Second subject examination room monitoring method and system based on digital twinning

Abstract

The invention discloses a second subject examination room monitoring method and system based on digital twinning, and relates to the field of motor vehicle driving, and the method comprises the steps: carrying out the serialization processing of collected multi-source heterogeneous data, and obtaining three-dimensional virtual data; according to the three-dimensional virtual data, a QEM algorithm is adopted to construct a subject II examination room model, and electronic fence data, camera position data, static element data and dynamic element data are marked in the constructed subject II examination room model; mapping the collected multi-source heterogeneous data to the constructed subject II examination room model to obtain a digital twin examination room model; and monitoring and detecting the examination of the subject II by using the digital twin examination room model. Aiming at the low invigilation efficiency of the second subject examination room, the method constructs the second subject examination room model through the QEM algorithm, maps the real-time data into the model to achieve virtual and real synchronization, and improves the invigilation efficiency of the second subject examination room.

Description

TECHNICAL FIELD

[0001] The application relates to the field of motor vehicle driving, and more particularly to a subject two test site monitoring method and system based on digital twinning. BACKGROUND

[0002] With the rapid development of the national economy and the vigorous rise of the automobile industry, automobiles as important means of transportation have been increasingly popular in social life, and more and more people choose to take the motor vehicle driving test. With the sharp increase in the number of people taking the driving test, the standardization, safety and efficiency of the driving test have become increasingly prominent. The subject two test site, as a key link in the driving test, directly affects the fairness and road traffic safety of the driving test. Therefore, how to effectively improve the monitoring efficiency and management level of the subject two test site has become an important issue to be solved.

[0003] Currently, the subject two test site monitoring mainly relies on traditional manual intervention to ensure the standardization and safety of the test process. Specifically, the invigilators need to manually check a large amount of video monitoring data, rely on the naked eye to judge whether the examinee has abnormal behaviors such as line pressure, boundary crossing and person-vehicle inconsistency, and manually record and handle various test violations. This traditional monitoring method has the following technical defects: first, it consumes a large amount of human and material resources, and the monitoring cost is high; second, the human judgment has subjective differences, and the error rate is relatively high, which is easy to cause misjudgment or omission; third, it cannot realize real-time and accurate monitoring of the whole area and whole period of the test site, and the monitoring coverage is limited; fourth, it lacks data-based and intelligent analysis means, and cannot quantitatively evaluate and trace the test process. These problems seriously restrict the improvement of the monitoring efficiency of the subject two test site, and bring great challenges to the safety, real-time and fairness of the test site. SUMMARY

[0004] In view of the low efficiency of the subject two test site monitoring, the application provides a subject two test site monitoring method and system based on digital twinning, which constructs a subject two test site model through a QEM algorithm, and maps real-time data to the model to realize virtual-real synchronization and improve the monitoring efficiency of the subject two test site.

[0005] One aspect of the application provides a subject two test site monitoring method based on digital twinning, comprising: collecting multi-source heterogeneous data; performing serialization processing on the collected multi-source heterogeneous data to obtain three-dimensional virtual data, the three-dimensional virtual data representing the three-dimensional spatial structure and physical environment of the test site; constructing a subject two test site model using a QEM algorithm according to the three-dimensional virtual data, and marking electronic fence data, camera position data, static element data and dynamic element data in the constructed subject two test site model; mapping the collected multi-source heterogeneous data to the constructed subject two test site model to obtain a digital twinning test site model; and monitoring the subject two test using the digital twinning test site model.

[0006] The three-dimensional virtual data refers to the digital three-dimensional space information formed after the serialization processing of the collected multi-source heterogeneous data, specifically including the space coordinate data in a predefined format converted from the GPS positioning data and the camera position data, and the data set obtained after serialization packaging with the three-dimensional coordinate data. The data can completely represent the three-dimensional space structure, physical environment layout, and spatial position relationship of various facilities of the subject two examination room, providing basic data support for subsequent construction of the digital twin examination room model.

[0007] The QEM algorithm is a calculation method for simplifying and optimizing a three-dimensional mesh model. In the present scheme, the algorithm constructs a three-dimensional mesh data structure of vertices, edges, and facets, calculates the Quadric error matrix of each vertex to quantify the error constraint of the facet on the vertex position, and through steps such as constructing an edge shrinkage priority queue and iteratively shrinking edges, realizes accurate simplification of the three-dimensional model, and finally constructs a high-precision and computationally efficient subject two examination room model.

[0008] The static element data refers to the data information of physical entities and facilities with fixed positions and relatively stable states in the subject two examination room, specifically including the layout data of the five test items of warehouse-in and warehouse-out, curve driving, ramp starting, side parking, and right-angle turning, as well as the three-dimensional position coordinates, size specifications, and orientation angles of the property of the examination room road, buildings, and monitoring equipment. These data serve as fixed reference objects in the examination room model, providing a spatial reference for dynamic monitoring.

[0009] The dynamic element data refers to the data information of moving targets with real-time changes in position and state in the subject two examination room, mainly including the real-time position, driving trajectory, and license plate information of the test vehicles (cars and motorcycles), the real-time position, attitude state, and identity information of the examinees and other personnel. These data are obtained by real-time collection through cameras, GPS modules, and vehicle-mounted terminals, and are continuously updated to the digital twin examination room model, realizing dynamic monitoring and abnormal detection of the examination process.

[0010] Further, the multi-source heterogeneous data includes personnel attitude data, license plate data, identity card data, vehicle position data, GPS positioning data, camera position data, and three-dimensional coordinate data.

[0011] Further, the static element data includes warehouse-in and warehouse-out, curve driving, ramp starting, side parking, and right-angle turning test item data, as well as the position data of the examination room road, buildings, and monitoring equipment.

[0012] The dynamic element data includes real-time state data of test vehicles, examinees, and other personnel.

[0013] Further, the collected multi-source heterogeneous data is serialized to obtain three-dimensional virtual data, including: converting the GPS positioning data and the camera position data into spatial coordinate data in a predefined format; and serializing and packaging the spatial coordinate data and the three-dimensional coordinate data to obtain three-dimensional virtual data representing the three-dimensional spatial structure and the physical environment of the examination room.

[0014] Further, according to the three-dimensional virtual data, a subject two examination room model is constructed using a QEM algorithm, including: constructing a three-dimensional grid according to the three-dimensional virtual data, the three-dimensional grid including vertices, edges, and facets; calculating plane equation coefficients according to a plane equation of each facet; calculating a Quadric error matrix of each vertex according to the plane equation coefficients, the Quadric error matrix being used to quantify error constraints of the facet on the vertex position; traversing all edges in the three-dimensional network, calculating a contraction cost of each edge, sorting the edges in the contraction cost from small to large to construct an edge contraction priority queue; iteratively contracting the edges according to the edge contraction priority queue, selecting the edge with the smallest cost in the queue to contract, generating a new vertex, updating the Quadric error matrix of the corresponding vertex and the contraction cost of the adjacent edges of the vertex, reinserting the updated edges into the priority queue, and repeating the contraction until a preset simplification rate is reached; removing isolated vertices and invalid facets to obtain a simplified three-dimensional model as the subject two examination room model.

[0015] The Quadric error matrix refers to a 4x4 matrix used to quantify the error constraints of the facet on the vertex position in the QEM algorithm. The specific calculation process is as follows: according to the plane equation ax+by+cz+d=0 of each facet, the normal vector n=[a,b,c] T and the distance parameter d are extracted, where a, b, and c represent the components of the plane normal vector in the x-axis, y-axis, and z-axis directions, respectively; T represents transposition; a four-dimensional vector q=[a,b,c,d] T is constructed, and then the Quadric matrix corresponding to the facet is obtained through matrix operation. For each vertex, the Quadric matrices of all adjacent facets are accumulated to obtain the Quadric error matrix of the vertex. This matrix can accurately describe the position constraint relationship of the vertex in the three-dimensional space, providing an error evaluation basis for subsequent edge contraction operations.

[0016] The contraction cost refers to the error loss value caused by the contraction of an edge (V i V j ) to a new vertex V' in the edge contraction process of the QEM algorithm. The smaller the contraction cost, the smaller the impact of the edge contraction on the overall geometric shape of the model, and therefore the higher the contraction priority in the edge contraction priority queue.

[0017] Pre-set simplification rate: refers to the model simplification target proportion pre-set in the three-dimensional model optimization process, used to control the iteration contraction termination condition of QEM algorithm. The parameter determines the reserved proportion of the final simplified three-dimensional model relative to the original model in terms of vertex number, face number or model complexity. When the model simplification reaches the pre-set simplification rate, the algorithm stops the edge contraction iteration process, ensures the geometric accuracy and visual effect of the subject two examination room model while meeting the calculation efficiency requirement, and realizes the balanced optimization of model performance and quality.

[0018] Further, the plane equation coefficients are calculated according to the plane equation of each face, including: establishing the plane equation of each face in the three-dimensional network: ax+by+cz+d=0; extracting the plane equation coefficients of the plane equation, including the normal vector n=[a,b,c] T and the distance parameter d;

[0019] Further, the Quadric error matrix of each vertex is calculated according to the plane equation coefficients, including: according to the normal vector n=[a,b,c] T and the distance parameter d, constructing a four-dimensional vector q=[a,b,c,d] T ; according to the four-dimensional vector q=[a,b,c,d] T , constructing the Quadric error matrix:

[0020]

[0021] Further, all edges in the three-dimensional network are traversed, the contraction cost of each edge is calculated, and the edge contraction priority queue is constructed in ascending order of the contraction cost, including: for the edge (V i V j ), the contraction cost to the new vertex V' is calculated as Cost(V')=V' T (Q Vi +Q Vj )V', wherein, and are the Quadric error matrices of the vertices V i and V j , respectively. Specifically, V i represents the starting vertex in the three-dimensional mesh, V j represents the terminal vertex in the three-dimensional mesh, V' represents the new vertex after the edge contraction operation, and Cost(V') represents the contraction cost function of the new vertex V'; the position of the optimal new vertex V' is calculated by solving ; all edges in the three-dimensional mesh are traversed, the contraction cost of each edge is calculated, and the edge contraction priority queue is formed in ascending order of the contraction cost.

[0022] Further, the collected multi-source heterogeneous data is mapped to the constructed driving test field model to obtain a digital twin driving test field model, including: performing secondary packaging and statistical analysis processing on the collected multi-source heterogeneous data through middleware; binding and mapping vehicle position data to virtual vehicles in the driving test field model; establishing a corresponding relationship between license plate number data and virtual vehicles; mapping and binding camera position data to camera hardware in the driving test field model to obtain the digital twin driving test field model.

[0023] The middleware includes: a message queue, data caching, streaming media conversion, an AI algorithm, message pushing, statistical analysis, a DB middleware, and streaming analysis.

[0024] The middleware refers to a collection of software service components between the data acquisition layer and the digital twin model layer in the digital twin driving test field control system, and is used for realizing the transmission, processing, and management functions of multi-source heterogeneous data. In the present scheme, the middleware specifically includes functional modules such as a message queue (responsible for data transmission and buffering), data caching (providing high-speed data access), streaming media conversion (processing video and audio data format conversion), an AI algorithm (realizing intelligent data analysis), message pushing (realizing abnormal alarm notification), statistical analysis (performing data statistical calculation), a DB middleware (managing database operations), and streaming analysis (realizing real-time data stream processing). These components work cooperatively to ensure that multi-source data can be efficiently and reliably transmitted and processed.

[0025] Secondary packaging refers to repackaging and packaging original data (such as GPS positioning data, license plate number data, and camera data) from different data sources according to unified data formats and protocol standards, so that the original data can be recognized and processed by the digital twin system. Statistical analysis processing refers to mathematical statistical calculation, trend analysis, correlation analysis, and other operations on the packaged data to extract valuable information features, provide analysis basis for subsequent data mapping, abnormal detection, and intelligent judgment, and ensure data quality and analysis accuracy.

[0026] Another aspect of the present application also provides a driving test field monitoring system based on digital twinning, including: a data acquisition module that acquires multi-source heterogeneous data; a data synchronization module that serializes the acquired multi-source heterogeneous data to obtain three-dimensional virtual data; a digital twin module that constructs a driving test field model by using a QEM algorithm according to the three-dimensional virtual data, and marks electronic fence data, camera position data, static element data, and dynamic element data in the constructed driving test field model; a data mapping module that maps the acquired multi-source heterogeneous data to the driving test field model to obtain a digital twin driving test field model; and a monitoring module that monitors a driving test by using the digital twin driving test field model.

[0027] Compared with the prior art, the present application has the following advantages:

[0028] (1) This application uses digital twin technology to synchronize the physical subject II examination room with the virtual digital model in real time. By collecting multi-source heterogeneous data such as personnel posture data, vehicle position data, GPS positioning data, and mapping these real-time data to the constructed digital twin examination room model, it realizes automated monitoring of the entire examination process.

[0029] (2) This application uses the QEM algorithm to optimize the three-dimensional grid. By calculating the vertex Quadric error matrix, building an edge shrinkage priority queue, iteratively shrinking edges and other technical steps, a high-precision subject two test site model is constructed to ensure the accurate correspondence between the virtual model and the physical test site. At the same time, through middleware technology, multi-source heterogeneous data such as vehicle location data, license plate data, and camera location data are integrated and analyzed, and AI algorithms are used for intelligent judgment. It can accurately identify various abnormal situations such as human intrusion, suspected line crossing, human-vehicle mismatch, and vehicle crossing the boundary, thereby improving the reliability of proctoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a diagram of the monitoring architecture of the subject 2 examination room based on digital twins for this application;

[0031] Figure 2 This is a flow chart of a method for monitoring the examination room for subject 2 based on digital twins in this application. DETAILED DESCRIPTION

[0032] The present application is described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] Example 1

[0034] like Figure 1 As shown, a digital twin-based subject two examination room monitoring mainly includes a data acquisition module for collecting multi-source heterogeneous data; a data synchronization module for serializing the collected multi-source heterogeneous data to obtain three-dimensional virtual data; a digital twin module for constructing a subject two examination room model based on the three-dimensional virtual data using the QEM algorithm, and marking the electronic fence data, camera position data, static element data and dynamic element data in the constructed subject two examination room model; a data mapping module for mapping the collected multi-source heterogeneous data to the subject two examination room model to obtain a digital twin examination room model; and a monitoring module for monitoring the subject two examination using the digital twin examination room model.

[0035] The data acquisition module is the foundation and hardware layer of the entire system. Its purpose is to collect relevant data such as personnel posture, body posture, license plate numbers, ID cards, and vehicle location information for subsequent data analysis, converting physical objects into digital information. It primarily consists of hardware devices such as cameras, transmitters, receivers, GPS modules, card readers, vehicle terminals, and alarms.

[0036] The data synchronization module is the transmission layer of the entire system. It is mainly used to transmit and serialize the data collected by the data acquisition module, and convert the collected data from GPS positioning, camera position information and other data into custom three-dimensional virtual data.

[0037] The digital twin module, based on the data synchronization module, performs secondary packaging and statistical analysis on the collected or packaged data. This allows for the construction of virtual 3D scenes and the mapping of positioning data. This includes binding positioning data within the 3D scene, the correspondence between virtual vehicles and real-world license plates, and the mapping and binding of hardware such as cameras. Data is encapsulated and transmitted through middleware, and through data fusion, integration, packaging, and analysis of relevant data information. Ultimately, intelligent judgment is used to determine whether a driver has passed the test, cheated, engaged in illegal driving practice, or engaged in illegal intrusion.

[0038] Specifically, the digital twin module includes digital twin, middleware, data fusion and intelligent evaluation, among which digital twin mainly refers to three-dimensional modeling, three-dimensional model library, data binding, model optimization, simulation algorithm, model loading, site combination, site positioning, lighting simulation, etc.; middleware includes message queue, data cache, streaming media conversion, AI algorithm, configuration management, message push, statistical analysis, DB middleware, streaming analysis, etc.; data fusion includes information transmission, information analysis, data integration, data encapsulation, etc.; intelligent evaluation includes operation evaluation, trajectory evaluation, electronic fence, trajectory playback, face recognition, body recognition, posture recognition, license plate recognition, data tracking, etc.

[0039] The monitoring module classifies and centrally displays the data output by the digital twin module, and manages the main functions in a modular manner. It is mainly divided into: area editing is a secondary editing of the virtual scene and the actual site, which can rebind the corresponding relationship and update this binding relationship; vehicle linkage is based on the real vehicle's GPS positioning information to provide real-time feedback to the visualization, so that the vehicle's driving trajectory is consistent with the real vehicle; when there is a person intrusion in the test room, the system provides an abnormal alarm function, locates the position and picture of the intruder, and provides a 20-second video slice for review and verification.

[0040] Example 2

[0041] like Figure 2As shown, the subject examination room is modeled and restored to the real physical environment using digital twin technology based on the three-dimensional data collected by the data acquisition module, and the examination area and the examination area are divided according to the actual environment. A real environment model is constructed by optimizing the model, combining the site, and simulating the light.

[0042] The QEM (Quadric Error Metrics) algorithm is mainly used to optimize the model, and the specific steps are as follows:

[0043] Initialize the grid data structure, build a three-dimensional grid data structure of vertices, edges, and facets, establish topological connection relationship, and store vertex coordinates, normal vector, color, and other attributes;

[0044] Calculate the vertex Quadric error matrix, calculate the square equation of each facet: ax+by+cz+d=0, and the facet plane equation coefficient is n=[a,b,c] T (normal vector) and d, then the Quadric matrix is represents the error contribution of the plane to the vertex;

[0045] Construct an edge contraction priority queue, and contract the edge (V i V j ) to the new vertex V' with a cost of The position of the optimal V' is calculated by solving ; traverse all edges, calculate the contraction cost (i.e. the minimum quadratic error) of each edge (V i V j ), and sort the priority queue by cost;

[0046] Iterative edge contraction, exclude the edge with the smallest cost in the queue for contraction, generate a new vertex V', update the Quadric matrix of the affected vertices and the contraction cost of the adjacent edges, reinsert the queue, and repeat until the target simplification rate is reached or the contraction cannot continue;

[0047] The Quadric matrix of the new vertex V' is

[0048] Post-processing and rendering update, finally remove isolated vertices and invalid facets, optimize network topology, update graphics rendering pipeline data, and output the simplified three-dimensional model.

[0049] The electronic fence of each item in the marking field is needed, that is, the position and range size of each library of the five items of subject two in the examination area are marked, and the marks of the scheduled starting and ending of the examination are marked. In addition, the positions, orientations, and visible areas of all the cameras in the subject two examination room are marked, and the label data of each camera is bound. Static and dynamic elements in the subject two examination room are marked, wherein the static elements include reverse parking, curve driving, hill starting, side parking, right-angle turning, roads around the examination room, buildings, and camera monitoring equipment in the examination room, etc. The dynamic elements include examination vehicles (including cars and motorcycles), examinees, and other personnel, etc.

[0050] The positions and states of static and dynamic elements are monitored by cameras, vehicle-mounted terminals, GPS, and other modules, and the examinee's face is compared.

[0051] The subject two examination is mainly divided into three stages: the vehicle boarding stage, the starting examination stage, and the examination ending stage. The vehicle boarding stage can be divided into four items: gate identity verification exception, not called number, wrong vehicle, and examination delay. The starting examination stage can be divided into ten items: personnel intrusion, suspected line pressing, marker, test vehicle license plate color abnormality, examination room irregular training, test vehicle irregular entry, person-vehicle inconsistency, AB vehicle (i.e. unrelated personnel using an examination vehicle to take the test), vehicle boundary crossing, vehicle stagnation, and examination overtime. The examination ending stage has one item: examinee irregular vehicle practice.

[0052] In the vehicle boarding stage, the face recognition is used to compare whether it is a called number examinee in the not called number item, and a warning prompt is given. The data analysis technology is used to identify whether it is abnormal in the gate identity verification exception, wrong vehicle, and examination delay, wherein the gate identity verification exception is determined to be abnormal when the number of consecutive identity verification failures detected by the gate in the examination room reaches a set value. The wrong vehicle is determined to be abnormal when the number of identity verification errors at the beginning of the examination exceeds a threshold value. The examination delay is determined to be abnormal when the vehicle has started the examination but has not entered the examination area (vehicle plate recognition and vehicle positioning) within a set time.

[0053] In the beginning of the examination stage, the personnel intrusion, suspected line compression, marker, test vehicle license plate color anomaly, test site irregular training, test vehicle irregular entry, person-vehicle mismatch, AB vehicle and other projects are judged to be abnormal by using AI algorithm. Among them, the personnel intrusion is judged by judging whether there is relevant personnel entering the examination area during the examination time period and there is no equipment failure at present; the suspected line compression is judged by judging whether the tire is compressed in the project judging line during the beginning of the examination to the end of the examination; the marker is judged by comparing the video difference before and after the examination of the designated area of the test site during the opening of the test; the test vehicle license plate color anomaly is judged by identifying the vehicle information (license plate recognition) of the yellow license plate during the test; the test site irregular training is judged by the test vehicle not being assigned for examination but driving in the test project in the test site which has started the examination (data tracking); the person-vehicle mismatch is judged by face snapshot comparison whether the test vehicle and the test personnel match; the AB vehicle is judged by using flow analysis to separate the test vehicle and the in-vehicle monitoring to judge the matching degree of the test personnel. The vehicle boundary crossing, vehicle stasis and test overtime are judged to be abnormal by data analysis technology, among which the vehicle boundary crossing is judged to be abnormal by whether the test vehicle crosses the electronic fence of the test area for more than a set time; the vehicle stasis is judged to be abnormal by analyzing the situation that the test vehicle stays in place in the test area for more than a set time; the test overtime is judged to be abnormal by analyzing the state in the non-waiting area for more than a set time.

[0054] In the end of the examination stage, the testee irregular practice is judged to be abnormal by behavior data analysis. One is that the test vehicle has ended the examination or has been reassigned, and stays in the test area for a long time, which is abnormal; the other is that the test vehicle has ended the examination or has been reassigned, and stays in the test item for a long time, which is abnormal.

[0055] As long as the abnormal situation occurs, the abnormal alarm service is immediately used to visually display the abnormal situation.

[0056] The staff will immediately check the alarm information, and determine whether the abnormal alarm is true by checking the video slice, video playback and real-time video stream. If it is true, a voice command is issued to guide the testee or the illegal intruder to leave the test area.

[0057] The above describes the application creation and its implementation mode schematically, which is not limited, and the application can be realized in other specific forms without departing from the spirit or essential characteristics of the application. The embodiment shown in the drawings is only one of the embodiments of the application, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by it, without departing from the spirit of the application, similar structure and embodiments can be designed without creative design, which shall belong to the protection scope of the application. In addition, the word "comprising" does not exclude other elements or steps, and the word "one" before the element does not exclude including "multiple" elements. The words "first", "second" and the like are used to represent names, and do not represent any specific order.

Claims

1. A method for monitoring the second test site based on digital twins, characterized in that: include: Collect multi-source heterogeneous data; Serializing the collected multi-source heterogeneous data to obtain three-dimensional virtual data, wherein the three-dimensional virtual data represents the three-dimensional spatial structure and physical environment of the examination room; Based on the three-dimensional virtual data, the QEM algorithm is used to construct the subject 2 examination room model, and the electronic fence data, camera position data, static element data and dynamic element data are marked in the constructed subject 2 examination room model; Map the collected multi-source heterogeneous data to the constructed subject 2 examination room model to obtain a digital twin examination room model; Use the digital twin examination room model to monitor the second subject examination.

2. The digital twin-based test center monitoring method for subject 2 according to claim 1 is characterized by: Multi-source heterogeneous data includes: personnel posture data, license plate number data, ID card data, vehicle location data, GPS positioning data, camera location data and three-dimensional coordinate data.

3. The digital twin-based test center monitoring method for subject 2 according to claim 2 is characterized by: Static element data includes test data for reverse parking, curve driving, ramp starting, parallel parking, and right-angle turns, as well as location data for test site roads, buildings, and monitoring equipment. Dynamic element data includes real-time status data of test vehicles, test takers, and other personnel.

4. The digital twin-based test center monitoring method for subject 2 according to claim 3 is characterized by: Serialize the collected multi-source heterogeneous data to obtain 3D virtual data, including: Convert GPS positioning data and camera position data into spatial coordinate data in a predefined format; The spatial coordinate data and the three-dimensional coordinate data are serialized and encapsulated to obtain three-dimensional virtual data representing the three-dimensional spatial structure and physical environment of the examination room.

5. The digital twin-based test center monitoring method for subject 2 according to claim 4 is characterized in that: The QEM algorithm is used to construct the subject 2 examination room model, including: Constructing a three-dimensional mesh according to the three-dimensional virtual data, wherein the three-dimensional mesh includes vertices, edges and facets; Calculate the plane equation coefficient according to the plane equation of each patch; Calculate the Quadric error matrix of each vertex according to the plane equation coefficients. The Quadric error matrix is ​​used to quantify the error constraints of the patch on the vertex position; Traverse all edges in the three-dimensional network, calculate the contraction cost of each edge, sort them from small to large in contraction cost, and build an edge contraction priority queue; Iteratively shrink edges according to the edge shrinkage priority queue, select the edge with the smallest cost in the queue for shrinkage, generate a new vertex, update the Quadric error matrix of the corresponding vertex and the shrinkage cost of the adjacent edges of the vertex, reinsert the updated edge into the priority queue, and repeat the shrinkage until the preset simplification rate is reached; Remove isolated vertices and invalid faces to obtain a simplified three-dimensional model, which will be used as the test room model for Subject 2.

6. The method for monitoring the subject 2 examination room based on digital twin according to claim 5 is characterized in that: Calculate the plane equation coefficients based on the plane equation of each patch, including: Establish the plane equation of each facet in the 3D network: ax+by+cz+d=0, where a, b and c represent plane parameters; Extract the plane equation coefficients of the plane equation, which include the normal vector n = [a, b, c] T and distance parameter d.

7. The digital twin-based test center monitoring method for subject 2 according to claim 6 is characterized by: Calculate the Quadric error matrix for each vertex based on the plane equation coefficients, including: According to the normal vector n = [a, b, c] T And the distance parameter d, construct the four-dimensional vector q = [a, b, c, d] T ; According to the four-dimensional vector q=[a,b,c,d] T , construct the Quadric error matrix Q f :

8. The digital twin-based test center monitoring method for subject 2 according to claim 6 is characterized by: Traverse all edges in the 3D network, calculate the contraction cost of each edge, sort them from smallest to largest contraction cost, and build an edge contraction priority queue, including: For the edge (V i V j ), the cost of shrinking to the new vertex V' is calculated as in, and Vertex V i and V j Quadric error matrix; By solving Calculate the position of the optimal new vertex V'; Traverse all edges in the three-dimensional grid, calculate the contraction cost of each edge, and sort them from small to large according to the contraction cost to form an edge contraction priority queue.

9. The digital twin-based test center monitoring method for subject 2 according to claim 8 is characterized by: Map the collected multi-source heterogeneous data to the constructed Subject 2 examination room model to obtain a digital twin examination room model, including: Perform secondary packaging and statistical analysis on the collected multi-source heterogeneous data through middleware; Bind and map the vehicle location data with the virtual vehicle in the subject 2 test site model; Establish a corresponding relationship between the license plate number data and the virtual vehicle; Map and bind the camera position data to the camera hardware in the subject 2 examination room model to obtain a digital twin examination room model; Among them, middleware includes: message queue, data cache, streaming media conversion, AI algorithm, message push, statistical analysis, DB middleware and streaming analysis.

10. A digital twin-based subject two examination room monitoring system, characterized in that: include: Data acquisition module, collecting multi-source heterogeneous data; The data synchronization module serializes the collected multi-source heterogeneous data to obtain three-dimensional virtual data; The digital twin module uses the QEM algorithm to construct a Subject 2 test site model based on 3D virtual data, and marks the electronic fence data, camera position data, static element data, and dynamic element data in the constructed Subject 2 test site model; The data mapping module maps the collected multi-source heterogeneous data to the subject 2 examination room model to obtain a digital twin examination room model; The monitoring module uses the digital twin examination room model to monitor the second subject examination.

Images