physical properties can include attributes such as insulation class, and each heat-sensitive zone can be affected by temperature in a different way. For example, the heat-sensitive zones can include a first zone, a second zone, and a third zone. The first zone can be the winding end, which has high temperature resistance; the second zone can be inside the winding slot, which has difficulty in heat dissipation; and the third zone can be the area affected by the permanent magnet, which has low temperature resistance.

[0093] Specifically, refer to Figure 9 As shown, for any one of multiple heat-sensitive zones, the components and their materials are determined based on the zone's range. The maximum temperature these components can withstand, and their susceptibility to temperature changes, are then determined based on the material properties of these components, resulting in the heat tolerance characteristics of that heat-sensitive zone. Safety control settings are then implemented based on these heat tolerance characteristics to obtain the temperature safety threshold for that heat-sensitive zone.

[0094] After obtaining the temperature safety threshold, the temperature data of the current temperature field at each key node is extracted to obtain the key node temperature data. For any heat-sensitive partition, based on the partition range of that heat-sensitive partition, the key nodes contained in that heat-sensitive partition are taken as partition nodes, and the highest temperature data and its corresponding spatial location index of that heat-sensitive partition are determined based on the key node temperature data of the partition nodes. After obtaining the highest temperature data, the highest temperature data is compared with the temperature safety threshold of that heat-sensitive partition. If the highest temperature data exceeds the temperature safety threshold, it indicates that the temperature safety margin of that heat-sensitive partition has been exhausted, and an abnormal situation may occur.

[0095] Furthermore, if any heat-sensitive zone has remaining tolerance time, a safety strategy is matched based on the remaining tolerance time and temperature safety threshold. The safety management strategy that matches the remaining tolerance time and temperature safety threshold is dynamically invoked, and the target motor is managed according to the safety management strategy to reduce the safety risk of any heat-sensitive zone and ensure that the target motor can work normally.

[0096] For example, safety management strategies may include instantaneous release strategies, precise derating strategies, and active cooling strategies. An instantaneous release strategy is a safety strategy invoked when the remaining tolerance time is greater than 0 and the spatial index of the highest temperature data is located in a high-temperature tolerance zone. In this case, the motor controller temporarily suspends current limiting for the target motor, allowing it to maintain overload output for a short period until the remaining tolerance time returns to zero. A precise derating strategy is a safety strategy invoked when the remaining tolerance time is close to 0. In this case, the motor controller linearly or non-linearly reduces the torque and current commands of the target motor based on a preset derating slope for any heat-sensitive zone. An active cooling strategy is a safety strategy invoked when the spatial index of the highest temperature data is located in a cooling blind zone. In this case, the motor controller pre-increases the cooling water pump speed or fan duty cycle to improve the cooling effect in the target motor.

[0097] This application also provides several embodiments for thermal management of target motors using the motor temperature prediction method provided in this application. The first embodiment is applied to complex road conditions, such as when an oil-cooled motor is operating at low speed and high torque during uphill climbing or muddy terrain. In these conditions, the target motor speed is extremely low, the oil cooling effect is poor, and the permanent magnets face extremely high reverse magnetic fields, posing a risk of demagnetization. Related technologies typically only monitor some of the highest temperature data in the target motor. However, if the stator winding temperature is high but not exceeding the limit, while the permanent magnet temperature is close to the demagnetization threshold, these technologies may not trigger an alarm, leading to irreversible demagnetization of the permanent magnets and permanent damage to the motor.

[0098] In this embodiment, the target motor is divided into multiple heat-sensitive zones, including a winding zone and a magnet zone. The preset temperature resistance of the winding zone is 180°C, and that of the magnet zone is 140°C. When the spatial location index of the highest temperature data is detected in the winding zone, the magnet temperature is relatively low, allowing for a relaxation of the current limit on the target motor to allow the winding to operate at 175°C. When the target motor is detected to be operating at low speed or experiencing concentrated eddy current losses, the spatial location index of the highest temperature data may shift to the magnet zone. In this case, the winding temperature may be only 150°C, while the magnet temperature may have reached 135°C. In this situation, derating is performed in advance based on the preset temperature resistance of 140°C for the magnet zone to prevent demagnetization of the permanent magnets. It can be seen that this embodiment can provide differentiated protection based on the physical limits of different components, reducing the probability of damage to high-value components and avoiding premature performance limitations in non-critical zones, which could restrict the vehicle's passability and reliability under complex operating conditions.

[0099] The second embodiment applies to electric vehicle motors with a mileage exceeding a preset mileage, such as 150,000 kilometers. At this mileage, due to prolonged high-load operation, the vehicle's cooling system typically suffers from problems such as water pump wear, scale buildup in the cooling channels, or aging of the thermal grease. Related technologies for temperature distribution prediction are usually based on fixed parameters calibrated for new vehicles. When the vehicle's cooling system experiences performance degradation, such as decreased heat dissipation capacity, the current temperature field obtained from temperature distribution prediction using fixed parameters may be lower than the actual temperature. This results in the inability to predict the temperature rise in the target motor due to aging, leading to unpredictable overheating failures during long uphill climbs or heavy-load conditions.

[0100] In this embodiment, a long-term optimization mechanism is designed for the target motor. During the steady-state operation phase, such as urban elevated highway cruising, the temperature prediction results and reference results are compared for continuous deviation trends. For example, when the predicted result is consistently lower than the reference result by approximately 10°C, and 10°C exceeds a preset long-term threshold, the inverse problem is initiated. Combining the current coolant flow rate and temperature data of the target motor, it is identified that the current global equivalent convective heat transfer coefficient may have decayed to 75% of the factory value. At this time, deviation calculations are performed based on the current operating parameters of the target motor, the current temperature field, and the reference temperature field to obtain motor loss parameters as aging correction factors. The thermal state evolution equation is then optimized and updated, ensuring that the thermal state evolution equation accurately reflects the thermal state after the heat dissipation capacity has decreased. In this embodiment, after the thermal state evolution equation is optimized and updated, the vehicle undergoes a full-load hill climb test. During this process, the thermal state evolution equation can predict in advance that the temperature will exceed the limit and issues a derating request earlier than when the vehicle is new.

[0101] The third embodiment applies to extreme operating scenarios in high-performance electric vehicles. In such scenarios, the current of the target motor fluctuates drastically within a short period, for example, jumping from 0A to 600A instantaneously, resulting in a huge transient thermal shock. Related technologies typically use NTC sensors for temperature measurement. However, due to the thermal capacity of the packaging and the thermal resistance of the insulation layer, traditional NTC sensors usually exhibit a physical lag of 5-15 seconds in temperature response. When the NTC sensor reading reaches the alarm threshold, the actual temperature of the stator windings in the target motor often significantly exceeds the safe temperature threshold, leading to a reduction in the lifespan of the insulation layer or even burnout. In other cases, the motor controller is forced to reserve an excessively large safety margin for safety requirements, preventing the vehicle from outputting full power and causing a decrease in the vehicle's 0-100 km / h acceleration performance.

[0102] In this embodiment, the initial thermal state coefficient output by the thermal state evolution equation is subjected to real-time condition judgment, thereby enabling real-time calculation of nonlinear copper loss and reconstruction of the effective temperature boundary within the stator slot. At 0.5 seconds after vehicle launch, although the end sensor reading only increases by 2°C, the predicted result in the current temperature field may instantly surge by 30°C, and due to drastic current fluctuations, pure data prediction may result in overshoot. At this time, the trend residual of the initial thermal state coefficient is calculated based on the current operating parameters and current physical parameters to obtain the thermodynamic physical residual of the initial thermal state coefficient. Real-time condition judgment is then performed based on the thermodynamic physical residual to impose real-time constraints on the thermal state evolution equation, promptly adjusting the output thermal state coefficient to ensure the accuracy of the output 180°C peak temperature.

[0103] Furthermore, in this embodiment, system state simulation and safety boundary search can be performed within a future time window based on the vehicle's current full throttle state to determine the remaining tolerance time of the insulation layer in the target motor. At this point, based on the remaining tolerance time, an instantaneous release strategy can be adopted. It is recommended that the motor controller not immediately limit power, allowing the target motor to maintain maximum torque output, and only intervene with linear derating in the last 0.5 seconds to maximize the short-term potential of the target motor.

[0104] Accordingly, please refer to Figure 10 This application provides a motor temperature prediction device, which includes: The current operating condition acquisition module 1010 is used to acquire the current operating condition parameters of the target motor at the current moment.

[0105] The thermal state coefficient prediction module 1020 is used to substitute the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation to perform thermal state deduction and obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient of the target motor over time.

[0106] The physical residual correction module 1030 is used to adjust the initial thermal state coefficient online based on the optimization objective of minimizing the physical residual when the initial thermal state coefficient does not meet the preset physical residual conditions, so as to obtain the corrected thermal state coefficient. The temperature field reconstruction module 1040 is used to perform spatial mapping based on the corrected thermal state coefficient and thermal mode basis function to obtain the current temperature field of the target motor. The thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

[0107] In some optional implementations, the thermal state coefficient prediction module 1020 includes: The historical coefficient calculation unit is used to project historical global temperature data onto the low-dimensional space of the thermal model basis function to obtain the historical thermal state coefficients corresponding to the historical time of the historical global temperature data.

[0108] The training coefficient fitting unit is used to substitute historical operating condition parameters and historical thermal state coefficients into the thermal state evolution equation to perform data fitting on the training coefficients in the thermal state evolution equation, thereby completing the training process of the thermal state evolution equation.

[0109] In some optional implementations, the system state adjustment module 1030 includes: The data acquisition unit is used to acquire the local nonlinear characteristics of the target motor at key nodes, as well as the thermodynamic physical residuals of the thermal state evolution equation at the current moment. The key nodes are the grid nodes in the target motor that are sensitive to temperature changes and physical parameters, and the local nonlinear characteristics are the physical parameters at the key nodes.

[0110] The iterative optimization unit is used to iteratively optimize the initial thermal state coefficient with the goal of minimizing the thermodynamic physical residual, so as to obtain the corrected thermal state coefficient.

[0111] In some alternative implementations, the iterative optimization unit includes: The baseline evolution calculation subunit is used to substitute the initial thermal state coefficient, current operating parameters, and local nonlinear characteristics into the thermal state evolution equation to calculate the baseline evolution rate of the initial thermal state coefficient.

[0112] The time-series change calculation subunit is used to calculate the time-series change of the initial thermal state coefficient based on the reference thermal state coefficient, and to calculate the time-series change rate of the initial thermal state coefficient.

[0113] The deviation calculation sub-unit is used to treat the deviation between the baseline evolution rate and the time-series change rate as a thermodynamic residual.

[0114] In some optional implementations, the data acquisition unit includes: The temperature coupling analysis subunit is used to perform temperature coupling analysis on historical physical parameters based on historical global temperature data, and to obtain the physical temperature characteristic parameters corresponding to each historical physical parameter.

[0115] The energy proportion screening subunit is used to extract feature models and calculate modal energy based on physical temperature characteristic parameters, obtain nonlinear modal vectors and their corresponding nonlinear modal energies, and screen the nonlinear modal vectors by energy proportion based on the nonlinear modal energies in order to construct nonlinear spatial basis functions.

[0116] The key node screening sub-unit is used to perform nonlinear residual calculation on all grid nodes of the target motor according to the nonlinear spatial basis function, obtain the nonlinear characteristic residual of all grid nodes, and screen all grid nodes according to the nonlinear characteristic residual to obtain the key nodes.

[0117] In some alternative implementations, the device further includes a security management module, comprising: The operating condition prediction unit is used to predict future operating condition parameters at subsequent times based on the current operating condition parameters.

[0118] The state simulation unit is used to perform thermal state deduction and spatial mapping of the target motor based on the corrected thermal state coefficient and future operating parameters, and to obtain the temperature evolution trajectory of the target motor in subsequent time moments.

[0119] The boundary exploration unit is used to search for the safety boundary of the target motor based on the temperature evolution trajectory and the tolerance limit curve of the target motor, and to obtain the remaining tolerance time of the target motor.

[0120] In some optional implementations, the security management module further includes: The corresponding setting unit is used to set the safety control for any heat-sensitive zone among multiple heat-sensitive zones, based on the material characteristics of any heat-sensitive zone, to obtain the temperature safety threshold of any heat-sensitive zone.

[0121] The fine management unit is used to invoke the corresponding safety management strategy to perform safety management on any heat-sensitive zone based on the remaining tolerance time and temperature safety threshold of any heat-sensitive zone.

[0122] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0123] In this embodiment, the motor temperature prediction device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0124] Please see Figure 11 , Figure 11 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application, such as... Figure 11As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 11 Take a processor 10 as an example.

[0125] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0126] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0127] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0128] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0129] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0130] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0131] This application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method of any embodiment of this application.

[0132] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0133] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0134] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0135] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0136] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0137] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0138] It should also be noted that 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 a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0139] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0140] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

[0141] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method of motor temperature prediction, characterized by, The method includes: Obtain the current operating parameters of the target motor at the current moment; Substituting the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation, thermal state deduction is performed to obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient of the target motor over time. When the initial thermal state coefficient does not meet the preset physical residual condition, based on the optimization objective of minimizing the physical residual, the initial thermal state coefficient is adjusted online by mechanism evolution to obtain the corrected thermal state coefficient. This includes: acquiring the local nonlinear characteristics of the target motor at key nodes, and the thermodynamic physical residual of the thermal state evolution equation at the current moment; wherein, the key node is a grid node in the target motor that has temperature change sensitivity and physical parameter sensitivity, and the local nonlinear characteristics are the physical parameters at the key nodes; the thermodynamic physical residual is calculated by substituting the initial thermal state coefficient, the current operating condition parameters, and the local nonlinear characteristics into the thermal state evolution equation to calculate the baseline evolution rate of the initial thermal state coefficient; calculating the time-series change rate of the initial thermal state coefficient based on the baseline thermal state coefficient; using the deviation between the baseline evolution rate and the time-series change rate as the thermodynamic residual; and iteratively optimizing the initial thermal state coefficient with the optimization objective of minimizing the thermodynamic physical residual to obtain the corrected thermal state coefficient. The current temperature field of the target motor is obtained by spatial mapping based on the modified thermal state coefficient and thermal mode basis function; wherein the thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

2. The method according to claim 1, characterized in that, The thermal state evolution equation is trained in the following manner: The historical global temperature data is projected onto the low-dimensional space of the thermal model basis function to obtain the historical thermal state coefficients corresponding to the historical time of the historical global temperature data. The historical operating condition parameters and the historical thermal state coefficients are substituted into the thermal state evolution equation to perform data fitting on the training coefficients in the thermal state evolution equation, thereby completing the training process of the thermal state evolution equation.

3. The method according to claim 1, characterized in that, The key nodes are obtained in the following way: Based on the historical global temperature data, temperature coupling analysis is performed on the historical physical parameters to obtain the physical temperature characteristic parameters corresponding to each of the historical physical parameters. Based on the physical temperature characteristic parameters, feature model extraction and modal energy calculation are performed to obtain nonlinear modal vectors and their corresponding nonlinear modal energies. The energy ratio of the nonlinear modal vectors is then filtered based on the nonlinear modal energies to construct nonlinear spatial basis functions. Nonlinear residuals are calculated for all grid nodes in the target motor based on the nonlinear spatial basis function to obtain the nonlinear characteristic residuals of all grid nodes. Then, the key nodes are obtained by filtering all grid nodes based on the nonlinear characteristic residuals.

4. The method according to claim 1, characterized in that, The method further includes: Predict future operating parameters based on the current operating parameters; Based on the modified thermal state coefficient and the future operating parameters, the thermal state evolution equation is used to perform thermal state deduction and spatial mapping on the target motor to obtain the temperature evolution trajectory of the target motor at the subsequent time. Based on the temperature evolution trajectory and the tolerance limit curve of the target motor, a safety boundary search is performed on the target motor to obtain the remaining tolerance time of the target motor.

5. The method according to claim 4, characterized in that, The target motor is divided into multiple heat-sensitive zones; the method further includes: For any one of the plurality of heat-sensitive zones, a safety control setting is performed based on the material properties of the heat-sensitive zone to obtain the temperature safety threshold of the heat-sensitive zone. Based on the remaining tolerance time of any heat-sensitive partition and the temperature safety threshold, the corresponding safety management strategy is invoked to perform safety management on any heat-sensitive partition.

6. A motor temperature prediction device, characterized in that, The device includes: The current operating condition acquisition module is used to acquire the current operating condition parameters of the target motor at the current moment; The thermal state coefficient prediction module is used to substitute the current operating parameters and the reference thermal state coefficient of the previous moment into the thermal state evolution equation to perform thermal state deduction and obtain the initial thermal state coefficient of the target motor at the current moment; wherein, the thermal state evolution equation is used to characterize the dynamic evolution law of the thermal state coefficient of the target motor over time. A physical residual correction module is used to adjust the initial thermal state coefficient online based on the optimization objective of minimizing the physical residual when the initial thermal state coefficient does not meet the preset physical residual conditions, thereby obtaining a corrected thermal state coefficient. This includes: acquiring the local nonlinear characteristics of the target motor at key nodes, and the thermodynamic physical residual of the thermal state evolution equation at the current moment; wherein the key node is a grid node in the target motor that is sensitive to temperature changes and physical parameters, and the local nonlinear characteristics are the physical parameters at the key nodes; calculating the thermodynamic physical residual by substituting the initial thermal state coefficient, the current operating parameters, and the local nonlinear characteristics into the thermal state evolution equation to calculate the baseline evolution rate of the initial thermal state coefficient; calculating the time-series change rate of the initial thermal state coefficient based on the baseline thermal state coefficient; using the deviation between the baseline evolution rate and the time-series change rate as the thermodynamic residual; and iteratively optimizing the initial thermal state coefficient with the optimization objective of minimizing the thermodynamic physical residual to obtain the corrected thermal state coefficient. The temperature field reconstruction module is used to perform spatial mapping based on the corrected thermal state coefficient and the thermal mode basis function to obtain the current temperature field of the target motor; wherein, the thermal mode basis function is obtained by offline spatial dimensionality reduction extraction based on the historical global temperature data and historical operating parameters of the target motor.

7. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the method of any one of claims 1 to 5.

Images