Comprehensive Test Method and System for Aerodynamic Characteristics of Multiple Cable-Stayed Bridges

Abstract

This invention provides a comprehensive testing method and system for the aerodynamic characteristics of multiple cable-stayed bridges, relating to the field of bridge inspection technology. By constructing a comprehensive aerodynamic characteristic model including the Reynolds number, this invention comprehensively considers the influence of factors such as Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity on the aerodynamic test results. This allows the aerodynamic analysis model to more comprehensively consider the impact of various factors on the test results, making the output results of the aerodynamic analysis model closer to the actual working conditions of cable-stayed bridges. This solves the technical problem of low accuracy in traditional wind tunnel testing methods and improves the accuracy of aerodynamic characteristic analysis of cable-stayed bridges.

Description

Technical Field

[0001] This invention relates to the field of bridge inspection technology, and in particular to a comprehensive testing method and system for the aerodynamic characteristics of multiple cable stays. Background Technology

[0002] With the rapid development of bridge engineering technology, cable-stayed bridges have been widely used in modern bridge construction due to their advantages such as lightweight structure and strong spanning capacity. As the main load-bearing component of a cable-stayed bridge, the aerodynamic performance of the stay cables has a significant impact on the overall safety and stability of the bridge. However, the aerodynamic characteristics of stay cables under wind conditions are complex and variable, influenced by factors such as wind speed, wind direction, cable geometry, and relative position. Therefore, accurately assessing the aerodynamic performance of stay cables is crucial for ensuring the safe operation of bridges.

[0003] Currently, the evaluation of the aerodynamic characteristics of cable-stayed bridges mainly relies on wind tunnel testing. Traditional wind tunnel testing methods only consider the aerodynamic performance of cable-stayed bridges under specific wind speeds, resulting in a single test parameter. However, in actual cable-stayed bridge applications, the aerodynamic performance of cable-stayed bridges is also affected by factors such as wind direction, air pressure, and temperature. The analytical results obtained from traditional wind tunnel testing methods with a single test parameter cannot be adapted to the actual conditions of cable-stayed bridges, resulting in low accuracy. Summary of the Invention

[0004] This invention provides a comprehensive testing method and system for the aerodynamic characteristics of multiple cable-stayed bridges, which can solve the technical problem of low accuracy in traditional wind tunnel testing methods and improve the accuracy of aerodynamic characteristic analysis of cable-stayed bridges.

[0005] In a first aspect, the present invention provides a comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges. The method includes: acquiring test condition data and test result data for various types of cable-stayed bridges under different test conditions, as well as the surface characteristics of each type of cable-stayed bridge; determining the Reynolds number under each test condition based on the test condition data, whereby the Reynolds number characterizes the relative magnitude of the gas inertial force and gas viscous force flowing through each cable-stayed bridge; generating comprehensive aerodynamic characteristics for each type of cable-stayed bridge under different test conditions based on the Reynolds number, test condition data, and surface characteristics, whereby the comprehensive aerodynamic characteristics include one or more of the following: Reynolds number, wind speed characteristics, gas density characteristics, cable-stayed bridge characteristics, and rainfall intensity; and performing data fitting and attention network training based on the comprehensive aerodynamic characteristics and test result data to obtain an aerodynamic analysis model, thereby achieving comprehensive testing of the aerodynamic characteristics of multiple cable-stayed bridges.

[0006] In one possible implementation, after performing data fitting and attention network training based on the comprehensive aerodynamic characteristics and test results data to obtain an aerodynamic analysis model and achieve comprehensive aerodynamic characteristic testing of multiple stay cables, the following steps are also included: acquiring meteorological data on the type of the stay cable to be tested and its location; determining the Reynolds number of the stay cable to be tested based on the meteorological data; generating the comprehensive aerodynamic characteristics of the stay cable to be tested based on the type of the stay cable to be tested, the meteorological data, and the Reynolds number; and determining the analysis results of the stay cable to be tested based on the comprehensive aerodynamic characteristics and the aerodynamic analysis model, including aerodynamic coefficients, vibration frequency, and amplitude.

[0007] In one possible implementation, the Reynolds number under each test condition is determined based on the test condition data, including: extracting wind speed, temperature, and air pressure under each test condition based on the test condition data; calculating the gas density under each test condition based on the temperature and air pressure; calculating the gas dynamic viscosity under each test condition based on the temperature; determining the characteristic length of each type of cable-stayed bridge based on the surface characteristics of each type of cable-stayed bridge; standardizing the units of the wind speed, gas density, gas dynamic viscosity, and characteristic length of the cable-stayed bridge under each test condition to obtain the processed wind speed, gas density, gas dynamic viscosity, and characteristic length; and calculating the Reynolds number under each test condition based on the processed wind speed, gas density, gas dynamic viscosity, and characteristic length.

[0008] In one possible implementation, based on Reynolds number, test condition data, and surface features, the comprehensive aerodynamic characteristics of various types of stay cables under different test conditions are generated, including: determining gas density, air pressure, temperature, humidity, and rainfall intensity under each test condition based on the test condition data; generating gas density characteristics based on the gas density, air pressure, temperature, and humidity under each test condition; determining wind speed, fan speed, duct cross-sectional area, and wind direction angle under each test condition based on the test condition data; generating wind speed characteristics based on the wind speed, fan speed, duct cross-sectional area, and wind direction angle under each test condition; determining the stay cable diameter, characteristic length, stay cable surface condition, stay cable cross-sectional shape, and stay cable spatial attitude of each stay cable based on surface features; generating stay cable characteristics based on the stay cable diameter, characteristic length, stay cable surface condition, stay cable cross-sectional shape, and stay cable spatial attitude of each stay cable; and generating comprehensive aerodynamic characteristics of various types of stay cables under different test conditions based on Reynolds number, wind speed characteristics, gas density characteristics, stay cable characteristics, and rainfall intensity.

[0009] In one possible implementation, based on comprehensive aerodynamic characteristics and test result data, data fitting and attention network training are performed to obtain an aerodynamic analysis model, realizing comprehensive testing of the aerodynamic characteristics of multiple cable-stayed structures. This includes: generating multiple training samples using comprehensive aerodynamic characteristics as input and test result data as output; performing data fitting based on multiple training samples and a pre-defined multinomial regression model to obtain a data fitting model; training an attention model using a self-attention mechanism based on multiple training samples and a pre-defined neural network model; and performing weighted fusion of multiple training samples, the data fitting model, and the attention model to obtain an aerodynamic analysis model.

[0010] In one possible implementation, data fitting is performed based on multiple training samples and a pre-defined multinomial regression model to obtain a data fitting model, including: Step 1, initializing the multinomial regression model; using the model parameters of the initialized multinomial regression model as the model parameters for the current iteration process, and setting the optimal mean squared error to infinity; Step 2, based on the model parameters for the current iteration process and multiple training samples, calculating the output error between the model output and the sample output for each training sample; Step 3, based on the output error between the model output and the sample output for each training sample, calculating the mean squared error between the model output and the sample output for the current iteration process; Step 4, if the mean squared error for the current iteration process is less than the optimal mean squared error, then setting the mean squared error for the current iteration process to the optimal mean squared error, and setting the model parameters for the current iteration process to the optimal model parameters; Step 5, determining whether the number of iterations is greater than the maximum number of iterations; if yes, then exiting the iteration process; if no, then incrementing the number of iterations by one, updating the model parameters, and repeating steps 2 to 5 until exiting the iteration process; Step 6, determining the data fitting model based on the optimal model parameters.

[0011] In one possible implementation, based on multiple training samples and a pre-defined neural network model, an attention model is trained using a self-attention mechanism, including: Step 1, for the training samples in the current iteration, the aerodynamic comprehensive features of the training samples in the current iteration are used as input features, and linearly transformed by the self-attention layer to obtain a query vector, a key vector, and a value vector; Step 2, based on the query vector and the key vector, the attention score of each feature in the aerodynamic comprehensive features in the current iteration is calculated; Step 3, based on the attention score of each feature in the aerodynamic comprehensive features, the attention weight of each feature is calculated; Step 4, based on the attention weight and value vector of each feature, the output vector of the self-attention layer is determined; Step 5, based on the output vector of the self-attention layer and the fully connected... Step 1: Determine the model output of the attention model; Step 6: Calculate the comprehensive loss function based on the model output and sample output of each training sample; Step 7: If the comprehensive loss function of the current iteration is less than the optimal value, then determine the optimal parameters of the attention model based on the model parameters of the current iteration; where the optimal value is the minimum value of the comprehensive loss function in all previous iterations; Step 8: Determine whether the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process; if no, increment the number of iterations by one, update the model parameters of the attention model, and repeat steps 1 to 8 until exiting the iteration process; The model parameters of the attention model include the linear transformation parameters from the attention layer and the fully connected parameters of the fully connected layer; Step 9: Determine the attention model based on the optimal parameters.

[0012] In one possible implementation, a weighted fusion of multiple training samples, a data fitting model, and an attention model is performed to obtain an aerodynamic analysis model, including: Step 1, initializing the weights of the data fitting model and the attention model, with the sum of the weights of the data fitting model and the attention model being 1; Step 2, determining multiple first model outputs based on multiple training samples and the data fitting model; Step 3, determining multiple second model outputs based on multiple training samples and the attention model; Step 4, calculating multiple comprehensive outputs based on the multiple first model outputs, the multiple second model outputs, and the weights of the data fitting model and the attention model; Step 5, based on the multiple comprehensive outputs... Step 1: Combine the output of multiple training samples and calculate the combined output error; Step 2: Determine whether the combined output error in the current iteration process is less than the optimal combined error. If so, determine the combined output error in the current iteration process as the optimal combined error and the weights in the current iteration process as the optimal weights; Step 3: Determine whether the number of iterations is greater than the maximum number of iterations. If so, exit the iteration process. If not, increment the number of iterations by one, update the weights of the data fitting model and the attention model, and repeat steps 1 to 7 until exiting the iteration process; Step 4: Based on the optimal weights, the data fitting model, and the attention model, determine the aerodynamic analysis model.

[0013] In one possible implementation, the method further includes: acquiring test condition data and test result data of various types of stay cables under different test conditions during a historical period, as well as the surface characteristics of each type of stay cable; generating aerodynamic comprehensive characteristics and test result characteristics for each moment based on the test condition data, test result data, and surface characteristics of the stay cables during the historical period; dividing the aerodynamic comprehensive characteristics and test result characteristics for each moment during the historical period into time windows to obtain aerodynamic comprehensive characteristics and test result characteristics for multiple time windows during the historical period; determining input features based on the aerodynamic comprehensive characteristics of any time window and the test result characteristics at the start time of that time window; using the test result characteristics of a set time window after that time window as output features; generating multiple time-series training samples based on the input features and output features corresponding to each time window; training a neural network based on the multiple time-series training samples to obtain an aerodynamic prediction model for the stay cable; and predicting the aerodynamic characteristics of the stay cable to be tested based on the aerodynamic prediction model for the stay cable.

[0014] Secondly, embodiments of the present invention provide a comprehensive testing device for the aerodynamic characteristics of multiple cable-stayed bridges. This device includes a communication module and a processing module. The communication module is used to acquire test condition data and test result data for various types of cable-stayed bridges under different test conditions, as well as the surface characteristics of each type of cable-stayed bridge. The processing module is used to determine the Reynolds number under each test condition based on the test condition data. The Reynolds number characterizes the relative magnitude of the gas inertial force and gas viscous force flowing through each cable-stayed bridge. Based on the Reynolds number, test condition data, and surface characteristics, it generates comprehensive aerodynamic characteristics of each type of cable-stayed bridge under different test conditions. These comprehensive aerodynamic characteristics include one or more of the following: Reynolds number, wind speed characteristics, gas density characteristics, cable-stayed bridge characteristics, and rainfall intensity. Based on the comprehensive aerodynamic characteristics and test result data, it performs data fitting and attention network training to obtain an aerodynamic analysis model, thereby realizing the comprehensive testing of the aerodynamic characteristics of multiple cable-stayed bridges.

[0015] Thirdly, embodiments of the present invention provide a comprehensive testing system for the aerodynamic characteristics of multiple cable-stayed bridges. The comprehensive testing system includes an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to perform the steps of the method as described in the first aspect and any possible implementation thereof.

[0016] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the steps of the method as described in the first aspect and any possible implementation thereof.

[0017] This invention provides a comprehensive testing method and system for the aerodynamic characteristics of multiple cable-stayed bridges. By constructing a comprehensive aerodynamic characteristic including the Reynolds number, this invention comprehensively considers the influence of factors such as Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity on the aerodynamic test results. This allows the aerodynamic analysis model to more comprehensively consider the influence of each factor on the test results, making the output results of the aerodynamic analysis model closer to the actual working conditions of the cable-stayed bridges. This solves the technical problem of low accuracy in traditional wind tunnel testing methods and improves the accuracy of aerodynamic characteristic analysis of cable-stayed bridges. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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.

[0019] Figure 1 This is a flowchart illustrating a comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges provided in an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of the structure of a comprehensive testing device for the aerodynamic characteristics of multiple cable-stayed bridges provided in an embodiment of the present invention;

[0021] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0022] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0023] In the description of this invention, unless otherwise stated, " / " means "or". For example, A / B can mean A or B. The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" and "more than one" refer to two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0024] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0025] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules is not limited to the steps or modules listed, but may optionally include other steps or modules not listed, or may optionally include other steps or modules inherent to such process, method, product, or device.

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0027] like Figure 1 As shown, this embodiment of the invention provides a comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges, which includes steps S101-S104.

[0028] S101. Obtain test condition data and test result data of various types of stay cables under different test conditions, as well as the surface characteristics of each type of stay cable.

[0029] In some embodiments, the test condition data includes one or more of the following: wind speed, fan speed, duct cross-sectional area, wind direction angle, gas density, temperature, air pressure, humidity, and rainfall intensity.

[0030] In some embodiments, the test result data includes aerodynamic coefficients, vibration frequency, and amplitude.

[0031] In some embodiments, the surface features of the stay cable include the stay cable diameter, the stay cable surface condition, the stay cable cross-sectional shape, and the stay cable spatial orientation.

[0032] S102. Based on the test condition data, determine the Reynolds number under each test condition.

[0033] In the embodiments of this application, the Reynolds number is used to characterize the relative magnitude of the gas inertial force and gas viscous force flowing through each cable.

[0034] As one possible implementation, step S102 can be specifically implemented as steps S1021-S1026.

[0035] S1021. Based on the test condition data, extract the wind speed, temperature and air pressure under each test condition.

[0036] S1022. Calculate the gas density under each test condition based on the temperature and pressure.

[0037] For example, embodiments of the present invention can determine the gas density for each test condition by modifying the ideal gas equation. The formula is shown below.

[0038] ;

[0039] in, Let M be the gas density under certain test conditions, M be the molar mass of air (typically taken as 28.9647 g / mol), and R be the universal gas constant (typically taken as 8.3144626 J / (mol·K)). The latent heat of vaporization of water is typically taken as 2.5 × 10⁻⁶. 6 J / kg, q is the specific humidity, calculated using humidity H; The specific heat capacity of air at constant pressure is generally taken as 1005 J / (kg·K), where P is the air pressure under the test conditions and T is the gas temperature under the test conditions.

[0040] The specific humidity q mentioned above can be calculated using the following formula.

[0041] ;

[0042] in, q is the saturated vapor pressure, expressed in hPa; q is the specific humidity; and H is the relative humidity of the air.

[0043] .

[0044] S1023. Calculate the gas dynamic viscosity under each test condition based on the temperature.

[0045] For example, the formula for calculating the dynamic viscosity of a gas is as follows.

[0046] ;

[0047] in, The gas dynamic viscosity under certain test conditions. Reference temperature The dynamic viscosity at 273.15 K is typically taken as 1.716 × 10⁻⁶. -5 Pa·s; S is the Sutherland constant, typically taken as 110.4 K. T is the air temperature under certain test conditions.

[0048] In the above formula This is a correction factor used to correct dynamic viscosity when the temperature range exceeds standard atmospheric conditions.

[0049] S1024. Based on the surface characteristics of each type of cable-stayed cable, determine the characteristic length of each type of cable-stayed cable.

[0050] In some embodiments, the characteristic length of a circular cross-section cable is its diameter D (unit: m).

[0051] In some embodiments, the irregularly shaped cross-section stay cables (such as elliptical or polygonal) adopt hydraulic diameter... As the feature length. Wherein, . Where A is the hydraulic diameter, A is the cross-sectional area of ​​the irregular cross-section cable, and L is the cross-sectional perimeter of the irregular cross-section cable.

[0052] In some embodiments, if the surface of the stay cable is relatively rough, the roughness of the surface affects the aerodynamic characteristics of the stay cable, and therefore the characteristic length of the stay cable can be equivalently modified.

[0053] ;

[0054] in, is the equivalent diameter of the stay cable, i.e., the corrected equivalent characteristic length. D is the characteristic length before correction. This refers to the surface roughness of the cable-stayed bridge.

[0055] S1025. The wind speed, gas density, gas dynamic viscosity, and characteristic length of the cable-stayed bridge under each test condition are standardized to obtain the processed wind speed, gas density, gas dynamic viscosity, and characteristic length.

[0056] For example, embodiments of the present invention can standardize the units of wind speed, gas density, gas dynamic viscosity, and characteristic length of the cable-stayed bridge under various test conditions according to the International System of Units (SI).

[0057] S1026. Based on the processed wind speed, gas density, gas dynamic viscosity, and characteristic length, calculate the Reynolds number under each test condition.

[0058] For example, the formula for calculating the Reynolds number is as follows.

[0059] ;

[0060] in, Let Reynolds number be the Reynolds number under a certain test condition. The gas density under these test conditions. The equivalent diameter of the stay cable. Let be the gas dynamic viscosity under the test conditions. Let U be the wind speed under the test conditions.

[0061] S103. Based on Reynolds number, test condition data, and surface features, generate comprehensive aerodynamic characteristics of various types of stay cables under different test conditions.

[0062] In this embodiment of the application, the comprehensive aerodynamic characteristics include one or more of the following: Reynolds number, wind speed characteristics, gas density characteristics, cable-stayed bridge characteristics, and rainfall intensity.

[0063] In some embodiments, wind speed characteristics include: gas wind speed, fan speed, duct cross-sectional area, and wind direction angle.

[0064] In some embodiments, gas density characteristics include: gas density, temperature, gas pressure, and humidity;

[0065] In some embodiments, the characteristics of the cable-stayed bridge include: cable diameter, characteristic length, cable surface condition, cable cross-sectional shape, and cable spatial orientation.

[0066] As one possible implementation, step S103 can be specifically implemented as steps S1031-S1037.

[0067] S1031. Based on the test condition data, determine the gas density, air pressure, temperature, humidity and rainfall intensity under each test condition.

[0068] S1032. Generate gas density characteristics based on gas density, pressure, temperature and humidity under various test conditions.

[0069] S1033. Based on the test condition data, determine the wind speed, fan speed, duct cross-sectional area, and wind direction angle under each test condition.

[0070] S1034. Based on the wind speed, fan speed, duct cross-sectional area and wind direction angle under each test condition, generate wind speed characteristics.

[0071] S1035. Based on surface features, determine the cable diameter, characteristic length, surface condition, cross-sectional shape, and spatial orientation of each cable.

[0072] In some embodiments, the surface condition of the stay cable can be the surface roughness of the stay cable. The cross-sectional shape can be a cross-sectional profile, such as circular, elliptical, or variable. The spatial attitude of the stay cable can be measured by tilt sensors, such as the pitch angle and azimuth angle of the stay cable.

[0073] S1036. Based on the cable diameter, characteristic length, surface condition, cross-sectional shape, and spatial attitude of each cable, generate cable features.

[0074] S1037. Based on Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity, the comprehensive aerodynamic characteristics of various types of cable stays under different test conditions are generated.

[0075] S104. Based on the comprehensive aerodynamic characteristics and test results data, data fitting and attention network training are performed to obtain an aerodynamic analysis model, thereby realizing the comprehensive testing of the aerodynamic characteristics of multiple cable-stayed bridges.

[0076] As one possible implementation, step S104 can be specifically implemented as steps S1041-S1047.

[0077] S1041. Using aerodynamic comprehensive features as sample input and test result data as sample output, generate multiple training samples.

[0078] S1042. Based on multiple training samples and a pre-defined multinomial regression model, perform data fitting to obtain a data fitting model.

[0079] For example, step S1042 can be specifically implemented as steps one through six.

[0080] Step 1: Initialize the multinomial regression model; use the model parameters of the initial multinomial regression model as the model parameters of the current iteration process, and set the optimal mean square error to infinity.

[0081] In some embodiments, the multinomial regression model can be expressed as the following formula.

[0082] ;

[0083] Where y represents the output of the polynomial regression model. n is the dimension of the comprehensive aerodynamic characteristics. The polynomial order k is 3. This represents the i-th feature in the comprehensive aerodynamic characteristics; This represents the j-th feature in the comprehensive aerodynamic characteristics; The parameter represents the first-order term of the i-th feature in the multinomial regression model. The quadratic term parameter represents the relationship between the i-th and j-th features in a multinomial regression model. The parameter represents the cubic term of the i-th feature in the multinomial regression model. This is the constant term in the multinomial regression model.

[0084] Step 2: Based on the model parameters of the current iteration process and multiple training samples, calculate the output error between the model output and the sample output for each training sample.

[0085] Step 3: Based on the output error between the model output and the sample output of each training sample, calculate the mean square error between the model output and the sample output in the current iteration process.

[0086] In some embodiments, the mean squared error is the average of the squares of the errors between the model output and the sample output for each sample.

[0087] Step 4: If the mean squared error in the current iteration is less than the optimal mean squared error, then set the mean squared error in the current iteration to the optimal mean squared error and set the model parameters in the current iteration to the optimal model parameters.

[0088] Step 5: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the model parameters, and repeat steps 2 to 5 until exiting the iteration process.

[0089] Step 6: Determine the data fitting model based on the optimal model parameters.

[0090] S1043. Based on multiple training samples and a pre-defined neural network model, an attention model is trained using a self-attention mechanism.

[0091] For example, step S1043 can be implemented as steps one through nine.

[0092] Step 1: For the training samples in the current iteration process, the aerodynamic comprehensive features of the training samples in the current iteration process are used as input features. After linear transformation by the self-attention layer, query vector, key vector and value vector are obtained.

[0093] For example, embodiments of the present invention can perform feature dimension mapping, such as generating a query vector Q, a key vector K, and a value vector V from the comprehensive aerodynamic features X∈Rm×n (m is the number of samples, n is the feature dimension) through three independent linear transformation layers:

[0094] Q=XW Q K=XW K V=XW V ;

[0095] Among them, W Q W K W V ∈Rn×d k Let d be the learnable parameter matrix, i.e., the linear transformation matrix. k For the attention dimension.

[0096] Step 2: Based on the query vector and key vector, calculate the attention score of each feature in the aerodynamic integrated features during the current iteration.

[0097] For example, for each sample i, compute the query vector q. i With key vector k j The dot product, scaled (divided by d) k This is to prevent gradient vanishing. The formula is shown below.

[0098] ;

[0099] in, This represents the attention score between the i-th feature and the j-th feature.

[0100] Step 3: Calculate the attention weight of each feature based on the attention score of each feature in the comprehensive aerodynamic features.

[0101] For example, embodiments of the present invention can calculate the attention weight of each feature based on the following formula.

[0102] ;

[0103] in, This represents the j-th element in the i-th row of the attention weight matrix A.

[0104] Step 4: Determine the output vector of the self-attention layer based on the attention weights and value vectors of each feature.

[0105] For example, embodiments of the present invention can determine the output vector based on the following formula.

[0106] Z = AV; where Z is the output vector, A is the attention weight matrix, and V is the value vector.

[0107] Step 5: Based on the output vector of the self-attention layer and the fully connected layer, determine the model output of the attention model.

[0108] Step 6: Calculate the comprehensive loss function based on the model output and sample output of each training sample.

[0109] In some embodiments, the comprehensive loss function includes a main loss function and a regularization term. The main loss function is calculated based on the error between the model output and the sample output, and the regularization term is calculated based on the model parameters of the attention model.

[0110] For example, for each training sample, the error between the model output and the sample output is calculated. The error is calculated using the mean squared error (MSE). That is, for the i-th training sample, the model output is... The sample output is Then the mean square error of a single sample is MSEi = , where N is the dimension of the output feature;

[0111] The mean squared error of all training samples is aggregated to obtain the main loss function. Where M is the number of training samples.

[0112] Calculate the regularization term ;in, , , The linear transformation parameters of the self-attention layer, Here are the parameters for the fully connected layer, and λ is the regularization coefficient, which is preset to 0.001.

[0113] Comprehensive loss function .

[0114] It should be noted that this application involves a regression problem (predicting continuous values ​​such as aerodynamic coefficients and vibration frequencies), therefore, mean squared error is used as the main loss function, suitable for measuring the difference between predicted and true values. The attention model includes self-attention layers and fully connected layers, and the regularization term covers the parameters of these layers to prevent overfitting. This application introduces L2 regularization (weight decay) to control model complexity; the regularization coefficient λ is preset based on experience and can also be adjusted through cross-validation.

[0115] Step 7: If the overall loss function of the current iteration is less than the optimal value, then the model parameters of the current iteration are determined as the optimal parameters of the attention model; where the optimal value is the minimum value of the overall loss function in each iteration before the current iteration.

[0116] Step 8: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the model parameters of the attention model, and repeat steps 1 to 8 until exiting the iteration process.

[0117] In some embodiments, the model parameters of the attention model include the linear transformation parameters of the self-attention layer and the fully connected parameters of the fully connected layer.

[0118] Step nine: Determine the attention model based on the optimal parameters.

[0119] S1044. Based on multiple training samples, a data fitting model, and an attention model, a weighted fusion is performed to obtain an aerodynamic analysis model.

[0120] For example, step S1044 can be implemented as steps one through eight.

[0121] Step 1: Initialize the weights of the data fitting model and the attention model. The sum of the weights of the data fitting model and the attention model is 1.

[0122] Step 2: Based on multiple training samples and the data fitting model, determine multiple first model outputs.

[0123] Step 3: Based on multiple training samples and the attention model, determine multiple second model outputs.

[0124] Step 4: Calculate multiple comprehensive outputs based on the outputs of multiple first models, multiple second models, and the weights of the data fitting model and the attention model.

[0125] Step 5: Calculate the comprehensive output error based on multiple comprehensive outputs and the sample outputs of multiple training samples.

[0126] Step 6: Determine whether the overall output error in the current iteration process is less than the optimal overall error. If so, determine the overall output error in the current iteration process as the optimal overall error and the weights in the current iteration process as the optimal weights.

[0127] Step 7: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the weights of the data fitting model and the attention model, and repeat steps 1 to 7 until exit the iteration process.

[0128] Step 8: Based on the optimal weights, as well as the data fitting model and attention model, determine the aerodynamic analysis model.

[0129] For example, the aerodynamic analysis model can be expressed as the following formula.

[0130] ;

[0131] in, This represents the output of the data fitting model. This represents the output of the attention model. The weights for the data fitting model. These are the weights for the attention model. This is the output of the aerodynamic analysis model.

[0132] This invention provides a comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges. By constructing a comprehensive aerodynamic characteristic including the Reynolds number, it comprehensively considers the influence of factors such as Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity on the aerodynamic test results. This allows the aerodynamic analysis model to more comprehensively consider the influence of each factor on the test results, making the output results of the aerodynamic analysis model closer to the actual working conditions of the cable-stayed bridges. This solves the technical problem of low accuracy in traditional wind tunnel testing methods and improves the accuracy of aerodynamic characteristic analysis of cable-stayed bridges.

[0133] Optionally, the comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges provided in this embodiment of the invention further includes steps S201-S204 after step S104.

[0134] S201. Obtain meteorological data on the type of the cable to be tested and the location of the cable to be tested.

[0135] In some embodiments, meteorological data include data such as wind speed, air pressure, temperature, humidity, and rainfall intensity.

[0136] S202. Based on meteorological data, determine the Reynolds number of the cable to be tested.

[0137] For example, step S202 can be implemented through steps S1021-S1026.

[0138] S203. Based on the type of the cable under test, meteorological data, and Reynolds number, generate the comprehensive aerodynamic characteristics of the cable under test.

[0139] For example, step S203 can be implemented through steps S1031-S1037.

[0140] S204. Based on the comprehensive aerodynamic characteristics of the cable under test and the aerodynamic analysis model, determine the analysis results of the cable under test.

[0141] In some embodiments, the analysis results include aerodynamic coefficients, vibration frequencies, and amplitudes.

[0142] Thus, embodiments of the present invention can perform aerodynamic characteristic analysis based on the actual working conditions of the cable-stayed bridge, thereby improving the accuracy of the aerodynamic characteristic analysis of the cable-stayed bridge.

[0143] Optionally, the comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges provided in this embodiment of the invention further includes steps S301-S308.

[0144] S301. Obtain test condition data and test result data of various types of stay cables under different test conditions during historical periods, as well as the surface characteristics of various types of stay cables.

[0145] S302. Based on the test condition data, test result data and surface characteristics of the cable at each moment in the historical period, generate the comprehensive aerodynamic characteristics and test result characteristics at each moment.

[0146] S303. Divide the aerodynamic comprehensive characteristics and test result characteristics at each moment in the historical period into time windows to obtain the aerodynamic comprehensive characteristics and test result characteristics of multiple time windows in the historical period.

[0147] S304. Determine the input features based on the comprehensive aerodynamic characteristics of any time window and the test result characteristics at the start time of that time window.

[0148] S305. The output feature is the test result feature of the set time window after this time window.

[0149] S306. Based on the input and output features corresponding to each time window, generate multiple time-series training samples.

[0150] S307. Based on multiple time-series training samples, a neural network is trained to obtain an aerodynamic prediction model for the cable-stayed bridge.

[0151] S308. Based on the aerodynamic prediction model of the cable-stayed bridge, the aerodynamic characteristics of the cable-stayed bridge under test are predicted.

[0152] In this way, the embodiments of the present invention can predict the aerodynamic characteristics of the cable-stayed bridge over time, which facilitates the protection of the cable-stayed bridge.

[0153] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0154] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0155] Figure 2 A schematic diagram of a comprehensive testing device for the aerodynamic characteristics of multiple cable-stayed bridges provided in an embodiment of the present invention is shown. The comprehensive testing device 400 includes a communication module 401 and a processing module 402.

[0156] The communication module 401 is used to acquire test condition data and test result data of various types of stay cables under different test conditions, as well as the surface characteristics of each type of stay cable.

[0157] The processing module 402 is used to determine the Reynolds number under each test condition based on the test condition data. The Reynolds number is used to characterize the relative magnitude of the gas inertial force and gas viscous force flowing through each cable. Based on the Reynolds number, test condition data, and surface features, it generates the comprehensive aerodynamic characteristics of each type of cable under different test conditions. The comprehensive aerodynamic characteristics include one or more of the following: Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity. Based on the comprehensive aerodynamic characteristics and test result data, it performs data fitting and attention network training to obtain an aerodynamic analysis model, thereby realizing the comprehensive testing of the aerodynamic characteristics of multiple cables.

[0158] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. For example... Figure 3 As shown, the electronic device 500 includes: a processor 501, a memory 502, and a computer program 503 stored in the memory 502 and executable on the processor 501. When the processor 501 executes the computer program 503, it implements the steps in the above-described method embodiments, for example... Figure 1 The steps S101-S104 are shown. Alternatively, when the processor 501 executes the computer program 503, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of the communication module 401 and the processing module 402 shown are illustrated.

[0159] For example, the computer program 503 can be divided into one or more modules / units, which are stored in the memory 502 and executed by the processor 501 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program 503 in the electronic device 500. For example, the computer program 503 can be divided into... Figure 2 The communication module 401 and the processing module 402 are shown.

[0160] The processor 501 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0161] The memory 502 can be an internal storage unit of the electronic device 500, such as a hard disk or memory of the electronic device 500. The memory 502 can also be an external storage device of the electronic device 500, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 500. Furthermore, the memory 502 can include both internal and external storage units of the electronic device 500. The memory 502 is used to store the computer program and other programs and data required by the terminal. The memory 502 can also be used to temporarily store data that has been output or will be output.

[0162] The above-described 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges, characterized in that, include: Acquire test condition data and test result data of various types of stay cables under different test conditions, as well as the surface characteristics of each type of stay cable; Based on the test condition data, determine the Reynolds number under each test condition; Based on the Reynolds number, the test condition data, and the surface features, aerodynamic comprehensive characteristics of various types of cable stays under different test conditions are generated. The aerodynamic comprehensive characteristics include one or more of the following: Reynolds number, wind speed characteristics, gas density characteristics, cable stay characteristics, and rainfall intensity. Based on the comprehensive aerodynamic characteristics and the test result data, data fitting and attention network training are performed to obtain an aerodynamic analysis model, realizing comprehensive testing of the aerodynamic characteristics of multiple cable-stayed bridges, including: using the comprehensive aerodynamic characteristics as sample input and the test result data as sample output to generate multiple training samples; Based on the multiple training samples and a preset multinomial regression model, data fitting is performed to obtain a data fitting model; based on the multiple training samples and a preset neural network model, an attention model is trained using a self-attention mechanism; based on the multiple training samples, the data fitting model, and the attention model, a weighted fusion is performed to obtain an aerodynamic analysis model.

2. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, After obtaining the aerodynamic analysis model by performing data fitting and attention network training based on the comprehensive aerodynamic characteristics and the test result data, and realizing the comprehensive aerodynamic characteristic testing of multiple cable-stayed bridges, the process further includes: Obtain meteorological data on the type of cable to be tested and its location; Based on the meteorological data, determine the Reynolds number of the cable to be tested; Based on the type of the cable under test, the meteorological data, and the Reynolds number, the comprehensive aerodynamic characteristics of the cable under test are generated. Based on the comprehensive aerodynamic characteristics of the cable under test and the aerodynamic analysis model, the analysis results of the cable under test are determined, including aerodynamic coefficients, vibration frequency, and amplitude.

3. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, The determination of the Reynolds number under each test condition based on the test condition data includes: Based on the test condition data, wind speed, temperature and air pressure under each test condition are extracted; Based on the temperature and air pressure under each test condition, calculate the gas density under each test condition; Based on the temperature under each test condition, calculate the gas dynamic viscosity under each test condition; Based on the surface characteristics of each type of cable, the characteristic length of each type of cable is determined; The wind speed, gas density, gas dynamic viscosity, and characteristic length of the cable under the various test conditions are standardized to obtain the processed wind speed, gas density, gas dynamic viscosity, and characteristic length. Based on the processed wind speed, gas density, gas dynamic viscosity, and characteristic length, the Reynolds number under each test condition is calculated.

4. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, Based on the Reynolds number, the test condition data, and the surface features, the aerodynamic comprehensive characteristics of various types of stay cables under different test conditions are generated, including: Based on the test condition data, the gas density, air pressure, temperature, humidity and rainfall intensity under each test condition are determined; Based on the gas density, pressure, temperature, and humidity under the aforementioned test conditions, a gas density characteristic is generated; Based on the test condition data, determine the wind speed, fan speed, duct cross-sectional area, and wind direction angle under each test condition; Based on wind speed, fan speed, duct cross-sectional area, and wind direction angle under various test conditions, wind speed characteristics are generated. Based on the surface features, the cable diameter, characteristic length, surface condition, cross-sectional shape, and spatial orientation of each cable are determined. Based on the cable diameter, characteristic length, surface condition, cross-sectional shape, and spatial orientation of each cable, cable features are generated. Based on the Reynolds number, wind speed characteristics, gas density characteristics, cable characteristics, and rainfall intensity, the comprehensive aerodynamic characteristics of various types of cable stays under different test conditions are generated.

5. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, The step of performing data fitting based on the multiple training samples and a preset multinomial regression model to obtain a data fitting model includes: Step 1: Initialize the multinomial regression model; use the model parameters of the initial multinomial regression model as the model parameters of the current iteration process, and set the optimal mean square error to infinity; Step 2: Based on the model parameters of the current iteration process and the multiple training samples, calculate the output error between the model output and the sample output for each training sample; Step 3: Based on the output error between the model output and the sample output of each training sample, calculate the mean square error between the model output and the sample output in the current iteration process; Step 4: If the mean square error in the current iteration is less than the optimal mean square error, then set the mean square error in the current iteration to the optimal mean square error and set the model parameters in the current iteration to the optimal model parameters. Step 5: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the model parameters, and repeat steps 2 to 5 until exit the iteration process. Step 6: Determine the data fitting model based on the optimal model parameters.

6. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, The process of training an attention model based on the multiple training samples and a pre-defined neural network model using a self-attention mechanism includes: Step 1: For the training samples in the current iteration process, take the comprehensive aerodynamic features of the training samples in the current iteration process as input features, and obtain the query vector, key vector and value vector through the linear transformation of the self-attention layer. Step 2: Based on the query vector and key vector, calculate the attention score of each feature in the comprehensive aerodynamic features during the current iteration. Step 3: Calculate the attention weight of each feature based on the attention score of each feature in the comprehensive aerodynamic features. Step 4: Based on the attention weights of each feature and the value vector, determine the output vector of the self-attention layer; Step 5: Based on the output vector of the self-attention layer and the fully connected layer, determine the model output of the attention model; Step 6: Calculate the comprehensive loss function based on the model output and sample output of each training sample; Step 7: If the overall loss function of the current iteration is less than the optimal value, then the model parameters of the current iteration are determined as the optimal parameters of the attention model; where the optimal value is the minimum value of the overall loss function in all previous iterations. Step 8: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the model parameters of the attention model, and repeat steps 1 to 8 until exit the iteration process. The model parameters of the attention model include the linear transformation parameters of the self-attention layer and the fully connected parameters of the fully connected layer. Step nine: Determine the attention model based on the optimal parameters.

7. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, The aerodynamic analysis model is obtained by weighted fusion of the multiple training samples, the data fitting model, and the attention model, including: Step 1: Initialize the weights of the data fitting model and the attention model, wherein the sum of the weights of the data fitting model and the attention model is 1; Step 2: Based on the multiple training samples and the data fitting model, determine multiple first model outputs; Step 3: Based on the multiple training samples and the attention model, determine multiple second model outputs; Step 4: Based on the multiple first model outputs, the multiple second model outputs, and the weights of the data fitting model and the attention model, calculate multiple comprehensive outputs; Step 5: Calculate the combined output error based on multiple combined outputs and the sample outputs of multiple training samples; Step 6: Determine whether the overall output error in the current iteration process is less than the optimal overall error. If so, determine the overall output error in the current iteration process as the optimal overall error and the weights in the current iteration process as the optimal weights. Step 7: Determine if the number of iterations is greater than the maximum number of iterations. If yes, exit the iteration process. If no, increment the number of iterations by one, update the weights of the data fitting model and the attention model, and repeat steps 1 to 7 until exit the iteration process. Step 8: Based on the optimal weights, the data fitting model, and the attention model, determine the aerodynamic analysis model.

8. The comprehensive testing method for the aerodynamic characteristics of multiple cable-stayed bridges according to claim 1, characterized in that, The method further includes: Acquire test condition data and test result data of various types of stay cables under different test conditions during historical periods, as well as the surface characteristics of various types of stay cables; Based on the test condition data, test result data and surface characteristics of the cable at each moment in history, the comprehensive aerodynamic characteristics and test result characteristics at each moment are generated. By dividing the comprehensive aerodynamic characteristics and test results at various moments within a historical period into time windows, we obtain the comprehensive aerodynamic characteristics and test results of multiple time windows within a historical period. The input features are determined by combining the aerodynamic characteristics of any time window with the test result characteristics at the start time of that time window. The output features are the test results characteristics of a set time window following this time window; Based on the input and output features corresponding to each time window, multiple time-series training samples are generated. Based on the aforementioned multiple time-series training samples, a neural network is trained to obtain an aerodynamic prediction model for the cable-stayed bridge. Based on the aerodynamic prediction model of the cable-stayed bridge, the aerodynamic characteristics of the cable-stayed bridge under test are predicted.

9. A comprehensive testing system for the aerodynamic characteristics of multiple cable-stayed bridges, characterized in that, The testing system includes an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 8.

Images