Optimization method, device, electronic device and storage medium for motor finite element model

Abstract

The present disclosure relates to an optimization method, device, electronic device and storage medium for a motor finite element model, and relates to the field of motors. The present disclosure obtains a pre-established motor finite element initial model and at least one target optimization variable specified by a user, obtains the sensitivity value of each of the structural variables, divides a plurality of the structural variables into a plurality of structural variable sets according to the sensitivity values, determines a target structural variable set from the plurality of structural variable sets, uses a plurality of preset algorithms to construct proxy models for the target structural variable sets respectively, and determines the optimization parameters corresponding to each structural variable according to the plurality of proxy models corresponding to the current target structural variable set. In this way, by stratifying a plurality of structural variables by sensitivity values, and determining the optimization parameters corresponding to each structural variable by a plurality of proxy models, the efficiency of motor optimization can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of motors, and in particular to a method, device, electronic device and storage medium for optimizing a finite element model of a motor. Background Art

[0002] After completing the initial structural design of the motor, the selected structural variables are optimized using a suitable algorithm to obtain the optimized target parameters. In the related technology, a mathematical model based on magnetic circuit calculation is generally used to perform optimization using an algorithm, such as finite element calculation.

[0003] In the related technology, finite element calculation is a method of motor simulation design. However, since there are many structural variables that affect the motor, it is very time-consuming and cumbersome to perform finite element calculation on multiple structural variables because there are many structural variables involved. Summary of the invention

[0004] In order to overcome the problems existing in the related art, the present disclosure provides a method, device, electronic device and storage medium for optimizing a finite element model of a motor.

[0005] According to a first aspect of an embodiment of the present disclosure, a method for optimizing a finite element model of a motor is provided, the method comprising:

[0006] Acquire a pre-established motor finite element initial model and at least one target optimization variable specified by a user, wherein the motor finite element initial model includes a plurality of structural variables;

[0007] Obtaining a sensitivity value of each of the structural variables;

[0008] Dividing the plurality of structural variables into a plurality of structural variable sets according to the sensitivity values;

[0009] Determining a target structural variable set from the plurality of structural variable sets, and using a plurality of preset algorithms to construct proxy models for the target structural variable set respectively, so as to obtain a plurality of proxy models corresponding to the target structural variable set;

[0010] The optimization parameter corresponding to each structural variable is determined according to the multiple proxy models corresponding to the current target structural variable set.

[0011] Optionally, determining the optimization parameter corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set includes:

[0012] Determining a target proxy model corresponding to the target structure variable set from the plurality of proxy models;

[0013] Determining multiple groups of optimization results corresponding to the target optimization variables through the target proxy model;

[0014] Determining a target optimization result from the multiple groups of optimization results;

[0015] Obtaining a target difference between the target optimization result and the optimization benchmark result;

[0016] In the case where the target difference is greater than or equal to the preset difference threshold, determining the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result; and updating the target structural variable set according to the sensitivity value to obtain the updated target structural variable set, and again executing the step of using multiple preset algorithms to construct proxy models for the target structural variable set respectively to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold;

[0017] When the target difference is less than the preset difference threshold, the undetermined parameter corresponding to each current structural variable is used as the optimization parameter corresponding to the structural variable.

[0018] Optionally, updating the target structure variable set according to the sensitivity value includes:

[0019] According to the sensitivity value of each structural variable, obtaining an average sensitivity value corresponding to each set of structural variables;

[0020] The structural variable set with the highest average sensitivity value among the multiple structural variable sets is used as the updated target structural variable set.

[0021] Optionally, dividing the plurality of structure variables into a plurality of structure variable sets according to the sensitivity values ​​comprises:

[0022] According to one or more preset sensitivity value thresholds, the plurality of structural variables are divided into a plurality of structural variable sets.

[0023] Optionally, the plurality of preset algorithms include at least two of a GRNN neural network, a response surface and Kriging, and the method further includes:

[0024] Obtaining a sensitive interval of each of the structural variables;

[0025] Accordingly, the adopting of multiple preset algorithms to construct proxy models for the target structure variable set respectively includes:

[0026] Determine a sample point set of the target structure variable set within the sensitive interval of the plurality of target structure variables corresponding to the target structure variable set by using an optimal hypercube sampling method, wherein the sample point set includes a first sampling parameter of each structure variable in the target structure variable set and a second sampling parameter of structure variables in other structure variable sets among the plurality of structure variable sets except the target structure variable set;

[0027] Determine a sample result set of the target structure variable set by finite element calculation according to the sample point set of the target structure variable set;

[0028] The proxy models of the target structure variable set are respectively constructed using the multiple preset algorithms according to the sample point set and the sample result set.

[0029] Optionally, determining a target proxy model corresponding to the target optimization set from the plurality of proxy models comprises:

[0030] Determine a determination coefficient of each of the proxy models according to the sample point set and the sample result set, wherein the determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model;

[0031] The target proxy model corresponding to the target optimization set is determined according to the determination coefficients of the plurality of proxy models.

[0032] Optionally, the method is applied to a permanent magnet synchronous motor;

[0033] The structural variables include at least two of the slot width, the slot depth, the pole arc coefficient, and the permanent magnet thickness;

[0034] The target optimization variable includes at least one of maximum torque, minimum cogging torque, minimum torque ripple, motor efficiency, and radial electromagnetic force.

[0035] According to a second aspect of an embodiment of the present disclosure, there is provided a device for optimizing a finite element model of a motor, the device comprising:

[0036] A first acquisition module is configured to acquire a pre-established motor finite element initial model and at least one target optimization variable specified by a user, wherein the motor finite element initial model includes a plurality of structural variables;

[0037] A second acquisition module is configured to acquire a sensitivity value of each of the structural variables;

[0038] A division module, configured to divide the plurality of structural variables into a plurality of structural variable sets according to the sensitivity values;

[0039] A proxy model construction module is configured to determine a target structural variable set from the plurality of structural variable sets, and to use a plurality of preset algorithms to respectively construct proxy models for the target structural variable set, so as to obtain a plurality of proxy models corresponding to the target structural variable set;

[0040] The optimization module is configured to determine the optimization parameter corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set.

[0041] Optionally, the optimization module includes:

[0042] A determination submodule is configured to determine a target proxy model corresponding to the target structure variable set from the plurality of proxy models;

[0043] A first acquisition submodule is configured to determine a plurality of groups of optimization results corresponding to the target optimization variable through the target proxy model; and determine a target optimization result from the plurality of groups of optimization results;

[0044] A second acquisition submodule is configured to obtain a target difference between the target optimization result and the optimization benchmark result;

[0045] The loop submodule is configured to determine the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result when the target difference is greater than or equal to a preset difference threshold; and update the target structural variable set according to the sensitivity value to obtain the updated target structural variable set, and again execute the step of using multiple preset algorithms to respectively construct proxy models for the target structural variable set to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold;

[0046] The result acquisition submodule is configured to use the undetermined parameter corresponding to each structural variable in the current target structural variable set as the optimization parameter corresponding to the structural variable when the target difference is less than the preset difference threshold.

[0047] Optionally, the loop submodule is further configured to:

[0048] According to the sensitivity value of each structural variable, obtaining an average sensitivity value corresponding to each set of structural variables;

[0049] The structural variable set with the highest average sensitivity value among the multiple structural variable sets is used as the updated target structural variable set.

[0050] Optionally, the division module is further configured to:

[0051] According to one or more preset sensitivity value thresholds, the plurality of structural variables are divided into a plurality of structural variable sets.

[0052] Optionally, the plurality of preset algorithms include at least two of GRNN neural network, response surface and Kriging, and the proxy model building module is further configured to:

[0053] Acquire a sensitive interval of each of the structural variables; determine a sample point set of the target structural variable set within the sensitive intervals of the multiple target structural variables corresponding to the target structural variable set using an optimal hypercube sampling method, wherein the sample point set includes a first sampling parameter of each of the structural variables in the target structural variable set and a second sampling parameter of structural variables in other structural variable sets among the multiple structural variable sets except the target structural variable set; determine a sample result set of the target structural variable set by finite element calculation based on the sample point set of the target structural variable set; and construct the proxy models of the target structural variable set respectively using the multiple preset algorithms based on the sample point set and the sample result set.

[0054] Optionally, the determining submodule is further configured to:

[0055] Determine a determination coefficient of each of the proxy models based on the sample point set and the sample result set, wherein the determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model; and determine a target proxy model corresponding to the target optimization set based on the determination coefficients of the multiple proxy models.

[0056] Optionally, the method is applied to a permanent magnet synchronous motor, the structural variables include at least two of the slot width, slot depth, pole arc coefficient and permanent magnet thickness; the target optimization variables include at least one of the maximum torque, minimum cogging torque, minimum torque fluctuation, motor efficiency and radial electromagnetic force.

[0057] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, the electronic device comprising:

[0058] a memory having a computer program stored thereon;

[0059] A processor is used to execute the computer program in the memory to implement the steps of the method described in any embodiment of the first aspect above.

[0060] According to a fourth aspect of an embodiment of the present disclosure, there is provided a non-temporary computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0061] The technical solution provided by the embodiments of the present disclosure may have the following beneficial effects:

[0062] The present disclosure first obtains a pre-established motor finite element initial model and at least one target optimization variable specified by a user, then obtains the sensitivity value of each of the structural variables, divides the multiple structural variables into multiple structural variable sets according to the sensitivity values, determines the target structural variable set from the multiple structural variable sets, and uses multiple preset algorithms to build proxy models for the target structural variable set respectively, so as to obtain multiple proxy models corresponding to the target structural variable set, and determines the optimization parameters corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set. In this way, multiple structural variables are stratified by sensitivity values, and the optimization parameters corresponding to each structural variable are determined by multiple proxy models, which can improve the efficiency of motor optimization.

[0063] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0064] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0066] Figure 1 The figure is a flow chart of a method for optimizing a finite element model of a motor according to an exemplary embodiment.

[0067] Figure 2 The figure is a flow chart of another method for optimizing a finite element model of a motor according to an exemplary embodiment.

[0068] Figure 3 It is a block diagram of a device for optimizing a finite element model of a motor according to an exemplary embodiment.

[0069] Figure 4 It is a block diagram of another device for optimizing a finite element model of a motor according to an exemplary embodiment.

[0070] Figure 5 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0071] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims, and it should be understood that the specific embodiments described herein are only used to illustrate and explain the present disclosure and are not used to limit the present disclosure.

[0072] It should be noted that all actions of acquiring signals, information or data in the present disclosure 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 given by the owner of the corresponding device.

[0073] First, the application scenario of the present disclosure is explained. The present disclosure is applied to the optimization scenario of the motor. The structural variables of the motor have an important influence on the performance of the motor. There are many structural variables that affect the performance of the motor, such as slot width, pole arc coefficient and permanent magnet thickness. The optimization of the structural variables of the motor can be carried out by prototype testing. With the development of computer simulation technology, computer simulation software can be used to obtain the influence of the structural variables of the motor on the final target optimization variables, such as the well-known ANSYS Maxwell software in the industry. The basic principle of this computer simulation software is FEA (Finite Element Analysis, finite element analysis), and its basic principle is: by discretizing the mathematical model using FEM (Finite Element Analysis, finite element method), the corresponding numerical model can be obtained; then the discrete equation is solved and the result is analyzed, which is finite element analysis. In the optimization scenario of the motor, since there are many structural variables that affect the performance of the motor, and the optimization of the motor is a typical multi-objective problem, there are also many target optimization variables that are concerned, which will cause the amount of calculation of the computer simulation software to be very large, affecting the efficiency of the optimization of the motor.

[0074] The present disclosure provides an optimization method, device, electronic device and storage medium for a motor finite element model, which can stratify multiple structural variables, select suitable proxy models, perform stratified optimization on multiple structural variables, reduce the computational workload of motor optimization, and improve the efficiency of motor optimization.

[0075] The present disclosure is described below in conjunction with specific embodiments.

[0076] Figure 1 is a flow chart of a finite element model optimization method according to an exemplary embodiment. Figure 1 As shown, the method may include the following steps:

[0077] In step S101 , a pre-established motor finite element initial model and at least one target optimization variable specified by a user are obtained.

[0078] Among them, the motor finite element initial model includes multiple structural variables, and the structural variables may include at least two of the slot width, slot depth, pole arc coefficient and permanent magnet thickness. The target optimization variable may include one or more of the maximum torque, the minimum cogging torque, the minimum torque fluctuation, the motor efficiency and the radial electromagnetic force. The maximum torque refers to the maximum torque that the motor can generate when the load is increased without causing a sudden drop in speed at the rated voltage and rated frequency. The cogging torque refers to the torque generated by the interaction between the permanent magnet and the stator core when the motor winding is not energized, which is caused by the tangential component of the interaction force between the permanent magnet and the armature tooth. The torque fluctuation refers to the fluctuation of the motor torque that not only causes the vibration of the motor body, but also causes the vibration of the components that are directly or indirectly in contact with the motor. The motor efficiency refers to the ratio of the output power of the motor to the input power. The radial electromagnetic force generally refers to the force acting on a cylindrical object, and the direction of the force passes through the center of the cross section of the object.

[0079] In step S102 , the sensitivity value of each structural variable is obtained.

[0080] Different structural variables have different degrees of influence on the target optimization variable, and the same structural variable has different degrees of influence on the target optimization variable in different value ranges. The degree of influence of the change of the structural variable on the optimization result of the target optimization variable is called the sensitivity value. The sensitivity value of the same structural variable in a certain value range is greater than or equal to the preset effective sensitivity threshold, and the value range can be called the sensitive interval.

[0081] In some possible implementations, the sensitivity value of the structural variable is determined through the following steps.

[0082] Step S11, determining a preset value range and a preset default value of each structural variable.

[0083] For example, the preset value range of the permanent magnet thickness may be 3 mm to 3.8 mm, and the default value is 3.4 mm.

[0084] Step S12: select a target structural variable, select a test structural variable from the preset value range, and take the preset default values ​​of the remaining structural parameters to form a target test variable set.

[0085] For example, the permanent magnet thickness can be selected from [3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8] mm, and the remaining slot widths, slot depths, and pole arc coefficients are preset default values, forming 9 target test variable sets, which traverse 9 different permanent magnet thickness values.

[0086] Step S13: using finite element calculation to obtain a target test result set corresponding to the above-mentioned multiple target test variable sets, wherein the target test result set includes multiple target optimization variables.

[0087] For example, still taking the thickness of the permanent magnet as an example, finite element calculation is used to determine 9 target optimization results corresponding to the 9 target test variables.

[0088] In some embodiments, the finite element calculation may set constraints, which may include:

[0089] The output power of the motor is greater than or equal to a preset power threshold; and / or,

[0090] The efficiency of the motor is greater than or equal to a preset efficiency threshold.

[0091] The above target optimization results need to be determined while satisfying the above constraints.

[0092] Step S14: Determine the sensitive interval of the structural variable according to the target test variable set and the target test result set.

[0093] In some possible implementations, the sensitivity interval of the corresponding structural variable can be determined by the change rate of the target test result of the target optimization variable in the target test result set and the benchmark test result. The benchmark test result can be the most undesirable result value of the target optimization variable in the target test result set. For example, when the target optimization variable includes motor efficiency, the benchmark test result is the lowest motor efficiency in the target test result set. When the change rate of the target test result of the target optimization variable in the target test result set compared with the benchmark test result is greater than or equal to a preset effective sensitivity threshold, it is determined that the target test variable corresponding to the target test result falls into the sensitive interval, and the interval formed by multiple target test variables falling into the sensitive interval is the sensitive interval.

[0094] When the target test result set contains multiple target optimization variables, the sensitive interval can be determined by weighted normalization of the multiple target optimization variables using preset weights. The specific technical solution can refer to the relevant technology and will not be repeated here.

[0095] Step S15: Calculate the sensitivity value of the structural variable within the sensitive interval.

[0096] For example, the sensitivity value can be expressed using Formula 1.

[0097]

[0098] Among them, R i is the result of the target optimization variable in the target test result set, R is the benchmark test result of the target optimization variable, N is the number of results of the target optimization variable in the target test result set, and S is the sensitivity value.

[0099] When the target test result set includes multiple target optimization variables, the sensitivity values ​​of the multiple target optimization variables can be weighted by preset weights to obtain a normalized sensitivity value, and the normalized sensitivity value is used as the sensitivity value.

[0100] Step S16: Repeat the above steps S11 to S15 to obtain the sensitivity values ​​and sensitive intervals of all structural variables.

[0101] In step S103 , the plurality of structural variables are divided into a plurality of structural variable sets according to the sensitivity values.

[0102] In this step, multiple structural variables are divided into multiple sets of structural variables according to sensitivity values, so as to first optimize the structural variables with higher sensitivity values. The structural variables with lower sensitivity values ​​can be fixed to preset default values, which can effectively improve the efficiency of motor optimization.

[0103] The number of the multiple structural variable sets may be two or more, and the present disclosure does not impose any limitation on this.

[0104] In some embodiments, a plurality of structural variables may be divided into a plurality of structural variable sets according to one or more preset sensitivity value thresholds.

[0105] For example, multiple structural variables can be divided into two structural variable sets by the following method: first, a preset sensitivity value threshold is obtained. When the sensitivity value of the structural variable is greater than or equal to the sensitivity value threshold, it can be determined that the structural variable belongs to a high sensitivity value structural variable set. When the sensitivity value of the structural variable is less than the sensitivity value threshold, it can be determined that the structural variable belongs to a low sensitivity value structural variable set. For another example, according to two different preset sensitivity value thresholds (for example, a first sensitivity value threshold and a second sensitivity value threshold, wherein the first sensitivity value threshold is less than the second sensitivity value threshold), multiple structural variables can be divided into three structural variable sets (for example, respectively, the sensitivity value is less than the first sensitivity value threshold, the sensitivity value is greater than or equal to the first sensitivity value threshold and less than the second sensitivity value threshold, and the sensitivity value is greater than or equal to the second sensitivity value threshold).

[0106] In step S104, a target structural variable set is determined from multiple structural variable sets, and multiple preset algorithms are used to construct proxy models for the target structural variable set, so as to obtain multiple proxy models corresponding to the target structural variable set.

[0107] For example, a target structure variable set may be determined from multiple structure variable sets by the following steps:

[0108] Step S21, obtaining the average sensitivity value of multiple structural variables in each structural variable set.

[0109] For example, the arithmetic mean of the sensitivity values ​​of multiple structural variables in each structural variable set can be used as the average sensitivity value of the structural variable set, or the weighted average of the sensitivity values ​​of multiple structural variables in each structural variable set can be used as the average sensitivity value of the structural variable set according to preset structural variable weights. The present disclosure does not impose any restrictions on this.

[0110] Step S22: determine the target structure variable set according to the average sensitivity value.

[0111] In some possible implementations, the target structure variable set can be determined using a queue data structure method. For example, the multiple structure variables can be combined and added to a structure variable set queue according to the average sensitivity value, and then the target structure variable set can be obtained from the head of the structure variable set queue.

[0112] After obtaining the target structural variable set, a plurality of preset algorithms can be used to construct proxy models for the target structural variable set respectively, wherein the proxy models include a GRNN neural network model, a response surface model or a Kriging model, and the plurality of preset algorithms include at least two of the GRNN neural network, the response surface and the Kriging.

[0113] For example, the basic principle is introduced by taking the response surface model as an example.

[0114] The response surface model is a model that simulates the real limit state surface through a series of multi-variable, deterministic "experiments". The results of the experiment using the response surface model can be expressed as the following formula 2.

[0115]

[0116] Among them, y i is the output parameter of the ith test, β is the coefficient, x ij is the jth group of input parameters for the i-th trial, ε i is the ith observation error.

[0117] It should be noted that the principle of the response surface model is to use the principle of Taylor expansion to fit any function using polynomials. Generally, at least constant terms, linear terms, and quadratic terms are considered. The above k does not represent the number of input parameters. ij It may be the value of a certain input parameter in the i-th experiment, or it may be the product of multiple input parameters. For details, please refer to the principle of the response surface model in the relevant technology, which will not be described here.

[0118] The above formula 2 is organized into a matrix form to form the following formula 3.

[0119] The formula 2 includes:

[0120] y=Xβ+ε(Formula 3)

[0121] in:

[0122]

[0123] The minimum variance of this test can be expressed as the following formula 4.

[0124]

[0125] Where L is the minimum variance.

[0126] When the partial derivative of L with respect to β is 0, the minimum variance takes the minimum value, and the final response surface model is obtained as shown in Formula 5.

[0127] in,

[0128] in, are the coefficients obtained based on the experimental data, X is the input parameter of the response surface model, X′ is the transpose of the input parameter matrix of the response surface model, is the output parameter of the response surface model, (X′X) -1 Represents the inverse matrix of matrix X′X.

[0129] In some implementations, a proxy model may be constructed for a target structure variable set by the following steps.

[0130] Step S31: Obtain the sensitive interval of each structural variable.

[0131] This step has been described in detail in step S102 and will not be repeated here.

[0132] Step S32: using the optimal hypercube sampling method to determine a sample point set of the target structure variable set within a sensitive interval of multiple target structure variables corresponding to the target structure variable set, the sample point set including a first sampling parameter of each structure variable in the target structure variable set and a second sampling parameter of structure variables in other structure variable sets among the multiple structure variable sets except the target structure variable set.

[0133] For example, the target structural variable set is the permanent magnet thickness and the slot width, and the pole arc coefficient and the slot depth other than the target structural variable set in the plurality of structural variables are determined as the preset second sampling parameters. For example, the pole arc coefficient is determined to be 0.89, and the slot depth is determined to be 3 mm.

[0134] In addition, for the structural variables in the target structural variable set, such as the permanent magnet thickness and the slot width, the Latin hypercube sampling method is used to obtain the first sampling parameter of each structural variable in the target structural variable set from the sensitive intervals corresponding to the permanent magnet thickness and the slot width, and the first sampling parameter and the second sampling parameter are combined to determine the sample point set of the target structural variable set.

[0135] Among them, Latin hypercube sampling is a type of Monte Carlo simulation method. It improves the sampling strategy to achieve higher sampling accuracy with a smaller sampling scale. Its core steps are stratified sampling and shuffle sorting. The sample values ​​of stratified sampling can cover the entire distribution range of the input random variable; compared with the simple random sampling of the Monte Carlo simulation method, the Latin hypercube sampling method produces samples with higher spatial coverage, and its essence is that the standard deviation of the samples is smaller. Optimal Latin hypercube sampling is to add a criterion to Latin hypercube sampling. This criterion can select the total mean square error to obtain the optimal design under this criterion. Compared with ordinary Latin hypercube, the randomness of sampling is reduced so that all test points are distributed as evenly as possible in the actual space.

[0136] Step S33: Determine a sample result set of the target structure variable set by finite element calculation according to the sample point set of the target structure variable set.

[0137] The sample result set includes sample results of multiple target optimization variables corresponding to the sample point set.

[0138] Step S34: constructing proxy models of the target structure variable set respectively using a plurality of preset algorithms according to the sample point set and the sample result set.

[0139] In this step, the sample point set and the sample result set are used to establish corresponding multiple proxy models. For example, the sample point set and the sample result set are used to establish a response surface proxy model through Formula 5. When there are multiple target optimization variables, multiple response surface proxy models can be established.

[0140] In step S105, the optimization parameter corresponding to each structural variable is determined according to the multiple proxy models corresponding to the current target structural variable set.

[0141] In some possible implementations, the optimization parameters corresponding to each structural variable are determined through the following steps.

[0142] Step S41: Determine a target proxy model corresponding to a target structure variable set from a plurality of proxy models.

[0143] In some possible implementations, the target proxy model corresponding to the target structure variable set may be determined through the following steps.

[0144] First, the coefficient of determination of each proxy model is determined based on the sample point set and the sample result set.

[0145] The determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model.

[0146] For example, the sample point set is substituted into the response surface model shown in Formula 5 to obtain the output result of the response surface proxy model, and the output result is used as the proxy sample result set of the response surface proxy model.

[0147] For example, the determination coefficient may be a complex correlation coefficient. When the sample result set includes a sample result of a target optimization variable, the complex correlation coefficient is obtained according to the sample point set, the sample result set, the proxy sample result set and formula six, and the complex correlation coefficient is converted into.

[0148]

[0149] Among them, R 2 is the multiple correlation coefficient, y i represents the sample result in the sample result set corresponding to the i-th sample point, represents the average value of k sample results in the sample result set, is the proxy sample result in the proxy sample result set corresponding to the i-th sample point.

[0150] In the case where the sample result set includes sample results of multiple target optimization variables, the weighted multiple correlation coefficient may be determined according to preset weights of the multiple target optimization variables and corresponding multiple correlation coefficients.

[0151] For example, the weighted complex correlation coefficient may be determined according to the preset weights of a plurality of target optimization variables, the corresponding complex correlation coefficients, and the following formula 7.

[0152]

[0153] in, is the weighted multiple correlation coefficient, N is the number of target optimization variables, w i The weight coefficient of the preset i-th objective optimization variable, Optimize the multiple correlation coefficient of the variables for the ith objective.

[0154] Then, the target proxy model corresponding to the target optimization set is determined according to the determination coefficients of the multiple proxy models.

[0155] For example, the determination coefficient may be a complex correlation coefficient, and according to the complex correlation coefficients of the multiple proxy models, the proxy model with the highest complex correlation coefficient is selected as the target proxy model. In the case where there are multiple target optimization variables, according to the weighted complex correlation coefficients of the multiple proxy models, the proxy model with the highest weighted complex correlation coefficient is selected as the target proxy model.

[0156] Step S42: Determine multiple groups of optimization results corresponding to the target optimization variables through the target proxy model.

[0157] In an exemplary embodiment, the target proxy model may be optimized using the NSGA-II algorithm to obtain a Pareto optimal solution set, and the Pareto optimal solution set is used as a plurality of groups of optimization results corresponding to the target optimization variable.

[0158] Among them, NSGA-II is a multi-objective genetic algorithm. The specific principle can be referred to the description in the relevant technology, which will not be repeated here.

[0159] The Pareto optimal solution is also called a non-dominated solution or an undominated solution. When there are multiple objective optimization variables, due to the conflict and incomparability between the objective optimization variables, a solution is the best in a certain objective optimization variable and may be the worst in other objective optimization variables. While improving any objective optimization variable, it will inevitably weaken at least one other objective optimization variable.

[0160] In some embodiments, constraints may be set for the process of determining multiple groups of optimization results corresponding to the target optimization variables. The constraints may include:

[0161] The output power of the motor is greater than or equal to a preset power threshold; and / or,

[0162] The efficiency of the motor is greater than or equal to a preset efficiency threshold.

[0163] The Pareto optimal solutions that do not meet the constraint condition are removed from the Pareto optimal solution set.

[0164] Step S43: determine a target optimization result from multiple groups of optimization results.

[0165] In some possible implementations, the MADM (Multi-attribution decision-making) method in the relevant technology can be used to determine the target optimization result from multiple groups of optimization results. The basic principle is to determine the target optimization result by considering the weights of multiple optimization results. For example, the MAVT (Multi-Attribute Value Theory) method in MADM can be used to determine the utility value of each optimization result to determine the target optimization result. For specific details, please refer to the relevant technology and will not be repeated here.

[0166] Step S44: Obtain a target difference between the target optimization result and the optimization benchmark result.

[0167] The optimization method of the motor finite element model disclosed in the present invention is a cyclic optimization process. For example, multiple structural variable sets are optimized in sequence according to the average sensitivity value. When optimizing the first structural variable set, the values ​​of multiple structural variables in the remaining multiple structural variable sets can be preset default values. After the optimization of the first structural variable set is completed, the second structural variable set with the highest average sensitivity value is selected from the remaining structural variable sets for optimization. At this time, the values ​​of the structural variables of the first structural variable set can be determined as the structural variable parameters of the first structural variable set corresponding to the target optimization result. By analogy, after all structural variable sets are completed, a new round of optimization can be continued from the first structural variable set. In an exemplary embodiment, it can be determined whether the optimization target is achieved by the MADM utility value of the target optimization result and the MADM utility value of the optimization benchmark result. When performing the first optimization, the sample result with the highest MADM utility value in the sample result set can be used as the optimization benchmark result. In the subsequent cyclic optimization, the last target optimization result can be used as the optimization benchmark result.

[0168] For example, the MADM utility value difference between the target optimization result and the optimization benchmark result may be obtained, and the MADM utility value difference may be used as the target difference.

[0169] Step S45, when the target difference is greater than or equal to the preset difference threshold, determine the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result; and update the target structural variable set according to the sensitivity value to obtain an updated target structural variable set, and again execute the step of using multiple preset algorithms to construct proxy models for the target structural variable set respectively to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold.

[0170] Among them, the target difference is greater than or equal to the preset difference threshold, indicating that the optimization of the motor has not yet reached the expected goal, and further optimization is needed.

[0171] Firstly, the undetermined parameters corresponding to each structural variable in the current target structural variable set are determined according to the target optimization result.

[0172] For example, the current target structural variable set includes permanent magnet thickness and slot width, and the values ​​of permanent magnet thickness and slot width corresponding to the target optimization result are used as the undetermined parameters, for example, the permanent magnet thickness is 3.5 mm, and the slot width is 1 mm.

[0173] Then, the target structure variable set is updated according to the sensitivity value, and multiple preset algorithms are used again to build proxy models for the target structure variable set respectively, so as to obtain multiple proxy models corresponding to the target structure variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold.

[0174] In some possible implementations, the target structure variable set can be updated according to the sensitivity value using a queue data structure method. For example, the current target structure variable set is moved to the end of the structure variable set queue, and the structure variable set at the head of the structure variable set queue is obtained from the structure variable set queue as the updated target structure variable set. Then, the steps of constructing proxy models for the target structure variable set using multiple preset algorithms are performed again to obtain multiple proxy models corresponding to the target structure variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold.

[0175] Step S46: When the target difference is less than the preset difference threshold, the undetermined parameter corresponding to each structural variable is used as the optimization parameter corresponding to the structural variable.

[0176] When the target difference is less than the preset difference threshold, it indicates that the optimization of the motor has reached the expected goal, and the undetermined parameters corresponding to each structural variable in each structural variable set are used as the optimization parameters corresponding to the result variable.

[0177] In this way, multiple structural variables are stratified by sensitivity values, and suitable proxy models are selected to perform hierarchical optimization on multiple structural variables, thereby reducing the computational workload of motor optimization and improving the efficiency of motor optimization.

[0178] Figure 2 is a flow chart of another finite element model optimization method according to an exemplary embodiment. Figure 2 As shown, the method may include the following steps:

[0179] In step S201 , a pre-established motor finite element initial model and at least one target optimization variable specified by a user are obtained.

[0180] The motor finite element initial model includes a plurality of structural variables, and the structural variables may include at least two of slot width, slot depth, pole arc coefficient and permanent magnet thickness. The target optimization variable may include one or more of maximum torque, minimum cogging torque, minimum torque ripple, motor efficiency and radial electromagnetic force.

[0181] In step S202 , the sensitivity value of each structural variable is obtained.

[0182] The step of obtaining the sensitivity value of each structural variable can refer to the description in step S102, which will not be repeated here.

[0183] In step S203 , the plurality of structural variables are divided into a plurality of structural variable sets according to the sensitivity values.

[0184] In this step, multiple structural variables are divided into multiple sets of structural variables according to sensitivity values, so as to first optimize the structural variables with higher sensitivity values. The structural variables with lower sensitivity values ​​can be fixed to preset default values, which can effectively improve the efficiency of motor optimization.

[0185] The number of the multiple structural variable sets may be two or more, and the present disclosure does not impose any limitation on this.

[0186] The specific step of dividing the multiple structural variables into multiple structural variable sets according to the sensitivity values ​​can be referred to step S103, which will not be repeated here.

[0187] In step S204, a current target structure variable set is determined from a plurality of structure variable sets.

[0188] In some possible implementations, the target structure variable set can be determined using a queue data structure method. For example, the multiple structure variables can be combined and added to a structure variable set queue according to the average sensitivity value, and then the target structure variable set can be obtained from the head of the structure variable set queue.

[0189] In step S205, multiple preset algorithms are used to construct proxy models for the target structure variable set respectively, so as to obtain multiple proxy models corresponding to the target structure variable set.

[0190] For example, after obtaining the target structure variable set, multiple preset algorithms can be used to construct proxy models for the target structure variable set respectively, where the proxy model includes a GRNN neural network model, a response surface model or a Kriging model, and the multiple preset algorithms include at least two of the GRNN neural network, response surface and Kriging.

[0191] In this step, by constructing multiple proxy models, the proxy model with the highest degree of fit can be selected as the target proxy model, thereby improving the performance of the motor optimization process. The principle of the specific preset algorithm has been described in step S104 using the response surface model as an example, and will not be repeated here.

[0192] Specifically, the sensitive interval of each structural variable is obtained, and the optimal hypercube sampling method is used to determine the sample point set of the target structural variable set within the sensitive intervals of multiple target structural variables corresponding to the target structural variable set. The sample result set of the target structural variable set is determined by finite element calculation based on the sample point set of the target structural variable set. The steps of respectively constructing proxy models of the target structural variable set using multiple preset algorithms based on the sample point set and the sample result set have been described in step S104 and will not be repeated here.

[0193] In step S206, a target proxy model corresponding to the target structural variable set is determined from the multiple proxy models.

[0194] In some possible implementations, the target proxy model corresponding to the target structure variable set may be determined through the following steps.

[0195] First, the coefficient of determination of each proxy model is determined based on the sample point set and the sample result set.

[0196] Among them, the determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model. By way of example, the determination coefficient can be a complex correlation coefficient. The specific description of the complex correlation coefficient can be found in Formula 6. When the sample result set contains sample results of multiple target optimization variables, the weighted complex correlation coefficient can be determined based on the preset weights of the multiple target optimization variables and the corresponding complex correlation coefficients.

[0197] The target proxy model corresponding to the target optimization set is determined according to the determination coefficients of the multiple proxy models.

[0198] For example, the determination coefficient may be a complex correlation coefficient, and according to the complex correlation coefficients of the multiple proxy models, the proxy model with the highest complex correlation coefficient is selected as the target proxy model. In the case where there are multiple target optimization variables, according to the weighted complex correlation coefficients of the multiple proxy models, the proxy model with the highest weighted complex correlation coefficient is selected as the target proxy model.

[0199] In step S207, multiple groups of optimization results corresponding to the target optimization variables are determined through the target proxy model.

[0200] In an exemplary embodiment, the target proxy model may be optimized using the NSGA-II algorithm to obtain a Pareto optimal solution set, and the Pareto optimal solution set is used as a plurality of groups of optimization results corresponding to the target optimization variable.

[0201] In some embodiments, constraints may be set for the process of determining multiple groups of optimization results corresponding to the target optimization variables. The constraints may include:

[0202] The output power of the motor is greater than or equal to a preset power threshold; and / or,

[0203] The efficiency of the motor is greater than or equal to a preset efficiency threshold.

[0204] The Pareto optimal solutions that do not meet the constraint condition are removed from the Pareto optimal solution set.

[0205] In step S208, a target optimization result is determined from the multiple groups of optimization results.

[0206] In some possible implementations, the MADM (Multi-attribution decision-making) method in the relevant technology can be used to determine the target optimization result from multiple groups of optimization results. The basic principle is to determine the target optimization result by considering the weights of multiple optimization results. For example, the MAVT (Multi-Attribute Value Theory) method in MADM can be used to determine the utility value of each optimization result to determine the target optimization result. For specific details, please refer to the relevant technology and will not be repeated here.

[0207] In step S209, a target difference between the target optimization result and the optimization benchmark result is obtained.

[0208] For example, whether the optimization target is achieved can be determined by the MADM utility value of the target optimization result and the MADM utility value of the optimization benchmark result. When performing the first optimization, the sample result with the highest MADM utility value in the sample result set can be used as the optimization benchmark result. In subsequent cyclic optimizations, the previous target optimization result can be used as the optimization benchmark result.

[0209] For example, the MADM utility value difference between the target optimization result and the optimization benchmark result may be obtained, and the MADM utility value difference may be used as the target difference.

[0210] In step S210, the undetermined parameters corresponding to each structural variable in the current target structural variable set are determined according to the target optimization result; and the target structural variable set is updated according to the sensitivity value.

[0211] When the target difference is greater than or equal to the preset difference threshold, it indicates that the optimization of the motor has not yet reached the expected goal, and further optimization is required.

[0212] Firstly, the undetermined parameters corresponding to each structural variable in the current target structural variable set are determined according to the target optimization result.

[0213] For example, the current target structural variable set includes permanent magnet thickness and slot width, and the values ​​of permanent magnet thickness and slot width corresponding to the target optimization result are used as the undetermined parameters, for example, the permanent magnet thickness is 3.5 mm, and the slot width is 1 mm.

[0214] Secondly, the target structure variable set is updated according to the sensitivity value, and steps S205 to S209 are re-executed.

[0215] In some possible implementations, a queue data structure method can be used to update the target structure variable set according to the sensitivity value. For example, the current target structure variable set is moved to the end of the structure variable set queue, and the structure variable set at the head of the structure variable set queue is obtained from the structure variable set queue as the updated target structure variable set.

[0216] In step 211, the undetermined parameters corresponding to each current structural variable are used as optimization parameters corresponding to the structural variable.

[0217] When the target difference is less than the preset difference threshold, it indicates that the optimization of the motor has reached the expected goal, and the undetermined parameters corresponding to each structural variable in each structural variable set are used as the optimization parameters corresponding to the result variable.

[0218] In this way, multiple structural variables are stratified by sensitivity values, and suitable proxy models are selected to perform hierarchical optimization on multiple structural variables, thereby reducing the computational workload of motor optimization and improving the efficiency of motor optimization.

[0219] Figure 3 is a block diagram of an optimization device for a motor finite element model according to an exemplary embodiment, such as Figure 3 As shown, the optimization device 300 of the motor finite element model may include:

[0220] The first acquisition module 301 is configured to acquire a pre-established motor finite element initial model and at least one target optimization variable specified by a user, wherein the motor finite element initial model includes a plurality of structural variables.

[0221] The second acquisition module 302 is configured to acquire the sensitivity value of each structural variable.

[0222] The division module 303 is configured to divide the multiple structural variables into multiple structural variable sets according to the sensitivity values.

[0223] The proxy model construction module 304 is configured to determine a target structural variable set from the multiple structural variable sets, and use multiple preset algorithms to construct proxy models for the target structural variable set respectively, so as to obtain multiple proxy models corresponding to the target structural variable set.

[0224] The optimization module 305 is configured to determine the optimization parameter corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set.

[0225] Figure 4 is a block diagram of another device for optimizing a finite element model of a motor according to an exemplary embodiment, such as Figure 4 As shown, the optimization module 305 may include:

[0226] The determination submodule 3051 is configured to determine a target proxy model corresponding to a target structural variable set from a plurality of proxy models.

[0227] The first acquisition submodule 3052 is configured to determine multiple groups of optimization results corresponding to the target optimization variables through the target proxy model; and determine the target optimization result from the multiple groups of optimization results.

[0228] The second acquisition submodule 3053 is configured to obtain a target difference between the target optimization result and the optimization benchmark result.

[0229] The loop submodule 3054 is configured to determine the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result when the target difference is greater than or equal to the preset difference threshold; and update the target structural variable set according to the sensitivity value to obtain an updated target structural variable set, and again execute the step of using multiple preset algorithms to construct proxy models for the target structural variable set respectively to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold.

[0230] The result acquisition submodule 3055 is configured to use the undetermined parameter corresponding to each current structural variable as the optimization parameter corresponding to the structural variable when the target difference is less than the preset difference threshold.

[0231] Optionally, the loop submodule 3054 is further configured to obtain an average sensitivity value corresponding to each of the structural variable sets according to the sensitivity value of each structural variable;

[0232] The structural variable set with the highest average sensitivity value among the multiple structural variable sets is used as the updated target structural variable set.

[0233] Optionally, the division module 303 is further configured to:

[0234] According to one or more preset sensitivity value thresholds, the multiple structural variables are divided into multiple structural variable sets.

[0235] Optionally, the plurality of preset algorithms include at least two of GRNN neural network, response surface and Kriging, and the proxy model building module 304 is further configured to:

[0236] Obtain the sensitive interval of each structural variable; use the optimal hypercube sampling method to determine the sample point set of the target structural variable set within the sensitive intervals of multiple target structural variables corresponding to the target structural variable set, the sample point set including the first sampling parameter of each structural variable in the target structural variable set and the second sampling parameters of the structural variables in other structural variable sets among the multiple structural variable sets except the target structural variable set; determine the sample result set of the target structural variable set through finite element calculation based on the sample point set of the target structural variable set; and use a variety of preset algorithms to respectively construct proxy models of the target structural variable set based on the sample point set and the sample result set.

[0237] Optionally, the determination submodule 3051 is further configured to:

[0238] The determination coefficient of each proxy model is determined according to the sample point set and the sample result set, and the determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model; the target proxy model corresponding to the target optimization set is determined according to the determination coefficients of multiple proxy models.

[0239] Optionally, the optimization device 300 of the motor finite element model is applied to a permanent magnet synchronous motor, and the structural variables include at least two of the slot width, slot depth, pole arc coefficient and permanent magnet thickness; the target optimization variables include at least one of the maximum torque, minimum cogging torque, minimum torque fluctuation, motor efficiency and radial electromagnetic force.

[0240] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0241] In the above technical solution, the input structural parameters are stratified by sensitivity, and a suitable proxy model is selected to perform stratified optimization on the input structural parameters, thereby reducing the computational workload of motor design and improving the efficiency of motor design.

[0242] Figure 5 FIG. 5 is a block diagram of an electronic device 500 according to an exemplary embodiment. Figure 5 As shown, the electronic device 500 may include: a processor 501 and a memory 502. The electronic device 500 may also include one or more of a multimedia component 503, an input / output (I / O) interface 504, and a communication component 505.

[0243] The processor 501 is used to control the overall operation of the electronic device 500 to complete all or part of the steps in the above-mentioned motor finite element model optimization method. The memory 502 is used to store various types of data to support the operation of the electronic device 500, and these data may include, for example, instructions for any application or method used to operate on the electronic device 500, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 502 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, disk or optical disk. The multimedia component 503 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 502 or sent through the communication component 505. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 504 provides an interface between the processor 501 and other interface modules, and the above-mentioned other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 505 is used for wired or wireless communication between the electronic device 500 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 505 may include: Wi-Fi module, Bluetooth module, NFC module, etc.

[0244] In another exemplary embodiment, a non-transitory computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned motor finite element model optimization method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 502 including program instructions, and the above-mentioned program instructions can be executed by the processor 501 of the electronic device 500 to complete the above-mentioned motor finite element model optimization method.

[0245] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the present disclosure. This application is intended to cover any variations, uses or adaptations of the present disclosure, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The specification and examples are to be considered as exemplary only, and the true scope and spirit of the present disclosure are indicated by the following claims.

[0246] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for optimizing a finite element model of a motor, characterized in that: The method comprises: Acquire a pre-established motor finite element initial model and at least one target optimization variable specified by a user, wherein the motor finite element initial model includes a plurality of structural variables; Obtaining a sensitivity value of each of the structural variables; Dividing the plurality of structural variables into a plurality of structural variable sets according to the sensitivity values; Determining a target structural variable set from the plurality of structural variable sets, and using a plurality of preset algorithms to construct proxy models for the target structural variable set respectively, so as to obtain a plurality of proxy models corresponding to the target structural variable set; Determine the optimization parameter corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set; Determining the optimization parameters corresponding to each structural variable according to the multiple proxy models corresponding to the current target structural variable set includes: Determining a target proxy model corresponding to the target structure variable set from the plurality of proxy models; Determining multiple groups of optimization results corresponding to the target optimization variables through the target proxy model; Determining a target optimization result from the multiple groups of optimization results; Obtaining a target difference between the target optimization result and the optimization benchmark result; In the case where the target difference is greater than or equal to the preset difference threshold, determining the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result; and updating the target structural variable set according to the sensitivity value to obtain the updated target structural variable set, and again executing the step of using multiple preset algorithms to construct proxy models for the target structural variable set respectively to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold; When the target difference is less than the preset difference threshold, the undetermined parameter corresponding to each current structural variable is used as the optimization parameter corresponding to the structural variable; The updating of the target structure variable set according to the sensitivity value comprises: According to the sensitivity value of each structural variable, obtaining an average sensitivity value corresponding to each set of structural variables; Taking the structure variable set with the highest average sensitivity value among the multiple structure variable sets as the updated target structure variable set; The plurality of preset algorithms include at least two of GRNN neural network, response surface and Kriging, and the method further includes: Obtaining a sensitive interval of each of the structural variables; Accordingly, the adopting of multiple preset algorithms to construct proxy models for the target structure variable set respectively includes: Determine a sample point set of the target structure variable set within the sensitive interval of the plurality of target structure variables corresponding to the target structure variable set by using an optimal hypercube sampling method, wherein the sample point set includes a first sampling parameter of each structure variable in the target structure variable set and a second sampling parameter of structure variables in other structure variable sets among the plurality of structure variable sets except the target structure variable set; Determine a sample result set of the target structure variable set by finite element calculation according to the sample point set of the target structure variable set; The proxy models of the target structure variable set are respectively constructed using the multiple preset algorithms according to the sample point set and the sample result set.

2. The method according to claim 1, characterized in that The dividing the plurality of structural variables into a plurality of structural variable sets according to the sensitivity values ​​comprises: According to one or more preset sensitivity value thresholds, the plurality of structural variables are divided into a plurality of structural variable sets.

3. The method according to claim 1, characterized in that Determining the target proxy model corresponding to the target optimization set from the plurality of proxy models comprises: Determine a determination coefficient of each of the proxy models according to the sample point set and the sample result set, wherein the determination coefficient is used to characterize the degree of fit between the sample result set corresponding to the sample point set and the proxy model; The target proxy model corresponding to the target optimization set is determined according to the determination coefficients of the plurality of proxy models.

4. The method according to any one of claims 1 to 3, characterized in that The method is applied to a permanent magnet synchronous motor; The structural variables include at least two of the slot width, the slot depth, the pole arc coefficient, and the permanent magnet thickness; The target optimization variable includes at least one of maximum torque, minimum cogging torque, minimum torque ripple, motor efficiency, and radial electromagnetic force.

5. An optimization device for a motor finite element model, characterized in that: The device comprises: A first acquisition module is configured to acquire a pre-established motor finite element initial model and at least one target optimization variable specified by a user, wherein the motor finite element initial model includes a plurality of structural variables; A second acquisition module is configured to acquire a sensitivity value of each of the structural variables; A division module, configured to divide the plurality of structural variables into a plurality of structural variable sets according to the sensitivity values; A proxy model construction module is configured to determine a target structural variable set from the plurality of structural variable sets, and to use a plurality of preset algorithms to respectively construct proxy models for the target structural variable set, so as to obtain a plurality of proxy models corresponding to the target structural variable set; An optimization module is configured to determine an optimization parameter corresponding to each structural variable according to a plurality of proxy models corresponding to the current target structural variable set; The optimization module comprises: A determination submodule is configured to determine a target proxy model corresponding to the target structure variable set from the plurality of proxy models; A first acquisition submodule is configured to determine a plurality of groups of optimization results corresponding to the target optimization variable through the target proxy model; and determine a target optimization result from the plurality of groups of optimization results; A second acquisition submodule is configured to obtain a target difference between the target optimization result and the optimization benchmark result; The loop submodule is configured to determine the undetermined parameters corresponding to each structural variable in the current target structural variable set according to the target optimization result when the target difference is greater than or equal to a preset difference threshold; and update the target structural variable set according to the sensitivity value to obtain the updated target structural variable set, and again execute the step of using multiple preset algorithms to respectively construct proxy models for the target structural variable set to obtain multiple proxy models corresponding to the target structural variable set, until the target difference between the target optimization result and the optimization benchmark result is obtained, until the target difference is less than the preset difference threshold; A result acquisition submodule is configured to use the undetermined parameter corresponding to each structural variable in the current target structural variable set as the optimization parameter corresponding to the structural variable when the target difference is less than the preset difference threshold; The loop submodule is further configured as follows: According to the sensitivity value of each structural variable, obtaining an average sensitivity value corresponding to each set of structural variables; Taking the structure variable set with the highest average sensitivity value among the multiple structure variable sets as the updated target structure variable set; The plurality of preset algorithms include at least two of GRNN neural network, response surface and Kriging, and the proxy model building module is further configured to: Acquire a sensitive interval of each of the structural variables; determine a sample point set of the target structural variable set within the sensitive intervals of the multiple target structural variables corresponding to the target structural variable set using an optimal hypercube sampling method, wherein the sample point set includes a first sampling parameter of each of the structural variables in the target structural variable set and a second sampling parameter of structural variables in other structural variable sets among the multiple structural variable sets except the target structural variable set; determine a sample result set of the target structural variable set by finite element calculation based on the sample point set of the target structural variable set; and construct the proxy models of the target structural variable set respectively using the multiple preset algorithms based on the sample point set and the sample result set.

6. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 4 are implemented.

Images