Curvature-driven dynamic compensation motor multi-physical field modeling method

Abstract

The application discloses a curvature driving dynamic compensation motor multi-physical field modeling method, and relates to the field of motor simulation modeling.The method comprises the following steps: constructing a bidirectional coupling basic model of an electromagnetic field, a temperature field and a structural mechanics field; correcting thermal parameter drift and mechanical deformation error in the electromagnetic field through a dynamic parameter compensator; based on local curvature characteristics of a parameter curve, adopting a curvature driving bimodal adaptive switching mechanism to smooth the electromagnetic parameter curve, and generating a dynamic weight factor according to the curvature distribution characteristics to fuse the output results of the two types of modes; generating a parameter response surface based on a Kriging surrogate model to replace nonlinear equations for calculating key physical fields by finite elements; and outputting the joint simulation results of electromagnetic torque ripple spectrum, winding temperature distribution and rotor vibration mode.The application scheme can solve the three-way contradiction among precision, efficiency and stability in traditional motor modeling, and has high application value.

Description

Technical Field

[0001] This invention relates to the field of motor simulation modeling, and in particular to a multiphysics modeling method for curvature-driven dynamic compensation motors. Background Technology

[0002] With the widespread application of high-precision electromechanical systems in aerospace, high-end equipment, and new energy fields, multi-physics coupling modeling of motors has become a core means of optimizing electromechanical performance and predicting thermomechanical reliability. Traditional modeling methods mainly rely on single-physics finite element analysis and parameter transfer through offline data tables. However, under high-speed, variable operating conditions, and long-term operation, significant electromagnetic, thermal, and mechanical multi-field dynamic coupling effects exist inside the motor. For example, centrifugal deformation of the rotor in a high-speed motor can lead to distortion of the air gap magnetic field, and winding temperature rise can cause material property drift. Existing methods, by neglecting the real-time interaction mechanism of multiple fields, struggle to accurately capture nonlinear behavior under transient conditions. Furthermore, the development needs of industrial digital twins and predictive maintenance further require models to possess both high fidelity and real-time performance, necessitating breakthroughs in the computational power and accuracy bottlenecks of traditional offline modeling.

[0003] Existing technologies suffer from three main systemic defects: 1. Mainstream methods generate inductance and flux linkage parameter curves through finite element scanning and then create offline data tables after smoothing using fast Fourier transform or least squares methods. However, these tables cannot correct for time-varying parameter errors caused by thermal drift and mechanical deformation during lookup, leading to significant degradation of model accuracy under high-speed transient conditions. 2. Existing modeling methods often employ unidirectional parameter transfer, lacking a two-way real-time feedback loop encompassing electromagnetic, thermal, and force fields, causing errors to accumulate across multiple physical domains. Especially at high speeds, rotor dynamic instability and local overheating create positive feedback, exacerbating the risk of model divergence. 3. To improve real-time performance, some studies introduce surrogate models to replace finite element calculations. However, sample sampling strategies rely on uniform distribution or single sampling, failing to focus on densely populated areas of actual operating conditions, resulting in a sharp increase in prediction errors in critical operating condition intervals. Furthermore, traditional order reduction methods, by ignoring spatial harmonic saturation effects, struggle to meet high-precision control requirements. Therefore, based on these challenges, this invention proposes a multi-physics modeling method for curvature-driven dynamic compensation motors. Summary of the Invention

[0004] Purpose of the invention

[0005] To address the aforementioned issues, the present invention aims to provide a multiphysics modeling method for curvature-driven dynamic compensation motors. Through an innovative data post-processing mechanism, numerical noise in inductance and flux linkage curves is eliminated, and a high-fidelity lumped parameter model is constructed, thereby improving the simulation accuracy and computational stability of electrical systems and meeting the high reliability requirements of motor models in fields such as aviation and rail transportation.

[0006] Technical solution

[0007] To achieve the above objectives, this invention provides a multiphysics modeling method for curvature-driven dynamic compensation motors. This method constructs a finite element model based on motor structural parameters, extracts the original inductance and flux linkage parameter curves by scanning current combinations under different operating conditions using equivalent current vectors, and performs fast Fourier transform analysis on the parameter curves using a dual-path error filtering mechanism to extract DC components and main harmonic components. For permanent magnet flux linkages, they need to be reconstructed by weighting and sorting according to harmonic orders. The weighted least squares method is used to ensure the continuity of the first derivative of the curves and solve the phase distortion problem. The smoothed parameters are also stored in an offline data table, and circuit equations with windings and conductors as units are constructed in a circuit simulator. The nonlinear equations of the electrical system are solved simultaneously based on the nodal voltage method, and the weak coupling iteration of the mechanical and circuit models is achieved through electromagnetic torque and position parameters.

[0008] In a first aspect, the present invention provides a multiphysics modeling method for a curvature-driven dynamic compensation motor, comprising:

[0009] A two-way coupled basic model of electromagnetic field, temperature field and structural mechanical field is constructed, wherein the temperature field outputs the material property change to the electromagnetic field in real time, and the structural mechanical field outputs the rotor deformation to the electromagnetic field in real time.

[0010] The thermal parameter drift and mechanical deformation error in the electromagnetic field are corrected online by a dynamic parameter compensator.

[0011] Parametric response surfaces are generated based on the Kriging proxy model to replace the nonlinear equations of key physical fields calculated by the finite element method.

[0012] The joint simulation results of electromagnetic torque ripple spectrum, winding temperature distribution and rotor vibration modes are output.

[0013] Furthermore, the bidirectional coupling of the multiphysics field is achieved through a weak coupling iterative paradigm;

[0014] The electromagnetic field outputs the iron loss power density and copper loss power density to the temperature field, and the temperature field feeds back the thermal conductivity matrix to the electromagnetic field.

[0015] The structural mechanical field outputs the rotor dynamic eccentricity to the electromagnetic field, and the electromagnetic field feeds back the radial electromagnetic force density to the structural mechanical field;

[0016] By using a real-time cross-physics interaction mechanism, the accumulation of errors caused by unidirectional coupling is suppressed, thereby improving the consistency of co-simulation of electromechanical systems.

[0017] Furthermore, the dynamic parameter compensator dynamically adjusts the resistivity parameter based on the winding temperature rise. The mapping relationship is calibrated through material aging experiments. The air gap permeability function is reconstructed based on the rotor centrifugal deformation to suppress the nonlinear offset of inductance caused by magnetic saturation, thereby eliminating the nonlinear distortion of electromagnetic parameters caused by thermal expansion and centrifugal deformation, and ensuring the continuity of inductance and flux prediction across the entire operating range.

[0018] Furthermore, the input variables of the Kriging proxy model include the magnitude of the stator current space vector, the flow rate of the cooling medium, and the rotor angular velocity, and the output response includes the permanent magnet eddy current loss density, the leakage inductance of the end winding, and the equivalent stress of the rotor sheath.

[0019] The model employs an adaptive sample density strategy to generate response surfaces with confidence levels greater than a certain threshold in the parameter space, thereby achieving real-time solution of multiphysics equations while preserving the characteristics of spatial harmonics and saturation effects.

[0020] Furthermore, the adaptive sample density strategy increases the sample point density in regions where the parameter gradient change rate is greater than a set threshold, and eliminates redundant sample points in the parameter space based on Euclidean distance constraints, generating a high-confidence response surface with the minimum sample size, avoiding the risk of overfitting by the surrogate model, and ensuring the prediction generalization ability.

[0021] Furthermore, the method uses a dynamic resource allocation mechanism to freeze the finite element mesh update in the stator region with stable boundary conditions, and uses a heterogeneous parallel architecture to allocate tasks such as electromagnetic field eddy current loss calculation, temperature field heat conduction solution, and structural field modal analysis to independent computing units. This can significantly reduce the computational resource requirements of high-fidelity multiphysics models and meet the real-time constraints of online simulation on embedded platforms.

[0022] Furthermore, when the method is applied to a high-speed permanent magnet motor, the bearing support stiffness coefficient is dynamically adjusted based on the critical speed offset through the rotor dynamics stability constraint module. When the eddy current loss density exceeds the material tolerance threshold, an electromagnetic load reconfiguration command is triggered, which simultaneously suppresses the risk of high-speed rotor dynamic instability and local overheating, and expands the reliability boundary of the model under extreme conditions.

[0023] Furthermore, the method also includes a curvature-driven dual-mode switching mechanism, which, based on the local curvature characteristics of the parametric curve, uses the least squares method to ensure the continuity of the first derivative in the low curvature region, and switches to an improved fast Fourier transform method in the high curvature region to retain key harmonic components, as shown in the following formula:

[0024]

[0025] In the formula, The weights are the least squares weights. The least squares method is used for full weighting; The sensitivity coefficient is set through calibration experiments, and its typical value is [value missing]. ; The second derivative of the parametric curve is the rate of change of curvature.

[0026] Furthermore, the method also includes a dynamic relaxation factor node voltage iteration method, which dynamically adjusts the iterative relaxation factor by evaluating the ill-conditioned degree of the Jacobian matrix in real time, as shown in the following formula:

[0027]

[0028] In the formula, It is a dynamic relaxation factor; It is the minimum relaxation boundary; This represents the maximum relaxation boundary. The attenuation coefficient; Let be the condition number of the Jacobian matrix.

[0029] Secondly, the present invention also provides a multiphysics modeling system for a curvature-driven dynamic compensation motor, the system being based on the method described in the first aspect above, comprising:

[0030] Multi-field coupled solution engine, integrating electromagnetic finite element solver, thermal network model solver and rotor dynamics equation solver;

[0031] A dynamic parameter database is used to store material property matrices and geometric topology data that are updated in real time with temperature and deformation;

[0032] The proxy model acceleration interface is used to call the Kriging response surface to perform online prediction of electromagnetic parameters.

[0033] Furthermore, the dynamic parameter database is associated with a material degradation assessment unit for recording the thermal aging index of insulating materials and predicting the remaining lifespan, and a deformation accumulation monitoring unit for counting rotor fatigue load cycles and outputting the deformation accumulation. By integrating long-term service degradation factors, the prediction accuracy of full life cycle parameter compensation is improved.

[0034] Furthermore, when the system is deployed on an embedded platform, the proxy model data volume is compressed through a topology pruning algorithm, and the motor controller operating status data packets are synchronized in real time using a time-sensitive network protocol, thereby achieving high-concurrency real-time simulation of the vehicle edge device and providing online decision support for predictive maintenance of the motor.

[0035] This invention constructs a finite element model based on parameters such as the stator and rotor structure and winding arrangement of the motor. It generates original inductance and flux linkage parameter curves by scanning current combinations under multiple operating conditions using equivalent current vector scanning. Fast Fourier transform analysis is performed on the parameter curves, extracting the DC component and main harmonic components for the inductance parameters. For the permanent magnet flux linkage parameters, a harmonic order weighted sorting mechanism is introduced to retain key harmonic characteristics. The weighted least squares method is used to force the continuity of the first derivative of the parameter curves, ensuring phase fidelity. The smoothed parameters are stored in an offline data table. Circuit equations are constructed in a circuit simulator using windings and conductors as units. The nonlinear equations of the electrical system are solved synchronously based on the nodal voltage method. Weak coupling iteration between the mechanical and circuit models is achieved through electromagnetic torque and position parameters.

[0036] This method effectively filters out numerical noise introduced by mesh discretization in finite element scanning through a smoothing mechanism, suppresses harmonic error propagation during parameter curve differentiation, and significantly reduces torque pulsation simulation deviation. Smoothing parameter curves ensures the convergence of the nodal voltage method iteration and reduces the risk of divergence under strongly nonlinear conditions. At the same time, the weakly coupled architecture reduces computational complexity by separating circuit and mechanical equation solving. Balancing harmonic feature preservation and dynamic process fidelity, the model has significant advantages in high-reliability scenarios such as aerospace motor fault analysis and rail transit system simulation.

[0037] Beneficial effects

[0038] By implementing the multiphysics modeling method for a curvature-driven dynamic compensation motor provided by the present invention, the following technical effects are achieved:

[0039] (1) By performing fast Fourier transform analysis on the inductance and flux linkage curves and introducing a harmonic order weighted sorting mechanism, the intrinsic harmonic characteristics and numerical noise are effectively separated. This method solves the problem of non-smooth curves caused by mesh discretization in finite element scanning. It enables the reconstructed parameter curves to retain the essential characteristics of spatial harmonics and saturation effects while eliminating spurious high-frequency components introduced by numerical calculations. This suppresses the transmission of harmonic errors in differentiation operations from the source and significantly improves the accuracy consistency of the model under steady-state and transient conditions.

[0040] (2) By using mathematical optimization constraints, the first derivatives of the inductance and flux linkage parameter curves are forced to be continuous and smooth, thus reconstructing the time-domain continuity characteristics of the parameter curves. Compared with the frequency domain reconstruction method, this technique significantly improves the phase fidelity of the parameter curves. In the simulation of dynamic processes such as motor acceleration and stall, it effectively suppresses the torque pulsation calculation deviation caused by phase distortion, enhances the prediction accuracy of the lumped parameter model for electromechanical transient processes, and provides convergence guarantee for the nodal voltage method solver.

[0041] (3) By driving smooth mode adaptive switching through curvature characteristics, the least squares method is used to ensure derivative continuity in the low curvature region, and the improved fast Fourier transform is used to retain key harmonic components in the high curvature region, thus achieving synergistic optimization of harmonic fidelity and derivative continuity. This method eliminates the inherent limitations of traditional single methods in inductance curve processing, significantly reduces the electromagnetic torque harmonic distortion rate in transient operating condition simulation, greatly attenuates the oscillation amplitude of the node voltage iterative convergence trajectory, and fundamentally avoids the risk of harmonic amplification of numerical errors in derivative calculations, providing a high-fidelity parameter basis for electromechanical coupling simulation.

[0042] (4) An adaptive balance mechanism between the stability and efficiency of node voltage iteration was established by adjusting the relaxation factor in real time through the condition number of the Jacobian matrix. Under strongly nonlinear conditions, the method automatically shrinks the step size to suppress divergence based on the ill-conditioning of the equation set; under well-conditioned conditions, the step size is expanded to accelerate convergence. This method simultaneously improves the solution stability under extreme transient conditions and the computational efficiency under normal conditions, significantly reduces the sensitivity of the electromechanical system co-simulation to the initial value, and elevates the robustness of the modeling method to the level of industrial applications. Attached Figure Description

[0043] To make the above-described multiphysics modeling method for a curvature-driven dynamic compensation motor of the present invention more obvious and understandable, the accompanying drawings used in the specific embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating the method described in this application;

[0045] Figure 2 This diagram illustrates bidirectional coupling.

[0046] Figure 3 This diagram illustrates the smoothing process for the parameter curve. Detailed Implementation

[0047] Example 1:

[0048] A multiphysics modeling method for curvature-driven dynamic compensation motors is provided, the method flow is as follows: Figure 1 As shown, it includes:

[0049] A two-way coupled basic model of electromagnetic field, temperature field and structural mechanical field is constructed, wherein the temperature field outputs the material property change to the electromagnetic field in real time, and the structural mechanical field outputs the rotor deformation to the electromagnetic field in real time.

[0050] The thermal parameter drift and mechanical deformation error in the electromagnetic field are corrected online by a dynamic parameter compensator.

[0051] Parametric response surfaces are generated based on the Kriging proxy model to replace the nonlinear equations of key physical fields calculated by the finite element method.

[0052] The joint simulation results of electromagnetic torque ripple spectrum, winding temperature distribution and rotor vibration modes are output.

[0053] The multi-physics bidirectional coupling is achieved through a weak coupling iterative paradigm, such as... Figure 2 As shown;

[0054] The electromagnetic field outputs the iron loss power density and copper loss power density to the temperature field, and the temperature field feeds back the thermal conductivity matrix to the electromagnetic field.

[0055] The structural mechanical field outputs the rotor dynamic eccentricity to the electromagnetic field, and the electromagnetic field feeds back the radial electromagnetic force density to the structural mechanical field.

[0056] The dynamic parameter compensator dynamically adjusts the resistivity parameter based on the winding temperature rise and reconstructs the air gap permeability function according to the rotor centrifugal deformation, suppressing the inductance nonlinearity offset caused by magnetic saturation. Tests showed that under transient conditions where the motor's rated load and speed suddenly increased from 3000 rpm to 15000 rpm, the prediction error of the air gap magnetic flux density fundamental amplitude was reduced from ±12% using the traditional offline lookup method to within ±3% after using this compensator, and the inductance parameter nonlinearity offset was reduced by over 80%.

[0057] Parametric curve smoothing, such as Figure 3 As shown, the input variables of the Kriging proxy model include the magnitude of the stator current space vector, the flow rate of the cooling medium, and the rotor angular velocity, and the output response includes the permanent magnet eddy current loss density, the leakage inductance of the end winding, and the equivalent stress of the rotor sheath.

[0058] The model employs an adaptive sample density strategy to generate response surfaces in the parameter space with confidence levels greater than a certain threshold.

[0059] The adaptive sample density strategy increases the sample point density in regions where the parameter gradient change rate is greater than a set threshold, and removes redundant sample points in the parameter space based on Euclidean distance constraints.

[0060] The method uses a dynamic resource allocation mechanism to freeze the finite element mesh update in the stator region with stable boundary conditions, and uses a heterogeneous parallel architecture to allocate tasks such as electromagnetic field eddy current loss calculation, temperature field heat conduction solution, and structural field modal analysis to independent computing units.

[0061] The proposed method was validated using a high-speed permanent magnet motor model with a rated power of 200 kW. As shown in Table 1, under various typical operating conditions, this method significantly outperforms traditional methods in both computational accuracy and efficiency.

[0062] Table 1. Performance Comparison between this Method and Traditional Methods

[0063]

[0064] When the method is applied to a high-speed permanent magnet motor, the bearing support stiffness coefficient is dynamically adjusted based on the critical speed offset through the rotor dynamics stability constraint module. When the eddy current loss density exceeds the material tolerance threshold, an electromagnetic load reconfiguration command is triggered.

[0065] A multiphysics modeling system for curvature-driven dynamic compensation motors is also provided, the system comprising:

[0066] Multi-field coupled solution engine, integrating electromagnetic finite element solver, thermal network model solver and rotor dynamics equation solver;

[0067] A dynamic parameter database is used to store material property matrices and geometric topology data that are updated in real time with temperature and deformation;

[0068] The proxy model acceleration interface is used to call the Kriging response surface to perform online prediction of electromagnetic parameters.

[0069] The dynamic parameter database is associated with a material degradation assessment unit for recording the thermal aging index of insulating materials and predicting the remaining life, and a deformation accumulation monitoring unit for counting the number of rotor fatigue load cycles and outputting the cumulative deformation.

[0070] When the system is deployed on an embedded platform, the proxy model data volume is compressed through a topology pruning algorithm, and the motor controller operating status data packets are synchronized in real time using a time-sensitive networking protocol.

[0071] Example 2:

[0072] To address the contradiction between harmonic preservation and phase continuity in traditional smoothing methods, a curvature-driven dual-mode switching mechanism is proposed. Based on the local curvature characteristics of the parametric curve, the least squares method is used to ensure the continuity of the first derivative in the low curvature region, while the improved fast Fourier transform method is switched to preserve key harmonic components in the high curvature region. An adaptive weighting function is introduced to dynamically allocate the contribution weights of the two methods, as shown in the following formula:

[0073]

[0074] In the formula, The weights are the least squares weights. The least squares method is used for full weighting; The sensitivity coefficient is set through calibration experiments, and its typical value is [value missing]. ; The second derivative of the parametric curve is the rate of change of curvature.

[0075] Calculate the second derivative of the original inductance and flux linkage curves. ;

[0076] when Weighted least squares method is used for smoothing;

[0077] when Enable the improved Fast Fourier Transform method:

[0078] Harmonic component selection criteria: In the formula, For the first Second harmonic components; These are the Fast Fourier Transform coefficients; For harmonic order;

[0079] Only keep The component reconstruction curve;

[0080] Final curve In the formula, This is the final fusion curve; This is the output curve for the weighted least squares method. To reconstruct the output curve using Fast Fourier Transform.

[0081] Verification shows that, while achieving a similar average error to the above embodiments, compared to the traditional single FFT reconstruction method, the root mean square relative error of harmonic amplitude is reduced from 5.2% to 2.1%, a reduction of approximately 59.6% in harmonic amplitude error, while simultaneously increasing the iteration convergence speed by 28%. The results demonstrate that the curvature-driven dual-mode switching mechanism achieves synergistic optimization of harmonic fidelity and derivative continuity. In simulation experiments, an improved Fast Fourier Transform (FFT) method is used for the high curvature region of the A-phase winding inductance curve of the permanent magnet motor to accurately capture spatial harmonic components; in the low curvature region, a weighted least squares method is switched to ensure continuous and smooth first-order derivatives. This method significantly suppresses the inherent contradictions of traditional single smoothing methods, avoiding phase distortion caused by FFT reconstruction and overcoming the smoothing effect of the least squares method on higher harmonics. In the simulation of transient conditions such as motor start-up and stall, the processed parameter curves reduce the harmonic distortion rate of the electromagnetic torque ripple spectrum to below an acceptable threshold, and the oscillation amplitude of the node voltage iterative convergence trajectory is reduced year-on-year. This confirms that it eliminates the risk of harmonic amplification of numerical errors in the derivative calculation from the root, providing a fundamental guarantee for high-precision electromechanical coupling simulation.

[0082] Example 3:

[0083] To address the slow convergence of the nodal voltage method in strongly nonlinear systems, a dynamic relaxation factor based on the condition number of the Jacobian matrix is ​​proposed. The iterative relaxation factor is dynamically adjusted by evaluating the ill-conditioning of the Jacobian matrix in real time, as shown in the following formula:

[0084]

[0085] In the formula, It is a dynamic relaxation factor; It is the minimum relaxation boundary; This represents the maximum relaxation boundary. The attenuation coefficient; Let be the condition number of the Jacobian matrix.

[0086] Calculate in each iteration condition number;

[0087] like The condition is serious. Small step sizes prevent divergence;

[0088] like Good state Larger step sizes accelerate convergence;

[0089] Voltage update: In the formula, For the first The node voltage vector of the next iteration; For the first The node voltage vector of the next iteration; This represents the residual of the circuit equation.

[0090] The results of the dynamic relaxation factor node voltage iteration method are shown in Table 2.

[0091] Table 2. Summary of the effects of the dynamic relaxation factor nodal voltage iteration method

[0092] Operating conditions Traditional node voltage method Dynamic relaxation factor node voltage iteration method Rated steady state 35-step convergence 22-step convergence Short circuit fault transient divergence 58 steps to convergence Calculation time 100% 62%

[0093] Verification shows that, while achieving a similar average error to the aforementioned embodiments, the convergence rate under strongly nonlinear conditions is improved by 100%, and the overall simulation speed is improved by 38%. Experimental data demonstrate that the dynamic relaxation mechanism based on the condition number of the Jacobian matrix overcomes the convergence bottleneck of the nodal voltage method under strongly nonlinear conditions. Simulation experiments constructed a motor system model containing short-circuit faults and load mutations. The dynamic relaxation factor automatically adjusts according to the ill-conditioning of the equation set; in highly ill-conditioned conditions, the step size is contracted to suppress divergence, while in low ill-conditioned conditions, the step size is expanded to accelerate convergence. This method achieves a simultaneous leap in stability and efficiency under both rated steady-state and extreme transient conditions. Especially in benchmark tests compared to traditional fixed relaxation factors, it transforms short-circuit fault simulation from iterative divergence to stable convergence, and the iterative step size sequence under rated conditions exhibits an exponential decay characteristic. The improved smoothness of the electromagnetic torque and speed dynamic response curves verifies that this method significantly reduces the sensitivity of the solution process to initial values, enabling the robustness of multi-physics co-simulation to reach industrial applicability levels.

Claims

1. A multiphysics modeling method for a curvature-driven dynamic compensation motor, characterized in that, include: A two-way coupled basic model of electromagnetic field, temperature field and structural mechanical field is constructed, wherein the temperature field outputs the material property change to the electromagnetic field in real time, and the structural mechanical field outputs the rotor deformation to the electromagnetic field in real time. The resistivity parameters are dynamically adjusted based on the winding temperature rise, and the air gap permeability function is reconstructed based on the rotor centrifugal deformation, so as to correct the thermal parameter drift and mechanical deformation error in the electromagnetic field online. Based on the local curvature characteristics of the inductance parameter curve and / or flux linkage parameter curve, a curvature-driven dual-mode adaptive switching mechanism is used to smooth the parameter curve. In the low curvature region, a time-domain continuous optimization mode is used to reconstruct the curve, and in the high curvature region, a frequency-domain harmonic preservation mode is used to reconstruct the curve. Dynamic weighting factors are generated according to the curvature distribution characteristics to fuse the output results of the two modes. Parametric response surfaces are generated based on the Kriging proxy model to replace the nonlinear equations of key physical fields calculated by the finite element method. The joint simulation results of electromagnetic torque ripple spectrum, winding temperature distribution and rotor vibration modes are output.

2. The method according to claim 1, characterized in that: The bidirectional coupling of the multiphysics field is achieved through a weak coupling iterative paradigm; The electromagnetic field outputs the iron loss power density and copper loss power density to the temperature field, and the temperature field feeds back the thermal conductivity matrix to the electromagnetic field. The structural mechanical field outputs the rotor dynamic eccentricity to the electromagnetic field, and the electromagnetic field feeds back the radial electromagnetic force density to the structural mechanical field.

3. The method according to claim 1, characterized in that: The input variables of the Kriging proxy model include the magnitude of the stator current space vector, the flow rate of the cooling medium, and the rotor angular velocity. The output response includes the permanent magnet eddy current loss density, the leakage inductance of the end winding, and the equivalent stress of the rotor sheath. The model employs an adaptive sample density strategy to generate response surfaces in the parameter space with confidence levels greater than a certain threshold.

4. The method according to claim 3, characterized in that: The adaptive sample density strategy increases the sample point density in regions where the parameter gradient change rate is greater than a set threshold, and removes redundant sample points in the parameter space based on Euclidean distance constraints.

5. The method according to claim 1, characterized in that: The method uses a dynamic resource allocation mechanism to freeze the finite element mesh update in the stator region with stable boundary conditions, and uses a heterogeneous parallel architecture to allocate tasks such as electromagnetic field eddy current loss calculation, temperature field heat conduction solution, and structural field modal analysis to independent computing units.

6. The method according to claim 1, characterized in that: When the method is applied to a high-speed permanent magnet motor, the bearing support stiffness coefficient is dynamically adjusted based on the critical speed offset through the rotor dynamics stability constraint module. When the eddy current loss density exceeds the material tolerance threshold, an electromagnetic load reconfiguration command is triggered.

7. A multiphysics modeling system for a curvature-driven dynamic compensation motor, characterized in that: The system is implemented based on the method of any one of claims 1-6, comprising: Multi-field coupled solution engine, integrating electromagnetic finite element solver, thermal network model solver and rotor dynamics equation solver; A dynamic parameter database is used to store material property matrices and geometric topology data that are updated in real time with temperature and deformation; The proxy model acceleration interface is used to call the Kriging response surface to perform online prediction of electromagnetic parameters.

8. The system according to claim 7, characterized in that: The dynamic parameter database is associated with a material degradation assessment unit for recording the thermal aging index of insulating materials and predicting the remaining life, and a deformation accumulation monitoring unit for counting the number of rotor fatigue load cycles and outputting the cumulative deformation.

9. The system according to claim 7, characterized in that: When the system is deployed on an embedded platform, the proxy model data volume is compressed through a topology pruning algorithm, and the motor controller operating status data packets are synchronized in real time using a time-sensitive networking protocol.

Images