Converter noise analysis method and device based on proxy model, equipment and medium

Abstract

The invention relates to the technical field of converters, and provides a converter noise analysis method, device and equipment based on an agent model and a medium, and the method comprises the steps: collecting the actual measurement noise sample data of a converter, and selecting a noise influence parameter corresponding to the actual measurement noise sample data; performing global sensitivity analysis on the noise influence parameters to obtain a sensitivity coefficient corresponding to each parameter in the noise influence parameters; selecting noise influence key parameters of the converter from the noise influence parameters according to the sensitivity coefficient; training a preset agent model by using the noise influence key parameters and the actually measured noise sample data, and calculating multiple correlation coefficients of the trained agent model; and selecting the agent model with the maximum multiple correlation coefficient as a target agent model, and performing noise analysis on the to-be-analyzed noise data by using the target agent model to obtain a noise analysis result visual graph. According to the invention, the efficiency of converter noise analysis can be improved.

Description

Technical Field

[0001] This invention relates to the field of converter technology, and in particular to a converter noise analysis method, apparatus, equipment and medium based on a surrogate model. Background Technology

[0002] As rail transit vehicles transition from safety and functionality to comfort and intelligence, the noise standards for electrical equipment are becoming increasingly stringent. Converters generally refer to the current conversion and transmission units in rail transit vehicles, including traction converters responsible for high-voltage power supply for vehicle traction and auxiliary converters responsible for low-voltage power supply for onboard lighting, air conditioning, etc. They are core electrical components in rail transit vehicles. In order to improve the efficiency of converter noise analysis, it is necessary to conduct convenient, fast, and effective analysis of converter noise results.

[0003] Existing noise analysis techniques are based on conducting on-site noise tests on already produced sample cabinets. This technique is simple and direct, but it is time-consuming and resource-intensive. Furthermore, the subsequent costs of duct optimization and fan selection are too high, the cycle is too long, and the efficiency is low, resulting in low efficiency when performing converter noise analysis. Summary of the Invention

[0004] To address the aforementioned problems, embodiments of the present invention provide a converter noise analysis method, apparatus, device, and medium based on a surrogate model.

[0005] In a first aspect, embodiments of the present invention provide a converter noise analysis method based on a surrogate model, comprising:

[0006] Collect measured noise sample data of the converter and select the noise impact parameters corresponding to the measured noise sample data;

[0007] A global sensitivity analysis was performed on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter.

[0008] Based on the sensitivity coefficient, the key parameters affecting the noise of the converter are selected from the noise impact parameters;

[0009] The preset proxy model is trained using the key parameters of noise impact and the measured noise sample data, and the multiple correlation coefficient of the trained proxy model is calculated.

[0010] The surrogate model with the largest multiple correlation coefficient is selected as the target surrogate model. The target surrogate model is used to perform noise analysis on the preset noise data to be analyzed, and the noise analysis results are visualized.

[0011] According to an embodiment of the present invention, the measured noise sample data of the converter includes:

[0012] Noise acquisition arrays are configured around the converter cabinet according to the preset measurement point distances;

[0013] The noise acquisition array is used to acquire the sound pressure level data of the converter, and the frequency data of the sound pressure level data is analyzed to obtain the frequency band data of the converter.

[0014] The average sound pressure level of the converter is calculated based on the sound pressure level data collected around the cabinet, and the average sound pressure level data and the frequency band data are used to construct a measured noise data set.

[0015] Update the converter and return to the step of configuring a noise acquisition array around the converter cabinet according to the preset measurement point distance, until the number of acquisitions reaches the preset acquisition number threshold.

[0016] When the number of data collections reaches the threshold, the measured noise data set is compiled into measured noise sample data.

[0017] According to an embodiment of the present invention, the step of performing a global sensitivity analysis on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters includes:

[0018] A response surface model is constructed based on a preset high-order polynomial response proxy model to establish the relationship between the noise impact parameters and the preset noise level.

[0019] Based on the response surface model, the noise impact parameters are decomposed and variated to obtain the variance of each parameter in the noise impact parameters.

[0020] The total variance of the noise impact parameters is calculated by using the variance corresponding to each parameter in the noise impact parameters.

[0021] The sensitivity coefficient for each parameter is calculated based on the total variance and the variance of each parameter in the noise impact parameters.

[0022] According to an embodiment of the present invention, the step of performing a global sensitivity analysis on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters includes:

[0023] A response surface model is constructed based on a preset high-order polynomial response proxy model to establish the relationship between the noise impact parameters and the preset noise level.

[0024] Based on the response surface model, the noise impact parameters are decomposed and variated to obtain the variance of each parameter in the noise impact parameters.

[0025] The total variance of the noise impact parameters is calculated by using the variance corresponding to each parameter in the noise impact parameters.

[0026] The sensitivity coefficient for each parameter is calculated based on the total variance and the variance of each parameter in the noise impact parameters.

[0027] According to an embodiment of the present invention, training a preset surrogate model using the key parameters affecting noise and the measured noise sample data includes:

[0028] Configure the fan parameters according to the key parameters affecting noise, and generate a sequence of proxy models for the preset proxy models;

[0029] The wind turbine parameters and the measured noise sample data are input one by one into the surrogate models in the surrogate model sequence to obtain the noise output level value;

[0030] Calculate the loss value between the noise output level value and the actual noise level value;

[0031] When the loss value is less than the preset loss threshold, the trained proxy model in the proxy model sequence is output.

[0032] According to an embodiment of the present invention, the step of using the target proxy model to perform noise analysis on preset noise data to be analyzed, and obtaining a visual graph of the noise analysis results, includes:

[0033] Obtain the key parameters of the noise impact to be analyzed corresponding to the preset noise data to be analyzed;

[0034] The target proxy model is used to perform noise analysis on the key parameters of the noise impact to be analyzed, and the average sound pressure level data and frequency band data of the converter corresponding to the noise data to be analyzed are obtained.

[0035] The average sound pressure level data and the frequency band data are visualized to obtain a visual graph of the noise analysis results.

[0036] According to an embodiment of the present invention, visualizing the average sound pressure level data and the frequency band data to obtain a visual graph of the noise analysis results includes:

[0037] The average sound pressure level data is used as the vertical axis attribute, and the frequency band data is used as the horizontal axis attribute;

[0038] Generate a correspondence between the average sound pressure data and the frequency band data;

[0039] A two-dimensional coordinate system is generated between the average sound pressure level data and the frequency band data based on the vertical axis attribute and the horizontal axis attribute.

[0040] The correspondence is added to the two-dimensional coordinates to obtain a visualization of the noise analysis results.

[0041] Secondly, embodiments of the present invention provide a converter noise analysis device based on a surrogate model, characterized in that it includes:

[0042] The measured noise sample data acquisition module is used to acquire measured noise sample data of the converter and select the noise impact parameters corresponding to the measured noise sample data.

[0043] The sensitivity coefficient calculation module is used to perform a global sensitivity analysis on the noise impact parameters and obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters.

[0044] The noise impact key parameter selection module is used to select the noise impact key parameters of the converter from the noise impact parameters based on the sensitivity coefficient;

[0045] The multiple correlation coefficient calculation module is used to train the preset surrogate model using the key parameters of noise influence and the measured noise sample data, and to calculate the multiple correlation coefficient of the trained surrogate model.

[0046] The noise analysis module is used to select the surrogate model with the largest multiple correlation coefficient as the target surrogate model, and use the target surrogate model to perform noise analysis on the preset noise data to be analyzed, and obtain a visual graph of the noise analysis results.

[0047] Thirdly, embodiments of the present invention provide an electronic device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the converter noise analysis method based on the surrogate model described above.

[0048] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the converter noise analysis method based on the surrogate model described above.

[0049] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial effects:

[0050] This invention, for the first time, analyzes noise using a surrogate model, enabling rapid and effective iteration that continuously improves the accuracy of noise analysis. Through sensitivity analysis of converter noise parameters, the core parameters most influencing converter noise levels can be identified from a large pool of parameters, improving subsequent noise reduction. Training calculations are performed in MATLAB, resulting in a user-friendly interface for converter noise analysis. This simple and intuitive interface facilitates the noise analysis process, making it operable and visual. It enables rapid noise assessment, duct optimization, and fan selection in the early stages of converter design, providing convenience and accuracy that cannot be achieved through on-site testing and simulation analysis. This has strong engineering guidance significance for converter design. Therefore, the surrogate model-based converter noise analysis method, device, equipment, and medium proposed in this invention can solve the problem of low efficiency in converter noise analysis. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 The flowchart of the converter noise analysis method based on the surrogate model according to Embodiment 1 of the present invention is shown.

[0053] Figure 2 A schematic diagram of the fan parameter configuration according to Embodiment 1 of the present invention is shown;

[0054] Figure 3 A graphical representation of the noise analysis results of Embodiment 1 of the present invention is shown.

[0055] Figure 4 The diagram shows the functional block diagram of the converter noise analysis device based on the surrogate model according to Embodiment 3 of the present invention;

[0056] Figure 5 This diagram shows the structural composition of an electronic device that implements the converter noise analysis method based on the proxy model according to Embodiment 4 of the present invention. Detailed Implementation

[0057] The present disclosure will be further described below with reference to the embodiments shown in the accompanying drawings.

[0058] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0059] This invention proposes a surrogate model-based converter noise analysis method. Based on the original high-precision model theory and combined with model training methods, a surrogate model for converter noise analysis is built, enabling convenient, rapid, and effective noise assessment. Compared to traditional methods, the surrogate model-based converter noise analysis technique is more efficient and reduces human subjectivity, showing great potential and application prospects in converter noise interpretation.

[0060] Example 1

[0061] like Figure 1 As shown, this invention proposes a converter noise analysis method based on a surrogate model, comprising the following steps:

[0062] S1. Collect measured noise sample data of the converter and select the noise impact parameters corresponding to the measured noise sample data.

[0063] In this embodiment of the invention, the measured noise sample data refers to the noise collection from a large number of locomotives, EMUs, and urban rail converters collected in the early stage. The noise data is required to be tested in accordance with GB3736, with the measuring point 1 meter away from the cabinet, to obtain the average sound pressure level data and 1 / 3 octave band data of the cabinet.

[0064] In this embodiment of the invention, the measured noise sample data of the converter includes:

[0065] Noise acquisition arrays are configured around the converter cabinet according to the preset measurement point distances;

[0066] The noise acquisition array is used to acquire the sound pressure level data of the converter, and the frequency data of the sound pressure level data is analyzed to obtain the frequency band data of the converter.

[0067] The average sound pressure level of the converter is calculated based on the sound pressure level data collected around the cabinet, and the average sound pressure level data and the frequency band data are used to construct a measured noise data set.

[0068] Update the converter and return to the step of configuring a noise acquisition array around the converter cabinet according to the preset measurement point distance, until the number of acquisitions reaches the preset acquisition number threshold.

[0069] When the number of data collections reaches the threshold, the measured noise data set is compiled into measured noise sample data.

[0070] In detail, noise data of the converter is collected around the converter cabinet. Specifically, noise acquisition devices are installed 1 meter away from the cabinet around the converter cabinet, forming a noise acquisition array. The sound pressure level data around the converter cabinet is collected by the noise acquisition array, and the average sound pressure level data is calculated from all the sound pressure level data collected around the converter cabinet. Sound pressure level is a physical quantity that measures the intensity of sound. It represents the logarithm of the pressure change produced by a sound wave at a certain location relative to a reference pressure. Sound pressure level quantifies the intensity of sound, allowing noise levels at different sound distances to be directly compared. Frequency analysis is performed on the collected sound pressure level data around the converter to obtain the frequency spectrum data of the converter. Spectral analysis of sound pressure level (SPL) data yields the SPL levels at different frequencies. The entire frequency range corresponding to the collected SPL data is divided into several 1 / 3 octave bands. 1 / 3 octave band data refers to the SPL data obtained through 1 / 3 octave band analysis. 1 / 3 octave band data is typically represented by frequency and SPL as parameters, where frequency represents the sound's frequency and SPL represents the sound's intensity level. For example, if the frequency range is 20Hz to 20,000Hz, 1 / 3 octave band analysis will divide this range into many narrower frequency bands, such as 20Hz-25Hz, 25Hz-31.5Hz, 31.5Hz-40Hz, etc. Then, the SPL within each frequency band will be calculated.

[0071] Specifically, the average sound pressure level and 1 / 3 frequency band data of the converter are collected as a set of sample data. For several converters, the corresponding average sound pressure level and 1 / 3 frequency band data are collected to obtain multiple sets of sample data. All the collected sample data are compiled into measured noise sample data. The measured noise sample data contains more than 50 sets of data. Through a large number of field tests in the early stage, more than 50 sets of converter noise data have been accumulated. By sorting and classifying all cabinet and fan parameters and noise test data, all parameters that may affect the converter noise are extracted.

[0072] Furthermore, based on the noise impact parameters corresponding to the measured noise sample data selected in the previous experiment, 15 key parameters were selected, including cabinet air duct type, inlet and outlet area, air duct length, number of air duct bends, fan static pressure, fan air volume, fan speed, number of fan blades, cabinet size, noise impact factors of cabinet reactor, cabinet weight, air inlet filter type, fan size, fan power, and fan type. These 15 noise impact parameters were then used as initial design variables.

[0073] Furthermore, among the 15 noise impact parameters initially selected, some parameters have a relatively small impact on noise. That is, changes in noise impact parameters with a relatively small impact will not have a significant impact on changes in the system response, thus reducing the required number of samples. Therefore, sensitivity analysis of noise impact parameters is required before selecting key noise impact parameters.

[0074] S2. Perform a global sensitivity analysis on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters.

[0075] In this embodiment of the invention, the sensitivity coefficient is a quantitative indicator that measures the degree to which each input parameter contributes to the variation of the output variable (system response, performance index, etc.).

[0076] In this embodiment of the invention, the step of performing a global sensitivity analysis on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters includes:

[0077] A response surface model is constructed based on a preset high-order polynomial response proxy model to establish the relationship between the noise impact parameters and the preset noise level.

[0078] Based on the response surface model, the noise impact parameters are decomposed and variated to obtain the variance of each parameter in the noise impact parameters.

[0079] The total variance of the noise impact parameters is calculated by using the variance corresponding to each parameter in the noise impact parameters.

[0080] The sensitivity coefficient for each parameter is calculated based on the total variance and the variance of each parameter in the noise impact parameters.

[0081] In detail, a high-order polynomial response surrogate model (RS-HDMR) is constructed to represent the relationship between noise influence parameters and preset noise levels. This leads to the response surface model (RS-HDMR). The RS-HDMR (S-HDMR) model is an approximate model used to simplify complex simulation or experimental models. It involves using 15 noise-affecting parameters as a parameter combination, conducting simulations or experiments to obtain the corresponding noise levels, and then fitting these data with higher-order polynomials (such as quadratic and cubic polynomials) to establish a response surface model (RSM) model relating the noise-affecting parameters to the noise levels. This model collects actual data from the converter's operation, including 15 key parameters and their corresponding noise levels. A large data sample is used to ensure the accuracy of the modeling. The RS-HDMR method is used to construct a model relating each input parameter (X_1, X_2, ..., X_{15}) to the output noise (Y). Then, based on the response surface model, the total variance is decomposed into the individual contributions of each input parameter. Specifically, the variance of each noise-affecting parameter is calculated according to the response surface model. The variance represents the contribution of each parameter to the noise level fluctuation when it changes independently. The total variance is obtained by summing the variances of each parameter and the interaction variances. Therefore, the sensitivity coefficient of each parameter is calculated as the variance of that parameter divided by the total variance.

[0082] Furthermore, by analyzing the relationship between the output response and input variables of the system model, the sensitivity coefficients of the input variables are calculated. Based on the magnitude of these values, the relationship between the input variables and the coupling between variables on the system response is determined, thereby achieving a preliminary understanding and analysis of the system model. Based on the sensitivity analysis results, design variables with very small sensitivity coefficients can be disregarded and set as constant values, because changes in these variables will not have a significant impact on changes in the system response, thus reducing the required number of samples. Therefore, it is necessary to select key noise impact parameters from the noise impact parameters.

[0083] S3. Select the key parameters of the noise impact of the converter from the noise impact parameters based on the sensitivity coefficient.

[0084] In this embodiment of the invention, the key parameters affecting noise refer to the parameters that have a significant impact on noise, selected based on the sensitivity coefficient. These parameters include cabinet duct type, inlet and outlet area, duct length, number of duct bends, fan static pressure, fan air volume, fan speed, number of fan blades, and cabinet dimensions.

[0085] In this embodiment of the invention, selecting the key noise impact parameters of the converter from the noise impact parameters based on the sensitivity coefficient includes:

[0086] The sensitivity coefficients are sorted in descending order to obtain a sensitivity coefficient sequence.

[0087] The target sensitivity coefficient is selected from the sensitivity coefficient sequence according to the preset key quantity threshold;

[0088] The key parameters affecting the noise of the converter are determined based on the selected target sensitivity coefficient.

[0089] In detail, the sensitivity coefficients of all parameters are sorted in descending order. Based on a preset critical quantity threshold, the parameters with the largest sensitivity coefficients are selected as critical parameters. The critical quantity threshold is 8. The eight parameters that contribute the most to output variation are selected. Based on the preset critical quantity threshold, a corresponding number of target sensitivity coefficients are selected from the sorted sensitivity coefficient sequence. Based on the selected target sensitivity coefficients, the key parameters affecting the converter's noise are determined. These parameters have large sensitivity coefficients and contribute significantly to output variation, so they are determined as key parameters. Therefore, the design parameters corresponding to the eight largest sensitivity coefficients are obtained as key parameters affecting noise. By analyzing a large number of parameters, the core parameters that most affect the converter's noise level are obtained, which can improve the noise reduction in later stages.

[0090] Furthermore, based on the selected key parameters of noise impact and measured noise sample data, the selection of the surrogate model and sample training can be calculated, thereby obtaining a surrogate model that can analyze converter noise.

[0091] S4. The preset proxy model is trained using the key parameters of noise influence and the measured noise sample data, and the multiple correlation coefficient of the trained proxy model is calculated.

[0092] In one practical application scenario of this invention, a method widely used in various industries over the past few decades, including in areas such as automotive collision analysis and nuclear explosion analysis, has emerged. Due to the numerous types of surrogate models, each with different characteristics for different problems, the effectiveness of surrogate models can vary even when using the same data. Therefore, this invention considers employing multiple surrogate modeling techniques to model the data and using a hybrid surrogate modeling technique to ensure the accuracy and reliability of the predicted data.

[0093] In this embodiment of the invention, the surrogate model refers to a model that simulates the original high-precision model by using a set of samples obtained from the original high-precision model and then using a certain method to extract the approximate relationship between the input parameters and output parameters from these samples. Therefore, the surrogate model is also called the "model of the model". After using the surrogate model, the problem space of solving complex systems is transformed from a physical problem in a professional field into a mathematical problem that is easy to operate and understand. Thus, feasible methods can be found in global optimization to obtain the feasible domain of the design space. The surrogate models include, but are not limited to, Kriging (Gaussian process regression), LSSVR (Least Squares Support Vector Regression), TSVR (Twin Support Vector Regression), NPSVR (Non-parametric Support Vector Regression), and WAS (Wavelet Analysis Surrogates).

[0094] In this embodiment of the invention, training a preset surrogate model using the key parameters affecting noise and the measured noise sample data includes:

[0095] Configure the fan parameters according to the key parameters affecting noise, and generate a sequence of proxy models for the preset proxy models;

[0096] The wind turbine parameters and the measured noise sample data are input one by one into the surrogate models in the surrogate model sequence to obtain the noise output level value;

[0097] Calculate the loss value between the noise output level value and the actual noise level value;

[0098] When the loss value is less than the preset loss threshold, the trained proxy model in the proxy model sequence is output.

[0099] In detail, the key parameters affecting noise are determined as the wind turbine parameters that need to be configured, and the surrogate models to be trained are merged into a surrogate model sequence (Kriging, LSSVR, TSVR, NPSVR, WAS). The wind turbine parameters and measured noise sample data are then input into each surrogate model in the surrogate model sequence for training. The noise output level value of each surrogate model can be obtained, and the loss value between the noise output value and the actual noise level value is calculated. When the loss value is less than the preset loss threshold, it indicates that the surrogate model can accurately simulate the impact of wind turbine parameters on the noise level. After the training and verification of all surrogate models are completed, the surrogate model with the output loss value less than the preset loss threshold is taken as the final trained surrogate model. The training and calculation of the data are all performed in MATLAB, and an operable interface for converter noise analysis based on MATLAB is completed, so that the noise analysis process can be completed through a simple and easy-to-understand interface.

[0100] Furthermore, after determining the input parameters, the surrogate model is selected and trained. Various surrogate models can be used, such as Kriging, LSSVR, TSVR, NPSVR, and WAS, etc., and the multiple correlation coefficient R is obtained by iterating through them. 2 The value of R 2 The larger the value, the more accurate the model. Therefore, the multiple correlation coefficients of the trained surrogate models should be calculated sequentially to select the trained surrogate model with the best performance.

[0101] In this embodiment of the invention, the multiple correlation coefficient is a statistic used to evaluate the analytical ability of multiple independent variables on a dependent variable, and can be used to evaluate the model accuracy of a trained surrogate model.

[0102] In this embodiment of the invention, calculating the multiple correlation coefficient of the trained surrogate model includes:

[0103] The multivariate correlation coefficients of the trained proxy models in the proxy model sequence are calculated one by one using the following formula:

[0104]

[0105] Among them, R 2 For the multiple correlation coefficient of the surrogate model, y i Let i be the actual noise level value at the i-th observation point. Let be the noise output level value at the i-th observation point. is the mean of the actual noise level, and n is the number of observation points.

[0106] In detail, by comparing the actual noise level values ​​in the measured noise sample data with the noise level values ​​output by the surrogate model, the multiple correlation coefficient of each trained surrogate model can be obtained. The multiple correlation coefficient R0 2 The value of is between 0 and 1. The closer it is to 1, the better the fitting effect of the surrogate model and the stronger the interpretability of the surrogate model to the data.

[0107] Furthermore, for each trained surrogate model, the surrogate model with the largest multiple correlation coefficient is selected as the final noise analysis model. Thus, the noise of the converter under analysis can be analyzed through the final noise analysis model, which can achieve convenient, fast and effective noise assessment.

[0108] S5. Select the surrogate model with the largest multiple correlation coefficient as the target surrogate model, and use the target surrogate model to perform noise analysis on the preset noise data to be analyzed, and obtain a visual graph of the noise analysis results.

[0109] In this embodiment of the invention, the target surrogate model refers to the trained surrogate model with the largest multiple correlation coefficient. The target surrogate model is then used to analyze the noise level of the converter to be analyzed. The noise analysis result visualization graph refers to the visualization of various data obtained in the noise analysis process in the form of charts. The graph is conducive to a more intuitive understanding and analysis of noise characteristics.

[0110] In this embodiment of the invention, the step of using the target proxy model to perform noise analysis on the preset noise data to be analyzed, and obtaining a visual graph of the noise analysis results, includes:

[0111] Obtain the key parameters of the noise impact to be analyzed corresponding to the preset noise data to be analyzed;

[0112] The target proxy model is used to perform noise analysis on the key parameters of the noise impact to be analyzed, and the average sound pressure level data and frequency band data of the converter corresponding to the noise data to be analyzed are obtained.

[0113] The average sound pressure level data and the frequency band data are visualized to obtain a visual graph of the noise analysis results.

[0114] In detail, the key parameters affecting the noise of the converter to be analyzed are determined based on its noise data. These key parameters include the type of air duct in the cabinet, the area of ​​the inlet and outlet, the length of the air duct, the number of bends in the air duct, the static pressure of the fan, the air volume of the fan, the fan speed, the number of fan blades, and the cabinet dimensions. Based on these key parameters, the fan parameters are configured, and then, according to the configured fan parameters, such as... Figure 2The diagram shown illustrates the configuration of fan parameters, including the duct type, inlet / outlet area, duct length, number of duct bends, fan static pressure, fan air volume, fan speed, number of fan blades, and cabinet dimensions.

[0115] Specifically, a target surrogate model is used to perform noise analysis on the noise data of the converter under analysis, thereby obtaining the average sound pressure level data and 1 / 3 octave band data of the converter under analysis. The average sound pressure level data and 1 / 3 octave band data are then visualized to obtain a visual representation of the noise analysis results, such as... Figure 3 The diagram shown is a visualization of the noise analysis results. Based on the configured fan parameters, the noise results of the converter to be analyzed are analyzed to obtain the sound pressure level data corresponding to each frequency band, thereby obtaining the sound pressure level results of the converter to be analyzed. Based on the sound pressure level results, the noise level of the converter can be evaluated, thus realizing a convenient, fast and effective noise assessment of the converter.

[0116] Example 2

[0117] To better understand the present invention, a second embodiment is provided below to further explain the present invention. In order to intuitively observe the noise analysis results, the average sound pressure level data and the frequency band data are visualized to obtain a visual graph of the noise analysis results.

[0118] In this embodiment of the invention, the noise intensity in different frequency bands can be seen intuitively through graphical display. By visualizing continuous time series data, the changing trend of noise level over time can be observed, which is beneficial for long-term monitoring and analysis of noise source changes, evaluation of the effectiveness of noise reduction measures, and analysis of future noise levels.

[0119] In this embodiment of the invention, visualizing the average sound pressure level data and the frequency band data to obtain a visual graph of the noise analysis results includes:

[0120] The average sound pressure level data is used as the vertical axis attribute, and the frequency band data is used as the horizontal axis attribute;

[0121] Generate a correspondence between the average sound pressure data and the frequency band data;

[0122] A two-dimensional coordinate system is generated between the average sound pressure level data and the frequency band data based on the vertical axis attribute and the horizontal axis attribute.

[0123] The correspondence is added to the two-dimensional coordinates to obtain a visualization of the noise analysis results.

[0124] In detail, the average sound pressure level (SPL) data and frequency band data represent the noise intensity and range of different frequency bands, respectively. Axis attributes are then defined: the horizontal axis (X-axis) represents the frequency band data (e.g., 100-200Hz, 200-300Hz, etc.), and the vertical axis (Y-axis) represents the average SPL data (e.g., 60dB, 65dB, etc.). A one-to-one correspondence is established between the average SPL data and the frequency band data, so that each frequency band has a corresponding SPL value. For example, the frequency band 100-200Hz corresponds to an average SPL of 60dB, and the frequency band 200-300Hz corresponds to an average SPL of 65dB. Based on this correspondence, these points can be plotted on a two-dimensional plane, with the horizontal axis representing the frequency band and the vertical axis representing the sound pressure level, thus providing a visual representation of the noise intensity distribution within different frequency bands.

[0125] This invention, for the first time, analyzes noise using a surrogate model, enabling rapid and effective iteration that continuously improves the accuracy of noise analysis. Through sensitivity analysis of converter noise parameters, the core parameters most influencing converter noise levels can be identified from a large pool of parameters, improving subsequent noise reduction. Training calculations are performed in MATLAB, resulting in a user-friendly interface for converter noise analysis. This simple and intuitive interface facilitates the noise analysis process, making it operable and visual. It enables rapid noise assessment, duct optimization, and fan selection in the early stages of converter design, providing convenience and accuracy that cannot be achieved through on-site testing and simulation analysis. This has strong engineering guidance significance for converter design. Therefore, the surrogate model-based converter noise analysis method, device, equipment, and medium proposed in this invention can solve the problem of low efficiency in converter noise analysis.

[0126] Example 3

[0127] like Figure 4 As shown in the figure, this embodiment also provides a functional block diagram of a converter noise analysis device based on a surrogate model.

[0128] The converter noise analysis device 100 based on the surrogate model described in this embodiment can be installed in an electronic device. Depending on the functions implemented, the converter noise analysis device 100 based on the surrogate model may include a measured noise sample data acquisition module 101, a sensitivity coefficient calculation module 102, a noise impact key parameter selection module 103, a multiple correlation coefficient calculation module 104, and a noise analysis module 105. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and which are stored in the memory of the electronic device.

[0129] In this embodiment, the functions of each module / unit are as follows:

[0130] The measured noise sample data acquisition module 101 is used to acquire measured noise sample data of the converter and select the noise impact parameters corresponding to the measured noise sample data.

[0131] The sensitivity coefficient calculation module 102 is used to perform a global sensitivity analysis on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise impact parameters.

[0132] The noise impact key parameter selection module 103 is used to select the noise impact key parameters of the converter from the noise impact parameters according to the sensitivity coefficient;

[0133] The multiple correlation coefficient calculation module 104 is used to train a preset surrogate model using the key parameters of noise influence and the measured noise sample data, and to calculate the multiple correlation coefficient of the trained surrogate model.

[0134] The noise analysis module 105 is used to select the surrogate model with the largest multiple correlation coefficient as the target surrogate model, and use the target surrogate model to perform noise analysis on the preset noise data to be analyzed, and obtain a visual graph of the noise analysis results.

[0135] In detail, each module in the converter noise analysis device 100 based on the surrogate model described in the embodiments of the present invention adopts the same technical means as the converter noise analysis method based on the surrogate model described in Embodiment 1 and Embodiment 2, and can produce the same technical effect, which will not be repeated here.

[0136] Example 4

[0137] like Figure 5 As shown, this embodiment also provides a computer electronic device, which may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a converter noise analysis program based on a proxy model.

[0138] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing a converter noise analysis program based on a proxy model) and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0139] The memory 11 includes at least one type of medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Furthermore, the memory 11 can include both internal and external storage devices. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of a converter noise analysis program based on a proxy model, but also to temporarily store data that has been output or will be output.

[0140] The communication bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0141] The communication interface 13 is used for communication between the aforementioned electronic device and other electronic devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), or, optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0142] The figure only shows an electronic device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0143] For example, although not shown, the electronic device may also include a power supply (such as a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0144] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0145] The converter noise analysis program based on the surrogate model stored in the memory 11 of the electronic device is a combination of multiple instructions. When run in the processor 10, it can achieve the following:

[0146] Collect measured noise sample data of the converter and select the noise impact parameters corresponding to the measured noise sample data;

[0147] A global sensitivity analysis was performed on the noise impact parameters to obtain the sensitivity coefficient corresponding to each parameter.

[0148] Based on the sensitivity coefficient, the key parameters affecting the noise of the converter are selected from the noise impact parameters;

[0149] The preset proxy model is trained using the key parameters of noise impact and the measured noise sample data, and the multiple correlation coefficient of the trained proxy model is calculated.

[0150] The surrogate model with the largest multiple correlation coefficient is selected as the target surrogate model. The target surrogate model is used to perform noise analysis on the preset noise data to be analyzed, and the noise analysis results are visualized.

[0151] Specifically, the specific implementation method of the processor 10 for the above instructions can be referred to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, and will not be repeated here.

[0152] Furthermore, if the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0153] Example 5

[0154] This embodiment provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the converter noise analysis method based on the surrogate model described above.

[0155] This program code can also be loaded onto a computer or other programmable data processing device, causing a series of operational steps to be executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 Steps of a specified function in one or more processes.

[0156] Computer-readable storage media include permanent and non-permanent, removable and non-removable media, which can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0157] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0158] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0159] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0160] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0161] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the claims are intended to be included within the invention.

[0162] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0163] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0164] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method of converter noise analysis based on a proxy model, characterized by, The method comprises: collecting measured noise sample data of the converter, and selecting noise influence parameters corresponding to the measured noise sample data; performing global sensitivity analysis on the noise influence parameters to obtain a sensitivity coefficient corresponding to each parameter in the noise influence parameters; selecting a noise influence key parameter of the converter from the noise influence parameters according to the sensitivity coefficient; training a preset proxy model by using the noise influence key parameter and the measured noise sample data, and calculating a multiple correlation coefficient of the trained proxy model; selecting a proxy model with the largest multiple correlation coefficient as a target proxy model, and performing noise analysis on preset noise data to be analyzed by using the target proxy model to obtain a noise analysis result visual graph.

2. The proxy model based power converter noise analysis method of claim 1, wherein, The collecting of the measured noise sample data of the converter comprises: configuring a noise collection array around the cabinet of the converter according to a preset distance of a measuring point; collecting sound pressure level data of the converter by using the noise collection array, and performing frequency analysis on the sound pressure level data to obtain frequency data of the converter; calculating average sound pressure level data of the converter according to the sound pressure level data collected around the cabinet, and constructing the average sound pressure level data and the frequency data into a measured noise data group; updating the converter, and returning to the step of configuring the noise collection array around the cabinet of the converter according to the preset distance of the measuring point until the number of collection reaches a preset collection number threshold; when the number of collection reaches the collection number threshold, the measured noise data groups are collected into the measured noise sample data.

3. The proxy model based power converter noise analysis method of claim 1, wherein, The global sensitivity analysis on the noise influence parameters to obtain the sensitivity coefficient corresponding to each parameter in the noise influence parameters comprises: constructing a response surface relationship model between the noise influence parameters and a preset noise level according to a preset high-order polynomial response proxy model; decomposing and varying the noise influence parameters according to the response surface relationship model to obtain a variance corresponding to each parameter in the noise influence parameters; calculating a total variance of the noise influence parameters through the variance corresponding to each parameter in the noise influence parameters; calculating the sensitivity coefficient corresponding to each parameter according to the total variance and the variance corresponding to each parameter in the noise influence parameters.

4. The proxy model based power converter noise analysis method of claim 1, wherein, The selecting of the noise influence key parameter of the converter from the noise influence parameters according to the sensitivity coefficient comprises: sorting the sensitivity coefficients in descending order to obtain a sensitivity coefficient sequence; selecting a target sensitivity coefficient in the sensitivity coefficient sequence according to a preset key quantity threshold; determining the noise influence key parameter of the converter according to the selected target sensitivity coefficient.

5. The proxy model based power converter noise analysis method of claim 1, wherein, The training of the preset proxy model by using the noise influence key parameter and the measured noise sample data comprises: configuring fan parameters according to the noise influence key parameter, and generating a proxy model sequence of the preset proxy model; inputting the fan parameters and the measured noise sample data into the proxy models in the proxy model sequence one by one to obtain noise output level values; calculating a loss value between the noise output level values and true noise level values; When the loss value is less than a preset loss threshold, output a trained surrogate model in the sequence of surrogate models.

6. The proxy model based power converter noise analysis method of claim 1, wherein, The noise analysis result visualization graph includes: Obtaining a noise influence key parameter corresponding to the preset noise data to be analyzed; Using the target surrogate model to perform noise analysis on the noise influence key parameter to obtain average sound pressure level data and frequency range data of the converter corresponding to the noise data to be analyzed; The average sound pressure level data and the frequency range data are visualized to obtain a noise analysis result visualization graph.

7. The proxy model based power converter noise analysis method of claim 6, wherein, The average sound pressure level data and the frequency range data are visualized to obtain a noise analysis result visualization graph. The average sound pressure level data and the frequency range data are visualized to obtain a noise analysis result visualization graph. The average sound pressure level data and the frequency range data are visualized to obtain a noise analysis result visualization graph. The average sound pressure level data and the frequency range data are visualized to obtain a noise analysis result visualization graph. The device includes:

8. A converter noise analysis device based on a surrogate model, characterized in that, The measured noise sample data acquisition module is configured to collect measured noise sample data of the converter and select a noise influence parameter corresponding to the measured noise sample data; The sensitivity coefficient calculation module is configured to perform global sensitivity analysis on the noise influence parameter to obtain a sensitivity coefficient corresponding to each parameter in the noise influence parameter; The noise influence key parameter selection module is configured to select a noise influence key parameter of the converter from the noise influence parameter according to the sensitivity coefficient; The multiple correlation coefficient calculation module is configured to train a preset surrogate model using the noise influence key parameter and the measured noise sample data, and calculate a multiple correlation coefficient of the trained surrogate model; The noise analysis module is configured to select a surrogate model with the largest multiple correlation coefficient as a target surrogate model, and use the target surrogate model to perform noise analysis on preset noise data to be analyzed to obtain a noise analysis result visualization graph. The processor executes the computer program to implement the steps of the converter noise analysis method based on the surrogate model in any one of claims 1 to 7.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory, wherein the computer program, when executed by the processor, is arranged to perform the method of any one of claims 1 to 8. The computer program is executed by the processor to implement the steps of the converter noise analysis method based on the surrogate model in any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, ​

Images