Transformer sound field prediction method, device, equipment, medium and program product

Abstract

The invention relates to a transformer sound field prediction method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring a transformer geometric model corresponding to the magnetic control transformer and material parameters related to the magnetic control transformer; according to the transformer geometric model and the material parameters, obtaining a model simulation parameter result of the magnetic control transformer corresponding to the target working condition; inputting the model simulation parameter result into a pre-trained XGBoost magnetic control transformer sound field prediction model to obtain a sound field prediction result of the magnetic control transformer for the target working condition; the sound field prediction result comprises a sound field prediction cloud picture of the magnetic control transformer. By adopting the method, the processing efficiency of simulation data can be improved through the XGBoost model, and the sound field of the transformer can be quickly calculated.

Description

Technical Field

[0001] The present application relates to the field of transformer sound field prediction, and in particular to a transformer sound field prediction method, device, computer equipment, computer-readable storage medium and computer program product. Background Art

[0002] Magnetically controlled transformers, also known as magnetically controlled distribution transformers, are a new type of power equipment that is widely used in power systems. However, the noise generated during the operation of magnetically controlled transformers may affect the daily lives of residents. Therefore, it is necessary to study the noise of magnetically controlled transformers in the design stage before the magnetically controlled transformers are put into operation.

[0003] As a typical multi-physical field coupling device, the magnetron transformer is affected by multiple physical fields during operation, forming a relatively complex coupling relationship. In the current research methods for magnetron transformers, the calculation time for the transformer acoustic field is relatively long. Summary of the invention

[0004] Based on this, it is necessary to provide a transformer sound field prediction method, device, computer equipment, computer readable storage medium and computer program product that can quickly calculate the transformer sound field in order to solve the above technical problems.

[0005] In a first aspect, the present application provides a transformer sound field prediction method, comprising:

[0006] Obtaining a transformer geometric model corresponding to a magnetic control transformer and material parameters related to the magnetic control transformer;

[0007] According to the transformer geometric model and the material parameters, a model simulation parameter result of the magnetic control transformer corresponding to a target working condition is obtained;

[0008] The model simulation parameter results are input into a pre-trained XGBoost magnetic control transformer sound field prediction model to obtain the sound field prediction results of the magnetic control transformer for the target working condition; the sound field prediction results include a sound field prediction cloud map of the magnetic control transformer.

[0009] In one embodiment, the step of obtaining the XGBoost magnetically controlled transformer acoustic field prediction model includes:

[0010] Obtain information on various preset working conditions;

[0011] According to the plurality of preset working condition information, the transformer geometric model and the material parameters, model simulation parameter results corresponding to different preset working conditions are obtained;

[0012] The model simulation parameter results corresponding to different preset working conditions are input into a pre-built acoustic finite element calculation model to obtain the noise field distribution information of the magnetic control transformer under different preset working conditions; wherein the acoustic finite element calculation model is used to obtain the noise field distribution information of the sound wave propagation behavior in the magnetic control transformer under different preset working conditions;

[0013] The model simulation parameter results corresponding to different preset working conditions are input as training data into the XGBoost magnetic control transformer sound field prediction model to be trained to obtain the sound field prediction results of the magnetic control transformer for different preset working conditions;

[0014] Obtaining a loss value according to the difference between the noise field distribution information and the sound field prediction result;

[0015] The loss value is used to train the XGBoost magnetically controlled transformer acoustic field prediction model to be trained.

[0016] In one embodiment, before obtaining the loss value according to the difference between the noise field distribution information and the sound field prediction result, the method further includes:

[0017] The noise field distribution information of the magnetic control transformer under different preset working conditions is normalized.

[0018] In one embodiment, before using the loss value to train the XGBoost magnetically controlled transformer acoustic field prediction model to be trained, the method includes:

[0019] Obtaining a residual according to the difference between the noise field distribution information and the sound field prediction result;

[0020] A preset number of regression trees are obtained, and the residuals are fitted using the regression trees.

[0021] In one embodiment, the step of obtaining the acoustic finite element calculation model includes:

[0022] According to the material parameters, the acoustic characteristics of various media in the magnetic control transformer are determined, and the boundary conditions of the acoustic finite element calculation model are set according to the acoustic characteristics; the boundary conditions include open boundaries, absorbing boundaries or fixed boundaries;

[0023] Acquiring the sound source type and sound source intensity defined for the acoustic finite element calculation model, meshing information of the acoustic finite element calculation model, and a solver for sound field analysis;

[0024] The acoustic finite element calculation model is constructed based on the transformer geometric model, the material parameters, the boundary conditions, the sound source type, the sound source intensity, the meshing information and the solver.

[0025] In one of the embodiments, the noise field distribution information includes a sound field finite element calculation cloud map.

[0026] In a second aspect, the present application also provides a transformer sound field prediction device, comprising:

[0027] A geometric model acquisition module, used to acquire a transformer geometric model corresponding to a magnetic control transformer and material parameters related to the magnetic control transformer;

[0028] A simulation parameter acquisition module, used for obtaining a model simulation parameter result of the magnetic control transformer corresponding to a target working condition according to the transformer geometric model and the material parameters;

[0029] A prediction result acquisition module is used to input the model simulation parameter results into a pre-trained XGBoost magnetic control transformer sound field prediction model to obtain the sound field prediction results of the magnetic control transformer for the target working condition; the sound field prediction results include a sound field prediction cloud map of the magnetic control transformer.

[0030] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described in the first aspect when executing the computer program.

[0031] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the method described in the first aspect.

[0032] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which implements the steps of the method described in the first aspect when executed by a processor.

[0033] The transformer sound field prediction method, device, computer equipment, computer-readable storage medium and computer program product first obtain the transformer geometric model corresponding to the magnetically controlled transformer and the material parameters related to the magnetically controlled transformer; then, according to the transformer geometric model and the material parameters, obtain the model simulation parameter results of the magnetically controlled transformer corresponding to the target working condition; finally, the model simulation parameter results are input into the pre-trained XGBoost magnetically controlled transformer sound field prediction model to obtain the sound field prediction results of the magnetically controlled transformer for the target working condition; the sound field prediction results include the sound field prediction cloud map of the magnetically controlled transformer; the present application uses the geometric model and material parameters of the transformer for simulation, so as to accurately simulate the sound field characteristics of the transformer under specific working conditions, and uses the pre-trained XGBoost model to process complex simulation data in a short time, so as to quickly obtain the sound field prediction results. Compared with the method of calculating the simulation data by the finite element method to obtain the sound field prediction results in the prior art, the present application improves the processing efficiency of the simulation data by the XGBoost model, and can quickly calculate the transformer sound field. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A schematic diagram of a flow chart of a transformer sound field prediction method in one embodiment;

[0036] Figure 2 A schematic flow chart of the steps of obtaining an XGBoost magnetically controlled transformer acoustic field prediction model in one embodiment;

[0037] Figure 3 It is a structural schematic diagram of a transformer geometric model of a magnetic control transformer in one embodiment;

[0038] Figure 4 A schematic diagram of the effect of sound field calculation obtained by finite element method calculation in one embodiment;

[0039] Figure 5 A schematic diagram of the effect of noise field prediction results in one embodiment;

[0040] Figure 6 is a structural block diagram of a transformer sound field prediction device in one embodiment;

[0041] Figure 7FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] In an exemplary embodiment, Figure 1 As shown, a transformer sound field prediction method is provided. This embodiment uses the method applied to a server as an example for illustration. It can be understood that the method can also be applied to a terminal, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0044] Step S202, obtaining a transformer geometric model corresponding to the magnetic control transformer and material parameters related to the magnetic control transformer.

[0045] Among them, the above-mentioned magnetic control transformer may refer to a special type of transformer that uses magnetic control electronic devices to control and adjust voltage. The magnetic control transformer uses magnetic control electronic devices such as magnetrons and magnetron triodes to control voltage. The magnetic control transformer may be composed of a magnetron and a transformer. The magnetron is used as a switching tube to adjust the output voltage by changing the on and off states of the magnetron. When the magnetron is turned on, the input voltage of the transformer will be transferred to the output end, thereby obtaining a high voltage output; and when the magnetron is turned off, the input voltage of the transformer will be isolated, thereby obtaining a low voltage output. The specific type of the magnetic control transformer can be determined according to actual needs and is not limited here.

[0046] Among them, the transformer geometric model can refer to the equivalent 3D model obtained by geometric modeling of the magnetic control transformer, that is, the model used to represent the structure and electrical characteristics of the transformer; the various parts of the transformer can be decomposed into different components by geometric modeling of the physical structure of the transformer, and the equivalent 3D model can be established according to their position, shape and material properties. In the transformer geometric model, the parts considered may include the core, winding, insulation layer, etc. By geometrically modeling the structure of the transformer, the electrical characteristics of the transformer, such as the resistance, inductance, mutual inductance and other parameters of the transformer, can be more accurately analyzed and predicted. By geometrically modeling the physical structure of the transformer, a more accurate equivalent circuit model can be established, which is helpful to analyze and optimize the performance of the transformer.

[0047] Among them, the material parameters related to the magnetic control transformer may refer to the physical properties related to the transformer constituent materials, such as density, elastic modulus, damping coefficient, sound velocity, etc. Material parameters may also refer to a series of parameters describing the properties of viscous media, which may include but are not limited to sound pressure parameters, flow rate parameters, temperature parameters, and viscosity parameters, etc., and may also include other parameters, such as density, specific heat capacity, thermal conductivity, etc. The various medium characteristic parameters can be specifically determined according to actual needs and are not limited here.

[0048] The above-mentioned viscous medium may refer to the insulating material used to fill the inside of the magnetron transformer, which can be used for insulation, cooling and transmission of magnetic field. The viscous medium may include mineral oil or synthetic oil in the magnetron transformer, and may also include insulating paper, tape, etc. A suitable viscous medium can improve the insulation performance, heat dissipation effect and arc resistance of the transformer, and ensure the normal operation and safety of the transformer.

[0049] Exemplarily, the server obtains the transformer geometric model corresponding to the magnetic control transformer and the material parameters related to the magnetic control transformer. Among them, the specific acquisition process of the transformer geometric model may include: geometric modeling of the magnetic control transformer, including information such as the transformer's outer dimensions, winding structure, and core shape. By measuring and analyzing the actual structure of the magnetic control transformer, an equivalent circuit model can be established to calculate parameters such as magnetic field distribution and electromagnetic induction. In addition, the material parameters of various media of transformer oil, such as dielectric constant, dielectric constant, conductivity, and magnetic permeability, can also be obtained. These parameters can be obtained by experimental measurement, reference, or simulation calculation. This process can help us obtain the geometric model of the magnetic control transformer and the characteristic parameters of the viscous medium. An accurate geometric model helps to establish an accurate electromagnetic field model, so that the magnetic field distribution in the magnetic control transformer can be accurately predicted, which is crucial for optimizing the design and operation parameters of the transformer. By understanding the characteristic parameters of the material, we can better evaluate the insulation between the winding and the core, thereby optimizing the insulation design and improving the safety and stability of the magnetic control transformer. Finally, by analyzing the geometric model and material characteristic parameters of the magnetic control transformer, its structure and working conditions can be optimized, and efficiency and performance can be improved.

[0050] Step S204, obtaining the model simulation parameter results of the magnetic control transformer corresponding to the target working condition according to the transformer geometric model and material parameters.

[0051] Among them, the model simulation parameter results can refer to the sound field simulation using computer simulation software based on the transformer geometry model and material parameters, and the distribution of the sound field generated by the transformer under specific working conditions (such as frequency, power, etc.) can be obtained through numerical simulation, including parameters such as sound pressure and sound intensity. The target working condition can refer to a specific pre-set operating condition or working state, such as different load conditions, ambient temperature, input voltage, etc.

[0052] Exemplarily, the server can import the created geometric model into the sound field simulation software to realize the sound field analysis; set the working parameters: set the working conditions of the target working conditions, such as input current frequency, voltage and load conditions; simulation run: run the sound field simulation, the software will calculate the sound field distribution and output the model simulation parameter results.

[0053] Step S206, input the model simulation parameter results into the pre-trained XGBoost magnetic control transformer sound field prediction model to obtain the sound field prediction results of the magnetic control transformer for the target working condition. The sound field prediction results include the sound field prediction cloud map of the magnetic control transformer.

[0054] Among them, XGBoost can refer to a machine learning algorithm based on boosting tree (Boosting) for classification and regression tasks. In this method, the XGBoost model is pre-trained and can predict the sound field of the transformer based on the input features (model simulation parameter results).

[0055] Among them, the sound field prediction cloud map can refer to the visual expression of the sound field prediction results, which can be a two-dimensional or three-dimensional heat map, indicating the distribution of sound pressure level, sound intensity or other acoustic indicators in space, which is convenient for analyzing the noise characteristics of the transformer. The above sound field prediction results are data and charts obtained after quantitative analysis of the sound distribution and propagation in the magnetic control transformer. These results may include sound pressure level distribution diagrams, sound pressure level curves and sound field diagrams, which are intended to show the distribution of sound in space. In the sound field prediction results of the magnetic control transformer, the results may also cover sound field distribution diagrams at different frequencies, showing the intensity and directional distribution of sound in space. Through the sound field prediction results, the acoustic performance of the magnetic control transformer can be evaluated, so as to optimize the design and adjust related parameters to improve the propagation effect and quality of sound. In addition, the sound field prediction results can also include sound pressure level distribution curves at different frequencies, sound field uniformity analysis and sound field attenuation analysis, etc., to comprehensively evaluate the overall performance of the acoustic system.

[0056] For example, the server can extract characteristic parameters such as sound pressure, sound intensity, frequency, etc. from the model simulation results, organize them into input characteristic data, and then input the extracted characteristic data into the XGBoost model to obtain the sound field prediction results of the transformer under the target working conditions. The sound field prediction results can be plotted into cloud maps through data visualization tools to show the sound pressure level or sound intensity distribution at different locations.

[0057] Exemplarily, the server can first preprocess the sound field data in the model simulation parameter results to ensure the quality and relevance of the input data. The first step of preprocessing is data cleaning, which checks whether there are missing values, outliers or noise interference in the data set, and performs corresponding processing. Missing values ​​can be processed by interpolation or deleting related records to avoid negative impact on model training. At the same time, outliers are identified and corrected to ensure the accuracy of the data. After completing data cleaning, feature selection is performed. The correlation between each feature and the target variable (noise level) is analyzed, and the features with the greatest impact on model prediction are identified through statistical analysis and visualization methods (such as correlation heat maps and scatter plots). The purpose of feature selection is to remove redundant and irrelevant features to reduce model complexity and improve the generalization ability of the model. After the data preprocessing is completed, the XGBoost model is constructed. First, the data set is divided, and the preprocessed data is divided into a training set and a test set to ensure that the model can be evaluated on unseen data. Then, the XGBoost model is trained using the training set. During the training process, the key parameters of the model (such as learning rate, tree depth, subsample ratio, etc.) were tuned, and the best parameter combination was found through cross-validation to improve the prediction accuracy of the model. After multiple rounds of training and verification, an XGBoost model with excellent performance was obtained. Finally, the trained model was used to predict the noise of the transformer under new working conditions and evaluate its noise performance under different working conditions. The accuracy of the model can be further verified by comparing with the actual measured values. This series of steps not only enhances the understanding of the noise characteristics of the transformer, but also provides solid data support for subsequent noise control and optimization design. By deeply analyzing the prediction results of the model, the key factors of noise generation can be identified, thereby providing targeted suggestions for improving transformer design and operation, and ultimately achieving the purpose of optimizing equipment performance.

[0058] In the above-mentioned transformer sound field prediction method, by using the geometric model and material parameters of the transformer for simulation, the sound field characteristics of the transformer under specific working conditions can be accurately simulated. By using the pre-trained XGBoost model, complex simulation data can be processed in a short time, thereby quickly obtaining the sound field prediction results. Compared with the method of calculating the simulation data by the finite element method to obtain the sound field prediction results in the prior art, the present application improves the processing efficiency of the simulation data through the XGBoost model, and can quickly calculate the transformer sound field.

[0059] In an exemplary embodiment, Figure 2 As shown, the steps of obtaining the XGBoost magnetically controlled transformer sound field prediction model involved in the above step S206 may specifically include:

[0060] Step S301, obtaining a variety of preset working condition information.

[0061] Among them, the preset operating condition information may refer to various operating scenarios (such as load current, operating frequency, rated voltage, etc.) set when performing transformer sound field testing. These operating conditions will affect the operating characteristics of the transformer and the sound field generated.

[0062] Step S302, obtaining model simulation parameter results corresponding to different preset working conditions according to a plurality of preset working condition information, transformer geometric model and material parameters.

[0063] The model simulation parameter results corresponding to different preset working conditions may refer to a set of acoustic parameters corresponding to the preset working conditions obtained through numerical simulation based on the transformer geometric model and material parameters.

[0064] Step S303, inputting the model simulation parameter results corresponding to different preset working conditions into a pre-built acoustic finite element calculation model to obtain the noise field distribution information of the magnetic control transformer under different preset working conditions.

[0065] The acoustic finite element calculation model is used to obtain noise field distribution information of the acoustic wave propagation behavior in the magnetic control transformer for different preset working conditions. The acoustic finite element calculation model may refer to a mathematical model that uses the finite element method (FEM) to simulate the propagation of acoustic waves inside and around the transformer.

[0066] The noise field distribution information may refer to information obtained through an acoustic finite element calculation model, which describes the distribution characteristics of the sound pressure, sound intensity or other acoustic parameters of the transformer in space under different preset working conditions.

[0067] Step S304, inputting the model simulation parameter results corresponding to different preset working conditions as training data into the XGBoost magnetically controlled transformer sound field prediction model to be trained, and obtaining the sound field prediction results of the magnetically controlled transformer for different preset working conditions.

[0068] Step S305: obtaining a loss value according to the difference between the noise field distribution information and the sound field prediction result.

[0069] Step S306: Use the loss value to train the XGBoost magnetically controlled transformer sound field prediction model to be trained.

[0070] Among them, the loss value can refer to an indicator used to measure the deviation between the model prediction results and the actual results.

[0071] Exemplarily, the server obtains a variety of preset working condition information, and then obtains corresponding simulation operating conditions based on the analysis of the various preset working condition information, then inputs the transformer geometric model and its material parameters into the simulation software, sets the corresponding simulation operating conditions, performs sound field simulation, and obtains model simulation parameter results; then the obtained model simulation parameter results are input into the constructed acoustic finite element calculation model, and the solver is run to obtain corresponding noise field distribution information; then the model simulation parameter results under different preset working conditions are organized into an input feature data set, labeled with the corresponding noise field distribution information, and the model is initialized using XGBoost, hyperparameters are set, and the feature data is input into the model as a training set for initial training. During the model training process, the loss value between the sound field results predicted by the model and the actual noise field distribution information is calculated; the back propagation algorithm is applied, and the model parameters are adjusted with the obtained loss value, and the XGBoost model weights are iteratively updated through optimization algorithms such as stochastic gradient descent, and finally a trained XGBoost magnetically controlled transformer sound field prediction model is obtained.

[0072] In this embodiment, model simulation parameter results corresponding to different preset working conditions are obtained through a variety of preset working condition information, and then the noise field distribution information under different preset working conditions is obtained through a pre-constructed acoustic finite element calculation model. The model simulation parameters are used as input and the noise field distribution information is used as a control to train the XGBoost magnetically controlled transformer sound field prediction model. That is, the calculation accuracy of the pre-constructed acoustic finite element calculation model is used to improve the output accuracy of the XGBoost magnetically controlled transformer sound field prediction model, thereby improving the prediction accuracy of the sound field prediction results.

[0073] In an exemplary embodiment, before the above step S305, the following step may be further included: normalizing the noise field distribution information of the magnetic control transformer under different preset working conditions.

[0074] Among them, normalization processing can refer to transforming data in different rows or different ranges to the same scale (for example, between 0 and 1) for easy comparison and calculation. Normalization can reduce the impact of different dimensions on model training and evaluation.

[0075] In this embodiment, the deviation caused by different dimensions and ranges of dependent variables can be reduced through normalization processing, making the model training more stable and improving the prediction accuracy of the model.

[0076] In an exemplary embodiment, before the above step S306, the following steps may also be included: obtaining residuals according to the difference between the noise field distribution information and the sound field prediction results; obtaining a preset number of regression trees, and fitting the residuals using the regression trees.

[0077] Among them, residual can refer to the difference between the actual observed value and the model predicted value in regression analysis, which can reflect the part of data variation that the model cannot explain. Regression tree can refer to a decision tree model, which is specifically used for regression tasks. Regression tree divides the input space and gradually optimizes a set of input features to minimize the mean square error of the output value in each leaf node. Each leaf node represents a range of predicted values. Fitting can refer to the process of learning the characteristics of a data set in machine learning. The model minimizes the error between its prediction of given data and the actual result by adjusting its parameters.

[0078] Exemplarily, the server prepares the actual noise field distribution information and the sound field prediction results of the model, and then calculates the residual for each sample to form a residual vector. Before training the XGBoost model, the number of regression trees is set, and the obtained residual is used as the target variable for regression tree training. XGBoost adaptively adjusts the structure of the tree to minimize the residual in each tree. In each step, XGBoost recalculates the residual based on the prediction results of the previous tree, and then builds a new tree to fit the current residual. This process is iterated until the preset number of trees is reached or other stopping conditions are met. After introducing each new regression tree, XGBoost updates the prediction by fitting the residual, thereby gradually reducing the error.

[0079] In this embodiment, by fitting the residuals with a regression tree, the model can better capture unexplained changes in the data, thereby improving the accuracy of the entire sound field prediction model.

[0080] In an exemplary embodiment, the step of obtaining the acoustic finite element calculation model involved in the above step S305 may specifically include:

[0081] Step S11, determining the acoustic properties of various media in the magnetic control transformer according to the material parameters, and setting the boundary conditions of the acoustic finite element calculation model according to the acoustic properties, wherein the boundary conditions include open boundaries, absorbing boundaries or fixed boundaries.

[0082] Among them, acoustic properties can refer to the performance of materials under the action of sound waves, including sound speed, acoustic impedance, attenuation characteristics, etc. Acoustic properties determine the way sound waves propagate in the material. Boundary conditions can refer to the restrictions or conditions imposed on the surface of the model in finite element analysis, which determine the behavior of sound waves at the boundary of the model. Open boundaries can mean that sound waves can propagate freely without reflection. Applicable to open spaces that are considered to be non-reflective. Absorbing boundaries: used to simulate the absorption of sound wave energy and reduce reflections to more realistically simulate the acoustic environment. Fixed boundaries: Fixed sound wave propagation conditions, usually used in situations where certain boundaries do not allow sound waves to propagate.

[0083] Exemplarily, the server collects relevant acoustic parameters of various materials in the magnetically controlled transformer, applies the material parameters to calculate the sound velocity, acoustic impedance and other characteristics of each medium to determine their acoustic behavior, and selects appropriate boundary conditions based on the calculated acoustic characteristics.

[0084] Step S12, obtaining the sound source type and sound source intensity defined for the acoustic finite element calculation model, the meshing information of the acoustic finite element calculation model, and the solver for sound field analysis.

[0085] Among them, the sound source type can refer to the nature of the sound source, such as point sound source, surface sound source, body sound source, etc.; the sound source intensity can indicate the size of the sound energy emitted by the sound source, usually in decibels (dB) or watts (W).

[0086] Among them, meshing can refer to the process of dividing a continuous medium into discrete small units in finite element analysis for numerical calculation. The quality and density of the mesh have an important influence on the accuracy of the calculation results. Solvers can be algorithms and systems used to solve finite element models, determine the sound field distribution and sound wave propagation behavior, and different solvers may have different numerical solution methods and algorithm characteristics.

[0087] Exemplarily, the server selects the type and intensity of the sound source according to the design requirements, uses numerical tools to generate the mesh, adjusts the mesh density according to the complexity of the model and the required accuracy, ensures high resolution in key areas, selects a suitable solver according to the nature of the problem, and sets corresponding solution parameters to meet specific accuracy and convergence requirements.

[0088] Step S13, constructing an acoustic finite element calculation model based on the transformer geometric model, material parameters, boundary conditions, sound source type, sound source intensity, meshing information and solver.

[0089] Exemplarily, the server integrates the geometric model, material parameters, boundary conditions, sound source type, sound source intensity, meshing information and selected solver settings of the transformer, builds an acoustic calculation model using finite element analysis software, draws geometry, applies material properties and boundary conditions, and configures the sound source.

[0090] In this embodiment, through accurate calculation of material parameters and acoustic properties, the acoustic finite element calculation model can more realistically reflect the propagation characteristics of sound waves in the magnetic control transformer, and can provide reliable data support for the subsequent training of the XGBoost magnetic control transformer sound field prediction model.

[0091] In an exemplary embodiment, the noise field distribution information includes a sound field finite element calculation cloud map.

[0092] The sound field finite element calculation cloud map may refer to: a visualization form of sound field calculation, which represents the distribution of sound field data through colors and patterns. In the cloud map, colors usually represent different values ​​of parameters such as sound pressure level and sound intensity.

[0093] In this embodiment, the sound field finite element calculation cloud map provides an intuitive visual representation of complex data, allowing researchers to quickly understand the distribution characteristics of the sound field and the changes in important parameters.

[0094] In an exemplary embodiment, the present application also provides a transformer sound field prediction method, which may specifically include:

[0095] Obtaining a transformer geometric model corresponding to the magnetic control transformer and material parameters related to the magnetic control transformer;

[0096] According to the transformer geometric model and material parameters, the model simulation parameter results of the magnetic control transformer corresponding to the target working conditions are obtained;

[0097] The model simulation parameter results are input into the pre-trained XGBoost magnetically controlled transformer sound field prediction model to obtain the sound field prediction results of the magnetically controlled transformer for the target working condition; the sound field prediction results include the sound field prediction cloud map of the magnetically controlled transformer.

[0098] Among them, the steps for obtaining the XGBoost magnetic control transformer sound field prediction model include:

[0099] Obtain a variety of preset working condition information; obtain model simulation parameter results corresponding to different preset working conditions according to a variety of preset working condition information, transformer geometric model and material parameters; input the model simulation parameter results corresponding to different preset working conditions into a pre-built acoustic finite element calculation model to obtain noise field distribution information for the magnetic control transformer under different preset working conditions; wherein the noise field distribution information includes an acoustic field finite element calculation cloud diagram; wherein the acoustic finite element calculation model is used to obtain noise field distribution information of the sound wave propagation behavior in the magnetic control transformer under different preset working conditions; input the model simulation parameter results corresponding to different preset working conditions as training data into the XGBoost magnetic control transformer acoustic field prediction model to be trained to obtain the acoustic field prediction results of the magnetic control transformer under different preset working conditions; normalize the noise field distribution information for the magnetic control transformer under different preset working conditions. According to the difference between the noise field distribution information and the acoustic field prediction results, obtain the loss value; according to the difference between the noise field distribution information and the acoustic field prediction results, obtain the residual; obtain a preset number of regression trees, and use the regression trees to fit the residual. The loss value is used to train the XGBoost magnetically controlled transformer acoustic field prediction model to be trained.

[0100] Among them, the steps of obtaining the acoustic finite element calculation model include: determining the acoustic characteristics of various media in the magnetic control transformer according to the material parameters, and setting the boundary conditions of the acoustic finite element calculation model according to the acoustic characteristics; the boundary conditions include open boundaries, absorbing boundaries or fixed boundaries; obtaining the sound source type and sound source intensity defined for the acoustic finite element calculation model, the meshing information of the acoustic finite element calculation model, and the solver for sound field analysis; constructing the acoustic finite element calculation model based on the transformer geometric model, material parameters, boundary conditions, sound source type, sound source intensity, meshing information and solver.

[0101] In an exemplary embodiment, the present application also provides a method for quickly calculating the acoustic field of a magnetically controlled transformer based on XGBoost, comprising the following steps: first, obtaining a geometric model of the transformer and related dielectric characteristic parameters, wherein the geometric model includes the structure of the winding, the core, and the insulating material (such as Figure 3As shown, an optional transformer geometric model of a magnetically controlled transformer obtained in this application is shown), and the medium characteristic parameters include physical properties such as transformer oil (such as dielectric constant, viscosity and density). Based on the above model and parameters, a noise prediction model is constructed, using the XGBoost (extreme gradient boosting) algorithm. This algorithm is particularly suitable for processing large-scale data and complex nonlinear relationships, and improves prediction accuracy by gradually optimizing the model and reducing errors. During the model construction process, the data set is divided into a training set and a test set to ensure the generalization ability of the model. During the training process, a decision tree is used as a base learner, and multiple trees are iteratively constructed to reduce the prediction error. The cross-validation method is used to evaluate different parameter combinations and find the best settings. After the training is completed, the model can predict the noise of the transformer under new working conditions, and generate a noise intensity distribution diagram and a noise level change curve, which are compared with the actual measured values ​​to verify the accuracy of the model.

[0102] In an exemplary embodiment, the present application also provides a method for establishing a finite element calculation model of the noise field of a magnetically controlled transformer under multiple operating conditions, comprising the following steps: establishing a geometric model, that is, a transformer geometric model. Based on the geometric model, the magnetic field of the magnetically controlled transformer can be transiently calculated. During the calculation, the solution domain is set to air and the transformer body, and the discretization of the linear unit is selected to reduce the amount of calculation. By performing a parametric scan, the transient electromagnetic field calculation with a winding current range of 1A-50A and a step size of 1A is calculated. The solution step size can be set to 0.0005s, and the calculation time is a total of 0.1s. The electromagnetic force on the winding obtained by the magnetic field calculation and the magnetization intensity in the iron core are used as excitations and added to structural mechanics and pressure acoustics as excitation sources for winding vibration and magnetostrictive vibration of the iron core, so as to perform acoustic-structural coupling. The noise distribution obtained by the finite element calculation method can be shown as follows Figure 4 As shown, the figure shows the calculated acoustic field effect of the magnetic control transformer.

[0103] In an exemplary embodiment, the present application also provides a method for rapid calculation and prediction of the sound field of a multifunctional magnetically controlled transformer based on XGBoost, comprising the following steps: deriving the principle of XGBoost, the gradient boosting decision tree (GBDT) is an additive model based on the idea of ​​boosting integration, and adopts the forward distribution algorithm for greedy learning during training. Each iteration learns a CART tree to fit the residual between the prediction results of the previous t-1 trees and the true value of the training sample. Since the accuracy of the prediction results of a single tree model cannot achieve good results, multiple tree models are used at the same time, and the complexity of the model is measured by defining the structure and complexity of the tree, thereby effectively controlling the overfitting problem of the model algorithm. The final prediction result is the sum of the scores of each tree model, and the objective function model can be written as:

[0104]

[0105] In the above formula For a collection of functions Functions in ; is the set of all regression trees in the model; K is the number of all regression trees in the model. The objective function of the algorithm is:

[0106]

[0107] Continuously add helps minimize the objective, where represents the predicted value of training sample i at the t-1th iteration, is the training loss function of the model, is the regularization term. Each round of iterative training is to minimize the loss function and make the training of each tree reach the target accuracy. The loss function of XGBoost not only measures the fitting error of the model, but also limits the complexity of the tree through the penalty term (regularization term) of the tree complexity, thereby effectively preventing the model from overfitting. Among them, the final prediction result can be as follows Figure 5 As shown, NIOISE-Predict represents the noise field prediction result, and pressure (pa) represents the decibel size of the noise.

[0108] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can 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, and 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 part of the steps or stages in other steps.

[0109] Based on the same inventive concept, the embodiment of the present application also provides a transformer sound field prediction device for implementing the transformer sound field prediction method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more transformer sound field prediction device embodiments provided below can refer to the limitations of the transformer sound field prediction method above, and will not be repeated here.

[0110] In an exemplary embodiment, Figure 6 As shown, a transformer sound field prediction device is provided, including: a geometric model acquisition module 901, a simulation parameter acquisition module 902 and a prediction result acquisition module 903, wherein:

[0111] A geometric model acquisition module 901 is used to acquire a transformer geometric model corresponding to a magnetic control transformer and material parameters related to the magnetic control transformer;

[0112] A simulation parameter acquisition module 902 is used to obtain the model simulation parameter results of the magnetic control transformer corresponding to the target working condition according to the transformer geometric model and the material parameters;

[0113] The prediction result acquisition module 903 is used to input the above-mentioned model simulation parameter results into the pre-trained XGBoost magnetically controlled transformer sound field prediction model to obtain the sound field prediction results of the above-mentioned magnetically controlled transformer for the above-mentioned target working conditions; the above-mentioned sound field prediction results include the sound field prediction cloud map of the above-mentioned magnetically controlled transformer.

[0114] In an exemplary embodiment, the transformer sound field prediction device further includes a model acquisition module for acquiring a plurality of preset working condition information; obtaining model simulation parameter results corresponding to different preset working conditions according to the plurality of preset working condition information, the transformer geometric model and the material parameters; inputting the model simulation parameter results corresponding to different preset working conditions into a pre-built acoustic finite element calculation model to obtain noise field distribution information for the magnetic control transformer under different preset working conditions; wherein the acoustic finite element calculation model is used to obtain noise field distribution information of sound wave propagation behavior in the magnetic control transformer for different preset working conditions; inputting the model simulation parameter results corresponding to different preset working conditions as training data into the XGBoost magnetic control transformer sound field prediction model to be trained to obtain sound field prediction results of the magnetic control transformer for different preset working conditions; obtaining a loss value according to the difference between the noise field distribution information and the sound field prediction results; and using the loss value to train the XGBoost magnetic control transformer sound field prediction model to be trained.

[0115] In an exemplary embodiment, the model acquisition module is further used to normalize the noise field distribution information of the magnetic control transformer under different preset working conditions.

[0116] In an exemplary embodiment, the model acquisition module is further used to obtain a residual according to the difference between the noise field distribution information and the sound field prediction result; obtain a preset number of regression trees, and use the regression trees to fit the residual.

[0117] In an exemplary embodiment, the above-mentioned model acquisition module is also used to determine the acoustic properties of various media in the above-mentioned magnetic control transformer according to the above-mentioned material parameters, and set the boundary conditions of the acoustic finite element calculation model according to the above-mentioned acoustic properties; the above-mentioned boundary conditions include open boundaries, absorbing boundaries or fixed boundaries; obtain the sound source type and sound source intensity defined for the above-mentioned acoustic finite element calculation model, the mesh division information of the above-mentioned acoustic finite element calculation model, and the solver for sound field analysis; based on the above-mentioned transformer geometric model, the above-mentioned material parameters, the above-mentioned boundary conditions, the above-mentioned sound source type, the above-mentioned sound source intensity, the above-mentioned mesh division information and the above-mentioned solver, construct the above-mentioned acoustic finite element calculation model.

[0118] Each module in the above transformer sound field prediction device can be implemented in whole or in part by software, hardware and their combination. Each module 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 corresponding operations of each module.

[0119] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a transformer sound field prediction method is implemented.

[0120] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0121] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0122] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0123] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0124] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0125] The technical features of the above embodiments may 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 application.

[0126] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A transformer sound field prediction method, characterized in that: The method comprises: Obtaining a transformer geometric model corresponding to a magnetic control transformer and material parameters related to the magnetic control transformer; According to the transformer geometric model and the material parameters, a model simulation parameter result of the magnetic control transformer corresponding to a target working condition is obtained; The model simulation parameter results are input into a pre-trained XGBoost magnetic control transformer sound field prediction model to obtain the sound field prediction results of the magnetic control transformer for the target working condition; the sound field prediction results include a sound field prediction cloud map of the magnetic control transformer.

2. The method according to claim 1, characterized in that The steps of obtaining the XGBoost magnetically controlled transformer acoustic field prediction model include: Obtain information on various preset working conditions; According to the plurality of preset working condition information, the transformer geometric model and the material parameters, model simulation parameter results corresponding to different preset working conditions are obtained; The model simulation parameter results corresponding to different preset working conditions are input into a pre-built acoustic finite element calculation model to obtain the noise field distribution information of the magnetic control transformer under different preset working conditions; wherein the acoustic finite element calculation model is used to obtain the noise field distribution information of the sound wave propagation behavior in the magnetic control transformer under different preset working conditions; The model simulation parameter results corresponding to different preset working conditions are input as training data into the XGBoost magnetic control transformer sound field prediction model to be trained to obtain the sound field prediction results of the magnetic control transformer for different preset working conditions; Obtaining a loss value according to the difference between the noise field distribution information and the sound field prediction result; The loss value is used to train the XGBoost magnetically controlled transformer acoustic field prediction model to be trained.

3. The method according to claim 2, characterized in that Before obtaining the loss value according to the difference between the noise field distribution information and the sound field prediction result, the method further includes: The noise field distribution information of the magnetic control transformer under different preset working conditions is normalized.

4. The method according to claim 2, characterized in that: Before using the loss value to train the XGBoost magnetically controlled transformer acoustic field prediction model to be trained, the method includes: Obtaining a residual according to the difference between the noise field distribution information and the sound field prediction result; A preset number of regression trees are obtained, and the residuals are fitted using the regression trees.

5. The method according to claim 2, characterized in that: The step of obtaining the acoustic finite element calculation model comprises: According to the material parameters, the acoustic characteristics of various media in the magnetic control transformer are determined, and the boundary conditions of the acoustic finite element calculation model are set according to the acoustic characteristics; the boundary conditions include open boundaries, absorbing boundaries or fixed boundaries; Acquiring the sound source type and sound source intensity defined for the acoustic finite element calculation model, meshing information of the acoustic finite element calculation model, and a solver for sound field analysis; The acoustic finite element calculation model is constructed based on the transformer geometric model, the material parameters, the boundary conditions, the sound source type, the sound source intensity, the meshing information and the solver.

6. The method according to claim 2, characterized in that The noise field distribution information includes a sound field finite element calculation cloud map.

7. A transformer sound field prediction device, characterized in that: The device comprises: A geometric model acquisition module, used to acquire a transformer geometric model corresponding to a magnetic control transformer and material parameters related to the magnetic control transformer; A simulation parameter acquisition module, used for obtaining a model simulation parameter result of the magnetic control transformer corresponding to a target working condition according to the transformer geometric model and the material parameters; A prediction result acquisition module is used to input the model simulation parameter results into a pre-trained XGBoost magnetic control transformer sound field prediction model to obtain the sound field prediction results of the magnetic control transformer for the target working condition; the sound field prediction results include a sound field prediction cloud map of the magnetic control transformer.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images