Motor performance index prediction method and device
By establishing a performance index prediction model based on the air gap magnetic flux density harmonic amplitude and phase labels of design parameter samples, the problem of inaccurate motor performance index prediction is solved, achieving high-precision and fast motor performance analysis, which is applicable to various motor structures.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CRRC IND INST CO LTD
- Filing Date
- 2025-12-30
- Publication Date
- 2026-05-08
AI Technical Summary
Existing motor performance prediction models have weak generalization ability and inaccurate predictions, failing to meet high-precision design requirements. In particular, traditional methods struggle to achieve rapid comparison of multiple schemes and optimization of motor design parameters, especially in high-speed and high-precision control applications.
By obtaining the design parameters of the motor, a performance index prediction model is established based on the design parameter samples and amplitude and phase labels of the air gap magnetic flux density harmonics. The motor performance index is calculated using the amplitude and phase of the radial and tangential magnetic flux density harmonics, and a neural network model is used for training and prediction.
It achieves accurate prediction of motor performance indicators, improves the model's generalization ability, shortens analysis time, is applicable to various motor structures, and meets high-precision design requirements.
Smart Images
Figure CN121997565A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of motor technology, and in particular to a method and apparatus for predicting motor performance indicators. Background Technology
[0002] In modern industrial fields such as rail transit equipment, low-altitude aircraft, new energy vehicles, precision transmission equipment, and smart home appliances, electric motors, as core power components, directly impact the competitiveness of the entire product due to their operational stability, energy efficiency, and noise and vibration characteristics. As electric motor applications continue to evolve towards higher efficiency, smaller size, and greater precision, the requirements for core performance indicators such as no-load back EMF and cogging torque are increasing, especially in high-speed and high-precision control applications, where even minor deviations in performance indicators can lead to system malfunctions. Therefore, accurate prediction of core performance indicators is essential during the design phase.
[0003] Traditional performance index prediction methods have significant shortcomings. Although finite element simulation can reflect the electromagnetic field distribution in detail, it is complex to model and time-consuming to calculate, making it difficult to meet the needs of rapid comparison of multiple schemes and optimization of motor design parameters. Although analytical calculation is more efficient, it usually requires simplification of the model and neglect of factors such as higher harmonics and edge effects, which limits the prediction accuracy and cannot support high-precision design requirements.
[0004] Existing improvement schemes still face key technical bottlenecks. Methods based on empirical formulas have poor universality and require refitting for different motor structures, resulting in high costs and limited accuracy. Traditional machine learning models, such as linear regression, attempt to establish a mapping between motor design parameters (such as geometric parameters and material parameters) and performance indicators, but fail to deeply characterize the harmonic features of the air gap magnetic field and do not adequately depict the amplitude and phase coupling effects of multi-order harmonics, leading to weak model generalization ability and large prediction errors in the range of unknown design parameters. Summary of the Invention
[0005] This invention provides a method and apparatus for predicting motor performance indicators, which solves the problems of weak generalization ability and inaccurate prediction in existing motor performance indicator prediction models.
[0006] This invention provides a method for predicting motor performance indicators, comprising the following steps: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters; The design parameters are input into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model. The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics. The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0007] According to the method for predicting motor performance indicators provided by the present invention, the amplitude and phase of the air gap magnetic flux density harmonics include: the amplitude and phase of the radial magnetic flux density harmonics, and / or the amplitude and phase of the tangential magnetic flux density harmonics. The amplitude and phase of the radial flux density harmonics include: the amplitude and phase of the fundamental order related to the radial flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics. The amplitude and phase of the tangential flux density harmonics include: the amplitude and phase of the fundamental order related to the tangential flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
[0008] According to the present invention, a method for predicting motor performance indicators is provided, wherein the performance indicator prediction model includes: a first prediction model and / or a second prediction model; The first prediction model is used to predict the amplitude and phase of the radial flux density harmonics based on the design parameters; The second prediction model is used to predict the amplitude and phase of the tangential flux density harmonics based on the design parameters.
[0009] According to the present invention, a method for predicting motor performance indicators is provided, wherein the performance indicators include at least one of the following: no-load back electromotive force waveform, cogging torque waveform, core loss, permanent magnet eddy current loss, electromagnetic excitation force, and vibration noise. The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics, including at least one of the following calculation methods: Based on the amplitude and phase of the radial flux density harmonics, the radial air gap flux density distribution function is synthesized. The radial air gap flux density distribution function is calculated by integration and differentiation to obtain the no-load back electromotive force waveform and effective value of the motor. By combining the harmonic characteristics of radial and tangential magnetic flux densities, the cogging torque waveform and its peak value of the motor are calculated using the Maxwell tensor method. Based on the amplitude and phase of the radial flux density harmonics, the core loss in the stator and rotor cores is calculated using the Steinmetz or Bessel function decomposition model. Based on the amplitude and frequency of the time-varying radial magnetic flux density harmonics induced in the permanent magnet region, combined with the conductivity and segmentation of the permanent magnet, the permanent magnet eddy current loss in the permanent magnet can be calculated. By combining the amplitudes and phases of the radial and tangential magnetic flux density harmonics, the radial electromagnetic force density distribution acting on the stator teeth is calculated according to Maxwell's stress formula. The radial electromagnetic force density distribution is then subjected to two-dimensional Fourier decomposition in the spatial and time domains to obtain the force wave amplitudes of each order and frequency. Based on the force wave amplitudes, the electromagnetic vibration and vibration noise of the motor are evaluated.
[0010] According to the method for predicting motor performance indicators provided by the present invention, the training method of the performance indicator prediction model is as follows: Obtain the design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics; The design parameter samples are input into the initial prediction model to obtain the amplitude and phase prediction values of the air gap magnetic flux density harmonics output by the initial prediction model. The amplitude prediction value and phase prediction value, along with the corresponding amplitude label and phase label, are substituted into the loss function. When the loss function is not convergent, backpropagation is performed to adjust the model parameters of the initial prediction model until the loss function converges or the preset number of iterations is reached, thus obtaining the trained performance index prediction model.
[0011] According to the present invention, a method for predicting motor performance indicators includes obtaining design parameter samples and amplitude and phase labels of the corresponding air gap flux density harmonics, comprising: The original design data of the motor is sampled to obtain design parameter samples with different parameter combinations; The air gap magnetic flux density data sample was obtained by performing finite element simulation on the design parameter sample. Based on the theory of air gap magnetic field adjustment, a fast Fourier transform is performed on the air gap magnetic flux density data sample to obtain the amplitude and phase labels of the air gap magnetic flux density harmonics.
[0012] According to the method for predicting motor performance indicators provided by the present invention, after obtaining the amplitude and phase labels of the air gap flux density harmonics, the method further includes: For each set of design parameter samples, the top N largest amplitude labels and corresponding N phase labels are selected from all the amplitude and phase labels of the corresponding air gap magnetic flux density harmonics to train the performance index prediction model, where N is greater than or equal to 2.
[0013] The present invention also provides a motor performance index prediction device, comprising the following modules: The design parameter acquisition module is used to acquire the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters. The model prediction module is used to input the design parameters into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model. The performance index calculation module is used to calculate the motor's performance index based on the amplitude and phase of the air gap magnetic flux density harmonics. The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the program to implement the motor performance index prediction method as described above.
[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the motor performance index prediction method as described above.
[0016] The method and apparatus for predicting motor performance indicators provided by this invention input various design parameters into a performance indicator prediction model. Since the performance indicator prediction model is trained based on design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics, the model establishes a mapping relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics. This mapping relationship can accurately reflect the complex nonlinear coupling relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics, improving the model's generalization ability. Therefore, it can accurately predict the amplitude and phase of the air gap flux density harmonics based on the design parameters, and then accurately calculate the motor's performance indicators based on the amplitude and phase of the air gap flux density harmonics, thereby achieving accurate prediction of motor performance indicators. Attached Figure Description
[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0018] Figure 1 This is a flowchart illustrating the method for predicting motor performance indicators provided by the present invention.
[0019] Figure 2 This is a schematic diagram illustrating the amplitude and phase prediction accuracy analysis of radial flux density harmonics in the motor performance index prediction method provided by this invention.
[0020] Figure 3 This is a schematic diagram illustrating the accuracy analysis of amplitude and phase prediction of tangential flux density harmonics in the motor performance index prediction method provided by this invention.
[0021] Figure 4 This is a schematic diagram comparing the no-load back EMF waveform predicted in the motor performance index prediction method provided by this invention with the no-load back EMF waveform obtained from finite element simulation.
[0022] Figure 5 This is a schematic diagram comparing the cogging torque waveform predicted in the motor performance index prediction method provided by this invention with the cogging torque waveform obtained from finite element simulation.
[0023] Figure 6 This is a schematic diagram of the motor performance index prediction device provided by the present invention.
[0024] Figure 7 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation
[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0026] The various performance indicators of a motor are subject to complex nonlinear coupling relationships with geometric parameters such as stator inner diameter, permanent magnet thickness, and air gap width, which are essentially determined by the distribution characteristics of the air gap magnetic field. However, most existing machine learning models directly establish the mapping relationship between the motor's design parameters and various performance indicators, failing to deeply characterize the harmonic characteristics of the air gap magnetic field. Furthermore, the air gap magnetic flux density contains both the fundamental wave and multiple higher harmonics, and the coupling effects between these harmonic components further increase the difficulty of performance prediction. Consequently, existing machine learning models cannot deeply characterize the harmonic features of the air gap magnetic field, lack sufficient characterization of the amplitude and phase coupling effects of multiple harmonics, exhibit weak generalization ability, and show significant prediction errors in the range of unknown design parameters.
[0027] To address the aforementioned technical problems, this invention provides a method for predicting motor performance indicators, the specific process of which is as follows: Figure 1 As shown, it includes steps S110 to S130.
[0028] Step S110: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters.
[0029] The geometric parameters of the motor include key parameters such as permanent magnet thickness, stator inner diameter, stator tooth width, stator slot opening, pole arc coefficient, and air gap width.
[0030] Material parameters are the material parameters of the materials selected for each component when manufacturing an electric motor.
[0031] Environmental parameters refer to parameters such as temperature and humidity of the motor's operating environment. For example, a certain motor is specifically designed for environments below 0°C.
[0032] Operating parameters refer to parameters such as current and voltage of the motor when it is working.
[0033] The above design parameters have been assigned appropriate values or ranges during the motor design stage. Different values of different design parameters will result in different motor design schemes. That is, each different motor design scheme corresponds to a set of design parameters with different values. Under different motor design schemes, the motor will achieve different performance indicators.
[0034] In this step, the motor's design parameters can be obtained by accessing a pre-built design parameter data table. It's important to understand that in this table, the values for each design parameter in each row are not identical, and each row corresponds to a specific motor design scheme.
[0035] Step S120: Input the design parameters into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonics of the motor output by the performance index prediction model. The performance index prediction model is trained based on design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics.
[0036] Specifically, the design parameters are input into the performance index prediction model, which means inputting the design parameters corresponding to any motor design scheme into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonics of the motor output by the performance index prediction model. Since the performance index prediction model is trained based on design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics, this performance index prediction model establishes a mapping relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics. It can accurately reflect the complex nonlinear coupling relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics, improve the model's generalization ability, and thus accurately predict the amplitude and phase of the air gap flux density harmonics based on the design parameters.
[0037] Step S130: Calculate the motor's performance indicators based on the amplitude and phase of the air gap magnetic flux density harmonics. Since the amplitude and phase of the air gap magnetic flux density harmonics were accurately predicted in step S120, the motor's performance indicators can be accurately calculated based on the amplitude and phase of the air gap magnetic flux density harmonics.
[0038] It should be noted that different performance indicators are calculated in different ways, and for a certain performance indicator, different design parameters have different contributions to its influence. Therefore, when predicting a certain performance indicator, the design parameters that contribute more to the performance indicator can be selected to train and predict the performance indicator prediction model.
[0039] The motor performance index prediction method in this embodiment inputs various design parameters into a performance index prediction model. Since the performance index prediction model is trained based on design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics, it establishes a mapping relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics. This mapping relationship can accurately reflect the complex nonlinear coupling relationship between the motor's design parameters and the amplitude and phase of the air gap flux density harmonics, improving the model's generalization ability. Thus, it can accurately predict the amplitude and phase of the air gap flux density harmonics based on the design parameters, and then accurately calculate the motor's performance index based on the amplitude and phase of the air gap flux density harmonics, thereby achieving accurate prediction of motor performance indexes.
[0040] Taking surface-mounted permanent magnet synchronous motors (PMSMs) as an example, they are widely used in many fields due to their compact structure, high power density, and good control performance. However, their air gap magnetic field is more significantly affected by the coupling effect of the permanent magnet topology and the stator slot effect, resulting in more complex harmonic components. Traditional methods struggle to accurately capture their magnetic field characteristics. Therefore, this embodiment proposes a motor performance index prediction method, which is particularly suitable for surface-mounted PMSMs. By accurately extracting the harmonic characteristics of the air gap magnetic flux density and constructing an efficient nonlinear mapping model, it achieves rapid and high-precision conversion from the design parameters of the surface-mounted PMSM to the air gap magnetic field characteristics and core performance, effectively shortening the motor performance index analysis time and providing technical support for motor design optimization and rapid parameter iteration. Because this performance index prediction model has strong generalization ability, it is applicable to surface-mounted inner and outer rotor structures and can be further extended to various rotor topologies such as built-in delta, single V-type, and double V-type. Furthermore, this performance index prediction model can also be applied to the performance prediction of various permanent magnet synchronous motors and asynchronous motors, such as those used in rail transit, automobiles, and aircraft.
[0041] In some embodiments, the amplitude and phase of the air gap magnetic flux density harmonics include: the amplitude and phase of the radial magnetic flux density harmonics, and / or the amplitude and phase of the tangential magnetic flux density harmonics.
[0042] The amplitude and phase of the radial flux density harmonics include: the amplitude and phase of the fundamental order related to the radial flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
[0043] The amplitude and phase of the tangential flux density harmonics include: the amplitude and phase of the fundamental order related to the tangential flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
[0044] Since different performance indicators are related to the amplitude and phase of radial flux density harmonics, or the amplitude and phase of tangential flux density harmonics, or both, and since both types of flux density harmonics contain fundamental waves and multiple higher-order harmonics, the coupling effect between the harmonic components further increases the difficulty of performance prediction. Therefore, in this embodiment, both radial and tangential flux density harmonics include the aforementioned multiple amplitudes and phases. That is, the performance indicator prediction model can predict the amplitude and phase corresponding to different orders of the two types of flux density harmonics respectively, achieving accurate prediction of the amplitude and phase corresponding to different orders of the two types of flux density harmonics. Of course, during the training phase of the performance indicator prediction model, the amplitude and phase labels of the air gap flux density harmonics also include the labels of the amplitude and phase corresponding to different orders of the radial and tangential flux density harmonics, respectively.
[0045] In order to achieve accurate prediction of the amplitude and phase corresponding to different orders of radial and tangential magnetic flux density harmonics, in some embodiments, the performance index prediction model includes: a first prediction model and / or a second prediction model.
[0046] The first prediction model is used to predict the amplitude and phase of the radial flux density harmonics based on the design parameters.
[0047] The second prediction model is used to predict the amplitude and phase of the tangential flux density harmonics based on the design parameters.
[0048] In this embodiment, the performance index prediction model is divided into two prediction models. The two prediction models do not interfere with each other and predict the amplitude and phase of the radial magnetic flux density harmonics and the amplitude and phase of the tangential magnetic flux density harmonics, respectively.
[0049] In some embodiments, the performance indicators include at least one of the following: no-load back EMF waveform of the motor, cogging torque waveform, core loss, permanent magnet eddy current loss, electromagnetic excitation force, and vibration noise.
[0050] Based on this, step S130 calculates the motor's performance indicators based on the amplitude and phase of the air gap magnetic flux density harmonics, specifically including at least one of the following calculation methods: Based on the amplitude and phase of the radial flux density harmonics, a radial air gap flux density distribution function is synthesized. The radial air gap flux density distribution function is then calculated by integration and differentiation to obtain the no-load back electromotive force waveform and its effective value. Specifically, based on the principle of spatial harmonic synthesis, the radial air gap flux density distribution function is obtained by superimposing each harmonic component with amplitude, phase, and order as parameters.
[0051] Combining the harmonic characteristics of radial and tangential magnetic flux densities, the cogging torque waveform and its peak value of the motor are calculated using the Maxwell tensor method. Specifically, based on the principle of spatial harmonic synthesis, the radial magnetic flux density function is obtained by superimposing the respective harmonic components of the radial magnetic flux density with amplitude, phase, and order as parameters, and the tangential magnetic flux density function is obtained by superimposing the respective harmonic components of the tangential magnetic flux density with amplitude, phase, and order as parameters. According to the Maxwell tensor method formula, using the square of the air gap radius, the core length, and the vacuum permeability as basic parameters, the product of the radial and tangential magnetic flux density functions is integrated over the entire circumferential angle range, thereby obtaining the waveform and peak value of the cogging torque.
[0052] Based on the amplitude and phase of the radial flux density harmonics, the core losses (hysteresis and eddy current losses) in the stator and rotor cores are calculated using Steinmetz or Bessel function decomposition models.
[0053] Based on the amplitude and frequency of the time-varying radial magnetic flux density harmonics induced in the permanent magnet region, combined with the conductivity and segmentation of the permanent magnet, the permanent magnet eddy current loss in the permanent magnet can be calculated.
[0054] By combining the amplitudes and phases of the radial and tangential magnetic flux density harmonics, the radial electromagnetic force density distribution acting on the stator teeth is calculated according to Maxwell's stress formula. The radial electromagnetic force density distribution is then subjected to two-dimensional Fourier decomposition in the spatial and time domains to obtain the force wave amplitudes of each order and frequency. Based on the force wave amplitudes, the electromagnetic vibration and vibration noise of the motor are evaluated.
[0055] In some embodiments, the performance metric prediction model is trained as follows: Obtain the design parameter samples and the amplitude and phase labels of the corresponding air gap magnetic flux density harmonics.
[0056] The design parameter samples are input into the initial prediction model to obtain the amplitude and phase prediction values of the air gap magnetic flux density harmonics output by the initial prediction model.
[0057] The amplitude prediction value and phase prediction value, along with the corresponding amplitude label and phase label, are substituted into the loss function. When the loss function is not convergent, backpropagation is performed to adjust the model parameters of the initial prediction model until the loss function converges or the preset number of iterations is reached, thus obtaining the trained performance index prediction model.
[0058] Specifically, the amplitude and phase labels of the air gap magnetic flux density harmonics include: amplitude and phase labels of the fundamental order related to radial magnetic flux density and matched with the number of motor pole pairs, amplitude and phase labels of the slot harmonic order matched with the number of stator slots, and amplitude and phase labels of the combined order of the fundamental and slot harmonics. They also include: amplitude and phase labels of the fundamental order related to tangential magnetic flux density and matched with the number of motor pole pairs, amplitude and phase labels of the slot harmonic order matched with the number of stator slots, and amplitude and phase labels of the combined order of the fundamental and slot harmonics.
[0059] For example, different orders correspond to different amplitude and phase labels. Different orders, as well as different combinations of the two orders, can be selected for the fundamental and slot harmonics. This results in 15 amplitude labels and 15 phase labels related to radial flux density, for a total of 30 labels. Similarly, 15 amplitude labels and 15 phase labels related to tangential flux density, also for a total of 30 labels, can be obtained. That is, there are 30 labels for the first prediction model and 30 labels for the second prediction model. The design parameter samples and their corresponding labels can be divided into training, validation, and test sets in a 7:1.5:1.5 ratio. For example, with 2000 sets of design parameter samples and their corresponding labels, the training set would have 1400 sets of data, the validation set 300 sets of data, and the test set 300 sets of data.
[0060] For example, the design parameters take six key geometric parameters: permanent magnet thickness, stator inner diameter, stator tooth width, stator slot opening, pole arc coefficient, and air gap width. Each set of design parameter samples includes 6 geometric parameters and corresponds to 30 labels.
[0061] For the first prediction model, a three-layer artificial neural network model is constructed: an input layer, a hidden layer, and an output layer. The input layer has 6 neurons, corresponding to the 6 geometric parameters of the motor. The hidden layer has 3 layers, with 224 neurons in the first layer, 112 neurons in the second layer, and 56 neurons in the third layer. The output layer has 30 neurons, corresponding to 30 harmonic features (i.e., the 30 labels mentioned above). The ReLU function is used as the activation function, and the Adam optimizer is selected. The loss function can be the mean squared error (MSE) loss function. The initial learning rate is set to 0.0006, and the number of iterations is set to 200.
[0062] The model is trained using the training set, and performance metrics are calculated using the validation set every 25 iterations. The learning rate employs a segmented decay strategy, decreasing to 60% of its original value every 40 iterations. L2 regularization is incorporated during training. λ =0.0004) and Dropout (probability 0.2) are used to prevent overfitting. After training, the model is validated using a test set. The model performance is evaluated by the determination coefficient of amplitude prediction, the root mean square error, the angle error of phase prediction, and the accuracy. Finally, the phase prediction error is corrected by the four-position method to further improve the phase accuracy.
[0063] according to Figure 2 As shown, this embodiment provides accuracy verification results for the amplitude prediction and phase prediction values of the radial flux density harmonics (Br harmonics, including the fundamental wave, slot harmonics, and the amplitude and phase of a combination of the two, totaling 15 harmonics) predicted by the first prediction model, compared with the actual amplitude and phase values, respectively. The left figure is the amplitude prediction curve. The predicted amplitude and the actual amplitude highly overlap, indicating that the neural network training method provided in this embodiment can reproduce the actual amplitude variation trend of the Br harmonics with extremely high accuracy. The calculated coefficient of determination R of this prediction result is... 2 The accuracy reached 0.9947, verifying the excellent fitting ability and reliability of this embodiment in amplitude prediction. The right figure shows the phase prediction curve. The trends of the predicted phase and the actual phase are basically consistent, indicating that the method of this embodiment can effectively track the dynamic changes of the phase. There is a slight deviation between the two curves, with an average error of 3.37°. This error value is within the allowable range for engineering applications, proving that this embodiment can achieve high-precision amplitude prediction while also completing phase prediction with high accuracy, meeting the need for synchronous prediction of harmonic amplitude and phase in practical applications.
[0064] For the second prediction model, a three-layer artificial neural network model is constructed: the input layer has 6 neurons, corresponding to the 6 geometric parameters of the motor; the hidden layer has 3 layers, with the first layer having 512 neurons, the second layer having 256 neurons, and the third layer having 128 neurons; and the output layer has 30 neurons, corresponding to 30 harmonic features (i.e., the 30 labels mentioned above). The activation function is the ReLU function, the optimizer is the Adam optimizer, the initial learning rate is set to 0.005, and the number of iterations is set to 200.
[0065] The model is trained using the training set, and performance metrics are calculated using the validation set every 20 iterations. The learning rate employs a segmented decay strategy, decreasing to 50% of its original value every 40 iterations. L2 regularization is incorporated during training. λ =0.001) and Dropout (0.1 for the first layer and 0.05 for the second layer) are used to prevent overfitting. After training, the model is validated using a test set. The model performance is evaluated by the coefficient of determination of amplitude prediction, the root mean square error, the angle error of phase prediction, and the accuracy under different thresholds. Finally, the phase prediction error is corrected by the four-position method to further improve the phase accuracy.
[0066] It should be noted that in the first and second prediction models mentioned above, the number of neurons in each hidden layer is obtained by optimizing the hyperparameters during training. This can be understood as first defining a range of how many hidden layers there are and how many neurons are in each hidden layer, and then finding the optimal combination within this range.
[0067] according to Figure 3 As shown, this embodiment provides the accuracy verification results of the amplitude prediction and phase prediction values of the tangential flux density harmonics (Bt harmonics, including the fundamental wave, slot harmonics, and the amplitude and phase of a combination of the two, totaling 15 harmonics) predicted by the second prediction model, compared with the actual amplitude and phase values, respectively. The left figure is the amplitude prediction curve. Although there are slight fluctuations in local time periods, the overall trend of the two curves is highly consistent. The coefficient of determination R of this prediction result is... 2 The result reached 0.9222, verifying that this embodiment can reliably reflect the amplitude variation law. The right figure is the phase prediction curve. The trend of the predicted phase and the actual phase change is basically synchronized, indicating that the method of this embodiment can effectively track the dynamic change of the phase. The error between the two curves is small, with an average error of 2.50°, further proving that the embodiment performs well in amplitude prediction and also has high accuracy and practicality in phase prediction, and can meet the requirement of synchronous prediction of Bt harmonic amplitude and phase.
[0068] In some embodiments, obtaining the design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics specifically includes: The original design data of the motor was sampled to obtain design parameter samples with different parameter combinations. Specifically, a two-dimensional model of the motor was established using finite element simulation software, and the value range of each design parameter was set, such as the value range of six geometric parameters including permanent magnet thickness and stator inner diameter. 2000 sets of different design parameter sample combinations were generated according to the Latin hypercube sampling method.
[0069] The air gap magnetic flux density data samples are obtained by performing finite element simulation on the design parameter samples. Specifically, the air gap magnetic flux density data samples corresponding to each combination of design parameter samples are obtained through finite element simulation, such as radial magnetic flux density data and / or tangential magnetic flux density data.
[0070] Based on the air gap magnetic field adjustment theory, a Fast Fourier Transform (FFT) is performed on the air gap magnetic flux density data samples to obtain the amplitude and phase labels of the air gap magnetic flux density harmonics. Specifically, the FFT extracts the amplitude and phase labels of the fundamental order matching the number of pole pairs, the slot harmonic order matching the number of stator slots, and the combined harmonic order of the radial and / or tangential magnetic flux density data, respectively.
[0071] In this embodiment, different combinations of design parameters were obtained using finite element design software, and the labels corresponding to each combination of design parameters were obtained by performing finite element simulation and fast Fourier transform on the combinations of design parameters, thereby realizing the rapid and accurate construction of model training data.
[0072] In some embodiments, after obtaining the amplitude and phase labels of the air gap flux density harmonics, the method further includes: for each set of design parameter samples, selecting the top N largest amplitude labels and corresponding N phase labels from all the amplitude and phase labels of the corresponding air gap flux density harmonics for training the performance index prediction model, where N is greater than or equal to 2, for example: N=15, for a total of 30 labels.
[0073] In this embodiment, by filtering the top N largest amplitude labels and their corresponding N phase labels, harmonic components that significantly contribute to the radial / tangential flux density can be retained, while harmonics with extremely small amplitudes can be eliminated, ultimately simplifying calculations and improving model accuracy. The amplitude threshold can be set according to actual conditions.
[0074] The experimental results of the method of the present invention will be illustrated below using the design parameters, including six geometric parameters: permanent magnet thickness 3.3 mm, stator inner diameter 46.5 mm, stator tooth width 5.9 mm, stator slot opening 2.0 mm, pole arc coefficient 0.88, air gap width 1.0 mm, and two performance indicators, namely the predicted no-load back electromotive force waveform and the cogging torque waveform.
[0075] The above geometric parameters are input into the first and second prediction models after training, and the amplitude and phase of each specific order harmonic of radial and tangential magnetic flux density are output.
[0076] Based on the harmonic characteristics of the output, the distribution relationship of the radial air gap magnetic flux density is first constructed: it is the superposition of various spatial harmonic components, and each harmonic component is determined by the corresponding amplitude, phase, spatial angle, and electric angular velocity. On this basis, the flux linkage of a certain phase is calculated: using the air gap radius, core length, number of series turns per phase, and winding distribution coefficient of the nth harmonic as basic parameters, the radial air gap magnetic flux density is integrated over the corresponding spatial angle range, and the result is the flux linkage of that phase. By performing a time derivative operation on the flux linkage, the no-load back electromotive force waveform of that phase can be obtained. Then, through the combination relationship of phase voltages, the no-load line back electromotive force is further obtained, and its effective value is calculated.
[0077] Meanwhile, combining radial and tangential magnetic flux densities, the cogging torque is calculated using the Maxwell tensor method. Based on the square of the air gap radius, the core length, and the vacuum permeability as basic parameters, the product of the radial and tangential magnetic flux densities is integrated over the entire circumferential angle range, thereby obtaining the waveform and peak value of the cogging torque.
[0078] Accuracy verification: Perform finite element simulation on the target motor and compare the results with the calculated results for reference. Figure 4 The comparison results of the no-load back EMF waveforms show that the peak value of the no-load back EMF calculated by the method of this invention is 14.79V, while the peak value of the no-load back EMF calculated by the finite element simulation is 16.64V, with a relative error of 11.12%. The calculated effective value of the no-load back EMF is 10.80V, while the effective value of the no-load back EMF calculated by the finite element simulation is 11.59V, with a relative error of 6.8%. The predicted effective value of the no-load back EMF waveform is within an acceptable error range. (Reference) Figure 5 The comparison of the cogging torque waveforms shows that the peak cogging torque calculated by the method of this invention is -9.30 mN∙m, while the peak cogging torque calculated by the finite element simulation is -8.08 mN∙m. The absolute error between the two is 1.22 mN∙m, and the relative error is 13.12%, both of which meet the engineering design requirements. Moreover, the entire process of amplitude and phase prediction and performance calculation of the air gap magnetic flux density harmonics takes only 5 minutes, which is about 60% more efficient than the finite element simulation (which takes about 13 minutes).
[0079] Therefore, the method of the present invention has the following beneficial effects on the scheme of predicting the no-load back EMF waveform and cogging torque waveform for the above six geometric parameters.
[0080] (1) It retains the high precision advantage of finite element simulation and greatly improves prediction efficiency. The prediction speed is more than 60% faster than traditional finite element simulation, which meets the needs of rapid comparison of multiple schemes in the early stage of motor design.
[0081] (2) The six core geometric parameters and key harmonic orders are clearly defined, the data preprocessing process is standardized, the model input and output dimensions are reasonably matched, the generalization ability is strong, and it can be adapted to the performance prediction of motors with different structural types, thus solving the problem of poor universality of traditional empirical formulas.
[0082] (3) Harmonics are extracted by fast Fourier transform and hyperparameters of artificial neural network model are optimized. The prediction accuracy is high. The prediction error of the effective value of no-load back EMF is less than 7%, and the prediction error of the peak value of cogging torque is less than 15%, which is significantly better than conventional machine learning models and analytical methods. It can accurately reflect the motor performance under actual working conditions.
[0083] (4) The modular architecture is clear and the process is standardized. It does not rely on professional finite element simulation platforms and complex operating skills, making it easy to integrate into the existing motor design process. This helps to reduce development costs, shorten the R&D cycle, and has good engineering practicality.
[0084] The motor performance index prediction device provided by the present invention is described below. The motor performance index prediction device described below can be referred to in correspondence with the motor performance index prediction method described above.
[0085] The motor performance index prediction device of this invention embodiment, such as Figure 6 As shown, it includes: The design parameter acquisition module 610 is used to acquire the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters.
[0086] The model prediction module 620 is used to input the design parameters into the performance index prediction model to obtain the amplitude and phase of the air gap magnetic flux density harmonic of the motor output by the performance index prediction model.
[0087] The performance index calculation module 630 is used to calculate the performance index of the motor based on the amplitude and phase of the air gap magnetic flux density harmonic.
[0088] The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0089] In some embodiments, the amplitude and phase of the air gap magnetic flux density harmonics include: the amplitude and phase of the radial magnetic flux density harmonics, and / or the amplitude and phase of the tangential magnetic flux density harmonics.
[0090] The amplitude and phase of the radial flux density harmonics include: the amplitude and phase of the fundamental order related to the radial flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
[0091] The amplitude and phase of the tangential flux density harmonics include: the amplitude and phase of the fundamental order related to the tangential flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
[0092] In some embodiments, the performance metric prediction model includes: a first prediction model and / or a second prediction model.
[0093] The first prediction model is used to predict the amplitude and phase of the radial flux density harmonics based on the design parameters.
[0094] The second prediction model is used to predict the amplitude and phase of the tangential flux density harmonics based on the design parameters.
[0095] In some embodiments, the performance indicators include at least one of the following: no-load back EMF waveform of the motor, cogging torque waveform, core loss, permanent magnet eddy current loss, electromagnetic excitation force, and vibration noise.
[0096] The performance indicator calculation module 630 specifically includes at least one of the following modules: The no-load back EMF calculation module is used to synthesize the radial air gap magnetic flux density distribution function based on the amplitude and phase of the radial magnetic flux density harmonics. By integrating and differentiating the radial air gap magnetic flux density distribution function, the waveform and effective value of the no-load back EMF of the motor are obtained.
[0097] The cogging torque calculation module is used to calculate the cogging torque waveform and its peak value of the motor by combining the harmonic characteristics of the radial and tangential magnetic flux densities and using the Maxwell tensor method.
[0098] The core loss calculation module is used to calculate the core loss in the stator and rotor cores based on the amplitude and phase of the radial flux density harmonics, using a Steinmetz or Bessel function decomposition model.
[0099] The permanent magnet eddy current loss calculation module is used to calculate the permanent magnet eddy current loss in a permanent magnet based on the amplitude and frequency of the time-varying radial magnetic flux density harmonics induced in the permanent magnet region, combined with the conductivity and segmentation of the permanent magnet.
[0100] The vibration index calculation module is used to combine the amplitude and phase of the radial and tangential magnetic flux density harmonics, calculate the radial electromagnetic force density distribution acting on the stator teeth according to Maxwell's stress formula, and perform two-dimensional Fourier decomposition of the radial electromagnetic force density distribution in the spatial and time domains to obtain the force wave amplitude of each order and frequency. Based on the force wave amplitude, the electromagnetic vibration and vibration noise of the motor are evaluated.
[0101] In some embodiments, the performance metric prediction model is trained as follows: Obtain the design parameter samples and the amplitude and phase labels of the corresponding air gap magnetic flux density harmonics.
[0102] The design parameter samples are input into the initial prediction model to obtain the amplitude and phase prediction values of the air gap magnetic flux density harmonics output by the initial prediction model.
[0103] The amplitude prediction value and phase prediction value, along with the corresponding amplitude label and phase label, are substituted into the loss function. When the loss function is not convergent, backpropagation is performed to adjust the model parameters of the initial prediction model until the loss function converges or the preset number of iterations is reached, thus obtaining the trained performance index prediction model.
[0104] In some embodiments, obtaining design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics includes: The original design data of the motor is sampled to obtain design parameter samples with different parameter combinations.
[0105] The air gap magnetic flux density data sample was obtained by performing finite element simulation on the design parameter sample.
[0106] Based on the theory of air gap magnetic field adjustment, a fast Fourier transform is performed on the air gap magnetic flux density data sample to obtain the amplitude and phase labels of the air gap magnetic flux density harmonics.
[0107] In some embodiments, after obtaining the amplitude and phase labels of the air gap flux density harmonics, the method further includes: For each set of design parameter samples, the top N largest amplitude labels and corresponding N phase labels are selected from all the amplitude and phase labels of the corresponding air gap magnetic flux density harmonics to train the performance index prediction model, where N is greater than or equal to 2.
[0108] Figure 7 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 7As shown, the electronic device may include: a processor 710, a communications interface 720, a memory 730, and a communication bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 can call logic instructions in the memory 730 to execute a motor performance index prediction method, which includes: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters.
[0109] The design parameters are input into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model.
[0110] The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics.
[0111] The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0112] Furthermore, the logical instructions in the aforementioned memory 730 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0113] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is able to execute the motor performance index prediction method provided by the above methods, the method comprising: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters.
[0114] The design parameters are input into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model.
[0115] The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics.
[0116] The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0117] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the motor performance index prediction method provided by the above methods, the method comprising: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters.
[0118] The design parameters are input into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model.
[0119] The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics.
[0120] The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
[0121] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0123] All actions involving the acquisition of signal information or data in this invention are carried out in compliance with the relevant data protection laws and policies of the country where the device is located, and with the authorization granted by the owner of the device.
[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for predicting motor performance indicators, characterized in that, include: Obtain the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters; The design parameters are input into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model. The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics. The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
2. The method for predicting motor performance indicators according to claim 1, characterized in that, The amplitude and phase of the air gap magnetic flux density harmonics include: the amplitude and phase of the radial magnetic flux density harmonics, and / or the amplitude and phase of the tangential magnetic flux density harmonics; The amplitude and phase of the radial flux density harmonics include: the amplitude and phase of the fundamental order related to the radial flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics. The amplitude and phase of the tangential flux density harmonics include: the amplitude and phase of the fundamental order related to the tangential flux density and matched with the number of pole pairs of the motor, the amplitude and phase of the slot harmonic order matched with the number of stator slots, and the amplitude and phase of the combined order of the fundamental and slot harmonics.
3. The method for predicting motor performance indicators according to claim 2, characterized in that, The performance indicator prediction model includes: a first prediction model and / or a second prediction model; The first prediction model is used to predict the amplitude and phase of the radial flux density harmonics based on the design parameters; The second prediction model is used to predict the amplitude and phase of the tangential flux density harmonics based on the design parameters.
4. The method for predicting motor performance indicators according to claim 2, characterized in that, The performance indicators include at least one of the following: no-load back EMF waveform of the motor, cogging torque waveform, core loss, permanent magnet eddy current loss, electromagnetic excitation force, and vibration noise. The performance indicators of the motor are calculated based on the amplitude and phase of the air gap magnetic flux density harmonics, including at least one of the following calculation methods: Based on the amplitude and phase of the radial flux density harmonics, the radial air gap flux density distribution function is synthesized. The radial air gap flux density distribution function is calculated by integration and differentiation to obtain the no-load back electromotive force waveform and effective value of the motor. By combining the harmonic characteristics of radial and tangential magnetic flux densities, the cogging torque waveform and its peak value of the motor are calculated using the Maxwell tensor method. Based on the amplitude and phase of the radial flux density harmonics, the core loss in the stator and rotor cores is calculated using the Steinmetz or Bessel function decomposition model. Based on the amplitude and frequency of the time-varying radial magnetic flux density harmonics induced in the permanent magnet region, combined with the conductivity and segmentation of the permanent magnet, the permanent magnet eddy current loss in the permanent magnet can be calculated. By combining the amplitudes and phases of the radial and tangential magnetic flux density harmonics, the radial electromagnetic force density distribution acting on the stator teeth is calculated according to Maxwell's stress formula. The radial electromagnetic force density distribution is then subjected to two-dimensional Fourier decomposition in the spatial and time domains to obtain the force wave amplitudes of each order and frequency. Based on the force wave amplitudes, the electromagnetic vibration and vibration noise of the motor are evaluated.
5. The method for predicting motor performance indicators according to any one of claims 1 to 4, characterized in that, The training method for the performance indicator prediction model is as follows: Obtain the design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics; The design parameter samples are input into the initial prediction model to obtain the amplitude and phase prediction values of the air gap magnetic flux density harmonics output by the initial prediction model. The amplitude prediction value and phase prediction value, along with the corresponding amplitude label and phase label, are substituted into the loss function. When the loss function is not convergent, backpropagation is performed to adjust the model parameters of the initial prediction model until the loss function converges or the preset number of iterations is reached, thus obtaining the trained performance index prediction model.
6. The method for predicting motor performance indicators according to claim 5, characterized in that, Obtain design parameter samples and the amplitude and phase labels of the corresponding air gap flux density harmonics, including: The original design data of the motor is sampled to obtain design parameter samples with different parameter combinations; The air gap magnetic flux density data sample was obtained by performing finite element simulation on the design parameter sample. Based on the theory of air gap magnetic field adjustment, a fast Fourier transform is performed on the air gap magnetic flux density data sample to obtain the amplitude and phase labels of the air gap magnetic flux density harmonics.
7. The method for predicting motor performance indicators according to claim 6, characterized in that, After obtaining the amplitude and phase labels of the air gap flux density harmonics, the following is also included: For each set of design parameter samples, the top N largest amplitude labels and corresponding N phase labels are selected from all the amplitude and phase labels of the corresponding air gap magnetic flux density harmonics to train the performance index prediction model, where N is greater than or equal to 2.
8. A device for predicting motor performance indicators, characterized in that, include: The design parameter acquisition module is used to acquire the design parameters of the motor, which include at least one of the following: the motor's geometric parameters, material parameters, environmental parameters, and operating condition parameters. The model prediction module is used to input the design parameters into the performance index prediction model to obtain the amplitude and phase of the air gap flux density harmonic of the motor output by the performance index prediction model. The performance index calculation module is used to calculate the motor's performance index based on the amplitude and phase of the air gap magnetic flux density harmonics. The performance index prediction model is trained based on the design parameter samples and the amplitude and phase labels of the air gap magnetic flux density harmonics corresponding to the design parameter samples.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the motor performance index prediction method as described in any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the motor performance index prediction method as described in any one of claims 1 to 7.