Method applied to motor topology design and neural network agent model system
Through parameterized modeling and neural network proxy model, combined with multi-objective optimization algorithm, the problems of long motor design cycle and high computational cost are solved, and more efficient design solution exploration is achieved.
Patent Information
- Application Number
- CN202510460878.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-14
- Publication Date
- 2025-08-15
AI Technical Summary
The existing motors have long design cycles, high computational costs and are difficult to explore a wider design solution space.
Parametric modeling, neural network proxy model and multi-objective optimization algorithm are adopted to train neural network models by identifying key parameters, data augmentation and preprocessing, and quickly evaluate design solutions with multi-objective optimization algorithms to reduce the dependence of finite element analysis.
Accelerate the design process, reduce manual operations, reduce calculation costs, improve design efficiency, and find better solutions in a shorter time.
Smart Images
Figure CN120493836A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of motor design, and in particular to a method and a neural network agent model system for motor topology design. Background Art
[0002] Motor topology design refers to the process of optimizing the motor's electromagnetic performance, such as torque, efficiency, and loss, by adjusting the motor's geometric structure, such as the stator slot shape and rotor pole layout, and the motor's electrical parameters, such as the number of winding turns and permanent magnet size.
[0003] Currently, in the design of motors, especially motor topology structures, traditional design methods rely on empirical formulas and finite element analysis.
[0004] This leads to the following problems:
[0005] Long design cycle: Traditional motor design relies on manual modeling, which is time-consuming and may lack innovation;
[0006] Long calculation time: The high computational complexity of finite element analysis leads to long simulation time;
[0007] Optimization is single and time-consuming: Design optimization is inefficient and makes it difficult to explore a wide range of design spaces. Summary of the Invention
[0008] In order to solve the problems of long design cycle, high computational cost and difficulty in exploring a wider range of possible design solutions in the existing technology, the present application provides a method and a neural network agent model system for motor topology design.
[0009] The embodiment of the present application is implemented as follows:
[0010] In a first aspect, the present application provides a method for motor topology design, comprising:
[0011] Define key motor parameters and build a parametric model;
[0012] Collect historical design data and perform data augmentation and preprocessing;
[0013] Train the neural network proxy model and optimize the network weights to minimize the prediction error;
[0014] A multi-objective optimization algorithm is used to explore the design space and output the optimal solution set.
[0015] In a possible implementation, defining key motor parameters and constructing a parameterized model further includes:
[0016] Identify key parameters: Determine the geometric and electrical parameters that affect motor performance, including the stator inner diameter and rotor outer diameter.
[0017] Automated modeling architecture: A flexible automated modeling system is established based on these parameters so that parameter changes can be quickly reflected in the model.
[0018] In one possible implementation, the collecting of historical design data and performing data enhancement and preprocessing further includes:
[0019] Data collection: Collect motor parameters and performance indicators from existing cases;
[0020] Data augmentation: Use techniques to generate more diverse training samples and enhance the model’s generalization capabilities;
[0021] Data cleaning and preprocessing: Clean up noise and outliers, and standardize or normalize the data to facilitate neural network training.
[0022] In one possible implementation, the training of the neural network proxy model and optimizing the network weights to minimize the prediction error further includes:
[0023] Selecting a network architecture: Choose an appropriate neural network architecture based on the motor design requirements to capture the complex relationship between input parameters and output performance;
[0024] Feature engineering: extracting key features that help improve prediction accuracy as neural network input;
[0025] Model training: Use the prepared dataset to train the selected neural network and optimize the weights to reduce prediction errors.
[0026] In one possible implementation, the step of exploring the design space using a multi-objective optimization algorithm and outputting an optimal solution set further includes:
[0027] Rapid evaluation: Use trained surrogate models to quickly evaluate new design options, reducing the number of detailed simulations.
[0028] Multi-objective optimization algorithm: Use strategies such as evolutionary algorithms to find the optimal or near-optimal design solution under multiple performance indicators.
[0029] In one possible implementation, the method of exploring the design space using a multi-objective optimization algorithm and outputting an optimal solution set further includes:
[0030] Feedback mechanism: Compare the actual simulation results with the proxy model predictions, adjust the model parameters, and continuously improve the model accuracy.
[0031] In a second aspect, the present application provides a neural network agent model system for motor topology design, comprising:
[0032] A parametric modeling module, configured to generate a parametric simulation model based on geometric parameters and electrical parameters of the motor, wherein the geometric parameters include the inner diameter of the stator and the outer diameter of the rotor;
[0033] Data preprocessing module, used to clean, standardize and enhance motor design data to generate training data sets;
[0034] A neural network proxy model module, using a multi-layer perceptron (MLP), convolutional neural network (CNN), or radial basis function network architecture, takes as input parameterized model parameters and outputs motor performance predictions;
[0035] Multi-objective optimization module, integrating genetic algorithm or particle swarm optimization algorithm.
[0036] In one possible implementation, the parametric modeling module supports parametric modeling of tapered slot geometry and embedded V-shaped rotor structure, with specific parameters including the number of slots, stator outer diameter, and permanent magnet thickness.
[0037] In one possible implementation, the input features of the neural network proxy model include stator yoke height, slot opening width, and pole V angle, and the output target variables are average torque (N·m), copper loss (W), and permanent magnet eddy current loss.
[0038] In one possible implementation, the multi-objective optimization module is implemented through the following process:
[0039] Initialize the population and generate random designs;
[0040] Call the proxy model to predict the performance of each solution;
[0041] The dominant individuals are selected based on the non-dominated sorting and crowding calculation of the NSGA-II algorithm;
[0042] Iterate the crossover and mutation operations until convergence.
[0043] The technical solution provided by this application can achieve at least the following beneficial effects:
[0044] The present application provides a method and a neural network proxy model system for motor topology design, which accelerates the design process and reduces manual operations through parametric modeling and automated design tools, introduces a machine learning-based proxy model to replace part of the finite element analysis, and simultaneously utilizes parallel computing and cloud computing technologies to speed up the calculation speed, applies a multi-objective optimization algorithm for efficient search, and uses a proxy model to quickly evaluate the design scheme, so as to find a better solution in a shorter time. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.
[0046] Figure 1 1 is a flow chart of a method for motor topology design according to an exemplary embodiment of the present application;
[0047] Figure 2 This is a structural diagram of a neural network agent model system applied to motor topology design, shown in an exemplary embodiment of the present application;
[0048] Figure 3 is a schematic diagram of a tapered groove geometry structure shown in an exemplary embodiment of the present application;
[0049] Figure 4 Schematic diagram of an embedded V-shaped geometric structure shown in an exemplary embodiment of the present application.
[0050] Reference numerals:
[0051] 1. Parametric modeling module; 2. Data preprocessing module; 3. Neural network proxy model module; 4. Multi-objective optimization module. DETAILED DESCRIPTION
[0052] In order to make the purpose, implementation methods and advantages of the present application clearer, the exemplary implementation methods of the present application will be clearly and completely described below in conjunction with the drawings in the exemplary embodiments of the present application. Obviously, the described exemplary embodiments are only part of the embodiments of the present application, not all of the embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.
[0053] It should be noted that the brief descriptions of terms in this application are only for the purpose of facilitating the understanding of the embodiments described below, and are not intended to limit the embodiments of this application. Unless otherwise specified, these terms should be understood according to their ordinary and usual meanings.
[0054] In the specification and claims of this application and the accompanying drawings, the terms "first," "second," "third," etc. are used to distinguish similar or similar objects or entities, and are not necessarily intended to limit a particular order or sequence, unless otherwise noted. It should be understood that the terms used in this manner are interchangeable under appropriate circumstances.
[0055] The terms "comprise," "include," and "have," and any variations thereof, are intended to cover but not exclude inclusion; for example, a product or device comprising a list of components is not necessarily limited to all the components expressly listed but may include other components not expressly listed or inherent to such product or device.
[0056] Before explaining the method for motor topology design provided in the embodiment of the present application, the application scenario and implementation environment of the embodiment of the present application are first introduced.
[0057] Motor topology design refers to the process of optimizing the motor's electromagnetic performance, such as torque, efficiency, and loss, by adjusting the motor's geometric structure, such as the stator slot shape and rotor pole layout, and the motor's electrical parameters, such as the number of winding turns and permanent magnet size.
[0058] Currently, in the design of motors, especially motor topology structures, traditional design methods rely on empirical formulas and finite element analysis.
[0059] This leads to the following problems:
[0060] Long design cycle: Traditional motor design relies on manual modeling, which is time-consuming and may lack innovation;
[0061] Long calculation time: The high computational complexity of finite element analysis leads to long simulation time;
[0062] Optimization is single and time-consuming: Design optimization is inefficient and makes it difficult to explore a wide range of design spaces.
[0063] Based on this, the present application provides a method and a neural network agent model system for motor topology design.
[0064] Parametric simulation modeling of motor geometry: Rapidly generate different design options by adjusting key motor geometry parameters and immediately observe their impact on motor performance.
[0065] Research on motor performance prediction based on surrogate models: Train neural network surrogate models to quickly and accurately predict motor performance under different design schemes, reducing reliance on time-consuming finite element analysis.
[0066] Neural network proxy model accelerates optimization: Using neural networks as proxy models can efficiently approximate complex functional relationships, thereby significantly reducing computing time and cost, while making it possible to explore a wider range of design spaces.
[0067] Next, the technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems will be described in detail through embodiments and in conjunction with the accompanying drawings. The various embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. Obviously, the described embodiments are only part of the embodiments of the present application, not all of them.
[0068] Figure 1 1 is a flow chart of a method for motor topology design shown in an exemplary embodiment of the present application.
[0069] In an exemplary embodiment, Figure 1 As shown, a method for motor topology design is provided. In this embodiment, the method may include the following steps:
[0070] Step 100: Define key motor parameters and build a parameterized model.
[0071] Step 200: Collect historical design data, perform data enhancement and preprocessing.
[0072] Step 300: Train the neural network proxy model and optimize the network weights to minimize the prediction error.
[0073] Step 400: Use a multi-objective optimization algorithm to explore the design space and output the optimal solution set.
[0074] In one possible implementation, the design method is specifically implemented as follows:
[0075] 1. Define key parameters and build a parameterized model
[0076] Identify key parameters: Determine the key geometric parameters (such as stator inner diameter, rotor outer diameter) and electrical parameters that affect motor performance.
[0077] Automated modeling architecture: A flexible automated modeling system is established based on these parameters so that parameter changes can be quickly reflected in the model.
[0078] 2. Dataset construction and preprocessing
[0079] Data collection: Collect motor parameters and their performance indicators from existing cases.
[0080] Data augmentation: Use techniques to generate more diverse training samples and enhance the model's generalization ability.
[0081] Data cleaning and preprocessing: Clean up noise and outliers, and standardize or normalize the data to facilitate neural network training.
[0082] 3. Design of Neural Network Agent Model
[0083] Selecting a network architecture: Choose an appropriate neural network architecture based on the motor design requirements to capture the complex relationship between input parameters and output performance.
[0084] Feature engineering: Extract key features that help improve prediction accuracy as neural network input.
[0085] Model training: Use the prepared dataset to train the selected neural network and optimize the weights to reduce prediction errors.
[0086] 4. Design Space Exploration and Optimization
[0087] Rapid evaluation: Use trained surrogate models to quickly evaluate new design options, reducing the number of detailed simulations.
[0088] Multi-objective optimization algorithm: Use strategies such as evolutionary algorithms to find the optimal or near-optimal design solution under multiple performance indicators.
[0089] Feedback mechanism: Compare the actual simulation results with the proxy model predictions, adjust the model parameters, and continuously improve the model accuracy.
[0090] It should be understood that, although the various steps in the flowcharts involved in the above-described embodiments are displayed in sequence according to the instructions, these steps are not necessarily executed in the order indicated. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.
[0091] Corresponding to the aforementioned embodiment of the method for motor topology design, and adopting the same technical concept, the present application also provides an embodiment of a neural network agent model system applied to motor topology design.
[0092] Figure 2 This is a structural diagram of a neural network agent model system applied to motor topology design, shown in an exemplary embodiment of the present application;
[0093] In an exemplary embodiment, Figure 2 As shown in FIG, the neural network agent model system applied to motor topology design includes:
[0094] A parametric modeling module 1 is used to generate a parametric simulation model based on the geometric parameters and electrical parameters of the motor, wherein the geometric parameters include the inner diameter of the stator and the outer diameter of the rotor;
[0095] Data preprocessing module 2 is used to clean, standardize and enhance the motor design data to generate a training data set;
[0096] Neural network proxy model module 3, which uses a multi-layer perceptron (MLP), convolutional neural network (CNN) or radial basis function network architecture, inputs parameterized model parameters and outputs motor performance prediction values;
[0097] Multi-objective optimization module 4 integrates genetic algorithm or particle swarm optimization algorithm.
[0098] In one possible implementation, the parametric modeling module supports parametric modeling of tapered slot geometry and embedded V-shaped rotor structure, with specific parameters including the number of slots, stator outer diameter, and permanent magnet thickness.
[0099] In one possible implementation, the input features of the neural network proxy model include stator yoke height, slot opening width, and pole V angle, and the output target variables are average torque (N·m), copper loss (W), and permanent magnet eddy current loss.
[0100] In one possible implementation, the multi-objective optimization module is implemented through the following process:
[0101] Initialize the population and generate random designs;
[0102] Call the proxy model to predict the performance of each solution;
[0103] The dominant individuals are selected based on the non-dominated sorting and crowding calculation of the NSGA-II algorithm;
[0104] Iterate the crossover and mutation operations until convergence.
[0105] Figure 3 is a schematic diagram of a tapered groove geometry structure shown in an exemplary embodiment of the present application, Figure 4 Schematic diagram of an embedded V-shaped geometric structure shown in an exemplary embodiment of the present application.
[0106] The design feature of the tapered groove is that the width of the groove is wider at the top and narrower at the bottom, such as Figure 3 The following table shows the geometric structure of the tapered groove and its related variable parameters in detail.
[0107]
[0108] like Figure 4 The following table shows in detail the geometric structure of the embedded V-type built-in rotor and its related variable parameters.
[0109]
[0110]
[0111] The sampling parameters and ranges of the built-in rotor are shown in the following table.
[0112]
[0113] The remaining parameters and expressions of the permanent magnet synchronous motor are shown in the following table.
[0114]
[0115]
[0116] The output target variable parameters are shown in the following table.
[0117]
[0118] Regarding the specific limitations of the neural network proxy model system applied to motor topology design, please refer to the limitations of the method applied to motor topology design above, which will not be repeated here. The various modules in the above-mentioned neural network proxy model system applied to motor topology design can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
[0119] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0120] The embodiments described above merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.
Claims
1. A method for motor topology design, characterized in that: include: Define key motor parameters and build a parametric model; Collect historical design data and perform data augmentation and preprocessing; Train the neural network proxy model and optimize the network weights to minimize the prediction error; A multi-objective optimization algorithm is used to explore the design space and output the optimal solution set.
2. The method for motor topology design according to claim 1, wherein: Defining key motor parameters and building a parameterized model further includes: Identify key parameters: Determine the geometric and electrical parameters that affect motor performance, including the stator inner diameter and rotor outer diameter. Automated modeling architecture: A flexible automated modeling system is established based on these parameters so that parameter changes can be quickly reflected in the model.
3. The method for motor topology design according to claim 1, wherein: The collecting of historical design data and performing data enhancement and preprocessing further includes: Data collection: Collect motor parameters and performance indicators from existing cases; Data augmentation: Use techniques to generate more diverse training samples and enhance the model’s generalization capabilities; Data cleaning and preprocessing: Clean up noise and outliers, and standardize or normalize the data to facilitate neural network training.
4. The method for motor topology design according to claim 1, wherein: The training of the neural network proxy model and optimizing the network weights to minimize the prediction error further includes: Selecting a network architecture: Choose an appropriate neural network architecture based on the motor design requirements to capture the complex relationship between input parameters and output performance; Feature engineering: extracting key features that help improve prediction accuracy as neural network input; Model training: Use the prepared dataset to train the selected neural network and optimize the weights to reduce prediction errors.
5. The method for motor topology design according to claim 1, wherein: The method of using a multi-objective optimization algorithm to explore the design space and output an optimal solution set further includes: Rapid evaluation: Use trained surrogate models to quickly evaluate new design options, reducing the number of detailed simulations. Multi-objective optimization algorithm: Use strategies such as evolutionary algorithms to find the optimal or near-optimal design solution under multiple performance indicators.
6. The method for motor topology design according to claim 5, wherein: The use of a multi-objective optimization algorithm to explore the design space and output an optimal solution set also includes: Feedback mechanism: Compare the actual simulation results with the proxy model predictions, adjust the model parameters, and continuously improve the model accuracy.
7. A neural network agent model system for motor topology design, characterized in that: include: A parametric modeling module, configured to generate a parametric simulation model based on geometric parameters and electrical parameters of the motor, wherein the geometric parameters include the inner diameter of the stator and the outer diameter of the rotor; Data preprocessing module, used to clean, standardize and enhance motor design data to generate training data sets; A neural network proxy model module, using a multi-layer perceptron (MLP), convolutional neural network (CNN), or radial basis function network architecture, takes as input parameterized model parameters and outputs motor performance predictions; Multi-objective optimization module, integrating genetic algorithm or particle swarm optimization algorithm.
8. The neural network agent model system for motor topology design according to claim 7, characterized in that: The parametric modeling module supports parametric modeling of tapered slot geometries and embedded V-shaped rotor structures, with specific parameters including the number of slots, stator outer diameter, and permanent magnet thickness.
9. The neural network agent model system for motor topology design according to claim 7, characterized in that: The input features of the neural network proxy model include stator yoke height, slot opening width, and magnetic pole V angle, and the output target variables are average torque (N·m), copper loss (W), and permanent magnet eddy current loss.
10. The neural network agent model system for motor topology design according to claim 7, characterized in that: The multi-objective optimization module is implemented through the following process: Initialize the population and generate random designs; Call the proxy model to predict the performance of each solution; The dominant individuals are selected based on the non-dominated sorting and crowding calculation of the NSGA-II algorithm; Iterate the crossover and mutation operations until convergence.