High-speed motor parameter optimization method and system based on parameterization analysis

Abstract

The invention discloses a high-speed motor parameter optimization system and method based on parameterization analysis, and relates to the technical field of data processing.The method comprises the steps that performance data of a high-speed motor under different high-speed motor parameter combinations are collected; preprocessing the data, extracting key features, and constructing a high-speed motor parameter feature matrix; constructing an initial multilayer perceptron model, dividing the high-speed motor parameter characteristic matrix into a high-speed motor parameter characteristic training matrix and a high-speed motor parameter characteristic test matrix, and setting a label matrix; performing training test and optimization on the initial multilayer perceptron model to obtain a final multilayer perceptron model; substituting the characteristic data of the high-speed motor performance to be predicted into the final multilayer perceptron model to obtain the high-speed motor performance prediction of the high-speed motor parameter combination in the high-speed motor parameter characteristic matrix; and selecting high-speed motor parameters corresponding to the optimal high-speed motor performance according to the high-speed motor performance prediction.

Description

Technical Field

[0001] The present invention belongs to the field of data processing, and in particular, relates to a high-speed motor parameter optimization method and system based on parametric analysis. Background Art

[0002] At present, the parameter optimization of high-speed motors mainly relies on traditional experimental tests and empirical adjustment methods. These methods usually involve changing various parameters of the motor (such as rotor resistance, inductance, magnetic flux, etc.) in a laboratory environment and observing their effects on motor performance (such as efficiency, temperature rise, noise, etc.). Subsequently, based on the experimental results, engineers use experience to fine-tune the parameters to achieve the optimization purpose. In addition, there are also some optimization methods based on simulation software, which simulate the motor performance under different combinations of high-speed motor parameters by establishing a mathematical model of the motor, thereby screening out the optimal parameter configuration.

[0003] The existing technical solutions still have the following disadvantages:

[0004] Time-consuming and labor-intensive: Traditional experimental testing methods require a lot of manpower, material resources and time costs, and are limited by experimental conditions, making it difficult to fully cover all possible high-speed motor parameter combinations.

[0005] Strong dependence on experience: The optimization results rely heavily on the engineer’s personal experience and judgment, lacking objectivity and accuracy.

[0006] Limited simulation accuracy: Although optimization methods based on simulation software can reduce the number of experiments, it is often difficult to balance the accuracy of the model and the simulation efficiency. In addition, it is difficult to establish a model for complex high-speed motor systems.

[0007] Low optimization efficiency: Whether it is experimental testing or simulation, there are problems such as large parameter search space and many iterations, which leads to a long and inefficient optimization process. Summary of the invention

[0008] The present invention proposes a high-speed motor parameter optimization method and system based on parametric analysis. The method combines advanced machine learning algorithms with high-precision simulation models to achieve fast and accurate optimization of high-speed motor parameters.

[0009] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0010] S1. Collect a large amount of performance data of high-speed motors under different high-speed motor parameter combinations, including experimental data and simulation data; pre-process the data by data cleaning and normalization, extract key features, and construct a high-speed motor parameter feature matrix;

[0011] S2, construct an initial multilayer perceptron model, and set the relevant configuration of the initial multilayer perceptron model; set the ratio of training data and test data, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix according to the ratio, and set a label matrix; input the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, a final multilayer perceptron model is obtained;

[0012] S3. Collect the motor parameter data required to predict the high-speed motor performance, extract features and construct a predicted high-speed motor parameter feature matrix, substitute the predicted high-speed motor parameter feature matrix into the final multi-layer perceptron model, and obtain the high-speed motor performance prediction of z kinds of high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix; according to the high-speed motor performance prediction, select the high-speed motor parameters corresponding to the optimal high-speed motor performance.

[0013] Preferably, the S1 comprises the following steps:

[0014] S11, collecting a large amount of historical performance data of high-speed motors under different high-speed motor parameter combinations to obtain original data, including experimental data and simulation data;

[0015] S12. For the original data, use the mean, median, mode or to fill missing values, identify and remove outliers, delete duplicate records in the data set, convert the data into a distribution with a mean of 0 and a standard deviation of 1, and obtain preprocessed data;

[0016] S13, extracting the main components from the preprocessed data, excluding irrelevant and redundant features, selecting the features most relevant to the motor performance, and constructing new features based on the existing features, including the combination and interaction of parameters;

[0017] S14, construct the high-speed motor parameter characteristic matrix A, as follows,

[0018]

[0019] Among them, A in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature matrix, and m represents that there are m high-speed motor parameter combinations in the high-speed motor parameter feature matrix;

[0020] By collecting a large amount of historical performance data of high-speed motors under different combinations of high-speed motor parameters, we can accumulate rich data resources to provide a basis for subsequent data analysis and model building; by preprocessing the original data, we can improve the accuracy and reliability of the data. When constructing the feature matrix, considering the combination and interaction of parameters can provide a more comprehensive understanding of the factors affecting the performance of high-speed motors and provide a basis for subsequent optimization and improvement.

[0021] Preferably, S2 comprises the following steps:

[0022] S21, constructing an initial multilayer perceptron model, the model includes an input layer, a hidden layer and an output layer, setting the number of input layer nodes of the initial multilayer perceptron model to n, the number of hidden layer nodes to a, and the number of output layer nodes to b; setting the initial weight of the multilayer perceptron model to c and the initial threshold to d;

[0023] S22, setting the training data ratio to q1 and the test data ratio to q2, dividing the high-speed motor parameter feature matrix A into a high-speed motor parameter feature training matrix B and a high-speed motor parameter feature test matrix C according to the training data ratio q1 and the test data ratio q2, as follows,

[0024]

[0025] Among them, B in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature training matrix, and o represents a total of o high-speed motor parameter combinations in the high-speed motor parameter feature training matrix; C in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature test matrix, and p represents that there are p high-speed motor parameter combinations in the high-speed motor parameter feature test matrix;

[0026] S23. Set the label matrix D as follows:

[0027]

[0028] Where D in represents the nth motor performance characteristic of the i-th high-speed motor parameter combination in the high-speed motor parameter characteristic training matrix B;

[0029] S24, inputting the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model;

[0030] S25, input the high-speed motor parameter feature test matrix into the trained multi-layer perceptron model for test optimization, and after the test optimization is completed, the final multi-layer perceptron model is obtained; Model construction and optimization: By constructing an initial multi-layer perceptron model, the high-speed motor parameter features can be learned and predicted based on the structure of the input layer, hidden layer and output layer. Setting the initial weights and thresholds can provide a starting point for model training, and continuously optimize the model parameters according to the training data.

[0031] By dividing the high-speed motor parameter feature matrix into a training matrix and a test matrix, the actual application scenario can be simulated to ensure the generalization ability of the model on unknown data; by setting the training data ratio and the test data ratio, the needs of model training and verification can be balanced; setting the label matrix to associate the high-speed motor parameter characteristics with the corresponding motor performance characteristics to provide target output for model training; inputting the training matrix and the label matrix into the initial multi-layer perceptron model for training, the model gradually learns the complex relationship between input features and output performance by adjusting weights and thresholds.

[0032] Preferably, the S24 comprises the following steps:

[0033] S241, setting a training error threshold; inputting the high-speed motor parameter feature training matrix and the label matrix into the multilayer perceptron model for training, and when the training error of the multilayer perceptron model is less than the training error threshold, stopping the training to obtain the trained multilayer perceptron model;

[0034] Setting the training error threshold can help control the training time; when the model reaches the expected accuracy, there is no need to continue training, which can save computing resources and improve training efficiency.

[0035] Preferably, the S25 comprises the following steps:

[0036] S251, inputting the high-speed motor parameter characteristic test matrix into the trained multi-layer perceptron model for testing, and obtaining the test accuracy e1 after the test is completed;

[0037] S252, setting an accuracy threshold e2; when e1≥e2, using the trained neural network prediction model as the final multilayer perceptron model; otherwise, adjusting and optimizing the initial weight c and initial threshold d of the trained multilayer perceptron model to obtain an optimized multilayer perceptron model, and using the optimized multilayer perceptron model as the final multilayer perceptron model;

[0038] By setting the accuracy threshold, you can stop training after the model reaches a certain accuracy, thereby avoiding overfitting; adjust and optimize the initial weights and initial thresholds of the model to find the optimal weights and thresholds, thereby improving the performance of the model.

[0039] Preferably, the step of optimizing the initial weights and initial thresholds of the trained multilayer perceptron model in S252 to obtain the optimized multilayer perceptron model comprises the following steps:

[0040] S2521, construct a fish school; set the size of the fish school to be f, then the fish school is represented by g = {g1, g2, ..., g i ,...,g f}, g i represents the i-th fish in the fish school; sets the maximum number of iterations to h; sets the search space dimension of the fish school to two dimensions;

[0041] S2522, randomly generate a fish school initial position set r={(r 11 ,r 12 ),(r 21 ,r 22 ),...,(r i1 ,r i2 ),...,(r f1 ,r f2 )}; r i1 and r i2 Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th fish in the fish school;

[0042] The fitness function of the fish school is set, which is recorded as the fitness function; the fitness function formula is as follows: In the formula, s1 represents the fitness function; v1 represents the bias;

[0043] S2523, start the iteration operation, set the current iteration number h1; in each round of iteration, update the position of each fish in the initial position set of the fish school according to the function formula of the fitness function; and obtain the global best individual and the global best fitness in the fish school in each round of iteration;

[0044] S2524, when h1≥h, the fish group stops the iterative operation and outputs the optimal weight j and the optimal threshold k;

[0045] S2525, substituting the optimal weight j and the optimal threshold k into the trained multilayer perceptron model to obtain the final multilayer perceptron model;

[0046] By optimizing the initial weights and initial thresholds of the trained multilayer perceptron model, the prediction performance and accuracy of the model can be further improved; using the fish school algorithm to optimize model parameters can simulate the behavior of fish schools in nature and search for the optimal solution through swarm intelligence; by setting the fitness function, the difference between the model prediction results and the actual performance characteristics can be quantified, providing a search target for the fish school algorithm; the setting of the fitness function can affect the search direction and speed of the algorithm, thereby affecting the optimization effect; through iterative operations, the fish school algorithm can continuously update the position of each fish in the fish school and gradually approach the global optimal solution; in each round of iteration, the position of each fish in the initial position set of the fish school is updated according to the functional formula of the fitness function, so as to obtain the global best individual in the fish school and the global best fitness.

[0047] S31, collect the motor parameter data required to predict the performance of the high-speed motor, extract the features and construct the predicted high-speed motor parameter feature matrix F, as follows:

[0048]

[0049] Among them, F in represents the nth feature of the ith high-speed motor parameter combination in the predicted high-speed motor parameter feature matrix, and z represents a total of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix;

[0050] S32, substituting the predicted high-speed motor parameter feature matrix F into the final multi-layer perceptron model to obtain the high-speed motor performance prediction of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix;

[0051] S33, selecting high-speed motor parameters corresponding to the optimal high-speed motor performance according to the high-speed motor performance prediction;

[0052] By substituting the predicted high-speed motor parameter feature matrix into the final multi-layer perceptron model, the high-speed motor performance prediction of the high-speed motor parameter combination is obtained; this helps to understand the motor performance under different parameter combinations and provide a basis for subsequent parameter selection; according to the high-speed motor performance prediction results, the impact of different parameter combinations on motor performance can be analyzed, and the high-speed motor parameters corresponding to the optimal high-speed motor performance can be selected.

[0053] A high-speed motor parameter optimization system based on parametric analysis includes a data collection module, a multi-layer perceptron model training and optimization module, and a multi-layer perceptron model application module.

[0054] The data collection module collects the performance data of the high-speed motor under different high-speed motor parameter combinations; pre-processes the data, extracts key features, and constructs a high-speed motor parameter feature matrix;

[0055] The multilayer perceptron model training and optimization module is used to construct an initial multilayer perceptron model, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix, and set a label matrix; train the initial multilayer perceptron model to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, obtain a final multilayer perceptron model;

[0056] The multi-layer perceptron model application module substitutes the characteristic data required to predict the high-speed motor performance into the final multi-layer perceptron model to obtain the high-speed motor performance prediction of the high-speed motor parameter combination in the high-speed motor parameter characteristic matrix; according to the high-speed motor performance prediction, the high-speed motor parameters corresponding to the optimal high-speed motor performance are selected;

[0057] The present invention has the following beneficial effects:

[0058] By constructing a feature matrix, the performance data of high-speed motors from different sources and with different parameter combinations can be integrated; at the same time, preprocessing steps such as data cleaning and normalization ensure the quality and consistency of the data, making subsequent analysis more accurate and reliable; through feature extraction and selection, the features most relevant to the performance of high-speed motors can be identified, and irrelevant and redundant features can be excluded; by constructing a feature matrix, the impact of different parameter combinations and interactions on the performance of high-speed motors can be analyzed.

[0059] By training the multi-layer perceptron model, the mapping relationship between high-speed motor performance and parameters can be established, thereby realizing performance prediction and optimized design; by setting the training error threshold and accuracy threshold, the training time can be controlled and overfitting can be avoided, thereby improving training efficiency; at the same time, optimizing and adjusting the model can improve the model's prediction accuracy and generalization ability.

[0060] By substituting the predicted high-speed motor parameter feature matrix into the final multi-layer perceptron model, the high-speed motor performance prediction of the high-speed motor parameter combination is obtained. According to the high-speed motor performance prediction result, the high-speed motor parameters corresponding to the optimal high-speed motor performance can be selected; this helps to find the best parameter combination in the design stage to achieve the optimal motor performance; by predicting the performance of the high-speed motor, the need for physical experiments can be reduced, thereby reducing R&D costs.

[0061] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0063] Figure 1 The present invention is a flow chart of a high-speed motor parameter optimization method and system based on parametric analysis. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0065] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0066] Embodiment 1:

[0067] See also Figure 1 The present invention discloses a high-speed motor parameter optimization method based on parametric analysis, comprising the following steps:

[0068] S1. Collect a large amount of performance data of high-speed motors under different high-speed motor parameter combinations, including experimental data and simulation data; pre-process the data by data cleaning and normalization, extract key features, and construct a high-speed motor parameter feature matrix;

[0069] The S1 comprises the following steps:

[0070] S11, collecting a large amount of historical performance data of high-speed motors under different high-speed motor parameter combinations to obtain original data, including experimental data and simulation data;

[0071] S12. For the original data, use the mean, median, mode or to fill missing values, identify and remove outliers, delete duplicate records in the data set, convert the data into a distribution with a mean of 0 and a standard deviation of 1, and obtain preprocessed data;

[0072] S13, extracting the main components from the preprocessed data, excluding irrelevant and redundant features, selecting the features most relevant to the motor performance, and constructing new features based on the existing features, including the combination and interaction of parameters;

[0073] S14, construct the high-speed motor parameter characteristic matrix A, as follows,

[0074]

[0075] Among them, A in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature matrix, and m represents that there are m high-speed motor parameter combinations in the high-speed motor parameter feature matrix;

[0076] S2, construct an initial multilayer perceptron model, and set the relevant configuration of the initial multilayer perceptron model; set the ratio of training data and test data, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix according to the ratio, and set a label matrix; input the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, a final multilayer perceptron model is obtained;

[0077] The S2 comprises the following steps:

[0078] S21, constructing an initial multilayer perceptron model, the model includes an input layer, a hidden layer and an output layer, setting the number of input layer nodes of the initial multilayer perceptron model to n, the number of hidden layer nodes to a, and the number of output layer nodes to b; setting the initial weight of the multilayer perceptron model to c and the initial threshold to d;

[0079] S22, setting the training data ratio to q1 and the test data ratio to q2, dividing the high-speed motor parameter feature matrix A into a high-speed motor parameter feature training matrix B and a high-speed motor parameter feature test matrix C according to the training data ratio q1 and the test data ratio q2, as follows,

[0080]

[0081] Among them, B in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature training matrix, and o represents a total of o high-speed motor parameter combinations in the high-speed motor parameter feature training matrix; C in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature test matrix, and p represents that there are p high-speed motor parameter combinations in the high-speed motor parameter feature test matrix;

[0082] S23. Set the label matrix D as follows:

[0083]

[0084] Where D in represents the nth motor performance characteristic of the i-th high-speed motor parameter combination in the high-speed motor parameter characteristic training matrix B;

[0085] S24, inputting the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model;

[0086] S25, inputting the high-speed motor parameter characteristic test matrix into the trained multi-layer perceptron model for test optimization, and after the test optimization is completed, obtaining the final multi-layer perceptron model;

[0087] The S24 comprises the following steps:

[0088] S241, setting a training error threshold; inputting the high-speed motor parameter feature training matrix and the label matrix into the multilayer perceptron model for training, and when the training error of the multilayer perceptron model is less than the training error threshold, stopping the training to obtain the trained multilayer perceptron model;

[0089] The S25 comprises the following steps:

[0090] S251, inputting the high-speed motor parameter characteristic test matrix into the trained multi-layer perceptron model for testing, and obtaining the test accuracy e1 after the test is completed;

[0091] S252, setting an accuracy threshold e2; when e1≥e2, using the trained neural network prediction model as the final multilayer perceptron model; otherwise, adjusting and optimizing the initial weight c and initial threshold d of the trained multilayer perceptron model to obtain an optimized multilayer perceptron model, and using the optimized multilayer perceptron model as the final multilayer perceptron model;

[0092] The step S252 of optimizing the initial weights and initial thresholds of the trained multilayer perceptron model to obtain the optimized multilayer perceptron model includes the following steps:

[0093] S2521, construct a fish school; set the size of the fish school to be f, then the fish school is represented by g = {g1, g2, ..., g i ,...,g f}, g irepresents the i-th fish in the fish school; sets the maximum number of iterations to h; sets the search space dimension of the fish school to two dimensions;

[0094] S2522, randomly generate a fish school initial position set r={(r 11 ,r 12 ),(r 21 ,r 22 ),...,(r i1 ,r i2 ),...,(r f1 ,r f2 )}; r i1 and r i2 Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th fish in the fish school;

[0095] The fitness function of the fish school is set, which is recorded as the fitness function; the fitness function formula is as follows: In the formula, s1 represents the fitness function; v1 represents the bias;

[0096] S2523, start the iteration operation, set the current iteration number h1; in each round of iteration, update the position of each fish in the initial position set of the fish school according to the function formula of the fitness function; and obtain the global best individual and the global best fitness in the fish school in each round of iteration;

[0097] S2524, when h1≥h, the fish group stops the iterative operation and outputs the optimal weight j and the optimal threshold k;

[0098] S2525, substituting the optimal weight j and the optimal threshold k into the trained multilayer perceptron model to obtain the final multilayer perceptron model;

[0099] S3, collecting motor parameter data required for predicting high-speed motor performance, extracting features and constructing a predicted high-speed motor parameter feature matrix, substituting the predicted high-speed motor parameter feature matrix into the final multi-layer perceptron model, and obtaining high-speed motor performance predictions for z combinations of high-speed motor parameters in the predicted high-speed motor parameter feature matrix; according to the high-speed motor performance prediction, selecting high-speed motor parameters corresponding to the optimal high-speed motor performance;

[0100] The S3 comprises the following steps:

[0101] S31, collect the motor parameter data required to predict the performance of the high-speed motor, extract the features and construct the predicted high-speed motor parameter feature matrix F, as follows:

[0102]

[0103] Among them, Fin represents the nth feature of the ith high-speed motor parameter combination in the predicted high-speed motor parameter feature matrix, and z represents a total of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix;

[0104] S32, substituting the predicted high-speed motor parameter feature matrix F into the final multi-layer perceptron model to obtain the high-speed motor performance prediction of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix;

[0105] S33, selecting high-speed motor parameters corresponding to the optimal high-speed motor performance according to the high-speed motor performance prediction;

[0106] Embodiment 2:

[0107] The present invention also discloses a high-speed motor parameter optimization system based on parametric analysis, which includes a data collection module, a multi-layer perceptron model training and optimization module, and a multi-layer perceptron model application module.

[0108] The data collection module collects the performance data of the high-speed motor under different high-speed motor parameter combinations; pre-processes the data, extracts key features, and constructs a high-speed motor parameter feature matrix;

[0109] The multilayer perceptron model training and optimization module is used to construct an initial multilayer perceptron model, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix, and set a label matrix; train the initial multilayer perceptron model to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, obtain a final multilayer perceptron model;

[0110] The multilayer perceptron model application module substitutes the characteristic data required to predict the high-speed motor performance into the final multilayer perceptron model to obtain the high-speed motor performance prediction of the high-speed motor parameter combination in the predicted high-speed motor parameter characteristic matrix; according to the high-speed motor performance prediction, the high-speed motor parameters corresponding to the optimal high-speed motor performance are selected.

[0111] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0112] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A high-speed motor parameter optimization method based on parametric analysis, characterized in that: The following steps are involved: S1. Collect a large amount of performance data of high-speed motors under different high-speed motor parameter combinations, including experimental data and simulation data; pre-process the data by data cleaning and normalization, extract key features, and construct a high-speed motor parameter feature matrix; S2, construct an initial multilayer perceptron model, and set the relevant configuration of the initial multilayer perceptron model; set the ratio of training data and test data, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix according to the ratio, and set a label matrix; input the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, a final multilayer perceptron model is obtained; S3. Collect the motor parameter data required to predict the high-speed motor performance, extract features and construct a predicted high-speed motor parameter feature matrix, substitute the predicted high-speed motor parameter feature matrix into the final multi-layer perceptron model, and obtain the high-speed motor performance prediction of z kinds of high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix; according to the high-speed motor performance prediction, select the high-speed motor parameters corresponding to the optimal high-speed motor performance.

2. A high-speed motor parameter optimization method based on parametric analysis according to claim 1, characterized in that: The S1 comprises the following steps: S11, collecting a large amount of historical performance data of high-speed motors under different high-speed motor parameter combinations to obtain original data, including experimental data and simulation data; S12. For the original data, use the mean, median, mode or to fill missing values, identify and remove outliers, delete duplicate records in the data set, convert the data into a distribution with a mean of 0 and a standard deviation of 1, and obtain preprocessed data; S13, extracting the main components from the preprocessed data, excluding irrelevant and redundant features, selecting the features most relevant to the motor performance, and constructing new features based on the existing features, including the combination and interaction of parameters; S14, construct the high-speed motor parameter characteristic matrix A, as follows, Among them, A in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature matrix, and m represents that there are m high-speed motor parameter combinations in the high-speed motor parameter feature matrix.

3. The high-speed motor parameter optimization method based on parametric analysis according to claim 1 is characterized in that: The S2 comprises the following steps: S21, constructing an initial multilayer perceptron model, the model includes an input layer, a hidden layer and an output layer, setting the number of input layer nodes of the initial multilayer perceptron model to n, the number of hidden layer nodes to a, and the number of output layer nodes to b; setting the initial weight of the multilayer perceptron model to c and the initial threshold to d; S22, setting the training data ratio to q1 and the test data ratio to q2, dividing the high-speed motor parameter feature matrix A into a high-speed motor parameter feature training matrix B and a high-speed motor parameter feature test matrix C according to the training data ratio q1 and the test data ratio q2, as follows, Among them, B in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature training matrix, and o represents a total of o high-speed motor parameter combinations in the high-speed motor parameter feature training matrix; C in represents the nth feature of the ith high-speed motor parameter combination in the high-speed motor parameter feature test matrix, and p represents that there are p high-speed motor parameter combinations in the high-speed motor parameter feature test matrix; S23. Set the label matrix D as follows: Where D in represents the nth motor performance characteristic of the i-th high-speed motor parameter combination in the high-speed motor parameter characteristic training matrix B; S24, inputting the high-speed motor parameter feature training matrix and the label matrix into the initial multilayer perceptron model for training operation to obtain a trained multilayer perceptron model; S25, inputting the high-speed motor parameter characteristic test matrix into the trained multi-layer perceptron model for test optimization, and after the test optimization is completed, obtaining the final multi-layer perceptron model.

4. A high-speed motor parameter optimization method based on parametric analysis according to claim 3, characterized in that: The S24 comprises the following steps: S241, setting a training error threshold; inputting the high-speed motor parameter feature training matrix and the label matrix into the multilayer perceptron model for training. When the training error of the multilayer perceptron model is less than the training error threshold, stopping the training to obtain the trained multilayer perceptron model.

5. The high-speed motor parameter optimization method based on parametric analysis according to claim 3 is characterized in that: The S25 comprises the following steps: S251, inputting the high-speed motor parameter characteristic test matrix into the trained multi-layer perceptron model for testing, and obtaining the test accuracy e1 after the test is completed; S252, setting an accuracy threshold e2; when e1≥e2, using the trained neural network prediction model as the final multilayer perceptron model; otherwise, adjusting and optimizing the initial weights c and initial threshold d of the trained multilayer perceptron model to obtain an optimized multilayer perceptron model, and using the optimized multilayer perceptron model as the final multilayer perceptron model.

6. A high-speed motor parameter optimization method based on parametric analysis according to claim 5, characterized in that: The step S252 of optimizing the initial weights and initial thresholds of the trained multilayer perceptron model to obtain the optimized multilayer perceptron model includes the following steps: S2521, construct a fish school; set the size of the fish school to be f, then the fish school is represented by g = {g1, g2, ..., g i ,...,g f }, g i represents the i-th fish in the fish school; sets the maximum number of iterations to h; sets the search space dimension of the fish school to two dimensions; S2522, randomly generate a fish school initial position set r={(r 11 ,r 12 ),(r 21 ,r 22 ),...,(r i1 ,r i2 ),...,(r f1 ,r f2 )}; r i1 and r i2 Respectively represent the first dimension coordinate and the second dimension coordinate of the initial position of the i-th fish in the fish school; The fitness function of the fish school is set, which is recorded as the fitness function; the fitness function formula is as follows: In the formula, s1 represents the fitness function; v1 represents the bias; S2523, start the iteration operation, set the current iteration number h1; in each iteration process, update the position of each fish in the initial position set of the fish school according to the function formula of the fitness function; and obtain the global best individual and the global best fitness in the fish school in each iteration process; S2524, when h1≥h, the fish group stops the iterative operation and outputs the optimal weight j and the optimal threshold k; S2525. Substitute the optimal weight j and the optimal threshold k into the trained multilayer perceptron model to obtain the final multilayer perceptron model.

7. The high-speed motor parameter optimization method based on parametric analysis according to claim 1 is characterized in that: The S3 comprises the following steps: S31, collect the motor parameter data required to predict the performance of the high-speed motor, extract the features and construct the predicted high-speed motor parameter feature matrix F, as follows: Among them, F in represents the nth feature of the ith high-speed motor parameter combination in the predicted high-speed motor parameter feature matrix, and z represents a total of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix; S32, substituting the predicted high-speed motor parameter feature matrix F into the final multi-layer perceptron model to obtain the high-speed motor performance prediction of z high-speed motor parameter combinations in the predicted high-speed motor parameter feature matrix; S33. According to the high-speed motor performance prediction, select high-speed motor parameters corresponding to the optimal high-speed motor performance.

8. A high-speed motor parameter optimization system based on parametric analysis according to any one of claims 1 to 7, characterized in that: Data collection module, multi-layer perceptron model training and optimization module, multi-layer perceptron model application module; The data collection module collects the performance data of the high-speed motor under different high-speed motor parameter combinations; pre-processes the data, extracts key features, and constructs a high-speed motor parameter feature matrix; The multilayer perceptron model training and optimization module is used to construct an initial multilayer perceptron model, divide the high-speed motor parameter feature matrix into a high-speed motor parameter feature training matrix and a high-speed motor parameter feature test matrix, and set a label matrix; train the initial multilayer perceptron model to obtain a trained multilayer perceptron model; input the high-speed motor parameter feature test matrix into the trained multilayer perceptron model for test optimization, and after the test optimization is completed, obtain a final multilayer perceptron model; The multilayer perceptron model application module substitutes the characteristic data required to predict the high-speed motor performance into the final multilayer perceptron model to obtain the high-speed motor performance prediction of the high-speed motor parameter combination in the predicted high-speed motor parameter characteristic matrix; according to the high-speed motor performance prediction, the high-speed motor parameters corresponding to the optimal high-speed motor performance are selected.

Images