Torque ripple optimization method and device, equipment and storage medium
By acquiring historical motor data and optimizing the motor magnetic field using finite element simulation and deep learning models, the problem of output power fluctuation caused by motor torque pulsation in electric vehicles was solved, and the accuracy and stability of motor torque control were improved.
Patent Information
- Application Number
- CN202510977483.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-16
- Publication Date
- 2025-12-05
AI Technical Summary
In existing technologies for distributed drive systems in electric vehicles, motor torque pulsation causes output power fluctuations, affecting vehicle driving stability. Furthermore, adjusting the drive current waveform is complex and makes it difficult to improve the accuracy of torque control.
By acquiring historical motor design and operation data, using finite element simulation and deep learning models, the magnetic field data of the motor is optimized, the motor optimization model is trained, and the motor to be optimized is predicted and adjusted to reduce torque ripple frequency and improve torque control accuracy.
It achieves precise control of motor torque, reduces the complexity of current waveform, and improves motor driving stability and torque control accuracy.
Smart Images
Figure CN121077348A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of electric machines, in particular to a torque ripple optimization method, device, equipment and storage medium. BACKGROUND
[0002] In the field of electric vehicles, the control of motor torque can greatly affect the driving experience of the vehicle. Especially for distributed drive systems, the driving of the vehicle requires accurate control of the torque of each motor. When torque ripple occurs, it is easy to cause the output power of the motor of the vehicle to fluctuate, thereby affecting the driving stability of the vehicle. Currently, a common solution is to adjust the waveform of the driving current to reduce the generation of torque ripple by accurately adjusting the waveform of the driving current. However, this method has high requirements for the current waveform. In vehicle driving, due to distributed driving, the required current waveform is complex, which can easily reduce the accuracy of the control of the torque of the motor.
[0003] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY
[0004] The main purpose of the present application is to provide a torque ripple optimization method, device, equipment and storage medium, which aims to improve the accuracy of the control of the torque of the motor and reduce the frequency of torque ripple.
[0005] To achieve the above purpose, the present application provides a torque ripple optimization method, which comprises the following steps:
[0006] Obtain the design data of the historical motor and record the running data of the historical motor in the use process, wherein the running data includes torque ripple data;
[0007] Determine the corresponding simulation magnetic field data of the historical motor according to the design data and finite element simulation;
[0008] Determine the motor optimization model according to the simulation magnetic field data and the running data;
[0009] Determine the target motor according to the motor optimization model and the motor to be optimized.
[0010] Optionally, the step of determining the corresponding simulation magnetic field data of the historical motor according to the design data and finite element simulation comprises:
[0011] Generate a corresponding motor geometric model according to the design data, and set corresponding material properties for each component of the motor geometric model to obtain a virtual motor model;
[0012] Set a preset excitation corresponding to the virtual motor model;
[0013] According to the solver, the magnetic field corresponding to the virtual motor model is calculated to obtain the simulation magnetic field data.
[0014] Optionally, the step of determining the motor optimization model according to the simulation magnetic field data and the operation data comprises:
[0015] According to the simulation magnetic field, the corresponding magnetic field feature data of the historical motor is determined.
[0016] According to the operation data, torque operation feature data related to torque pulsation in the operation process is determined.
[0017] According to the magnetic field feature data and the torque operation feature data, a deep learning model is trained to obtain the motor optimization model.
[0018] Optionally, the step of determining the corresponding magnetic field feature data of the historical motor according to the simulation magnetic field comprises:
[0019] According to the simulation magnetic field and a preset viewing angle, a magnetic induction space distribution map corresponding to the historical motor is determined.
[0020] According to the simulation magnetic field, first type feature data of a target region is calculated, the first type feature data
[0021] According to the first type feature data and the magnetic induction space distribution map, the magnetic field feature data is determined.
[0022] Optionally, the deep learning model comprises a time sequence attention layer, a feature extraction layer and an embedding layer, and the step of training the deep learning model according to the magnetic field feature data and the torque operation feature data to obtain the motor optimization model comprises:
[0023] The first type feature data and the magnetic induction space distribution map are input into the deep learning model.
[0024] The first type feature data is transmitted to the embedding layer, and the magnetic induction space distribution map data is transmitted to the time sequence attention layer, after determining the image weight, the magnetic induction space distribution map data and the image weight are transmitted to the feature extraction layer.
[0025] The first output result processed by the feature extraction layer and the second output result of the embedding layer are output to a full connection layer, and the full connection layer outputs predicted torque feature data.
[0026] According to the predicted torque feature data and the torque operation feature data, the parameters of the deep learning model are adjusted to obtain the motor optimization model.
[0027] Optionally, the step of determining the target motor according to the motor optimization model and the motor to be optimized comprises:
[0028] calculating a torque ripple prediction result corresponding to the motor to be optimized according to the motor optimization model;
[0029] adjusting the motor to be optimized according to the torque ripple prediction result to obtain the target motor.
[0030] Optionally, the design data comprises permanent magnet system data, stator data, rotor assembly data and manufacturing tolerance parameters.
[0031] In addition, to achieve the above object, the present application also provides a torque ripple optimization device, characterized in that the torque ripple optimization device comprises:
[0032] an acquisition module, configured to acquire design data of a historical motor and record running data of the historical motor in use, wherein the running data comprises torque ripple data;
[0033] a simulation module, configured to determine simulation magnetic field data corresponding to the historical motor according to the design data and finite element simulation;
[0034] a training module, configured to determine a motor optimization model according to the simulation magnetic field data and the running data.
[0035] an optimization module, configured to determine a target motor according to the motor optimization model and a motor to be optimized.
[0036] In addition, to achieve the above object, the present application also provides a torque ripple optimization device, characterized in that the torque ripple optimization device comprises a memory, a processor and a torque ripple optimization program stored in the memory and executable on the processor, wherein the torque ripple optimization program is configured to implement the steps of the torque ripple optimization method according to any one of the above.
[0037] In addition, to achieve the above object, the present application also provides a storage medium, characterized in that the storage medium stores a torque ripple optimization program, wherein the torque ripple optimization program is executed by a processor to implement the steps of the torque ripple optimization method according to any one of the above.
[0038] The application provides a torque ripple optimization method, simulation magnetic field data corresponding to the historical motor are determined through the design data and finite element simulation; a motor optimization model is determined according to the simulation magnetic field data and the operation data; compared with optimizing the control current, the model can accurately train the precise prediction motor performance, and a target motor is determined according to the motor optimization model and the motor to be optimized, the motor is controlled, the complexity of the current waveform is reduced, and the precision of the motor torque control is improved. BRIEF DESCRIPTION OF DRAWINGS
[0039] Figure 1 is a torque ripple optimization device structure schematic diagram of a hardware running environment related to the embodiment scheme of the application.
[0040] Figure 2 is a flowchart of the first embodiment of the torque ripple optimization method of the application.
[0041] Figure 3 is a flowchart of the second embodiment of the torque ripple optimization method of the application.
[0042] Figure 4 is a flowchart of the third embodiment of the torque ripple optimization method of the application.
[0043] The implementation, functional features and advantages of the application will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION
[0044] It should be understood that the specific embodiments described herein are only used to explain the application, and are not used to limit the application.
[0045] Reference Figure 1 , Figure 1 is a torque ripple optimization device structure schematic diagram of a hardware running environment related to the embodiment scheme of the application.
[0046] As Figure 1As shown in the figure, the torque ripple optimization device can include a processor 1001, such as a central processing unit (CPU), a communication bus 1002, an interaction device 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to realize the connection communication between the components. The interaction device 1003 can include a display screen, an input unit such as a keyboard, and can also be connected to the communication bus through a standard wired interface or a wireless interface. The network interface 1004 can optionally include a standard wired interface or a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 can be a high-speed random access memory (RAM) memory or a stable non-volatile memory (NVM), such as a disk memory. The memory 1005 can also be a storage device independent of the aforementioned processor 1001.
[0047] Those skilled in the art can understand that Figure 1 The structure shown in the figure does not constitute a limitation on the torque ripple optimization device, and can include more or fewer components than the figure, or combine certain components, or different component arrangements.
[0048] As Figure 1 As shown in the figure, the memory 1005 as a storage medium can include an operating system, a data storage module, a network communication module, a user interface module, and a torque ripple optimization program.
[0049] In Figure 1 In the torque ripple optimization device shown in the figure, the network interface 1004 is mainly used for data communication with other devices; the interaction device 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the torque ripple optimization device of the application can be arranged in the torque ripple optimization device, and the torque ripple optimization device calls the torque ripple optimization program stored in the memory 1005 through the processor 1001, and executes the torque ripple optimization method provided by the embodiment of the application.
[0050] The embodiment of the application provides a torque ripple optimization method, which refers to Figure 2 , Figure 2 The flowchart of a torque ripple optimization method according to the first embodiment of the application.
[0051] In this embodiment, the torque ripple optimization method includes:
[0052] Step S1, obtaining design data of a historical motor, and recording operation data of the historical motor in use, the operation data including torque ripple data;
[0053] In the embodiment, the historical motor can be an axial magnetic field alternating pole permanent magnet motor. It should be noted that the historical motor is not limited to the type, and the historical motor obtained should be of the same type. In general, the historical motors are generally of the same type, and the design data of the historical motor existing before the current time is obtained, and the operation data of the historical motor in use is recorded, for example, whether there is torque ripple in use. In general, when torque ripple occurs, the current waveform will be distorted, and the operation data generally includes current harmonic data obtained by monitoring. In addition, the operation data can also include mechanical vibration data of the motor. In general, the motor is provided with a vibration sensor, and when torque ripple occurs, the amplitude of axial vibration will generally increase.
[0054] Step S2, determining simulation magnetic field data corresponding to the historical motor according to the design data and finite element simulation;
[0055] In the embodiment, the design data can be original motor design data. In addition to the structure data of the motor, the design data also includes material data of each component of the motor and standard current waveform of the motor in use. This is because the motor needs to avoid some magnetic field defects during design, so the corresponding optimized control current needs to be set. Therefore, the corresponding optimized current of different designs of the motor is not the same.
[0056] Step S3, determining a motor optimization model according to the simulation magnetic field data and the operation data;
[0057] It should be noted that in the embodiment, the number of historical motors is multiple, and each has corresponding design data. A common type of motor often has multiple different versions due to different optimization directions. The operation data is preferably obtained by a quality inspection test platform during testing of the historical motor, thereby ensuring the accuracy of the obtained data. Common tests include motor static performance test and motor dynamic performance test. Torque ripple is generally measured by a high-precision torque meter, noise data of the motor during operation is detected by setting a volume sensor in a semi-anechoic chamber, and vibration data can also be determined by an accelerometer. The simulation magnetic field data and the operation data are used to train a preset model to obtain a motor optimization model.
[0058] Step S4, determining a target motor according to the motor optimization model and a motor to be optimized.
[0059] The motor to be optimized herein is generally a newly developed motor that has not been produced and experimented, and in the design stage, the corresponding design process data of the motor to be optimized is input into the motor optimization model, so as to obtain the prediction result of the data model, and according to the prediction result, the design of the motor to be optimized is adjusted, so as to obtain the target motor.
[0060] In the embodiment, the simulation magnetic field data corresponding to the historical motor is determined through the design data and finite element simulation; the motor optimization model is determined according to the simulation magnetic field data and the operation data; compared with the optimization control current, the model for accurately predicting the motor performance can be accurately trained, and the target motor is determined according to the motor optimization model and the motor to be optimized, the complexity of the current waveform is reduced for the complex control of the motor, and the accuracy of the control of the motor torque is improved.
[0061] Further, based on the first embodiment, the second embodiment of the torque ripple optimization method of the application is proposed, in which Figure 3 The step of determining the simulation magnetic field data corresponding to the historical motor according to the design data and finite element simulation comprises:
[0062] In step S21, a corresponding motor geometric model is generated according to the design data, and corresponding material properties of each component of the motor geometric model are set to obtain a virtual motor model.
[0063] In the embodiment, specifically, the corresponding design data file is loaded into the corresponding analysis software, and generally, the process of loading can determine the material properties of each component according to the design data file, and in some embodiments, the properties of the components can also be directly adjusted in the analysis software. Common magnetic field analysis software can be ANSYS Maxwell, JMAG, etc. The above steps obtain the motor model.
[0064] In step S22, a preset excitation corresponding to the virtual motor model is set.
[0065] According to the virtual motor model, the corresponding electromagnetic boundary setting is determined, and the current and voltage are loaded. Generally, the current and voltage herein are the same as the actual operation data of the historical motor, for example, the amplitude, phase, frequency, etc. of the current. Thus, the waveform data of the excitation can be obtained.
[0066] In step S23, the magnetic field corresponding to the virtual motor model is calculated according to the solver, and the simulation magnetic field data is obtained.
[0067] Optionally, different mesh requirements are set for different regions using only the mesh partitioning method, specifically, high-density layered mesh is generally required for the air gap region, and curvature adaptive encryption mesh is required for the permanent magnet conversion.
[0068] In the embodiment, a virtual motor model is obtained by generating a corresponding motor geometric model according to the design data and setting corresponding material properties for each component of the motor geometric model; a preset excitation corresponding to the virtual motor model is set; a magnetic field corresponding to the virtual motor model is calculated according to a solver to obtain the simulation magnetic field data, thereby effectively analyzing a historical motor to obtain accurate magnetic field data, and thereby improving the accuracy of the magnetic field data.
[0069] Further, based on the first embodiment or the second embodiment, a third embodiment of the torque ripple optimization method of the application is proposed, which is described with reference to Figure 4 In the embodiment, the step of determining a motor optimization model according to the simulation magnetic field data and the operation data includes:
[0070] In step S31, the corresponding magnetic field feature data of the historical motor is determined according to the simulation magnetic field;
[0071] Specifically, the number of historical motors here is multiple, when the simulation magnetic field is obtained, the magnetic field feature data can be determined by extracting the data of the simulation magnetic field, preferably, in the embodiment, the features of each region of the simulation magnetic field can be extracted as the magnetic field feature data, in addition, the density data of the magnetic field distribution under different viewing angles can also be extracted.
[0072] In step S32, torque operation feature data related to torque ripple in the operation process is determined according to the operation data;
[0073] In the embodiment, it should be noted that the torque operation feature data here is not limited to the features of the torque data, but can also be other types of data, common operation data here is current data, vibration data and noise data, etc., the torque operation feature data is determined by extracting the features of the current harmonics, current frequency, current phase, vibration frequency and noise fluctuation in the above data,
[0074] In step S33, a deep learning model is trained according to the magnetic field feature data and the torque operation feature data to obtain the motor optimization model.
[0075] In the embodiment, preferably, in the embodiment, the magnetic field feature data and the torque operation feature data are taken as a data set, and the data set is divided into a training set and a verification set, in the embodiment, the motor optimization model is determined according to the steps.
[0076] Further, the step of determining the corresponding magnetic field feature data of the historical motor according to the simulated magnetic field comprises:
[0077] determining the corresponding magnetic induction space distribution map of the historical motor according to the simulated magnetic field and a preset viewing angle;
[0078] In this embodiment, the magnetic induction space distribution map corresponding to the historical motor is extracted by setting multiple standard preset viewing angles. It should be noted that, in order to ensure the consistency of data in the training process, the standard of electromagnetic field rendering should be the same in the analysis process. For example, the magnetic induction space distribution map can be a cloud chart. Specifically, the cloud chart actually divides the magnetic field distribution area into multiple small units, and different colors or gray values are assigned according to the size of the magnetic field physical quantity in each unit. The change of color or gray value reflects the change of the size of the magnetic field physical quantity. Generally speaking, the depth or warmth of color is used to distinguish the strength of the magnetic field. This representation method is the same for each magnetic induction space distribution map.
[0079] calculating first type feature data of a target region according to the simulated magnetic field, the first type feature data;
[0080] In this embodiment, the first type feature data of the historical motor in the pre-set target region can also be extracted. Common first type feature data can include air gap magnetic density, maximum magnetic density of the core, leakage coefficient, etc.
[0081] using the first type feature data and the magnetic induction space distribution map as the magnetic field feature data.
[0082] using the first type feature data and the magnetic induction space distribution map as the magnetic field feature data. In addition, it should be noted that, since the simulated magnetic field includes both static field and transient field, the magnetic field feature data herein can include the magnetic field feature data corresponding to the static field and the transient field respectively.
[0083] In this embodiment, by extracting the magnetic induction space distribution map and the first type feature data of the simulated magnetic field, the amount of data required to input the model can be effectively reduced, i.e. the overall data obtained by finite element analysis does not need to be input into the model for training, thereby reducing the computing resources required for training.
[0084] Further, based on any of the above embodiments, a fourth embodiment of the torque ripple optimization method of the present application is proposed. In this embodiment, the deep learning model comprises a time sequence attention layer, a feature extraction layer and an embedding layer. The step of training the deep learning model according to the magnetic field feature data and the torque running feature data to obtain the motor optimization model comprises:
[0085] inputting the first type of feature data and the magnetic induction spatial distribution map into the deep learning model;
[0086] The first type of feature data is passed to an embedding layer, and the magnetic induction spatial distribution map data is passed to a time attention layer. After determining the image weight, the magnetic induction spatial distribution map data and the image weight are passed to the feature extraction layer;
[0087] The first output result processed by the feature extraction layer and the second output result of the embedding layer are output to a fully connected layer, and the fully connected layer outputs predicted torque feature data.
[0088] According to the predicted torque feature data and the torque operation feature data, the parameters of the deep learning model are adjusted to obtain the motor optimization model.
[0089] In the embodiment, the deep learning model further includes an input layer that directly receives the input first type of feature data and the magnetic induction spatial distribution map. The data types received by the input layer are passed to different intermediate layers, for example, the first type of feature data is passed to an embedding layer, and the magnetic induction spatial distribution map data is passed to a time attention layer. In the embodiment, since the magnetic induction spatial distribution map is actually an image reflecting whether the magnetic field is uniform, it is necessary to determine the magnetic field map with unevenness in multiple images. X = (x1, C t1 ,x2, C2,..., x t ,C t ) is input data arranged according to the time of magnetic field change, x t is the magnetic induction spatial distribution map, and C t is the average color change gradient. Specifically, for each pixel in the magnetic induction spatial distribution map, the color difference between each pixel and its adjacent pixel is calculated. The color difference is specifically obtained by calculating the Euclidean distance of the color vectors of the two pixels.
[0090] The calculation formula in the time attention layer is:
[0091]
[0092] c tFor the image weight, tanh is an activation function, and the image weight is passed to a feature extraction layer after being calculated, and the feature extraction layer extracts the features of the image through the image weight. In this embodiment, the specific setting mode of the feature extraction layer is not limited, and a conventional bias calculation formula and an activation function are set. The type of the output of the model is the same as the torque running feature data, so that the parameters of the deep learning model can be adjusted according to the predicted torque feature data and the torque running feature data, and in this embodiment, the number of training sets accounts for 80% of the data set. Thus, the training data is improved.
[0093] In this embodiment, a deep learning model provided with a self-attention layer is used, and the training data used in this application includes the magnetic induction space distribution map. In the training process, the model focuses on the image with large magnetic field change by the average color change gradient of the image, thereby improving the training effect and further improving the accuracy of the model.
[0094] Further, based on any one of the above embodiments, a fifth embodiment of the torque ripple optimization method of the application is proposed. The step of determining a target motor according to the motor optimization model and the motor to be optimized includes:
[0095] calculating a torque ripple prediction result corresponding to the motor to be optimized according to the motor optimization model;
[0096] adjusting the motor to be optimized according to the torque ripple prediction result to obtain the target motor.
[0097] Specifically, the data corresponding to the motor to be optimized is input into the motor optimization model, the data output by the motor optimization model is obtained, and the motor to be optimized is adjusted according to the output data to obtain the target motor. Of course, in some cases, when the data output by the motor optimization model reaches the preset target design index, the design data of the motor to be optimized can be used as the design data of the target motor, that is, the motor to be optimized is used as the target motor.
[0098] Further, the historical motor is an axial magnetic field alternating pole permanent magnet motor, and the design data includes permanent magnet system data, stator data, rotor assembly data and manufacturing tolerance parameters.
[0099] In addition, the embodiment of the application also proposes a torque ripple optimization device, which comprises:
[0100] The acquisition module is configured to acquire the design data of the historical motor and record the running data of the historical motor in the use process, and the running data includes torque ripple data.
[0101] a simulation module configured to determine simulation magnetic field data corresponding to the historical motor according to the design data and finite element simulation;
[0102] a training module configured to determine a motor optimization model according to the simulation magnetic field data and the operation data.
[0103] an optimization module configured to determine a target motor according to the motor optimization model and a motor to be optimized.
[0104] In addition, the embodiment of the present application also proposes a torque ripple optimization device, which comprises a memory, a processor and a torque ripple optimization program stored in the memory and executable on the processor, and the torque ripple optimization program is configured to implement the steps of the embodiment of the torque ripple optimization method of any one of the above.
[0105] In addition, the embodiment of the present application also proposes a storage medium, which stores a torque ripple optimization program, and the torque ripple optimization program implements the steps of the embodiment of the torque ripple optimization method of any one of the above when executed by a processor.
[0106] It should be noted that in this document, the term "comprising" or "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or system. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, method, article or system including the element.
[0107] The above-mentioned embodiment numbers of the present application are only for description, not representing the advantages and disadvantages of the embodiments.
[0108] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by means of software and necessary general hardware platforms, of course, they can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server or network device, etc.) execute the method described in each embodiment of the present application.
[0109] The above merely describes the preferred embodiments of the present application, and is not intended to limit the patent scope of the present application, and any equivalent structure or equivalent process conversion, or direct or indirect application in other related technical fields, which are made by using the content of the present application specification and drawings, are also included in the patent protection scope of the present application.
Claims
1. A method of optimizing torque pulsation, characterized by, The optimization method of the torque ripple comprises the following steps: Obtain the design data of a historical motor and record the operation data of the historical motor during use, the operation data comprising torque ripple data; Determine the simulated magnetic field data corresponding to the historical motor according to the design data and finite element simulation; Determine a motor optimization model according to the simulated magnetic field data and the operation data; Determine a target motor according to the motor optimization model and a motor to be optimized.
2. The method of claim 1, wherein, The step of determining the simulated magnetic field data corresponding to the historical motor according to the design data and finite element simulation comprises: Generate a corresponding motor geometric model according to the design data, and set corresponding material properties for each component of the motor geometric model to obtain a virtual motor model; Set a preset excitation corresponding to the virtual motor model; Calculate the magnetic field corresponding to the virtual motor model according to a solver to obtain the simulated magnetic field data.
3. The method of claim 1, wherein the step of determining the torque ripple is performed by a computer program. The step of determining a motor optimization model according to the simulated magnetic field data and the operation data comprises: Determine the magnetic field feature data corresponding to the historical motor according to the simulated magnetic field; Determine the torque operation feature data related to torque ripple during operation according to the operation data; Train a deep learning model according to the magnetic field feature data and the torque operation feature data to obtain the motor optimization model.
4. The method of claim 3, wherein the step of determining the torque ripple is performed by a computer program. The step of determining the magnetic field feature data corresponding to the historical motor according to the simulated magnetic field comprises: Determine the magnetic induction spatial distribution map corresponding to the historical motor according to the simulated magnetic field and a preset viewing angle; Calculate the first type of feature data of a target region according to the simulated magnetic field, the first type of feature data Take the first type of feature data and the magnetic induction spatial distribution map as the magnetic field feature data.
5. The method of claim 4, wherein the step of determining the torque ripple is performed by the steps of: determining a torque ripple of the motor; and determining a torque ripple of the motor with the motor operating at a different speed. The deep learning model comprises a time sequence attention layer, a feature extraction layer and an embedding layer, and the step of training a deep learning model according to the magnetic field feature data and the torque operation feature data to obtain the motor optimization model comprises: Input the first type of feature data and the magnetic induction spatial distribution map into the deep learning model; Pass the first type of feature data to the embedding layer and pass the magnetic induction spatial distribution map data to the time sequence attention layer, determine the image weight after the magnetic induction spatial distribution map data and the image weight are passed to the feature extraction layer; Output the first output result processed by the feature extraction layer and the second output result of the embedding layer to a fully connected layer, and the fully connected layer outputs predicted torque feature data. Adjust the parameters of the deep learning model according to the predicted torque feature data and the torque operation feature data to obtain the motor optimization model.
6. The method of optimizing torque pulsation according to any one of claims 1 to 5, characterized in that, The step of determining a target motor according to the motor optimization model and a motor to be optimized comprises: Calculate the torque ripple prediction result corresponding to the motor to be optimized according to the motor optimization model; Adjust the motor to be optimized according to the torque ripple prediction result to obtain the target motor.
7. The method of optimizing torque pulses according to any one of claims 1 to 5, characterized in that, The design data comprises permanent magnet system data, stator data, rotor assembly data and manufacturing tolerance parameters.
8. A device for optimizing torque pulsations, characterized in that The optimization device of the torque ripple comprises: An acquisition module is configured to acquire design data of a historical motor and record operation data of the historical motor during use, the operation data including torque ripple data; An simulation module is configured to determine simulation magnetic field data corresponding to the historical motor according to the design data and finite element simulation; A training module is configured to determine a motor optimization model according to the simulation magnetic field data and the operation data. An optimization module is configured to determine a target motor according to the motor optimization model and a motor to be optimized.
9. A torque ripple optimization device characterized by, The torque ripple optimization device includes a memory, a processor, and a torque ripple optimization program stored in the memory and executable on the processor, and the torque ripple optimization program is configured to implement the steps of the torque ripple optimization method according to any one of claims 1 to 7.
10. A storage medium, characterized by The storage medium stores a torque ripple optimization program, and the torque ripple optimization program is executed by the processor to implement the steps of the torque ripple optimization method according to any one of claims 1 to 7.