A method for constructing a high-precision motor performance database based on a dual-model strategy

Abstract

The present invention discloses a method for constructing a high-precision motor performance database based on a dual-model strategy. First, an initial motor performance data set is obtained through a Latin hypercube sampling design, and the data set is randomly divided into two groups. One group establishes an optimized proxy model for predicting motor performance. Then, the proxy model is used to predict another set of data sets, and the predicted value is compared with the true value, and the relative error of the proxy model predicting the data set is calculated. The relative error calculated by the proxy model predicting each data point is used as the output response value, and the Euclidean distance of each component of the data point with the data point for constructing the proxy model is taken as the input variable, and a support vector machine regression model is constructed as an error model for judging whether the response of the proxy model to a certain design is a high-error response. The method proposed by the present invention can reduce the time required to construct a high-precision motor performance database, and has good practicality and economy.

Description

Technical Field

[0001] The invention relates to a method for constructing a high-precision motor performance database based on a dual-model strategy, and belongs to the technical field of motor body simulation design. Background Art

[0002] Most engineering design problems require simulation analysis experiments to evaluate the target results when different design parameters are used. However, for many practical problems, a single simulation analysis with a high confidence level may take minutes, hours, or even days to complete. Optimization design, sensitivity analysis, and what-if analysis require a lot of simulation analysis calculations, which takes a lot of time. To improve this situation, a proxy model can be used to simulate the model between the design parameters and the target results. The proxy model is often used for optimization design in motor design. Its calculation results are very close to the simulation model, but the solution calculation is small.

[0003] In the article "Research on Commonly Used Surrogate Models in Multidisciplinary Design Optimization", Mu Xuefeng et al. mentioned that the optimization based on the surrogate model often obtains the initial sampling points through experimental design to train the initial surrogate model, and then searches for the demand design through the optimization algorithm, solves the demand design and adds the results to the sampling points, repeating this operation until the design is good enough. However, this method may only converge to the local optimal design and cannot find the global optimal design.

[0004] In the article "Performance Analysis and Optimization Design of High-speed Permanent Magnet Motors for Electric Turbochargers", Gu Jialei proposed a multi-objective robustness optimization method for motors based on local surrogate models. In terms of experimental design, some deterministic optimization data are selected to build a surrogate model. The surrogate model is used to predict the motor performance results to obtain the optimal design for robustness optimization. The optimal design is then simulated and analyzed, and the results are added to the surrogate model training data set. This operation is repeated until the termination condition set by the experiment is met. Although the surrogate model trained by this method is very accurate, it still produces many redundant sampling designs.

[0005] Sacks proposed a statistical method to explain the mean square error of the Kriging model in the article "Design and analysis of computer experiments". However, in practical applications, the error estimation method unique to the Kriging surrogate model cannot be extended to the application of other surrogate models, and the error estimation value of the Kriging model is sometimes far from the actual error in sensitive designs, making it difficult to apply it in actual error prediction. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide a method for constructing a high-precision motor performance database based on a dual-model strategy, thereby improving the accuracy of the proxy model while being able to judge high-error designs, and solving the problem that the proxy model takes a lot of time to reach the required accuracy.

[0007] The present invention adopts the following technical solutions to solve the above technical problems:

[0008] A method for constructing a high-precision motor performance database based on a dual-model strategy, the method is specifically as follows:

[0009] An initial data set of motor performance is obtained, and the initial data set of motor performance is randomly divided into an S data set and an X data set. A proxy model is established through the motor input variables and output response values ​​in the S data set to fit the mapping relationship between motor parameters and motor performance; the motor performance corresponding to the motor parameters in the X data set is predicted by the proxy model, the predicted value is compared with the true value, and the relative error of the proxy model in predicting the X data set is calculated, and the relative error calculated by the proxy model in predicting each data point in the X data set is used as the output response value of the proxy model;

[0010] Constructing the input variables of the error model, training the support vector machine regression model as a judgment agent model to judge whether the response of the motor design is a high error response error model;

[0011] The error model is used to screen out the high-error motor design of the proxy model, and the high-error motor design is calculated using finite element simulation analysis, and the calculation results are added to the S data set, the proxy model is updated, and the motor performance corresponding to the motor parameters in the X data set is predicted according to the updated proxy model, and the relative error of the motor parameters in the X data set predicted by the updated proxy model is calculated as the output response of the updated error model, and the input variables of the error model are updated, and the error model is retrained; the proxy model and the error model are continuously updated until the accuracy of the proxy model reaches the required accuracy or the set number of proxy model updates is met.

[0012] As a preferred solution of the present invention, the initial data set of motor performance is obtained by performing finite element model simulation analysis and calculation on the motor through Latin hypercube sampling.

[0013] As a preferred solution of the present invention, the motor design to be judged is obtained by random search, Latin hypercube sampling or each generation decision solution of the multi-objective optimization algorithm NSGA-II.

[0014] As a preferred solution of the present invention, the process of establishing and updating the proxy model is as follows:

[0015] The grid search method is used to tune the hyperparameters of the proxy model. The error model is used to guide the proxy model to obtain a high-error motor design. The motor performance of the high-error motor design is calculated through finite element simulation analysis. The calculation results are added to the dataset for training the proxy model, and the proxy model is continuously updated until the proxy model meets the accuracy requirements.

[0016] As a preferred solution of the present invention, the standard for the error model to judge the high error motor design is:

[0017] If the relative error value of a motor design is greater than the average relative error of the X data set predicted by the proxy model + p×the standard deviation of the relative error of the X data set predicted by the proxy model, then the motor design is judged to be a high-error motor design, and p is the pre-designed value.

[0018] As a preferred solution of the present invention, the input variables of the construction error model are specifically constructed as follows:

[0019] Normalize the S data set and the X data set, and for the jth component x of the data point x in the X data set j , select the jth component of all data points in the S data set that is consistent with x j The data point s corresponding to the jth component with the smallest difference * , and calculate the Euclidean distance between the two data points as the jth component of the new data point x′, and calculate the distance between the data point x and the S data set as the last component of the new data point x′; perform the same processing on each data point in the X data set, and perform the second normalization on the new data point set obtained after the processing as the input variable of the error model;

[0020] The jth component of the new data point x′ is as follows:

[0021]

[0022] Among them, d(x, s * ) represents the data points x and s * The Euclidean distance between them, n is the dimension of each data point in the X data set, d(x, S) represents the minimum distance between the data point x and the S data set, that is, d(x, S) = inf{d(x, s)|s∈S}, s1, ..., s n represents the component of data point s in the S data set, x′ j represents the jth component of the new data point x′;

[0023] The normalization formula for all data sets is:

[0024]

[0025] Among them, y jis the jth component of the data point before normalization, y′ j is the jth component of the normalized data point' is the minimum value of the jth component of all data points in the data set, is the maximum value of the jth component of all data points in the data set.

[0026] As a preferred solution of the present invention, the updating method of the error model is as follows:

[0027] The updated proxy model is used to predict the motor performance corresponding to the motor parameters in the X data set, and the predicted value is compared with the true value to calculate the relative error of the updated proxy model in predicting the X data set. The relative error calculated for each data point in the X data set predicted by the updated proxy model is used as the output response value of the updated proxy model. The input variable construction method of the error model remains unchanged, and the error model is retrained to obtain an updated error model.

[0028] A computer device comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor. When the processor executes the computer program, the steps of the method for constructing a high-precision motor performance database based on a dual-model strategy are implemented.

[0029] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for constructing a high-precision motor performance database based on a dual-model strategy are implemented.

[0030] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0031] 1. The present invention obtains an initial data set by calling a finite element model for simulation analysis and calculation according to an experimental design, and then randomly divides the data set into two groups, one group is used to train a proxy model to establish a mapping relationship between motor design parameters and motor performance, and the other group is used to train errors to establish a mapping relationship between the replacement motor design parameters and the relative errors of the proxy model predicting the motor design. This method can effectively identify the motor parameter design with a high prediction error of the proxy model, and the process is very simple and easy to understand. After the error model identifies the high-error motor design predicted by the proxy model, the finite element model can be continued to be called on the high-error design of the proxy model for simulation analysis and calculation, thereby reducing the error of the proxy model in the identified motor design, and reducing unnecessary and extremely time-consuming finite element model analysis and calculation when establishing a higher-precision database.

[0032] 2. The present invention identifies the high-error motor design of the proxy model through the error model, and continuously calls the finite element model for simulation analysis and calculation, adds the results to the data set for training the proxy model, and updates the proxy model until the accuracy or number of updates set according to the application requirements is met. The final proxy model training data set is derived from the high-error motor design of the proxy model identified by the initial experimental design and the error model. The proxy model established by the present invention can approximate the mapping relationship between motor parameters and performance, and the calculation time of the proxy model is much shorter than the simulation analysis calculation time of the finite element model. It can be used in scenarios such as motor parameter optimization design, hypothesis analysis experiments, etc. that require large-scale finite element analysis and calculation. It has strong flexibility and is suitable for the establishment of performance databases for a variety of motors. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of a method for constructing a high-precision motor performance database based on a dual-model strategy according to the present invention;

[0034] Figure 2 48-slot stator and 8-pole double-V permanent magnet flat wire motor of the embodiment;

[0035] Figure 3 is a distribution diagram of the torque waveform of the 8-slot 8-pole permanent magnet flat wire drive motor of Example 40;

[0036] Figure 4 is a distribution diagram of the core loss of the 8-slot 8-pole permanent magnet flat wire drive motor of Example 40;

[0037] Figure 5 is a distribution diagram of eddy current loss of the 8-slot 8-pole permanent magnet flat wire drive motor of Example 40;

[0038] Figure 6 is a distribution diagram of copper loss of a permanent magnet flat wire drive motor with 8 slots and 8 poles in Example 40;

[0039] Figure 7 It is a schematic diagram of the prediction of the error model for the high error design of the surrogate model;

[0040] Figure 8 It is a line graph of the relative error of the proxy model prediction results and the number of updates. DETAILED DESCRIPTION

[0041] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0042] Motor parameter design is inevitable before the motor is put into production. In the simulation analysis of motor models, the finite element analysis model has a high confidence level, but its calculation is extremely time-consuming. In scenarios where large-scale finite element model analysis and calculation are required, using a proxy model to fit the mapping relationship between the input and output of the finite element model is a more commonly used method. However, if the proxy model is to meet higher accuracy requirements, it is necessary to collect enough finite element model simulation analysis and calculation sample data to train the proxy model. The technical solution adopted by the present invention minimizes the number of finite element model simulation analysis calculations, identifies the motor parameter design with higher errors when the proxy model predicts motor performance, and only performs finite element model simulation analysis calculations on the motor design with higher errors when updating the proxy model. The higher-precision proxy model finally obtained does not need to be too time-consuming.

[0043] Taking the process of establishing the performance database of a stator 48-slot rotor 8-pole double V permanent magnet flat wire motor as an example, the implementation process of the present invention is described in detail. Figure 2 This is the structural diagram of the motor.

[0044] The overall implementation process of the present invention is as follows Figure 1 As shown, first, an initial motor performance data set is collected through experimental design and randomly divided into two groups, namely, an S data set and an X data set. A proxy model K0 is established through the S data set to predict motor performance; then, the proxy model K0 is used to predict another set of data sets, and the predicted values ​​are compared with the true values. The relative error e of the proxy model predicting the data set is calculated, and a support vector machine regression model is trained as an error model C0 for judging whether the response of the proxy model K0 to a certain design is a high-error response; finally, the error model C0 is used to continuously screen out the proxy model high-error motor design, and the finite element simulation analysis is used to calculate the high-error motor design, and the result is added to the proxy model training set S, and the proxy model is updated. The error model is updated as the proxy model is updated until the accuracy of the proxy model reaches the required accuracy or exceeds the set number of proxy model updates.

[0045] In the dual-model strategy proposed in the invention, first, the motor parameter input variables and motor performance output responses of the proxy model database to be constructed are determined. In this embodiment, the selected motor parameter input variables and initial parameters are shown in Table 1.

[0046] Table 1 Input variables and initial parameters of stator 48 slot rotor 8 pole double V permanent magnet flat wire motor

[0047] Serial number Input variable description Variable designation Initial Value 1 Stator inner diameter / mm R 73.5 2 Stator slot width / mm Wslot_open 1 3 Stator slot height / mm Wtip 1 4 Stator slot width-flat wire length / mm Wslot_Wcoil 0.9 5 Stator slot depth-flat wire width*8 / mm Hslot_Lcoil 0.5 6 Distance between stator outer diameter and slot body / mm Rso_slot 20 7 L1 Air gap length in permanent magnet slot / mm d_FB1(d_pm_1-FB_W1) 2 8 L1 permanent magnet width / mm Wpm_1 4.5 9 L1 permanent magnet length / mm Lpm_1 20 10 L1 permanent magnet bridge width / mm FB_W1 2 11 <![CDATA[Width of the L2 permanent magnet magnetic bridge / mm]]> FB_W2 2 12 <![CDATA[Air gap length in the L2 permanent magnet slot / mm]]> d_FB2(d_pm_2-FB_W2) 1.77 13 <![CDATA[Width of L2 permanent magnet / mm]]> Wpm_2 3.5 14 <![CDATA[Length of L2 permanent magnet / mm]]> Lpm_2 10 15 Flat wire width / mm Wcoil 4.3 16 Flat wire width: Flat wire length W_L_ratio(Wcoil / Lcoil) 2.5 17 Stator slot depth / mm dcoil 0.5 18 Motor stack thickness / mm lstk 150

[0048] By using Latin hypercube sampling (LHS), the values ​​of the input variables are changed, and the finite element model simulation analysis is performed to obtain the initial data set of motor performance. In this embodiment, the motor performance obtained is the torque mean value during the stable period of the motor operation condition (e.g. Figure 3 The total loss includes the motor core loss, eddy current loss and copper loss (as shown in Figure 4-Figure 6 A total of 100 sets of data were collected.

[0049] Then, the proxy model and error model are constructed through the initial data set. That is, the initial motor performance data set is randomly divided into two groups, namely the S data set and the X data set. The data set is normalized to the maximum value. The normalization formula is:

[0050]

[0051] Among them, y j is the jth component of the data point before normalization, y′ j is the jth component of the normalized data point' is the minimum value of the jth component of all data points in the data set, is the maximum value of the jth component of all data points in the data set.

[0052] The proxy model K0 is established by the motor input variable values ​​and output response values ​​in the S data set, which is used to fit the mapping relationship between the motor parameters and the motor performance. In this embodiment, the Kriging model is used to construct the proxy model; then the motor performance corresponding to the motor input variables in another set of data sets is predicted by the proxy model K0, and the predicted value is compared with the true value to calculate the relative error e of the proxy model predicting the data set. The relative error calculation formula is:

[0053]

[0054] The relative error e calculated by the proxy model K0 for predicting each data point is used as the output response value, and the Euclidean distance between the normalized data point x∈X and the data point s∈S of the constructed proxy model is taken as the input variable x′, and the relative error e as the output response value and the input variable x′ are normalized for the second time. The support vector machine regression model is trained as the error model C0 for judging whether the response of the motor design is a high error response. The jth component x′ of the error model input variable x′ j The calculation formula is:

[0055]

[0056] Where n is the dimension of the data point x, d(x, s* ) represents the data points x and s * The Euclidean distance between * It is the data point with the smallest absolute value of the difference between the jth component and the jth component of x in the data set S. d(x, S) represents the minimum distance between the data point x and the set S, that is, d(x, S) = inf{d(x, s)|s∈S}. The grid search method is used to tune the hyperparameters of the error model C0, and the error model C0 with a more accurate judgment of the design of the high-error motor of the surrogate model is obtained.

[0057] The motor design of the surrogate model to be judged can be obtained by random search, Latin hypercube sampling, or each generation of decision solutions of the multi-objective optimization algorithm NSGA-II.

[0058] like Figure 7 As shown, the first 50 data points on the left side of the dotted line are data points in the X data set, and the 30 data points on the right side of the dotted line are randomly set to test the prediction effect of the error model, wherein the data points that meet the condition (error ≥ the average value of the relative error of the motor design data set predicted by the proxy model + p×the standard deviation of the relative error of the motor design data set predicted by the proxy model) are set as the proxy model high error motor design data points, and in this embodiment, p is 0.

[0059] Finally, the error model is used to continuously screen out the high-error motor designs of the proxy model, and the high-error motor designs are calculated using finite element simulation analysis, and the results are added to the proxy model training set S, and the proxy model is updated. The motor performance corresponding to the motor parameters in the training set X is predicted based on the updated proxy model, and the relative error of the motor parameters in the training set X predicted by the updated proxy model is calculated as the output response of the updated error model, and the input variables of the updated error model are recalculated according to the above formula, and the error model is retrained. In this process, the data set X used to train the error model remains unchanged. The proxy model and the error model are continuously updated until the accuracy of the proxy model reaches the required accuracy or the set number of proxy model updates is met. Figure 8 As shown, in this embodiment, the number of updates is set to 3 times, and the average relative error of the proxy model remains basically unchanged each time, while the maximum relative error is constantly decreasing.

[0060] Based on the same inventive concept, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the aforementioned method for constructing a high-precision motor performance database based on a dual-model strategy are implemented.

[0061] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the aforementioned method for constructing a high-precision motor performance database based on a dual-model strategy are implemented.

[0062] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0063] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0064] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0066] The above implementation modes are only exemplary descriptions of the present invention and do not limit its protection scope. Those skilled in the art may also make partial changes thereto. For example, the motor type and motor performance to be optimized may be selected according to actual conditions, the proxy model type used to fit the mapping relationship between motor parameter design and motor performance may be selected according to actual conditions, the proxy model may be defined according to actual conditions to predict high-error motor design, and the appropriate error model type may be selected according to actual conditions. Any form of equivalent substitution that conforms to the purpose of the invention falls within the protection scope of the present invention.

Claims

1. A method for constructing a high-precision motor performance database based on a dual-model strategy, characterized in that: The method is specifically as follows: An initial data set of motor performance is obtained, and the initial data set of motor performance is randomly divided into an S data set and an X data set. A proxy model is established through motor input variables and output response values ​​in the S data set to fit the mapping relationship between motor parameters and motor performance. The motor performance corresponding to the motor parameters in the X data set is predicted by the proxy model, the predicted value is compared with the true value, the relative error of the proxy model in predicting the X data set is calculated, and the relative error calculated for each data point in the X data set predicted by the proxy model is used as the output response value of the proxy model; Constructing the input variables of the error model, training the support vector machine regression model as a judgment agent model to judge whether the response of the motor design is a high error response error model; The error model is used to screen out the high-error motor design of the proxy model, and the high-error motor design is calculated using finite element simulation analysis, and the calculation results are added to the S data set, the proxy model is updated, and the motor performance corresponding to the motor parameters in the X data set is predicted according to the updated proxy model, and the relative error of the motor parameters in the X data set predicted by the updated proxy model is calculated as the output response of the updated error model, and the input variables of the error model are updated, and the error model is retrained; the proxy model and the error model are continuously updated until the accuracy of the proxy model reaches the required accuracy or the set number of proxy model updates is met.

2. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The motor performance initial data set is obtained by performing finite element model simulation analysis and calculation on the motor through Latin hypercube sampling.

3. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The motor design to be judged is obtained by random search, Latin hypercube sampling or each generation decision solution of the multi-objective optimization algorithm NSGA-II.

4. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The establishment and update process of the proxy model is as follows: The grid search method is used to tune the hyperparameters of the proxy model. The error model is used to guide the proxy model to obtain a high-error motor design. The motor performance of the high-error motor design is calculated through finite element simulation analysis. The calculation results are added to the dataset for training the proxy model, and the proxy model is continuously updated until the proxy model meets the accuracy requirements.

5. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The error model determines the standard of high error motor design as follows: If the relative error value of a motor design is greater than the average relative error of the X data set predicted by the proxy model + p×the standard deviation of the relative error of the X data set predicted by the proxy model, then the motor design is judged to be a high-error motor design, and p is the pre-designed value.

6. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The input variables of the construction error model are specifically: Normalize the S data set and the X data set, and for the jth component x of the data point x in the X data set j , select the jth component of all data points in the S data set that is consistent with x j The data point s corresponding to the jth component with the smallest difference * , and calculate the Euclidean distance between the two data points as the jth component of the new data point x′, and calculate the distance between the data point x and the S data set as the last component of the new data point x′; perform the same processing on each data point in the X data set, and perform the second normalization on the new data point set obtained after the processing as the input variable of the error model; The jth component of the new data point x′ is as follows: Among them, d(x,s * ) represents the data points x and s * The Euclidean distance between them, n is the dimension of each data point in the X data set, d(x,S) represents the minimum distance between the data point x and the S data set, that is, d(x,S) = inf{d(x,s)|s∈S}, s1,...,s n represents the component of data point s in the S data set, x′ j represents the jth component of the new data point x′; The normalization formula for all data sets is: Among them, y j is the jth component of the data point before normalization, y′ j is the jth component of the normalized data point, is the minimum value of the jth component of all data points in the data set, is the maximum value of the jth component of all data points in the data set.

7. The method for constructing a high-precision motor performance database based on a dual-model strategy according to claim 1, characterized in that: The updating method of the error model is as follows: The updated proxy model is used to predict the motor performance corresponding to the motor parameters in the X data set, and the predicted value is compared with the true value to calculate the relative error of the updated proxy model in predicting the X data set. The relative error calculated for each data point in the X data set predicted by the updated proxy model is used as the output response value of the updated proxy model. The input variable construction method of the error model remains unchanged, and the error model is retrained to obtain an updated error model.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that: When the processor executes the computer program, the steps of the method for constructing a high-precision motor performance database based on a dual-model strategy as described in any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for constructing a high-precision motor performance database based on a dual-model strategy as described in any one of claims 1 to 7 are implemented.

Images