A method and device for analyzing turbulence characteristics of CFD data

Abstract

The present application provides a method and device for analyzing turbulence characteristics of CFD data, which are applied to the field of data analysis technology. The method comprises: target annotating a cross-sectional visualization image of three-dimensional flow field data according to a target detection and annotation tool to obtain a turbulence target training sample set; model training a target detection model according to a training set in the turbulence target training sample set to obtain a candidate turbulence target detection model; model evaluation of the candidate turbulence target detection model based on a test set in the turbulence target training sample set to obtain a turbulence characteristic target detection model; simulation test solution calculations are performed on computational domain flow field data to obtain simulation test flow field data; target detection and analysis processing is performed on the turbulence characteristics contained in the simulation test flow field data according to the turbulence characteristic target detection model to obtain turbulence characteristic analysis results; the beneficial effects of the present application are: effectively improving the automation level of data analysis and processing, data analysis efficiency and data analysis accuracy.

Description

Technical Field

[0001] The present application relates to the technical field of flow field data analysis, and in particular to a method and device for analyzing turbulence characteristics of CFD data. Background Art

[0002] CFD (Computational Fluid Dynamics) simulation software mainly analyzes the internal laws and dynamic physical characteristics of the fluid of industrial design schemes through simulation experiments, and then improves and optimizes the design scheme. It is widely used in aerospace, shipbuilding, automobile, meteorology, medicine and other fields. It is a conventional method to use computational fluid dynamics simulation software to visualize and analyze the three-dimensional flow field data. Through the visualization cloud map, contour map and streamline map of scalar in the flow field, the physical characteristics of the flow field can be visually observed and analyzed, and the laws of change of physical characteristics can be identified and analyzed. Then, the advantages and disadvantages of the design scheme are analyzed, and the optimization and improvement of the design scheme are proposed. At present, the analysis of three-dimensional flow field data is generally based on the experience of analysts, using visualization tools to generate visualization images of flow field data, and then identifying the physical characteristics based on experience, such as whether there are turbulent characteristics in the flow field and the laws of change of turbulent characteristics. Since the analysis and processing of three-dimensional flow field data is complicated, it is necessary to manually select various visualization tools, set multiple parameters for repeated image generation, and then extract the physical features through visual observation and analysis; and the analysis and processing depends on the experience of the analyst, who needs to be familiar with the use of various visualization analysis tools and the accumulation of experience in flow field data analysis of design schemes. Therefore, how to use artificial intelligence methods to automatically perform feature detection and analysis to improve the intelligence level and analysis efficiency of data analysis is one of the problems that need to be solved urgently. Summary of the invention

[0003] In view of this, an embodiment of the present application provides a CFD data turbulence feature analysis method, which can automatically identify turbulence features and analyze and process turbulence change laws of CFD simulation test data, effectively improving the intelligence level and efficiency of data analysis and processing.

[0004] In a first aspect, an embodiment of the present application provides a method for analyzing turbulence characteristics of CFD data, comprising:

[0005] Performing target annotation on the cross-sectional visualization image of the three-dimensional flow field data according to the target detection and annotation tool to obtain a turbulent target training sample set, wherein the target training sample set is divided into a training set and a test set;

[0006] Performing model training on a target detection model according to a training set in the turbulence target training sample set to obtain a candidate turbulence target detection model;

[0007] Performing model evaluation on the candidate turbulence target detection model based on a test set in the turbulence target training sample set to obtain a turbulence feature target detection model;

[0008] Under the design scheme working conditions, the flow field data in the computational domain is simulated and calculated to obtain the flow field data of the simulation test;

[0009] According to the turbulence characteristic target detection model, target detection and analysis processing are performed on the turbulence characteristics contained in the simulation test flow field data to obtain turbulence characteristic analysis results.

[0010] In combination with the first aspect, the embodiment of the present application provides a first possible implementation of the first aspect, wherein, before the target is annotated on the cross-sectional visualization image of the three-dimensional flow field data, the method further includes:

[0011] Obtain the three-dimensional flow field data of the flow field in the calculation domain of the historical simulation test;

[0012] Using a slicing tool of a visualization library to slice the three-dimensional flow field data to obtain a plurality of cross-sectional data sets of the three-dimensional flow field cutting planes;

[0013] The data visualization analysis tool in the visualization class library is used to perform visualization processing on each section plane in the section data set to obtain a section visualization image corresponding to each section plane, wherein the section visualization image includes a cloud map, a contour map, and a streamline map.

[0014] In combination with the first possible implementation of the first aspect, the embodiment of the present application provides a second possible implementation of the first aspect, wherein target annotation is performed on a cross-sectional visualization image of three-dimensional flow field data according to a target detection and annotation tool to obtain a turbulent target training sample set, including:

[0015] Performing turbulence feature analysis on the cross-sectional visualization image to determine turbulence feature objects in the cross-sectional visualization image;

[0016] The turbulence feature objects are labeled according to the target detection and labeling tool to obtain a turbulence target training sample set.

[0017] In combination with the first possible implementation manner or the second possible implementation manner of the first aspect, the embodiment of the present application provides a third possible implementation manner of the first aspect, wherein the target detection model is trained according to the training set in the turbulent target training sample set to obtain a candidate turbulent target detection model, including:

[0018] The training set is input into a target detection model of a convolutional neural network for model training, and a candidate turbulence target detection model is obtained after training, wherein the preset training parameters of the flow field training set include: number of iterations, learning rate, momentum, number of categories, and NMS threshold (Non maximum suppression, NMS non-maximum suppression).

[0019] In combination with the first possible implementation manner or the second possible implementation manner of the first aspect, the embodiment of the present application provides a fourth possible implementation manner of the first aspect, wherein the candidate turbulent target detection model is evaluated based on a test set in the turbulent target training sample set to obtain a turbulent feature target detection model, including:

[0020] According to the detection evaluation index, the candidate turbulence target detection model is evaluated using the test set, wherein the detection evaluation index includes: an accuracy evaluation index and a speed evaluation index;

[0021] If the detection evaluation index of the candidate turbulent target detection model is greater than a preset threshold, the candidate turbulent target detection model meets the detection accuracy, and a turbulent characteristic target detection model is obtained and stored in the model library.

[0022] In combination with the first possible implementation manner or the second possible implementation manner of the first aspect, the embodiment of the present application provides a fifth possible implementation manner of the first aspect, wherein, under the design scheme working condition, a simulation test is performed on the flow field data in the computational domain to solve and calculate, and the simulation test flow field data is obtained, including:

[0023] According to the computational fluid dynamics model, under the design scheme working conditions, simulation tests are performed on the fluid pressure, velocity and turbulence characteristic parameters of the computational domain flow field data to solve and calculate, and the simulation test flow field data of the flow field outside the computational fluid dynamics model is obtained.

[0024] In combination with the first possible implementation or the second possible implementation of the first aspect, the embodiment of the present application provides a sixth possible implementation of the first aspect, wherein target detection analysis is performed on the turbulence features contained in the simulation test flow field data according to the turbulence feature target detection model to obtain a turbulence feature analysis result, including:

[0025] Performing target detection and analysis processing on the turbulence characteristics contained in the simulation test flow field data in the Z-axis direction of the coordinate system according to the turbulence characteristic target detection model;

[0026] The slicing tool of the visualization library is used to slice the three-dimensional flow field data with the Z axis as the tangent plane normal vector to obtain several cross-sectional data sets of the Z-axis three-dimensional flow field cutting plane;

[0027] Using a data visualization analysis tool in a visualization library, data visualization processing is performed on each section plane in the section data set to obtain a section visualization image corresponding to each section plane;

[0028] According to the target detection algorithm, the visual enhanced image corresponding to each section plane is subjected to target detection of turbulence features;

[0029] If turbulence features are detected in the visualization enhanced image, adding the visualization enhanced image of the section plane to the cross-sectional data set;

[0030] For each section plane in the cross-sectional data set, the visualization image of each duration from the start time to the end time of the test is used, and the turbulence characteristic animation data of each section plane position is generated in chronological order using the animation production tool;

[0031] The detected cross-section visualization images containing turbulence feature objects and the generated turbulence feature animation data are taken as turbulence feature analysis results.

[0032] In combination with the first possible implementation manner or the second possible implementation manner of the first aspect, the embodiment of the present application provides a seventh possible implementation manner of the first aspect, wherein the turbulence characteristic analysis result is visualized, including:

[0033] The analyzed turbulence characteristic change animation data results are visualized in a graphic interactive window, wherein the visualization includes: animation play operation, step operation, zoom in operation, and zoom out operation.

[0034] In a second aspect, an embodiment of the present application further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the CFD data turbulence characteristic analysis method according to any one of the above claims when executing the computer program.

[0035] The embodiment of the present application provides a method for analyzing turbulence characteristics of CFD data. The method performs target annotation on a visualization image of a three-dimensional flow field cross-section data according to a target detection annotation tool to obtain a turbulence target training sample set, inputs the training set in the turbulence target training sample set into a target detection model for model training, obtains a candidate turbulence target detection model, and evaluates the turbulence target detection model based on a test set in the target training sample set to obtain a turbulence characteristic target detection model that meets the detection accuracy. Under the design scheme working condition, the flow field data in the computational domain is simulated and solved to obtain simulation test flow field data, and the turbulence characteristics contained in the simulation test flow field data are subjected to target detection analysis and turbulence feature animation generation processing according to the turbulence feature target detection model, and the generated turbulence feature animation data is visualized and displayed. The beneficial effect of this scheme is that it can automatically identify turbulence characteristics of the three-dimensional flow field data of the CFD simulation test based on the target detection model, and generate turbulence feature animations, which can effectively improve the intelligence level and data analysis efficiency of CFD data analysis and processing.

[0036] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] Figure 1 A schematic flow chart of a CFD data turbulence characteristic analysis method provided in an embodiment of the present application is shown.

[0039] Figure 2 A schematic diagram of the process of target labeling of a cross-sectional visualization image in a CFD data turbulence feature analysis method provided in an embodiment of the present application is shown.

[0040] Figure 3 A schematic diagram of the process of performing target detection and analysis on turbulence characteristics in a CFD data turbulence characteristic analysis method provided in an embodiment of the present application is shown.

[0041] Figure 4 A schematic diagram of the structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than embodiments. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0043] At present, the analysis and processing of three-dimensional flow field data in CFD simulation experiments is generally done by using slicing tools to generate several sections, and then generating cloud maps and contour maps of scalar quantities in the flow field in each section, and then analyzing and obtaining the physical characteristics and laws therein. This manual operation analysis and processing method based on a visualization tool set requires analysts to manually select and use various tools based on experience, set various analysis parameters, generate different visualization images, and then identify the physical characteristics and laws in the images through manual observation. The existing method has the following disadvantages:

[0044] First, the analysis and processing work is complicated, requiring manual selection of various visualization tools, setting of multiple parameters for repeated image generation, and then extraction of physical features through visual observation and analysis, which is time-consuming and inefficient.

[0045] Second, it depends on the analyst's experience. The analyst needs to be familiar with the use of various visual analysis tools and have experience in analyzing flow field data of design solutions. Otherwise, it is difficult to draw scientific and accurate test conclusions.

[0046] Third, the same type of simulation tests often have certain correlations, and historical data can be used to train and build feature models, and then automatic feature recognition processing can be performed based on the feature models to improve the intelligent level of data processing. However, the current technical solution does not consider the collection of historical test data.

[0047] In view of the above problems at this stage, we can collect historical similar test data according to the target detection intelligent processing method, establish a turbulence feature annotation data set, train and establish a turbulence feature target detection model, and automatically perform turbulence feature recognition processing on CFD data based on the target detection model. Based on this, the embodiment of the present application provides a CFD data turbulence feature analysis method, which is described below through an embodiment.

[0048] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0049] Figure 1 FIG. 4 is a flow chart showing a method for analyzing turbulence characteristics of CFD data provided in an embodiment of the present application; Figure 1 As shown, the specific steps include:

[0050] Step S10, target annotation is performed on the cross-sectional visualization image of the three-dimensional flow field data according to the target detection and annotation tool to obtain a turbulent target training sample set, wherein the target training sample set is divided into a training set and a test set.

[0051] In the specific implementation of step S10, before the cross-sectional visualization image of the three-dimensional flow field data is subjected to target annotation, the step also includes opening a project menu in the visualization window of the computational fluid dynamics software, selecting a project file to be analyzed from the executed project menu, reading the three-dimensional flow field data of the historical simulation test of the automobile external flow field in the project file through the visualization class library, and parsing the read three-dimensional flow field data, and then using the slicing tool of the visualization class library to take multiple variables on the Z axis, and slicing several cross-sectional planes in the parsed three-dimensional flow field data with the Z axis as the normal vector of the tangent plane to obtain a cross-sectional data set of several three-dimensional flow field cross-sectional planes, and then using the data visualization analysis tool in the visualization class library to visualize each cross-sectional plane in the cross-sectional data set to generate a cross-sectional visualization image of the velocity scalar of each cross-sectional plane, and the above cross-sectional visualization image can be a cloud map, a contour map, and a streamline map; then, according to the target detection and annotation tool, target annotation is performed on the turbulence feature objects in the image to obtain an annotated target training sample set, and the annotated target training sample set is saved in the annotation database, wherein the annotation file of the above turbulence target training sample set follows the XML file in the PASCAL VOC format.

[0052] Step S20, performing model training on the target detection model according to the training set in the turbulence target training sample set to obtain a candidate turbulence target detection model.

[0053] In the specific implementation of step S20, two-thirds of the sample sets are selected from the turbulence target training sample set as the training set, the selected training set is input into the target detection model of the convolutional neural network, and the target detection model is trained according to the training parameters of the training set until the model reaches the number of iterations to obtain a candidate turbulence target detection model; commonly used turbulence target detection models include R-CNN model (Selective Search+CNN+SVM), Fast-R-CNN model (Selective Search+CNN+ROI)), Faster-R-CNN model (Selective Search+CNN+ROI), YOLO model (You Only Look Once), and SSD model (Single Shot MultiBox Detector).

[0054] Step S30 , performing model evaluation on the candidate turbulent target detection model based on the test set in the turbulent target training sample set to obtain a turbulent target detection model.

[0055] In the specific implementation of step S30, one-third of the sample set is selected from the turbulence target training sample set as the test set, and the model accuracy and model speed of the candidate turbulence target detection model are evaluated based on the test set. If the detection accuracy of the candidate turbulence target detection model reaches the threshold, the candidate turbulence target detection model is saved in the model library to obtain a turbulence feature target detection model that meets the detection accuracy.

[0056] Step S40, under the design scheme working condition, a simulation test is performed on the flow field data in the calculation domain to solve and calculate, and obtain the simulation test flow field data.

[0057] During the specific implementation of step S40, under the computational domain flow field conditions of the industrial design solution, based on the fluid dynamics solver of momentum and mass conservation at the bottom of the computational fluid dynamics model, combined with the time change, the turbulence characteristics at a specific time point of each local position of the computational domain flow field data are simulated and calculated, and the fluid dynamics solver matches the current simulation test operation according to the characteristics of the law of change of the turbulence characteristics, and the solution process is completed through the simulation test operation to obtain the simulation test flow field data.

[0058] Step S50, performing target detection and analysis processing on the turbulence characteristics contained in the simulation test flow field data and the turbulence characteristic animation data according to the turbulence characteristic target detection model to obtain a turbulence characteristic analysis result.

[0059] When step S50 is implemented, the turbulence characteristics of the Z axis, X axis, and Y axis corresponding to the simulation test flow field data are automatically identified according to the turbulence characteristic target detection model. If turbulence characteristics exist, corresponding turbulence characteristic animation data is generated as the turbulence characteristic analysis result.

[0060] In one possible implementation, Figure 2 A schematic diagram of a process for target labeling of a cross-sectional visualization image in a CFD data turbulence feature analysis method provided in an embodiment of the present application is shown; in the above step S10, target labeling is performed on the cross-sectional visualization image of the three-dimensional flow field data according to the target detection and labeling tool to obtain a turbulence target training sample set, including:

[0061] Step S101 : using a target detection and annotation tool to perform turbulence feature analysis on the cross-sectional visualization image in a manual recognition form, and determining turbulence feature objects in the cross-sectional visualization image.

[0062] Step S102: labeling the turbulence feature objects according to the target detection and labeling tool to obtain a turbulence target training sample set.

[0063] In the specific implementation of steps S101 and S102, according to the annotation function of the target detection and annotation tool, the turbulence features on the cross-sectional visualization image are annotated to obtain an annotated turbulence target training sample set, and the annotated turbulence target training sample set is saved in the annotation database. The above turbulence training sample set is a training sample set of the same type of CFD simulation test, such as the training sample set of the automobile external flow field simulation test in this embodiment, and other types of training sample sets can also be established according to the test category.

[0064] In a feasible implementation scheme, in the above step S20, the target detection model is trained according to the training set in the turbulent target training sample set to obtain a candidate turbulent target detection model, including:

[0065] Step 201, input the training set into the target detection model of the convolutional neural network for model training, and obtain a candidate turbulent target detection model after training, wherein the preset training parameters of the flow field training set include: number of iterations, learning rate, momentum, number of categories, and NMS threshold.

[0066] In the specific implementation of step 201, two-thirds of the sample sets are selected from the turbulence target training sample set as the training set, the training set is input into the convolutional neural network target detection model, and the model training calculation is performed according to the set training parameters to obtain the candidate turbulence target detection model. The training parameters include: number of iterations, learning rate, momentum, number of categories, and NMS threshold.

[0067] In a feasible implementation scheme, in the above step S30, the candidate turbulence target detection model is evaluated based on the test set in the turbulence target training sample set to obtain the turbulence feature target detection model, including:

[0068] Step 301 , based on the detection evaluation index, a test set is used to perform a model evaluation on the candidate turbulence target detection model, wherein the detection evaluation index includes: an accuracy evaluation index and a speed evaluation index.

[0069] Step 302: If the detection evaluation index of the candidate turbulence target detection model is greater than a preset threshold, the candidate turbulence target detection model meets the detection accuracy, and a turbulence characteristic target detection model is obtained and stored in the model library.

[0070] In the specific implementation of steps 301 and 302, after the model training is completed, one-third of the sample set is selected from the turbulence target training sample set as the test set, and the test set is input into the candidate turbulence target detection model, and the candidate turbulence target detection model is evaluated according to the accuracy evaluation index and the speed evaluation index. If the detection accuracy of the candidate turbulence target detection model reaches the threshold, it is saved in the model library for actual turbulence feature target detection.

[0071] In a feasible implementation scheme, in the above step S40, under the design scheme working condition, the flow field data in the computational domain is subjected to simulation test solution calculation to obtain the simulation test flow field data, including:

[0072] Step 401 , according to the computational fluid dynamics model under the design scheme working condition, the fluid pressure, velocity and turbulence characteristic parameters of the computational domain flow field data are simulated and calculated to obtain the simulation test flow field data of the flow field outside the computational fluid dynamics model.

[0073] During the specific implementation of step 401, according to the fluid dynamics solver of the underlying conservation of momentum and mass of the computational fluid dynamics model, the NS equations (Navier-Stokes equations, viscous fluid dynamics) and the continuity equation are used to perform simulation test solutions and calculations on the fluid characteristics in the computational domain flow field data within the three-dimensional region, such as fluid pressure, velocity, turbulence characteristic parameters, etc. The NS equations are used to split the Cartesian coordinate system, and the velocity vector is decomposed along each coordinate axis to obtain three velocity variables respectively. The above-mentioned continuity equation is combined to form a set of equations to simulate the flow process of the turbulence analysis data to be processed, and the simulation test flow field data is obtained.

[0074] In one possible implementation, Figure 3A schematic diagram of a process for target detection and analysis of turbulence features in a CFD data turbulence feature analysis method provided in an embodiment of the present application is shown; in the above step S50, target detection and analysis processing is performed on the turbulence features contained in the simulation test flow field data according to the turbulence feature target detection model to obtain turbulence feature analysis results, including:

[0075] Step S501 , using the slicing tool of the visualization library, with the Z-axis direction as the tangent plane normal vector, to slice the simulation test flow field data to obtain a plurality of cross-sectional data sets of the Z-axis three-dimensional flow field cutting planes.

[0076] Step S502 : using a data visualization analysis tool in a visualization class library, performing data visualization processing on each section plane in the section data set to obtain a section visualization image corresponding to each section plane.

[0077] Step S503 : performing image enhancement processing on the cross-sectional visualization image corresponding to each section plane to obtain a visualization enhanced image.

[0078] Step S504: According to the target detection algorithm, target detection and analysis processing of turbulence features are performed on the visualization enhanced image corresponding to each cross-section plane.

[0079] Step S505: if turbulence features are detected in the visualization enhanced image, the visualization enhanced image of the section plane is added to the cross-section data set.

[0080] Step S506, for each section plane in the cross-sectional data set, for each duration of the visualization image from the start time to the end time of the test, use an animation tool to generate turbulence feature animation data at each section plane position in chronological order.

[0081] Step S507: taking the detected cross-section visualization image containing the turbulence feature object and the generated turbulence feature animation data as the turbulence feature analysis result.

[0082] In the specific implementation of steps S501, S502, S503, S504, S505, S506, and S507, a slicing tool of a visualization class library is used to uniformly take multiple variables on the Z axis, and a plurality of cross-sectional planes in the parsed three-dimensional flow field data are sliced ​​with the Z axis as the normal vector of the tangent plane to obtain a cross-sectional data set of a plurality of three-dimensional flow field cross-sectional planes. Then, a data visualization analysis tool in the visualization class library is used to visualize each cross-sectional plane in the cross-sectional data set to generate a cross-sectional visualization image of the velocity scalar of each cross-sectional plane; the engineering drawing color template of the cross-sectional visualization image is set to a grayscale type, the number of color tables is set, the grayscale level of the cross-sectional visualization image is calculated based on an algorithm in the spatial domain, and the transformation coefficient value of the image is corrected in the image change domain based on an algorithm in the frequency domain. Then, the cross-sectional visualization image is smoothed and sharpened based on the mean filtering algorithm and the median filtering algorithm to obtain a visualization enhanced image; the turbulence feature target detection model is loaded from the model library, and the visualization enhanced image corresponding to each cross-sectional plane is subjected to turbulence feature target detection according to the target detection algorithm. If turbulence features are detected, the visualization image of the cross-sectional plane is added to the cross-sectional data set; the animation production tool is used to generate turbulence feature animation data for each cross-sectional plane position according to the visualization image of each time length from the start time to the end time of the test in the cross-sectional data set in chronological order; the turbulence feature parameters of the X-axis and Y-axis are respectively subjected to turbulence feature recognition and turbulence change animation generation and analysis processing, so as to obtain the turbulence feature analysis results of the CFD flow field data.

[0083] In a feasible implementation scheme, the above steps are used to visualize the analysis results of turbulence characteristics, including:

[0084] Step 601 , visually displaying the analyzed turbulence characteristic change animation data results in a graphic interactive window, wherein the visual display includes: animation play operation, step operation, zoom in operation, and zoom out operation.

[0085] To further illustrate the present solution, the present application provides a specific application embodiment of establishing a turbulence feature detection model in a CFD data turbulence feature analysis method, which specifically includes the following contents:

[0086] Step A1, collecting historical flow field data of automobile external flow field simulation test;

[0087] Step A2, read the flow field data of the historical simulation test of the external flow field of the automobile from the visualization class library, slice the flow field data on several cross-sectional planes according to the filter tool, and then generate the cross-sectional visualization image of the velocity scalar, such as cloud map, contour map, and streamline map, through the data visualization analysis tool, set the cross-sectional visualization image engineering drawing color template to grayscale mode, set the number of color tables to 20, and perform image enhancement processing on the cross-sectional visualization image;

[0088] Step A3: using a target detection and annotation tool to annotate the turbulence features in the cross-sectional visualization image, and save the annotated data in an annotation database;

[0089] Step A4: Repeat the above steps 2-3 until all historical test data are marked;

[0090] Step A5: Randomly select N labeled data from the labeled database as a sample set for model training, use 2 / 3 of the data as a training set, and the remaining 1 / 3 of the data as a test set;

[0091] Step A6: input the training set into the target detection model for model training until the model converges or the number of iterations is reached, and the training ends to obtain a candidate turbulence target detection model;

[0092] Step A7: perform model evaluation on the candidate turbulence target detection model according to the test set, and calculate the detection accuracy of the turbulence target detection model. If the detection accuracy reaches a threshold (>90%), the turbulence feature target detection model is obtained and saved in the model library.

[0093] To further illustrate the present solution, the present application provides a specific application embodiment of intelligent analysis of turbulence characteristics based on a computational fluid dynamics model in a CFD data turbulence characteristic analysis method, which specifically includes the following contents:

[0094] Step B1, open the project menu in the visualization window, and select the project file to be analyzed from the executed project menu;

[0095] Step B2, read the flow field data in the engineering file through the visualization class library, use the filter tool of the visualization class library to evenly take multiple variables on the Z axis, slice the analyzed flow field data with the Z axis as the normal vector of the tangent plane, and obtain the cross-sectional data set Sz;

[0096] Step B3, using the data visualization analysis tool in the visualization class library, generating a velocity scalar cross-sectional visualization image SCzi according to each slice data in the cross-sectional data set Sz;

[0097] Step B4, setting the engineering drawing color template of the cross-section visualization image SCzi to grayscale type, setting the number of color tables to 20, and performing image enhancement processing on the cross-section visualization image to make the velocity contour line clear;

[0098] Step B5, using the YOLO target detection model, loading the computational fluid dynamics model from the model library to detect the turbulence feature object in each image of the cross-sectional visualization image SCzi, if the cross-sectional visualization image SCzi has turbulence features, then adding it to the cross-sectional image set SCTz of turbulence features;

[0099] Step B6, if the cross-sectional image set SCTz of turbulence characteristics is not empty, it means that there is turbulence characteristics in the z-axis direction, and according to a certain section cross-sectional visualization image SCzi in the cross-sectional image set SCTz of turbulence characteristics, a dynamic picture Tz of the change law of the cross-sectional turbulence characteristics is generated in chronological order by using an animation production tool;

[0100] Step B7, analyzing the turbulence characteristics in the X-axis direction and the Y-axis direction respectively according to steps 2-6, and if turbulence exists, generating corresponding turbulence characteristic animation data Tx, Ty;

[0101] Step B8: Display the obtained dynamic results of turbulence feature prediction in a visualization window. The corresponding animation operation human-computer interaction functions include: animation play operation, step operation, zoom in operation, and zoom out operation.

[0102] Corresponds to Figure 1 The CFD data turbulence characteristic analysis method in the present application embodiment also provides a computer device 70, Figure 4 ,like Figure 4 As shown, the device includes a memory 701, a processor 702, and a computer program stored in the memory 701 and executable on the processor 702, wherein the processor 702 implements the method when executing the computer program.

[0103] The cross-sectional visualization image of the three-dimensional flow field data is annotated with a target according to the target detection and annotation tool to obtain a turbulent target training sample set, wherein the target training sample set is divided into a training set and a test set;

[0104] The target detection model is trained according to the training set in the turbulence target training sample set to obtain a candidate turbulence target detection model;

[0105] Based on the test set in the turbulence target training sample set, the candidate turbulence target detection model is evaluated to obtain the turbulence characteristic target detection model;

[0106] Under the design scheme working conditions, the flow field data in the computational domain is simulated and calculated to obtain the flow field data of the simulation test;

[0107] According to the turbulence characteristic target detection model, the turbulence characteristics contained in the simulation test flow field data are subjected to target detection analysis and processing to obtain the turbulence characteristic analysis results.

[0108] Based on the above analysis, it can be seen that compared with manual identification of turbulent dynamic physical characteristics in CFD simulation test flow field data by analysts, the method provided in the embodiment of the present application is first based on historical test data, establishes a training sample set through turbulence feature annotation, and performs model training to obtain a turbulence feature target detection model; then, based on the turbulence feature target detection model, turbulence features of CFD simulation test flow field data are automatically identified, and turbulence feature change animation data is generated, which can effectively improve the intelligence level and data analysis efficiency of data analysis and processing.

[0109] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of devices or units can be electrical, mechanical or other forms.

[0110] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0111] In addition, each functional unit in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0112] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0113] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and are not to be understood as indicating or implying relative importance.

[0114] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present application, which are used to illustrate the technical solutions of the present application, rather than to limit them. The protection scope of the present application is not limited thereto. Although the present application is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed in the present application, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should all be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for analyzing turbulence characteristics of CFD data, characterized in that: include: Performing target annotation on the cross-sectional visualization image of the three-dimensional flow field data according to the target detection and annotation tool to obtain a turbulent target training sample set, wherein the target training sample set is divided into a training set and a test set; Performing model training on a target detection model according to a training set in the turbulence target training sample set to obtain a candidate turbulence target detection model; Performing model evaluation on the candidate turbulence target detection model based on a test set in the turbulence target training sample set to obtain a turbulence feature target detection model; Under the design scheme working conditions, the flow field data in the computational domain is simulated and calculated to obtain the flow field data of the simulation test; Performing target detection and analysis processing on the turbulence characteristics contained in the simulation test flow field data according to the turbulence characteristic target detection model to obtain turbulence characteristic analysis results; According to the turbulence feature target detection model, target detection and analysis processing is performed on the turbulence features contained in the simulation test flow field data to obtain turbulence feature analysis results, including: The slicing tool of the visualization library is used to slice the simulation test flow field data with the Z-axis direction as the tangent plane normal vector to obtain several cross-sectional data sets of the Z-axis three-dimensional flow field cutting plane; Using a data visualization analysis tool in a visualization library, each section plane in the section data set is visualized to obtain a section visualization image corresponding to each section plane; Performing image enhancement processing on the cross-sectional visualization image corresponding to each section plane to obtain a visualization enhanced image; According to the target detection algorithm, the visual enhanced image corresponding to each section plane is subjected to target detection and analysis processing of turbulence characteristics; If turbulence features are detected in the visualization enhanced image, adding the visualization enhanced image of the section plane to the cross-sectional data set; For each section plane in the cross-sectional data set, the visualization image of each duration from the start time to the end time of the test is used, and the turbulence characteristic animation data of each section plane position is generated in chronological order using the animation production tool; The detected cross-section visualization images containing turbulence feature objects and the generated turbulence feature animation data are taken as turbulence feature analysis results.

2. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: Before target annotation of the cross-sectional visualization image of the three-dimensional flow field data, the following steps are also included: Obtain the three-dimensional flow field data of the flow field in the calculation domain of the historical simulation test; Using a slicing tool of a visualization library, the three-dimensional flow field data is sliced ​​to obtain a plurality of cross-sectional data sets of the three-dimensional flow field cutting planes; The data visualization analysis tool in the visualization class library is used to perform visualization processing on each section plane in the section data set to obtain a section visualization image corresponding to each section plane, wherein the section visualization image includes a cloud map, a contour map, and a streamline map.

3. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: The cross-sectional visualization image of the three-dimensional flow field data is annotated with the target detection and annotation tool to obtain a turbulent target training sample set, including: Performing turbulence feature analysis on the cross-sectional visualization image to determine turbulence feature objects in the cross-sectional visualization image; The turbulence feature objects are labeled according to the target detection and labeling tool to obtain a turbulence target training sample set.

4. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: The target detection model is trained according to the training set in the turbulent target training sample set to obtain a candidate turbulent target detection model, including: The training set is input into a target detection model of a convolutional neural network for model training, and a candidate turbulence target detection model is obtained after training, wherein the preset training parameters of the flow field training set include: number of iterations, learning rate, momentum, number of categories, and NMS threshold.

5. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: The candidate turbulence target detection model is evaluated based on a test set in the turbulence target training sample set to obtain a turbulence feature target detection model, including: According to the detection evaluation index, the candidate turbulence target detection model is evaluated using the test set, wherein the detection evaluation index includes: an accuracy evaluation index and a speed evaluation index; If the detection evaluation index of the candidate turbulent target detection model is greater than a preset threshold, the candidate turbulent target detection model meets the detection accuracy, and a turbulent characteristic target detection model is obtained and stored in the model library.

6. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: Under the design scheme conditions, the flow field data in the computational domain is simulated and calculated to obtain the simulation test flow field data, including: According to the computational fluid dynamics model, under the design scheme working conditions, simulation tests are performed on the fluid pressure, velocity and turbulence characteristic parameters of the computational domain flow field data to solve and calculate, and the simulation test flow field data of the flow field outside the computational fluid dynamics model is obtained.

7. The CFD data turbulence characteristic analysis method according to claim 1, characterized in that: Visualize the turbulence feature analysis results, including: The analyzed turbulence characteristic change animation data results are visualized in a graphic interactive window, wherein the visualization includes: animation play operation, step operation, zoom in operation, and zoom out operation.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 7 are implemented.

Images