Virtualized training live-action three-dimensional model generation system and method based on digital twinning

Abstract

The invention discloses a virtual training live-action three-dimensional model generation system and method based on digital twinning. The system comprises a data acquisition module, a digital twinning model establishment module, a live-action three-dimensional model establishment module, a model fusion module, a display module, a preliminary use module, a feedback suggestion module, a model optimization module and a model export module. The data acquisition module is used for collecting physical entity data related to a training scene, namely entity physical data; the digital twinborn model building module is used for building a digital twinborn model of a physical entity in a virtual space by utilizing the collected entity physical data; and the live-action three-dimensional model establishing module is used for establishing a three-dimensional model of a real scene by utilizing a live-action three-dimensional technology, namely obtaining a live-action three-dimensional model. According to the invention, the live-action three-dimensional model with higher accuracy can be generated, so that the training effect is better improved.

Description

Technical Field

[0001] The present invention relates to the field of three-dimensional model generation, and in particular to a system and method for generating a real-scene three-dimensional model for virtual training based on digital twins. Background Art

[0002] Virtual training is a kind of artificial virtual environment with real-time and three-dimensional information generated by virtual reality technology. Trainees enter the environment by using certain equipment and various sensory stimulations of the corresponding environment, and can control the environment, operate tools and objects through a variety of interactive devices as needed;

[0003] In virtual training, 3D models play a vital role. These models are created through professional 3D modeling software, which can accurately restore objects, scenes or processes in the real world, providing students with an immersive and highly simulated learning environment. In the process of building real-scene 3D models, real-scene 3D model generation systems and methods are used.

[0004] The existing real-scene 3D model generation system and method adopt a relatively traditional modeling method, which results in the generated model being not accurate enough, affecting the actual training effect and bringing certain impacts on the use of the real-scene 3D model generation system and method. Therefore, a real-scene 3D model generation system and method for virtual training based on digital twins is proposed. Summary of the invention

[0005] In view of the defects in the prior art, the present invention provides a system and method for generating a real-scene three-dimensional model for virtual training based on digital twins.

[0006] The present invention provides a system and method for generating a real-scene three-dimensional model for virtual training based on digital twins, comprising a data acquisition module, a digital twin model building module, a real-scene three-dimensional model building module, a model fusion module, a display module, a preliminary use module, a feedback suggestion module, a model optimization module and a model export module;

[0007] The data acquisition module is used to collect physical entity data related to the training scenario, i.e., entity physical data;

[0008] The digital twin model building module is used to build a digital twin model of the physical entity in the virtual space using the collected physical data of the entity;

[0009] The real-scene 3D model building module is used to build a 3D model of a real scene using real-scene 3D technology, that is, to obtain a real-scene 3D model;

[0010] The model fusion module is used to fuse the digital twin model with the real-scene three-dimensional model to obtain a preliminary application three-dimensional model;

[0011] The display module is used to display the preliminary application three-dimensional model as a three-dimensional model;

[0012] The preliminary use module is used to try out the preliminary application three-dimensional model;

[0013] The feedback suggestion module is used for users to provide feedback on abnormalities in the initial application of the three-dimensional model and feedback on opinions during the trial application of the three-dimensional model during the initial display process, that is, to obtain display feedback information and application feedback information;

[0014] The model optimization module is used to process the display feedback information and the application feedback information, generate model optimization information and optimize the model;

[0015] The model export module is used to export the optimized model after the model optimization is completed.

[0016] Furthermore, the specific process of establishing the digital twin model is as follows:

[0017] Data collation and preprocessing: collate the collected physical object data and test the data integrity and consistency;

[0018] After the data integrity and consistency are normal, the data is preprocessed to obtain the preprocessed data;

[0019] Extract features related to physical objects from the preprocessed data to obtain a feature set;

[0020] Select the feature set according to the modeling requirements, and then select the modeling method according to the characteristics of the physical object and the modeling requirements;

[0021] Based on the extracted feature set and the selected modeling method, a preliminary digital twin model is established;

[0022] Model verification and optimization: Use the verification data set to verify the established preliminary digital twin model and obtain the verification results. If the verification optimization results are normal and meet user requirements, the model can be directly exported.

[0023] If the verification and optimization results are abnormal or do not meet user requirements, the model is optimized and the verification and optimization process is repeated until the verification results are normal and meet user requirements, and the generated digital twin model is exported.

[0024] Furthermore, in the process of establishing the digital twin model based on the extracted feature set and the selected modeling method, at least two are simultaneously selected to simultaneously establish the model, and at least two preliminary digital twin models are obtained;

[0025] During the model verification and optimization process of at least two preliminary digital twin models, the model with the least number of verification issues is recorded for subsequent verification and optimization process.

[0026] Furthermore, the specific process of constructing a three-dimensional model of a real scene using the real-scene three-dimensional technology is as follows:

[0027] Data collection: collect data on the real scene area that needs to be modeled to obtain real scene modeling data;

[0028] Data preprocessing: preprocess the collected real-scene modeling data to obtain preprocessed modeling data;

[0029] Data import: import pre-processed modeling data into 3D modeling software;

[0030] 3D modeling: Based on the imported pre-processed modeling data, modeling is performed in 3D modeling software to obtain a solid 3D model;

[0031] Model optimization: Optimize the solid 3D model and export the model after meeting user needs.

[0032] Furthermore, the specific process of data collection is as follows: using drones, 3D scanning and laser radar to collect data on the modeled entity, and obtaining three entity data;

[0033] The three entity data are transformed into similar data, and then the three data of each feature point in the similar data are extracted, and the evaluation parameters of each feature point are obtained by performing calculations one by one;

[0034] Then, the evaluation parameters of each feature point are extracted. When the evaluation parameters of each feature point are normal, the entity data collected by any method is selected as the subsequent preprocessing data;

[0035] When there are abnormalities in the evaluation parameters of each feature point, the entity data is selected for preprocessing according to the actual situation.

[0036] Furthermore, the process of obtaining the evaluation parameters of each feature point and the process of abnormality determination are as follows:

[0037] Extract the collected data of the same type and mark the three entity data in the same type as Qi, Mi and Fi;

[0038] The data of the same points in Qi, Mi and Fi are Qi-Mi=Qmi, Qi-Fi=Qfi, Mi-Fi=Mfi, that is, the evaluation parameters Qmi, Qfi and Mfi of a single feature point are obtained;

[0039] When Qmi, Qfi and Mfi are all less than the preset values, it means that the evaluation parameters of this point are normal; when any one of Qmi, Qfi and Mfi is greater than the preset value, it means that the evaluation parameters of this point are abnormal;

[0040] When the evaluation parameters of each feature point are abnormal, the specific process of selecting entity data for preprocessing according to the actual situation is as follows: extract Qmi, Qfi and Mfi, and when any one of Qmi, Qfi and Mfi is greater than the preset value, but the remaining two are less than the preset value, the volume data corresponding to any one of the two values ​​less than the preset value is selected as the subsequent preprocessing data;

[0041] When Qmi, Qfi and Mfi are all greater than the preset values, the entity data is collected again.

[0042] Furthermore, the model optimization module processes the display feedback information and the application feedback information to generate model optimization information to optimize the model as follows:

[0043] Extract the collected display feedback information and application feedback information. The display feedback information is the model abnormality information uploaded by the preliminary application 3D model personnel during the display process of the preliminary application 3D model. The model abnormal point information is the model abnormal content during the display process.

[0044] Analyze and process all model abnormal contents during the display process of all uploaded content, obtain the feedback times of each abnormal content and mark it as Zv;

[0045] Then, application feedback information is extracted. The application feedback information is abnormal information in the model application process uploaded by the personnel who try out the preliminary application of the three-dimensional model during the trial process. The abnormal information in the model application process is abnormal content in the application process.

[0046] Process all the abnormal contents in the uploaded application process, obtain the number of feedbacks of abnormal contents in each application process, and mark them as Ev;

[0047] Then, the part of Zv and Ev that is greater than the preset value a1 is extracted, the modeling data is re-collected, the corresponding model of this part is rebuilt, and the part of Zv that is less than or equal to the preset value a2 is extracted to directly optimize the model.

[0048] Furthermore, when analyzing and processing the abnormal model content in all uploaded display processes, it is necessary to determine whether the abnormal model content in each display process actually exists. After confirmation, the number of feedback Zv of all abnormal content that actually exists is recorded;

[0049] When processing abnormal contents in all uploaded application processes, it is also necessary to conduct the same real verification as the model abnormal content analysis process in the display process, and then record the number of feedbacks of all abnormal contents that actually exist in each application process.

[0050] The method for generating a real-scene 3D model for virtual training based on digital twins includes the following steps:

[0051] Step 1: Collect physical entity data related to the training scenario, i.e., physical entity data;

[0052] Step 2: Using the collected physical data of the entity, a digital twin model of the physical entity is built in the virtual space;

[0053] Step 3: Use real-scene 3D technology to construct a 3D model of the real scene, that is, obtain the real-scene 3D model;

[0054] Step 4: Integrate the digital twin model with the real-scene 3D model to obtain a preliminary application 3D model;

[0055] Step 5: The display module is used to display the preliminary application three-dimensional model;

[0056] Step 6: The preliminary use module is used to test the preliminary application of the 3D model. The test is conducted on a small scale and the number of testers does not exceed the preset value.

[0057] Step 7: During the initial display process, users provide feedback on abnormalities in the initial application of the 3D model and feedback on opinions during the initial application of the 3D model, and obtain display feedback information and application feedback information;

[0058] Step 8: Process the display feedback information and the application feedback information, generate model optimization information and optimize the model;

[0059] Step 9: After the model optimization is completed, export the optimized model.

[0060] Furthermore, in the process of model optimization, the model needs to be optimized at least twice, and after each optimization, it needs to be displayed and tried again, and display feedback information and application feedback information are collected to continue model optimization;

[0061] After the model optimization is completed and before the optimized model is exported, the model needs to be tested for standards to monitor whether it meets the preset standards. When the model meets the preset standards, it is exported. When the model does not meet the preset standards, it is optimized again until it meets the preset standards.

[0062] The beneficial effects of the present invention are embodied in: the system and method for generating a real-life 3D model for virtualized training based on digital twins. The digital twin technology can quickly and accurately establish digital models of objects and processes, accurately restore objects, and then the real-life 3D model generated by the system can make the training content closer to the real environment and improve the training effect. By simulating and testing in a virtual environment, the number and cost of experiments in a real environment can be reduced. The real-life 3D model based on digital twins has high flexibility and customizability. Different scenes and models can be customized according to different training needs to meet diverse training needs. At the same time, the model can also be modified and updated at any time to adapt to the ever-changing environment and needs, and then meet different training needs, which can improve training effects, reduce costs, improve efficiency and safety, and make the system more worthy of promotion and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0064] Figure 1 It is a system block diagram of the present invention. DETAILED DESCRIPTION

[0065] The following embodiments of the technical solution of the present invention are described in detail in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and are therefore only used as examples, and cannot be used to limit the protection scope of the present invention.

[0066] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in this application should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0067] like Figure 1 As shown, this embodiment provides a technical solution: a virtual training real-scene three-dimensional model generation system based on digital twins, including a data acquisition module, a digital twin model building module, a real-scene three-dimensional model building module, a model fusion module, a display module, a preliminary use module, a feedback suggestion module, a model optimization module and a model export module;

[0068] The data acquisition module is used to collect physical entity data related to the training scenario, that is, physical data. It is necessary to collect various data of physical entities related to the training scenario, including design data, operation data, maintenance data, etc. These data are the basis for establishing a digital twin model;

[0069] The digital twin model building module is used to use the collected physical data of the entity to build a digital twin model of the physical entity in the virtual space. This process requires the use of physical models, sensor updates, operation history and other data, and integrates multi-disciplinary, multi-physical, multi-scale, and multi-probability simulation processes;

[0070] The real-scene 3D model building module is used to build a 3D model of a real scene using real-scene 3D technology, that is, to obtain a real-scene 3D model, which usually includes using a drone equipped with a camera to take pictures of the ground object at one vertical and four inclined angles at the same time, directly reflecting the appearance, position, height and other attributes of the ground object;

[0071] The model fusion module is used to fuse the digital twin model with the real-scene 3D model to obtain a preliminary application 3D model, and fuse the digital twin model with the real-scene 3D model to form a 3D model that contains both physical entity information and reflects the real scene. This process may require adjusting and optimizing the parameters and properties of the two models to ensure their matching and consistency, such as by importing the data and parameters of the digital twin model into the 3D modeling software, or by implementing data sharing and interaction through API interfaces;

[0072] The display module is used to display the preliminary applied 3D model and use advanced visualization technology to display the fused 3D model to the user in an intuitive way. At the same time, it provides functions for users to interact with the model, such as roaming, query, analysis and other operations;

[0073] The preliminary use module is used to try out the preliminary application of the three-dimensional model;

[0074] The feedback and suggestion module is used for users to provide feedback on abnormalities in the initial application of the 3D model and feedback on opinions during the trial of the initial application of the 3D model during the initial display process, that is, to obtain display feedback information and application feedback information;

[0075] The model optimization module is used to process the display feedback information and the application feedback information, generate model optimization information and optimize the model;

[0076] The model export module is used to export the optimized model after the model optimization is completed;

[0077] Model maintenance and update: As physical entities change and update, digital twin models and real-life 3D models need to be constantly maintained and updated. This includes adding new data, adjusting model parameters, fixing model errors, etc.

[0078] Through the above process, a real-life three-dimensional model of virtual training can be established based on digital twin technology, providing students with a real, intuitive and interactive learning environment.

[0079] The specific process of establishing a digital twin model is as follows:

[0080] Data collation and preprocessing: collate the collected physical object data, test the data integrity and consistency, and process the collected data including data cleaning, denoising, missing value filling, etc. to ensure the accuracy and completeness of the data;

[0081] After the data integrity and consistency are normal, the data is preprocessed to obtain the preprocessed data;

[0082] Extract features related to physical objects from preprocessed data and obtain feature sets to reduce model complexity and improve accuracy. These features may include physical properties, behavior patterns, etc.

[0083] According to the modeling requirements, select the feature set, then select the modeling method according to the characteristics of the physical object and the modeling requirements, and select the appropriate model type and algorithm according to the application field and objectives, such as physical model, statistical model, neural network, etc. Then, based on the collected data and features, establish the structure and parameters of the digital twin model;

[0084] According to the characteristics of the physical object and the modeling requirements, select appropriate modeling methods and tools. This may include physical modeling, statistical modeling, machine learning modeling, etc.

[0085] Based on the extracted features and the selected modeling method, a digital twin model is established. During the modeling process, it may be necessary to define the model's input, output, constraints, etc.

[0086] Based on the extracted feature set and the selected modeling method, a preliminary digital twin model is established;

[0087] Model verification and optimization: Use the verification data set to verify the preliminary digital twin model and obtain the verification results. If the verification optimization results are normal and meet user requirements, the model can be directly exported to verify and optimize the established model, including model testing, parameter adjustment, etc. This can be achieved by comparing and verifying with actual data to ensure the accuracy and reliability of the model;

[0088] If the verification and optimization results are abnormal or do not meet user requirements, the model is optimized and the verification and optimization process is repeated until the verification results are normal and meet user requirements, and the generated digital twin model is exported.

[0089] Based on the extracted feature set and the selected modeling method, in the process of establishing the digital twin model, at least two are selected to simultaneously establish the model, and at least two preliminary digital twin models are obtained;

[0090] During the model verification and optimization process of at least two preliminary digital twin models, the model with the least number of verification issues is recorded for subsequent verification and optimization process.

[0091] The specific process of using real-scene 3D technology to build a 3D model of a real scene is as follows:

[0092] Data collection: collect data on the real scene area that needs to be modeled to obtain real scene modeling data;

[0093] Data preprocessing: preprocess the collected real-scene modeling data to obtain preprocessed modeling data;

[0094] Data import: import pre-processed modeling data into 3D modeling software;

[0095] 3D modeling: Based on the imported pre-processed modeling data, modeling is performed in 3D modeling software to obtain a solid 3D model;

[0096] Model optimization: Optimize the solid 3D model and export the model after meeting user needs.

[0097] The specific process of data collection is as follows: using drones, 3D scanning and LiDAR to collect data on modeled entities, and obtaining data of three entities;

[0098] The three entity data are transformed into similar data, and then the three data of each feature point in the similar data are extracted, and the evaluation parameters of each feature point are obtained by performing calculations one by one;

[0099] Then, the evaluation parameters of each feature point are extracted. When the evaluation parameters of each feature point are normal, the entity data collected by any method is selected as the subsequent preprocessing data;

[0100] When the evaluation parameters of each feature point are abnormal, the entity data is selected for preprocessing according to the actual situation;

[0101] Use drones, 3D scanning, LiDAR and other equipment and technical means to collect 3D coordinate information of targets such as the surface, buildings, roads, as well as detailed information such as texture and color. These photos should be high-resolution, clear and without too much distortion to improve the quality of the photos and make them more suitable for modeling.

[0102] Preprocess the collected data, including eliminating data noise and unnecessary information, and adjusting and optimizing the data for subsequent processing steps.

[0103] Matching key points: Matching key points in different photos, usually done using feature descriptors (such as FLANN, Brute-Force, etc.). The matched key points will be used to calculate the camera pose and model structure.

[0104] Based on the matched key points and known camera parameters (such as focal length, principal point coordinates, etc.), the camera pose (position and rotation) of each photo is solved.

[0105] According to the camera posture and the features in the photo, a three-dimensional model is reconstructed through triangulation algorithms (such as Backward Tracking, Layered Tracking, etc.).

[0106] Mapping texture images onto 3D models makes the 3D models more realistic.

[0107] Optimize the 3D model, including deleting redundant vertices, optimizing the model's topological structure, etc., to improve the rendering efficiency of the model.

[0108] The process of obtaining the evaluation parameters of each feature point and the process of abnormality determination are as follows:

[0109] Extract the collected data of the same type and mark the three entity data in the same type as Qi, Mi and Fi;

[0110] The data of the same points in Qi, Mi and Fi are Qi-Mi=Qmi, Qi-Fi=Qfi, Mi-Fi=Mfi, that is, the evaluation parameters Qmi, Qfi and Mfi of a single feature point are obtained;

[0111] When Qmi, Qfi and Mfi are all less than the preset values, it means that the evaluation parameters of this point are normal; when any one of Qmi, Qfi and Mfi is greater than the preset value, it means that the evaluation parameters of this point are abnormal;

[0112] When the evaluation parameters of each feature point are abnormal, the specific process of selecting entity data for preprocessing according to the actual situation is as follows: extract Qmi, Qfi and Mfi, and when any one of Qmi, Qfi and Mfi is greater than the preset value, but the remaining two are less than the preset value, the volume data corresponding to any one of the two values ​​less than the preset value is selected as the subsequent preprocessing data;

[0113] When Qmi, Qfi and Mfi are all greater than the preset values, the entity data is collected again, i is a positive integer, and i is the number information of the feature points;

[0114] Through the above process, the accuracy of the collected modeling data can be discovered in a timely manner. The data can be collected in multiple ways at the same time, and the data can be converted into similar data for comparison, so that it can be intuitively understood whether the data used for modeling is accurate.

[0115] The model optimization module processes the display feedback information and the application feedback information, generates model optimization information, and optimizes the model as follows:

[0116] Extract the collected display feedback information and application feedback information. The display feedback information is the model abnormality information uploaded by the preliminary application 3D model personnel during the display process of the preliminary application 3D model. The model abnormal point information is the model abnormal content during the display process.

[0117] Analyze and process all model abnormal contents during the display process of all uploaded content, obtain the feedback times of each abnormal content and mark it as Zv;

[0118] Then, application feedback information is extracted. The application feedback information is abnormal information in the model application process uploaded by the personnel who try out the preliminary application of the three-dimensional model during the trial process. The abnormal information in the model application process is abnormal content in the application process.

[0119] Process all the abnormal contents in the uploaded application process, obtain the number of feedbacks of abnormal contents in each application process, and mark them as Ev;

[0120] Then, the part of Zv and Ev greater than the preset value a1 is extracted, the modeling data is re-collected, and the corresponding model of this part is rebuilt. The part of Zv less than or equal to the preset value a2 is extracted to directly optimize the model. v is the number of abnormal model contents fed back, and a1<a2.

[0121] When analyzing and processing all the model abnormal contents uploaded during the display process, it is necessary to determine whether the model abnormal contents in each display process actually exist. After confirmation, the number of feedback Zv of all the abnormal contents that actually exist is recorded;

[0122] When processing abnormal contents in all uploaded application processes, it is also necessary to conduct the same real verification as the model abnormal content analysis process in the display process, and then record the number of feedbacks of all abnormal contents that actually exist in each application process;

[0123] Through the above process, it is ensured that the problem to be optimized is real, thereby avoiding the occurrence of problems where the model optimization efficiency is affected by abnormal problems that do not exist in reality.

[0124] The method for generating a real-scene 3D model for virtual training based on digital twins includes the following steps:

[0125] Step 1: Collect physical entity data related to the training scenario, i.e., physical entity data;

[0126] Step 2: Using the collected physical data of the entity, a digital twin model of the physical entity is built in the virtual space;

[0127] Step 3: Use real-scene 3D technology to construct a 3D model of the real scene, that is, obtain the real-scene 3D model;

[0128] Step 4: Integrate the digital twin model with the real-scene 3D model to obtain a preliminary application 3D model;

[0129] Step 5: The display module is used to display the preliminary application three-dimensional model;

[0130] Step 6: The preliminary use module is used to test the preliminary application of the 3D model. The test is conducted on a small scale and the number of testers does not exceed the preset value.

[0131] Step 7: During the initial display process, users provide feedback on abnormalities in the initial application of the 3D model and feedback on opinions during the initial application of the 3D model, and obtain display feedback information and application feedback information;

[0132] Step 8: Process the display feedback information and the application feedback information, generate model optimization information and optimize the model;

[0133] Step 9: After the model optimization is completed, export the optimized model.

[0134] During the model optimization process, the model needs to be optimized at least twice. After each optimization, it needs to be demonstrated and tried again, and the demonstration feedback information and application feedback information need to be collected to continue model optimization.

[0135] After the model optimization is completed and before the optimized model is exported, the model needs to be tested for standards to monitor whether the model meets the preset standards. When the model meets the preset standards, it is exported. When the model does not meet the preset standards, it is optimized again until it meets the preset standards.

[0136] Through the above process, the model can be better optimized, thereby ensuring that the final exported model has better authenticity and higher accuracy.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A virtual training real-scene 3D model generation system based on digital twins, characterized in that: It includes data acquisition module, digital twin model building module, real-scene 3D model building module, model fusion module, display module, preliminary use module, feedback and suggestion module, model optimization module and model export module; The data acquisition module is used to collect physical entity data related to the training scenario, i.e., entity physical data; The digital twin model building module is used to build a digital twin model of the physical entity in the virtual space using the collected physical data of the entity; The real-scene 3D model building module is used to build a 3D model of a real scene using real-scene 3D technology, that is, to obtain a real-scene 3D model; The model fusion module is used to fuse the digital twin model with the real-scene three-dimensional model to obtain a preliminary application three-dimensional model; The display module is used to display the preliminary application three-dimensional model as a three-dimensional model; The preliminary use module is used to try out the preliminary application three-dimensional model; The feedback suggestion module is used for users to provide feedback on abnormalities in the initial application of the three-dimensional model and feedback on opinions during the trial application of the three-dimensional model during the initial display process, that is, to obtain display feedback information and application feedback information; The model optimization module is used to process the display feedback information and the application feedback information, generate model optimization information and optimize the model; The model export module is used to export the optimized model after the model optimization is completed.

2. The virtual training real-scene 3D model generation system based on digital twin according to claim 1 is characterized in that: The specific process of establishing the digital twin model is as follows: Data collation and preprocessing: collate the collected physical object data and test the data integrity and consistency; After the data integrity and consistency are normal, the data is preprocessed to obtain the preprocessed data; Extract features related to physical objects from the preprocessed data to obtain a feature set; Select the feature set according to the modeling requirements, and then select the modeling method according to the characteristics of the physical object and the modeling requirements; Based on the extracted feature set and the selected modeling method, a preliminary digital twin model is established; Model verification and optimization: Use the verification data set to verify the established preliminary digital twin model and obtain the verification results. If the verification optimization results are normal and meet user requirements, the model can be directly exported. If the verification and optimization results are abnormal or do not meet user requirements, the model is optimized and the verification and optimization process is repeated until the verification results are normal and meet user requirements, and the generated digital twin model is exported.

3. The virtual training real-scene 3D model generation system based on digital twin according to claim 1 is characterized in that: In the process of establishing the digital twin model based on the extracted feature set and the selected modeling method, at least two are selected simultaneously to establish the model simultaneously, and at least two preliminary digital twin models are obtained; During the model verification and optimization process of at least two preliminary digital twin models, the model with the least number of verification issues is recorded for subsequent verification and optimization process.

4. The virtual training real-scene 3D model generation system based on digital twin according to claim 1 is characterized in that: The specific process of using the real scene 3D technology to construct a 3D model of a real scene is as follows: Data collection: collect data on the real scene area that needs to be modeled to obtain real scene modeling data; Data preprocessing: preprocess the collected real-scene modeling data to obtain preprocessed modeling data; Data import: import pre-processed modeling data into 3D modeling software; 3D modeling: Based on the imported pre-processed modeling data, modeling is performed in 3D modeling software to obtain a solid 3D model; Model optimization: Optimize the solid 3D model and export the model after meeting user needs.

5. The virtual training real-scene 3D model generation system based on digital twin according to claim 4 is characterized in that: The specific process of data collection is as follows: using drones, 3D scanning and laser radar to collect data on the modeled entity, and obtaining three entity data; The three entity data are transformed into similar data, and then the three data of each feature point in the similar data are extracted, and the evaluation parameters of each feature point are obtained by performing calculations one by one; Then, the evaluation parameters of each feature point are extracted. When the evaluation parameters of each feature point are normal, the entity data collected by any method is selected as the subsequent preprocessing data; When there are abnormalities in the evaluation parameters of each feature point, the entity data is selected for preprocessing according to the actual situation.

6. The virtual training real-scene 3D model generation system based on digital twin according to claim 5 is characterized in that: The acquisition process of the evaluation parameters of each feature point and the abnormality determination process are as follows: Extract the collected data of the same type and mark the three entity data in the same type as Qi, Mi and Fi; The data of the same points in Qi, Mi and Fi are Qi-Mi=Qmi, Qi-Fi=Qfi, Mi-Fi=Mfi, that is, the evaluation parameters Qmi, Qfi and Mfi of a single feature point are obtained; When Qmi, Qfi and Mfi are all less than the preset values, it means that the evaluation parameters of this point are normal; when any one of Qmi, Qfi and Mfi is greater than the preset value, it means that the evaluation parameters of this point are abnormal; When the evaluation parameters of each feature point are abnormal, the specific process of selecting entity data for preprocessing according to the actual situation is as follows: extract Qmi, Qfi and Mfi, and when any one of Qmi, Qfi and Mfi is greater than the preset value, but the remaining two are less than the preset value, the volume data corresponding to any one of the two values ​​less than the preset value is selected as the subsequent preprocessing data; When Qmi, Qfi and Mfi are all greater than the preset values, the entity data is collected again.

7. The virtual training real-scene 3D model generation system based on digital twin according to claim 1 is characterized in that: The model optimization module processes the display feedback information and the application feedback information to generate model optimization information to optimize the model as follows: Extract the collected display feedback information and application feedback information. The display feedback information is the model abnormality information uploaded by the preliminary application 3D model personnel during the display process of the preliminary application 3D model. The model abnormal point information is the model abnormal content during the display process. Analyze and process all model abnormal contents during the display process of all uploaded content, obtain the number of feedbacks for each abnormal content and mark it as Zv; Then, application feedback information is extracted. The application feedback information is abnormal information in the model application process uploaded by the personnel who try out the preliminary application of the three-dimensional model during the trial process. The abnormal information in the model application process is abnormal content in the application process. Process all the abnormal contents in the uploaded application process, obtain the number of feedbacks of abnormal contents in each application process, and mark them as Ev; Then, the part of Zv and Ev that is greater than the preset value a1 is extracted, the modeling data is re-collected, the corresponding model of this part is rebuilt, and the part of Zv that is less than or equal to the preset value a2 is extracted to directly optimize the model.

8. The virtual training real-scene 3D model generation system based on digital twin according to claim 1 is characterized by: When analyzing and processing all the model abnormal contents uploaded during the display process, it is necessary to determine whether the model abnormal contents in each display process actually exist. After confirmation, the number of feedback Zv of all the abnormal contents that actually exist is recorded; When processing abnormal contents in all uploaded application processes, it is also necessary to conduct the same real verification as the model abnormal content analysis process in the display process, and then record the number of feedbacks of all abnormal contents that actually exist in each application process.

9. A method for generating a real-scene three-dimensional model for virtual training based on digital twins, the method being based on the generation information described in any one of claims 1 to 8, characterized in that: The following steps are involved: Step 1: Collect physical entity data related to the training scenario, i.e., physical entity data; Step 2: Using the collected physical data of the entity, a digital twin model of the physical entity is built in the virtual space; Step 3: Use real-scene 3D technology to construct a 3D model of the real scene, that is, obtain the real-scene 3D model; Step 4: Integrate the digital twin model with the real-scene 3D model to obtain a preliminary application 3D model; Step 5: The display module is used to display the preliminary application three-dimensional model; Step 6: The preliminary use module is used to test the preliminary application of the 3D model. The test is conducted on a small scale and the number of testers does not exceed the preset value. Step 7: During the initial display process, users provide feedback on abnormalities in the initial application of the 3D model and feedback on opinions during the initial application of the 3D model, and obtain display feedback information and application feedback information; Step 8: Process the display feedback information and the application feedback information, generate model optimization information and optimize the model; Step 9: After the model optimization is completed, export the optimized model.

10. The method for generating a real-scene three-dimensional model for virtual training based on digital twins according to claim 9, characterized in that: In the process of model optimization, the model needs to be optimized at least twice. After each optimization, it needs to be displayed and tried again, and display feedback information and application feedback information are collected to continue model optimization. After the model optimization is completed and before the optimized model is exported, the model needs to be tested for standards to monitor whether it meets the preset standards. When the model meets the preset standards, it is exported. When the model does not meet the preset standards, it is optimized again until it meets the preset standards.

Images