Wind tunnel wind field adjustment method and system based on CFD technology and deep learning

Through CFD technology and deep learning, a wind field adjustment model is constructed to predict the optimal layout of rough elements and sharp split devices in the wind tunnel, which solves the problems of low efficiency and low accuracy of wind field adjustment in traditional wind tunnel experiments, and realizes wind Efficient and automated adjustment of hole tests.

CN119026316APending Publication Date: 2024-11-26GUANGDONG PROVINCIAL ACAD OF BUILDING RES GRP CO LTD
View PDF 0 Cites 1 Cited by

Patent Information

Application Number
CN202410941909.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-15
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

In traditional wind tunnel experiments, wind farm adjustment depends on the engineer's experience and multiple tests, resulting in inefficiency and low accuracy, especially when simulating complex wind farms, multiple parameters need to be adjusted accurately.

Method used

Combining computational fluid mechanics (CFD) technology and deep learning, by obtaining wind tunnel experimental parameters, performing numerical simulation and parameter adjustment, building a wind field adjustment model, predicting the optimal layout of rough elements and sharp split devices, and realizing the wind tunnel wind field Intelligent adjustment and optimization.

Benefits of technology

It improves the efficiency and accuracy of wind tunnel tests, reduces the number and time of tests, reduces labor costs, makes wind tunnel wind farm adjustments more automated and controllable, and improves the practicality and operability of the tests.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119026316A_ABST
    Figure CN119026316A_ABST
Patent Text Reader

Abstract

The invention discloses a CFD technology and deep learning-based wind tunnel wind field adjustment method and system. The method comprises the steps of S1, obtaining various flow field experiment parameters of a wind tunnel; s2, according to the geometric model, the boundary conditions and the experimental parameters of the wind tunnel, performing numerical simulation on the wind field of the wind tunnel by using a CFD technology, adjusting simulation parameters until an output simulation result is matched with a result obtained after the actual wind field of the wind tunnel is adjusted, and collecting the simulation result; s3, training the collected simulation results under different working conditions in combination with a deep learning model, and constructing a wind field adjustment model; and S4, in an actual wind tunnel test, inputting the target wind field parameters into the wind field adjustment model, predicting the optimal layout of the roughness element and the wedge device in the wind tunnel, and adjusting the positions of the roughness element and the wedge device in the wind tunnel according to the optimal layout to realize the target wind field parameters. According to the invention, intelligent adjustment and optimization of the wind field of the wind tunnel are realized by accurately controlling the wind tunnel experiment parameters, and the wind tunnel experiment efficiency and accuracy are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to a wind tunnel wind field adjustment method and system based on CFD technology and deep learning. Background Art

[0002] Wind tunnel experiments are widely used in aerospace, automotive engineering, architectural design and other fields to evaluate the impact of wind on structures or objects, optimize design solutions, etc. A wind tunnel is a device used to simulate air flow in an atmospheric environment in a wind tunnel experiment.

[0003] In traditional wind tunnel experiments, adjusting the wind field of the wind tunnel requires gradual adjustment based on the experience of engineers and multiple test runs. This adjustment process is not only time-consuming but also requires a large number of tests. In order to improve the efficiency and accuracy of wind tunnel experiments, the introduction of computational fluid dynamics (CFD) technology has become an important method for adjusting wind tunnels. CFD technology uses numerical calculation methods to simulate the movement of fluids, and can quickly predict and analyze the airflow in wind tunnel experiments on computers.

[0004] Although CFD technology has the advantages of high efficiency and flexibility, when simulating complex wind tunnel wind fields, multiple parameters need to be precisely adjusted to make the results of CFD simulation consistent with the results of actual wind tunnel experiments. This process usually requires experienced engineers to conduct repeated trials and adjustments, which leads to low efficiency and low accuracy of wind tunnel tests. Summary of the invention

[0005] The first object of the present invention is to provide a wind tunnel wind field adjustment method based on CFD technology and deep learning, which can realize rapid adjustment and optimization of the wind tunnel wind field and improve the efficiency and accuracy of wind tunnel tests.

[0006] The first object of the present invention is achieved by the following technical measures: a wind tunnel wind field adjustment method based on CFD technology and deep learning, characterized in that it includes the following steps:

[0007] S1. Obtain various flow field experimental parameters of the wind tunnel;

[0008] S2. According to the geometric model, boundary conditions and experimental parameters of the wind tunnel, the wind field of the wind tunnel is numerically simulated using CFD technology, and the simulation parameters are adjusted until the output simulation results match the adjusted results of the actual wind tunnel wind field, and the simulation results are collected;

[0009] S3. Combine the deep learning model to train the collected simulation results under different working conditions and build a wind farm adjustment model;

[0010] S4. In the actual wind tunnel test, the target wind field parameters are input into the wind field adjustment model, the optimal layout of the roughness elements and the wedge devices in the wind tunnel is predicted, and the positions of the roughness elements and the wedge devices in the wind tunnel are adjusted according to the optimal layout to achieve the target wind field parameters.

[0011] The present invention is an advanced wind tunnel wind field adjustment method that combines computational fluid dynamics with deep learning. The present invention aims to achieve intelligent adjustment and optimization of the wind tunnel wind field by precisely controlling wind tunnel experimental parameters, thereby improving the efficiency and accuracy of wind tunnel experiments. Compared with traditional manual adjustment methods, the present invention reduces the number of experiments and experimental time during the experimental adjustment process, reduces labor costs, and simultaneously utilizes deep learning technology to make the wind tunnel wind field adjustment process more automated and controllable, thereby improving the practicability and operability of wind tunnel experiments.

[0012] The experimental parameters in step S1 of the present invention and the target wind field parameters in step S4 are wind speed, temperature and turbulence intensity.

[0013] In the step S2 of the present invention, the simulation parameters are the grid resolution, turbulence model and boundary conditions; the simulation results are the wind speed distribution, pressure distribution and turbulence intensity; the simulation results match the actual wind tunnel wind field adjusted results, which means that the error between the two is 10 -3 the following.

[0014] In the step S1 of the present invention, the experimental parameters are preprocessed to remove abnormal values ​​and fill missing values, and then the process proceeds to step S2.

[0015] In step S3 of the present invention, the simulation results under different working conditions as the training set are labeled, and each data point is marked as a wind field that meets the requirements or does not meet the requirements according to the preset target wind field parameters. At the same time, according to the actual experimental adjustment results, a part of the data of the training set is manually labeled for verification and evaluation of the deep learning model.

[0016] After completing the wind tunnel test, the present invention collects final wind field data, compares and analyzes the data with the prediction results of the wind field adjustment model, adjusts the parameters and structure of the wind field adjustment model according to the analysis results, and optimizes its prediction capability.

[0017] A second object of the present invention is to provide a system using the above-mentioned wind tunnel wind field adjustment method based on CFD and deep learning.

[0018] The second object of the present invention is achieved by the following technical measures: a system using the wind tunnel wind field adjustment method based on CFD technology and deep learning, characterized by comprising:

[0019] Parameter acquisition module, used to obtain various flow field experimental parameters of the wind tunnel;

[0020] The CFD numerical simulation module performs numerical simulation of the wind tunnel wind field according to the geometric model, boundary conditions and experimental parameters of the wind tunnel, and adjusts the simulation parameters until the output simulation results match the adjusted results of the actual wind tunnel wind field;

[0021] High-performance computing module, used to handle large-scale CFD numerical simulation calculations;

[0022] Deep learning module, used to train simulation results under different working conditions and build a wind farm adjustment model;

[0023] A prediction module, used to predict the optimal layout of the rough element and the wedge device through the wind field adjustment model according to the input target wind field parameters;

[0024] Output module, used to output prediction results.

[0025] The system of the present invention also includes a graphical user interface module, which is used for users to input target wind field parameters, and displays the predicted placement information of rough elements and sharp boundaries, and provides interactive operation and wind tunnel experiment parameter adjustment functions.

[0026] The system of the present invention includes a graphical user interface module and a database module. The graphical user interface module is used for users to input target wind field parameters, display the predicted optimal layout information of rough elements and wedge devices, and provide interactive operation and wind tunnel experiment parameter adjustment functions; the database module is used to store and manage simulation results, experimental parameters and wind field adjustment models and provide data query, analysis and backup functions; the parameter acquisition module obtains the experimental parameters of the wind tunnel through a sensor used for real-time monitoring of the internal environmental parameters of the wind tunnel. The sensor collects the environmental parameters and updates them to the parameter acquisition module in real time.

[0027] The system of the present invention includes a real-time monitoring module, a safety control module and an interface module. The real-time monitoring module is used to monitor the wind tunnel experiment process and the wind field adjustment results in real time, and display the real-time data of the experiment and the adjusted wind field simulation results through a graphical user interface; the safety control module is used to monitor the safety status of the wind tunnel experiment, and automatically stop the experiment and warn the user when a safety problem is detected; the interface module can be connected with other experimental equipment for data exchange and integration.

[0028] Compared with the prior art, the present invention has the following significant effects:

[0029] The present invention can realize intelligent adjustment and optimization of the wind tunnel wind field, and improve the efficiency and accuracy of wind tunnel tests. Compared with the traditional manual adjustment method, the present invention reduces the number of tests and experimental time in the experimental adjustment process, reduces labor costs, and uses deep learning technology to make the wind tunnel wind field adjustment process more automated and controllable, improving the practicality and operability of wind tunnel tests. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Figure 1 It is an overall flow chart of the method of the present invention. DETAILED DESCRIPTION

[0032] The present invention is further illustrated below through the description of specific implementation methods, but this is not a limitation of the present invention. Those skilled in the art can make various modifications or improvements based on the basic idea of ​​the present invention, but as long as they do not deviate from the basic idea of ​​the present invention, they are all within the protection scope of the present invention.

[0033] like Figure 1 As shown, a wind tunnel wind field adjustment method based on CFD technology and deep learning of the present invention comprises the following steps:

[0034] S1. Obtain various flow field experimental parameters of the wind tunnel, including wind speed, temperature, turbulence intensity, etc., and preprocess these data to remove outliers and fill in missing values ​​to ensure the accuracy and reliability of subsequent analysis;

[0035] S2. According to the geometric model, boundary conditions and pre-processed experimental parameters of the wind tunnel, the wind field of the wind tunnel is numerically simulated using CFD technology, and simulation parameters such as grid resolution, turbulence model and boundary conditions are adjusted to output simulation results such as wind speed distribution, pressure distribution and turbulence intensity until the output simulation results match the adjusted wind speed distribution, pressure distribution and turbulence intensity of the wind tunnel wind field in the actual wind tunnel experiment, that is, the error between the two meets 10 -3 The following; collects the simulation results;

[0036] S3. Combine the deep learning model to train the collected simulation results under different working conditions and build a wind farm adjustment model;

[0037] The simulation results under different working conditions collected in step S2 are used as training sets, and these data are annotated. The annotation can mark each data point as a wind field that meets or does not meet the requirements according to the preset target wind field parameters. At the same time, according to the actual experimental adjustment results, a part of the data can be manually annotated for verification and evaluation of the deep learning model.

[0038] Use the labeled data set to train the deep learning model. You can choose appropriate algorithms, such as support vector machine (SVM), random forest (Random Forest), convolutional neural network (CNN), etc., to train and optimize the model. The goal of the training is to predict the corresponding wind farm adjustment plan by inputting experimental parameters.

[0039] S4. In the actual wind tunnel test, the target wind field parameters are input into the wind field adjustment model to predict the optimal layout of the roughness elements and wedge devices in the wind tunnel. According to the optimal layout, the positions of the roughness elements and wedge devices in the wind tunnel are adjusted to achieve the target wind field parameters.

[0040] The trained wind field adjustment model is applied to the actual wind tunnel test. The user inputs the target wind field parameters, such as the turbulence intensity and wind speed of the wind profile of the wind tunnel test section, and the wind field adjustment model predicts the optimal layout positions of the roughness element and the wedge device. According to the prediction results of the wind field adjustment model, the positions of these devices in the wind tunnel are adjusted to achieve the target wind field parameters.

[0041] With the accumulation of more experimental data and the practical application of the model, the training model can be continuously updated and optimized to improve the accuracy and effectiveness of the model.

[0042] After the wind tunnel test is completed, the final wind field data (wind speed, turbulence intensity, temperature, etc.) is collected and compared with the results predicted by the wind field adjustment model. Based on the analysis results, the parameters and structure of the deep learning model are further adjusted to optimize its prediction ability. In addition, this data can also be used to verify and calibrate the simulation software to improve the accuracy and efficiency of future simulations.

[0043] A system using the wind tunnel wind field adjustment method based on CFD technology and deep learning, comprising:

[0044] Parameter acquisition module, used to obtain various flow field experimental parameters of the wind tunnel, including wind speed, temperature, turbulence intensity and other parameters;

[0045] The CFD numerical simulation module performs numerical simulation of the wind tunnel wind field according to the geometric model, boundary conditions and experimental parameters of the wind tunnel, and adjusts the simulation parameters until the output simulation results match the adjusted results of the actual wind tunnel wind field;

[0046] High-performance computing module, used to process large-scale CFD numerical simulation calculations and provide computing resource support;

[0047] Deep learning module, used to train simulation results under different working conditions and build a wind farm adjustment model;

[0048] A prediction module, used to predict the optimal layout of the rough element and the wedge device through the wind field adjustment model according to the input target wind field parameters;

[0049] The output module is used to output the prediction results, that is, to output the predicted optimal layout information of the rough elements and wedge devices in the wind tunnel, so as to achieve rapid and accurate adjustment of the target wind field.

[0050] The wind field adjustment model constructed by the deep learning module and the prediction module is modeled and optimized based on the training data of various wind tunnel experimental parameters and CFD simulation results using adaptive algorithms, neural networks, deep learning or other artificial intelligence methods to achieve precise control and adjustment of the wind tunnel wind field.

[0051] The deep learning module uses deep learning architectures such as multi-layer perceptron (MLP), convolutional neural network (CNN), recurrent neural network (RNN), long short-term memory network (LSTM) or graph neural network (GNN) to process CFD simulation data. By learning from a large amount of historical data, the module is able to identify and extract key features in wind tunnel wind field adjustment, including velocity distribution, pressure field, turbulence characteristics, etc., and how they are affected by changes in the internal layout of the wind tunnel. The learning process includes feedforward calculation, backpropagation and weight update, which optimizes model performance by gradually reducing prediction errors. In addition, the module can also apply transfer learning technology to quickly adapt models trained on one or more source wind tunnel experiments to new wind tunnel experiment configurations, reducing the amount of data required and training time.

[0052] The present invention also includes a graphical user interface module, a database module, a safety control module and an interface module, wherein the graphical user interface module is used for users to input target wind field parameters, and displays the predicted placement information of rough elements and wedge devices, and provides interactive operation and wind tunnel experiment parameter adjustment functions. The database module is used to store and manage relevant data such as CFD simulation results, experimental parameters and wind field adjustment models, and provides data query, analysis and backup functions. The real-time monitoring module is used to monitor the wind tunnel experiment process and wind field adjustment results in real time, and displays the real-time data of the experiment and the adjusted wind field simulation results through a graphical user interface, so that users can perform real-time analysis and decision-making. The safety control module is used to monitor the safety status of the wind tunnel experiment, including but not limited to monitoring excessive temperature, abnormal pressure changes and other potential safety risks, and automatically stopping the experiment and warning the user when any safety problem is detected. The interface module enables the system to exchange and integrate data with other experimental equipment such as flow meters, temperature sensors, humidity sensors, etc., to improve the automation level and efficiency of wind tunnel experiments.

[0053] The present invention uses a variety of sensors to monitor the environmental parameters inside the wind tunnel in real time, such as temperature, humidity, wind speed and pressure, and updates them to the parameter acquisition module in real time to ensure the accuracy of wind tunnel experiment parameters and the reliability of experimental results.

[0054] The present invention regularly updates and optimizes the model to adapt to new wind tunnel test data and improve the accuracy and efficiency of wind field adjustment, including strategies such as online learning, transfer learning or incremental learning. In order to adapt to new wind tunnel test data and improve the accuracy and efficiency of wind field adjustment, the update and optimization of the model can adopt the latest progress in the field of deep learning, such as adaptive learning rate adjustment algorithms (such as Adam, RMSprop), regularization techniques (such as Dropout, BatchNormalization) to reduce overfitting, and use generative adversarial networks (GAN) for data enhancement, thereby improving the generalization ability of the model. Regularly updating the model also includes regular retraining using newly collected experimental data, or using online learning strategies to update model parameters in real time, and using incremental learning methods to gradually add new knowledge without completely forgetting old knowledge.

[0055] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A wind tunnel wind field adjustment method based on CFD technology and deep learning, characterized in that The following steps are involved: S1. Obtain various flow field experimental parameters of the wind tunnel; S2. According to the geometric model, boundary conditions and experimental parameters of the wind tunnel, the wind field of the wind tunnel is numerically simulated using CFD technology, and the simulation parameters are adjusted until the output simulation results match the adjusted results of the actual wind tunnel wind field, and the simulation results are collected; S3. Combine the deep learning model to train the collected simulation results under different working conditions and build a wind farm adjustment model; S4. In the actual wind tunnel test, the target wind field parameters are input into the wind field adjustment model to predict the optimal layout of the roughness elements and the wedge devices in the wind tunnel. According to the optimal layout, the positions of the roughness elements and the wedge devices in the wind tunnel are adjusted to achieve the target wind field parameters.

2. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 1, characterized in that: The experimental parameters are wind speed, temperature and turbulence intensity.

3. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 2 is characterized in that: The simulation parameters are grid resolution, turbulence model and boundary conditions.

4. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 3 is characterized in that: The simulation results are wind speed distribution, pressure distribution and turbulence intensity. The simulation results match the actual wind tunnel wind field adjusted results, which means that the error between the two is 10 -3 the following.

5. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 4 is characterized in that: In step S1, the experimental parameters are preprocessed to remove outliers and fill in missing values, and then the process proceeds to step S2.

6. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 5 is characterized in that: In step S3, the simulation results under different working conditions used as training sets are labeled, and each data point is marked as a wind field that meets or does not meet the requirements according to preset target wind field parameters. At the same time, according to actual experimental adjustment results, a part of the data of the training set is manually labeled for verification and evaluation of the deep learning model.

7. The wind tunnel wind field adjustment method based on CFD technology and deep learning according to claim 6, characterized in that: After completing the wind tunnel test, the final wind field data is collected and compared with the prediction results of the wind field adjustment model. According to the analysis results, the parameters and structure of the wind field adjustment model are adjusted to optimize its prediction ability.

8. A system using the wind tunnel wind field adjustment method based on CFD technology and deep learning as claimed in claim 1, characterized in that include: Parameter acquisition module, used to obtain various flow field experimental parameters of the wind tunnel; The CFD numerical simulation module performs numerical simulation of the wind tunnel wind field according to the geometric model, boundary conditions and experimental parameters of the wind tunnel, and adjusts the simulation parameters until the output simulation results match the adjusted results of the actual wind tunnel wind field; High-performance computing module, used to handle large-scale CFD numerical simulation calculations; Deep learning module, used to train simulation results under different working conditions and build a wind farm adjustment model; A prediction module, used to predict the optimal layout of the rough element and the wedge device through the wind field adjustment model according to the input target wind field parameters; Output module, used to output prediction results.

9. The system according to claim 8, characterized in that: The system includes a graphical user interface module and a database module. The graphical user interface module is used for users to input target wind field parameters, display the predicted optimal layout information of rough elements and wedge devices, and provide interactive operation and wind tunnel experiment parameter adjustment functions; the database module is used to store and manage simulation results, experimental parameters and wind field adjustment models and provide data query, analysis and backup functions; the parameter acquisition module obtains the experimental parameters of the wind tunnel through a sensor used for real-time monitoring of the internal environmental parameters of the wind tunnel. The sensor collects the environmental parameters and updates them to the parameter acquisition module in real time.

10. The system according to claim 9, characterized in that: The system includes a real-time monitoring module, a safety control module and an interface module. The real-time monitoring module is used to monitor the wind tunnel experiment process and the wind field adjustment results in real time, and display the real-time data of the experiment and the adjusted wind field simulation results through a graphical user interface; the safety control module is used to monitor the safety status of the wind tunnel experiment, and automatically stop the experiment and warn the user when a safety problem is detected; the interface module can be connected with other experimental equipment for data exchange and integration.

Citation Information

Cited By

  • Low-altitude flight micrometeorological intelligent decision-making system based on multi-source sensing and AI fusion

    CN120577898A