A traffic situation analysis system based on holographic intersections

Abstract

This invention belongs to the field of intelligent transportation technology, specifically relating to a traffic situation analysis system based on holographic intersections. It includes edge devices, a data access module, an intersection modeling unit, a data analysis module, a data storage module, and an anomaly backtracking module. The edge devices collect cross-sectional flow, speed, queue length, vehicle trajectory, and traffic signal control data; the data access module receives this information. The intersection modeling unit constructs a detailed 3D digital model including lanes, stop lines, zebra crossings, and other elements. The data analysis module calculates multiple situation indicators such as intersection overflow degree, traffic imbalance index, and congestion degree based on traffic data and the model. The data storage module stores traffic data, the model, and records of anomalies exceeding thresholds. The anomaly backtracking module can trace back to the state before the anomaly occurred based on the anomaly records, supporting the retrieval of monitoring videos, trajectory data, and the situation results at the corresponding time, assisting in the analysis of the causes of the anomalies.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, specifically a traffic situation analysis system based on holographic intersections. Background Technology

[0002] In today's rapidly developing society, automobiles have become an indispensable part of daily life. Currently, China has 353 million vehicles, with an increasingly sophisticated urban transportation system and a massive market size. Existing traffic management systems typically possess the following functions:

[0003] First, basic modeling of intersection scenarios is performed to support applications such as vehicle trajectory projection, but the model's precision is limited. Second, the data reported by the collection devices is processed superficially, such as calculating vehicle speed, traffic flow, and queue length, which can only achieve basic judgment of traffic conditions (such as congestion judgment during peak hours). Third, although some systems have the ability to issue alarms for abnormal traffic conditions and can push alarm information to the user interface, they lack in-depth mining and further application of alarm data.

[0004] The current technical solutions have the following shortcomings: First, road modeling relies on simple point data and only outlines the lane contours. The modeling of intersection entities is usually limited to motor vehicles, making it difficult for users to fully grasp the real-time status of the intersection. Especially when there is a lack of real-time video monitoring, it is impossible to achieve immersive intersection situational awareness.

[0005] Secondly, the types of data collected are limited, the analysis methods are relatively simple, and the value of the data is not fully explored. This not only wastes computing and data collection resources, but also limits users' accurate judgment of the traffic situation at intersections, making it difficult to form effective decision support.

[0006] In addition, although the system can generate alarm information (including event alarms and situation warnings), it generally lacks in-depth analysis and systematic utilization of alarm information, which restricts its practical role in traffic management optimization. Summary of the Invention

[0007] The purpose of this invention is to provide a traffic situation analysis system based on holographic intersections, in order to solve the problems of simple modeling data, limited types of reported data, and lack of in-depth analysis of alarm information in the prior art mentioned in the background.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0009] A traffic situation analysis system based on holographic intersections includes:

[0010] Edge devices are used to collect traffic data at intersections. The traffic data includes at least cross-sectional flow data, cross-sectional speed data, cross-sectional queue data, vehicle trajectory data, and traffic signal control data.

[0011] The data access module communicates with the edge device and is used to receive traffic data collected by the edge device.

[0012] The intersection modeling unit is used to construct a 3D digital model of the intersection. The 3D digital model includes at least the spatial information of the lanes, stop lines, and zebra crossing areas.

[0013] The data analysis module is connected to the data access module and the intersection modeling unit respectively. It is used to calculate at least one traffic situation indicator based on traffic data and 3D digital model. The traffic situation indicator is selected from intersection overflow degree, traffic flow anomaly index, traffic imbalance index, intersection traffic flow imbalance index, intersection empty space degree, and congestion degree.

[0014] The data storage module, which communicates with the data analysis module, is used to store traffic data, 3D digital model data, and abnormal records generated when traffic situation indicators exceed thresholds.

[0015] The anomaly backtracking module is connected to the data storage module and the intersection modeling unit, respectively, and is used to trigger the state backtracking before the anomaly occurred based on the time information of the anomaly record.

[0016] According to the above technical solution, the edge device includes at least two or more of the following: radar, camera, traffic flow detection integrated machine, event evidence collection integrated machine, and intersection intelligent integrated machine.

[0017] According to the above technical solution, the 3D digital model constructed by the intersection modeling unit also includes spatial information of the intersection center area, waiting area, infrastructure area, median strip area, road area, lane lines, lane arrow indicators, and lane channelization method;

[0018] The spatial information of a lane includes at least the lane number, lane entrance / exit direction, lane width, lane gradient, and lane length.

[0019] The spatial information of lane markings includes at least the lane marking pattern, lane marking width, and lane marking color.

[0020] The spatial information of lane arrow indicators should include at least latitude and longitude, arrow direction, and arrow type; the spatial information of lane channelization methods should include at least channelization type and channelization road locations.

[0021] According to the above technical solution, the steps for the data analysis module to calculate the degree of intersection overflow include:

[0022] Step 1: Obtain the lane length from the 3D digital model of the intersection modeling unit;

[0023] Step 2: Extract the vehicle queue length for each lane from the traffic data;

[0024] Step 3: Calculate the ratio of vehicle queue length to lane length for each lane, and take the maximum value among all ratios as the degree of intersection overflow.

[0025] Step 4: Classify the overflow status according to the maximum value: maximum value ≤ 0.7 is normal status, 0.7 < maximum value ≤ 0.8 is mild overflow status, 0.8 < maximum value ≤ 0.9 is moderate overflow status, and maximum value > 0.9 is severe overflow status.

[0026] According to the above technical solution, the steps for the data analysis module to calculate the traffic flow anomaly index include:

[0027] Select the current time period as the statistical period, and calculate the historical traffic data S1, S2, ..., within the same period of one month from the data storage module. Traffic S during the current period C ;

[0028] The historical average flow rate is calculated as follows:

[0029]

[0030] in, This represents the average historical traffic volume for the same period within the past month, where n is the number of historical traffic data samples.

[0031] The root mean square error of historical flow rates is calculated as follows:

[0032]

[0033] in, is the mean squared error of historical traffic; n is the number of historical traffic data samples;

[0034] Calculate the flow rate S for the current time period. C Then calculate the current average traffic, as shown in the following formula:

[0035]

[0036] in, This represents the current average traffic, and n+1 is the total number of data items.

[0037] The mean square error of the current flow rate is calculated as follows:

[0038]

[0039] in, This represents the mean squared error of the total traffic volume, including the traffic volume for the current time period.

[0040] The traffic flow anomaly index is calculated as follows:

[0041] Traffic flow anomaly index =

[0042] If the index is less than 0, the traffic is normal; if the index is greater than or equal to 0, the traffic is abnormal.

[0043] According to the above technical solution, the steps for the data analysis module to calculate the traffic imbalance index include:

[0044] First, extract the phase, phase sequence, cycle duration, and green ratio of the intersection from the traffic data;

[0045] The saturation of each direction of rotation is calculated as shown in the following formula:

[0046]

[0047] Among them, X i The saturation of directional turning is represented by Qᵢ, which is the directional turning traffic volume, and CAPᵢ is the directional turning traffic capacity.

[0048] The imbalance index of the same-direction import lane is calculated as follows:

[0049]

[0050] in, The same direction import imbalance index, This represents the maximum saturation value for different steering angles at the entrance. This represents the minimum saturation value for different steering directions at the entrance;

[0051] The imbalance index of the opposite inlet is calculated as follows:

[0052]

[0053] in, Let j be the imbalance index of the j-th opposing import lane. Let i be the saturation level from inlet channel i to inlet channel j. The saturation level from inlet channel j to inlet channel i; Indicates taking and The maximum value in the range is used for normalization;

[0054] Data is collected m times at 5-minute intervals to form an m×n saturation matrix, where n is the number of phases. The matrix is ​​then dimensionless, as shown in the following formula:

[0055]

[0056] in, This is the original value of the index in the i-th row and j-th column. For all The minimum value in;

[0057] Calculate the standard deviation ω of each column of the matrix. J (J=1,2,…,n), for ω J Normalization is performed to obtain ω J *;

[0058] The traffic imbalance index is calculated as follows:

[0059]

[0060] in, The traffic imbalance index, Let J be the dimensionless mean saturation value of the J-th column, where J = 1, 2, ..., n. This is the normalized result of the standard deviation of saturation in column J.

[0061] According to the above technical solution, the steps for the data analysis module to calculate the intersection traffic flow imbalance index include:

[0062] Obtain lane type information of the approach lane from the intersection modeling unit to determine whether the approach lane contains both a straight lane and a left-turn lane.

[0063] If included, calculate the straight lane saturation SL. 直 and left-turn lane saturation SL 左 ;

[0064] The intersection traffic flow imbalance index is calculated using the following formula:

[0065]

[0066] Wherein, γ represents the intersection traffic flow imbalance index.

[0067] According to the above technical solution, the steps for the data analysis module to calculate the degree of vacancy at intersections include:

[0068] Spatial information of lanes is obtained from intersection modeling units, and the vehicle passage dataset of lane p within the statistical period is extracted from traffic data.

[0069] Extract the set of vehicle headway times for lane p. When the time distance of the xth train head is greater than a preset threshold, the time distance of that train head is used as the dividing point. Divided into multiple periodic vehicle sets;

[0070] Extract the first vehicle passing time T of the periodic vehicle set in lane p during period j. fqk And the first vehicle passage time T of the next release phase in the corresponding direction of that lane. fpj ;

[0071] The green light duration for lane p is calculated using the following formula:

[0072]

[0073] Where T wpj n represents the duration of the green light being left idle. p d represents the number of vehicles passing through lane p during this period. sat This is the saturation headway.

[0074] The green light utilization rate of lane p is calculated as follows:

[0075]

[0076] Wherein, σ represents the green light utilization rate of the signal cycle. The average value of σ over multiple cycles within the statistical period is taken as the green light utilization rate of lane p. The maximum value of the green light utilization rate of all lanes is taken as the green light utilization rate of the intersection. If the maximum value is greater than the preset threshold, the degree of idle light is low. If the maximum value is less than or equal to the preset threshold, the degree of idle light is high.

[0077] According to the above technical solution, the steps for the data analysis module to calculate the degree of congestion include:

[0078] First, extract the average vehicle speed V of the road segment from the traffic data;

[0079] Divide the set time into equal intervals of ≤15 minutes, calculate the average vehicle speed for each interval, and extract the interval vehicle speed data within a historical period.

[0080] Sort the interval vehicle speed data within the historical time period from largest to smallest, and take the average of the first 1 / 10 of the data as the free-flow speed V. f ;

[0081] Compare V and V f The larger the difference, the higher the congestion level; the smaller the difference, the lower the congestion level.

[0082] According to the above technical solution, the backtracking steps of the anomaly backtracking module include:

[0083] Extract the time information T of the abnormal record from the data storage module, and determine the backtracking start time as T-15 seconds;

[0084] The control streaming media module calls the video playback interface to rewind the real-time video of the intersection to the start time of the rewind and play it.

[0085] Extract the vehicle trajectory dataset from the data storage module back to the start time T, and divide the trajectory data into different frames according to the time order;

[0086] The framed trajectory data is sent to the intersection modeling unit, where the 3D digital model projects the vehicle trajectory frame by frame.

[0087] Traffic data from the start time to T is extracted from the data storage module, and the traffic situation indicators for that period are recalculated by the data analysis module and displayed on the 3D digital model.

[0088] Compared with the prior art, the present invention has the following beneficial effects:

[0089] The system in this invention has stronger and more refined intersection modeling capabilities. It not only performs detailed modeling of infrastructure such as lanes, zebra crossings, and turning areas at intersections, but also covers dynamic elements such as pedestrians and obstacles. In addition, the system in this invention also has stronger data analysis capabilities. Through multiple analysis and calculation tasks in the data analysis module, it uses the traffic data reported by the device to perform a very comprehensive analysis of the traffic situation at the intersection, fully exploring the value of the data.

[0090] To fully demonstrate the data analysis capabilities of the system in this invention and to browse the causes and consequences of traffic anomalies, this invention innovatively establishes a powerful anomaly backtracking mechanism. For anomalies, it can not only present the monitoring video before the anomaly occurred, but also review the trajectory data stream and the traffic situation analysis results at the corresponding time, which helps to analyze the causes of traffic anomalies or some traffic accidents in detail. Attached Figure Description

[0091] Figure 1 This is a schematic diagram of the installation of the edge device of the traffic situation analysis system of the present invention;

[0092] Figure 2 This is a flowchart of the backtracking mechanism of the traffic situation analysis system of the present invention. Detailed Implementation

[0093] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0094] Example 1

[0095] To address the shortcomings of existing technologies, such as coarse intersection modeling, superficial data analysis, and inability to trace anomalies, this invention aims to achieve three objectives: First, to construct a refined model covering all elements of the intersection, including infrastructure, pedestrians, and obstacles, to meet the needs of immersive situational awareness; second, to establish a multi-dimensional traffic situation analysis system to deeply mine the value of traffic data and quantify traffic operation status; and third, to design an anomaly tracing mechanism that integrates "video + trajectory + indicators" to reproduce the anomaly occurrence process and provide a complete basis for traffic decision-making and incident handling.

[0096] The holographic intersection-based traffic situation analysis system of this invention comprises eight core modules: video access module, data access module, streaming media module, data analysis module, basic component module, message middleware module, data storage module, and edge device. The system achieves real-time traffic situation analysis and anomaly backtracking through five key steps, as detailed below:

[0097] Edge devices are the core of data acquisition, including a traffic flow detection unit (TRVB-1), an event evidence collection unit (TRV8-2), an intersection intelligent unit (EC211), traffic light detectors, and high-definition cameras. These devices collect data such as traffic flow at intersection cross-sections, vehicle speed, queue length, vehicle trajectory, traffic light phase, and green light ratio, and push this data to the message middleware module via the network. The message middleware module, deployed based on EMQX, pre-agrees on data push topics with the edge devices. After receiving data reported by the devices, it forwards it to the data analysis module, achieving efficient and stable data flow. The streaming media module, by deploying the zlmediakit and wvp software packages and configuring a unified port, supports protocols such as RTSP / RTMP / GB28181, completing the pulling, transcoding, pushing, and playback of video streams, providing support for real-time video monitoring and playback.

[0098] The data storage module adopts a multi-database collaborative architecture. MySQL is used to store structured data such as user information and permission configurations; TDengine is dedicated to storing time-series data such as vehicle trajectories and traffic changes; Redis serves as a cache for storing frequently accessed data; and the NAS storage system is used to persistently store videos, images, and files, ensuring secure storage and fast retrieval of different data types. The basic component module includes a service monitoring unit, a service registration unit (implemented based on Nacos), a service configuration unit, a task scheduling unit (implemented based on snail-job), and a distributed tracing unit (implemented based on Skywalking), comprehensively ensuring the stable operation, dynamic configuration, and problem localization of all system services.

[0099] The data analysis module is the core processing unit of the system. Based on traffic data collected by edge devices and intersection modeling data, it calculates six major indicators—intersection overflow degree, traffic flow anomaly index, traffic imbalance index, intersection traffic flow imbalance index, intersection vacancy degree, and congestion degree—using specific algorithms to quantify the traffic situation. Simultaneously, this module is also responsible for triggering anomaly alarms and a backtracking mechanism. When an indicator exceeds a preset threshold, it immediately pushes an alarm message and stores relevant records. The video access module and data access module are responsible for receiving video streams and structured traffic data, respectively, achieving seamless integration between edge devices, streaming media modules, and data storage modules.

[0100] This embodiment provides specific implementation steps for a traffic situation analysis system, as follows:

[0101] like Figure 1 As shown, system implementation begins with the installation of equipment at the intersection. Field personnel install the edge devices at designated locations at the intersection (e.g., 3.5m above the lane, ensuring the entire lane is covered). After completing network access for the devices, the data acquisition program is started, and the devices are configured to push the collected traffic data to the message middleware server. After the equipment installation is complete, the message middleware server is set up. A server accessible via the network by the edge devices is selected, and the EMQX software is installed and started on the server. A data push topic (e.g., "traffic / data") is agreed upon with the edge devices to ensure that the data collected by the devices is accurately pushed to this topic.

[0102] Next, a streaming media server is set up. First, zlmediakit is deployed and its configuration file is modified to specify an available port (e.g., 1935). Then, the wvp package is pulled, and its configuration file is modified to ensure that the port is consistent with zlmediakit. The wvp package is then packaged, deployed to the server, and run to implement the video stream processing flow. After the front end requests the playback URL, the system first performs a security check. If the check passes, a SIP signaling request is sent to push the video stream. The edge device pushes the video stream and passes the security authentication. The streaming media module receives the stream, transcodes it, and finally pushes the video stream to the front end for playback. The streaming stops when the page is closed.

[0103] The platform was then built. First, a data persistence module was constructed, and MySQL, TDengine, Redis, and NAS storage systems were deployed and initialized. Next, basic component services were started, including service monitoring, service registration, service configuration, and task scheduling units, to ensure the system's basic support capabilities. Finally, the various sub-services of the software platform were run, covering data access, data analysis, data storage, authentication services, and digital dashboard support services. At the same time, a load balancing cluster was built using Nginx + Keepalived, and Sentinel was used to implement service rate limiting, circuit breaking, and degradation to ensure high availability of the platform.

[0104] Finally, intersection modeling is performed. After the system starts, the scene management function is used to create a new intersection, and 11 types of key information are entered in sequence: the latitude and longitude of the intersection center area, the range of the waiting area, the zebra crossing area (width and length), the infrastructure area (such as the location of traffic lights), the median strip area (width and length), the road area boundary, lane information (number, entrance and exit directions, width, slope, length), lane line information (line type, width, color), lane arrow indicators (latitude and longitude, arrow direction, arrow type), stop line information (width, coordinates), and lane channelization method (channelization type, channelization road location). After the modeling is completed, the traffic data collected by the edge device will be projected onto the 3D digital screen in real time, and the data analysis module will start traffic situation analysis simultaneously.

[0105] The calculation of the overflow level at an intersection requires first obtaining the lane length from the intersection modeling data, then combining it with the queuing data reported by the equipment to obtain the vehicle queue length for each lane, and finally taking the maximum value as the overflow level of the intersection by calculating the ratio of the queue length to the lane length, and classifying the levels according to the range of values.

[0106] The traffic flow anomaly index is calculated with a 5-minute timeframe, extracting historical traffic flow data (denoted as S1 to S2) from the traffic flow data reported by the equipment for the same week and time period within the past month. First, calculate the historical average flow rate. Mean squared deviation of historical flow ; Calculate the historical average flow As shown in the following formula:

[0107]

[0108] in, This represents the average historical traffic volume for the same period within the past month, where n is the number of historical traffic data samples.

[0109] The root mean square error of historical flow rates is calculated as follows:

[0110]

[0111] in, is the mean squared error of historical traffic; n is the number of historical traffic data samples;

[0112] Then obtain the current time period's traffic value S. C Calculate the average flow rate including the current value. Mean squared deviation of current flow Calculate the current average traffic flow as shown in the following formula:

[0113]

[0114] The mean square error of the current flow rate is calculated as follows:

[0115]

[0116] The final traffic anomaly index is the ratio of (σ2-σ1) to σ1. An index < 0 indicates that the current traffic is normal; otherwise, it indicates that the traffic is abnormal.

[0117] The calculation of the traffic imbalance index requires first obtaining the phase, phase sequence, cycle duration, and green ratio of the intersection, and then calculating the saturation X in each direction. i (X) i = Traffic flow in that direction / Traffic capacity in that direction); Same-direction approach imbalance index passed ( )and The ratio is obtained ( This represents the maximum value of the steering saturation at different turns for this inlet. The imbalance index of the opposing inlet (which is the minimum value of the saturation of different turns of the inlet) is obtained by the ratio of the absolute value of the saturation difference of the opposing inlet to the maximum value of the opposing saturation.

[0118] We collect m data points within a 5-minute interval to form a matrix. Since one phase period has n phases, all the data will form a matrix like this:

[0119]

[0120] The matrix is ​​dimensionless, and the processing method is as follows:

[0121]

[0122] in, This is the original value of the index in the i-th row and j-th column. For all The minimum value in;

[0123] Calculate the standard deviation ω of each column of the matrix. J (J=1,2,…,n), for ω J Normalization is performed to obtain ω J * (J = 1, 2, ..., n); Finally, the traffic imbalance index is obtained by summing the products of the normalized standard deviation and the mean of the column. The traffic imbalance index is calculated as follows:

[0124]

[0125] in, The traffic imbalance index, Let J be the dimensionless mean saturation value of the J-th column, where J = 1, 2, ..., n. This is the normalized result of the standard deviation of saturation in column J.

[0126] The calculation of the intersection traffic flow imbalance index is first based on the intersection modeling information to determine whether each approach direction simultaneously contains both straight-through lanes and left-turn lanes. If so, the saturation level SL of the left-turn lane is calculated separately. 左 Saturation of straight lanes (SL) 直 The traffic flow imbalance index for that approach direction is obtained by comparing the absolute value of the difference between the two values ​​with their maximum value. The traffic flow imbalance index at the intersection is calculated as follows:

[0127]

[0128] Wherein, γ represents the intersection traffic flow imbalance index.

[0129] Calculating the vacancy rate at an intersection requires first obtaining road modeling information, statistically analyzing the vehicle passing dataset for lane p within a specified time period, and extracting the headway set for vehicles in lane p within that time period. ; Traversal When the time distance to a certain vehicle head exceeds a preset threshold, this will be used as the dividing point. The vehicle is divided into multiple periodic vehicle sets; the first vehicle passing time T in the periodic vehicle set corresponding to period j of lane p is recorded. fqk And the first vehicle passage time T of the next release phase corresponding to lane p in the direction phase. fpj The green light duration for lane p is calculated using the following formula:

[0130]

[0131] Among them, T wpj n represents the duration of the green light being left idle. p d represents the number of vehicles passing through lane p during this period. sat This is the saturation headway.

[0132] Calculate the green light idle time; then calculate the green light utilization rate for this signal cycle using the following formula:

[0133]

[0134] in, This represents the green light utilization rate for that signal cycle. The average green light utilization rate of multiple signal cycles within the statistical period is taken as the green light utilization rate of lane p. The green light utilization rate of the entire intersection is the maximum value of the green light utilization rate of all lanes. When the green light utilization rate is greater than the preset threshold, it indicates that the intersection has a low degree of idleness and the signal control is reasonable. Conversely, it indicates that the intersection has a high degree of idleness and the signal control needs to be optimized.

[0135] The calculation of intersection congestion begins by using cross-sectional statistical data to obtain the average vehicle speed V of the road segment. The time period from 6:00 to 24:00 each day is divided into equal intervals of no more than 15 minutes, and the average vehicle speed within each interval is calculated. The interval vehicle speed data from the past 30 days is extracted, sorted from largest to smallest, and the average of the first 1 / 10 of the data is taken as the free-flow velocity V. f ; through V and V f The ratio of congestion levels is used to classify congestion levels, which satisfy the following relationship:

[0136] Table 1. Relationship between Congestion Levels

[0137]

[0138] When any traffic situation analysis indicator exceeds the preset threshold, the system will immediately push an alarm message to the user and store the alarm record. The record includes the time of the anomaly, the corresponding indicator value, and the information of the associated equipment.

[0139] like Figure 2 As shown, in the anomaly backtracking module, to achieve perfect trajectory reversal and restoration, precise calculations are required. Before executing the algorithm, several initial parameters need to be determined, including the playback duration T, playback start time Ts, playback redundancy time deviation Tm, and frames per second f (this parameter can be set to the device's reported data frequency). Based on the backtracking duration T, playback start time Ts, and playback redundancy time deviation Tm, the data query time range is determined as (Ts-Tm, Ts+T). After determining the time range for obtaining data, the trajectory dataset (in reverse chronological order) needs to be obtained from the database. Additionally, based on the backtracking duration T and frames per second f, the total number of frames Ftotal and the time span Tf of each frame are calculated. The total number of frames Ftotal is calculated as follows:

[0140] Ftotal=T*f

[0141] The time span Tf of each frame is calculated as follows:

[0142] Tf = (T * 1000) / Ftotal

[0143] Next, a frame list with a length of Ftotal should be initialized. Then, the data is packetized into each frame in chronological order. The specific steps are as follows: traverse the trajectory dataset, obtain the timestamp Ttemp of each data point, and subtract the playback start time Ts from the timestamp Ttemp of each data point. If the difference is greater than or equal to 0, proceed to the normal frame data processing flow. Divide the difference by the time span of each frame to obtain the frame index number into which the data point will be inserted. Retrieve the corresponding frame from the frame list according to this index number (this frame is the one into which the data point will be inserted), insert the data into that frame, and then insert the data into the next frame. If the timestamp of the data is less than the start time, initialize the negative offset frame list. Similarly, calculate the frame index using the difference, determine the specific frame using the index, and then insert the data into that negative offset frame. Continue until there is no data for that frame, then insert that frame into the negative offset frame list. Repeat this process. After the loop ends, flip the negative offset frame list and merge the positive and negative frame lists to form the final frame set dataset. This dataset is given to the front end through an interface. The front end uses a carousel component to play the divided frame dataset according to time. The target trajectory is then re-projected onto the holographic screen, and the screen begins to play the target trajectory, thus realizing trajectory retracing.

[0144] Compared with existing technologies, the advantages of this invention are significant and comprehensive: First, the modeling accuracy is greatly improved, covering not only infrastructure such as lanes, zebra crossings, and turning areas, but also pedestrians and obstacles. Users can immerse themselves in the real-time status of intersections through a 3D digital screen, effectively compensating for the blind spots in situational awareness when there is no real-time video. Second, the data analysis capability is more in-depth, quantifying traffic conditions through six dimensions of indicators, fully exploring the value of data such as cross-sectional flow, vehicle speed, trajectory, and signal control, avoiding resource waste, and providing precise guidance for decisions such as traffic signal optimization and lane planning. Finally, the anomaly backtracking mechanism realizes the multi-dimensional information collaborative reproduction of video, trajectory, and indicators, and can rewind to 15 seconds before the anomaly occurred to fully reproduce the process, helping to quickly locate the cause of the anomaly, providing direct evidence for traffic accident liability determination and signal control scheme adjustment, and has extremely high practical application value.

[0145] Example 2

[0146] This implementation, based on a real-world application scenario at the intersection of a main road and a secondary road in a certain city, details the implementation process of this invention:

[0147] The edge devices selected include the TRVB-1 traffic flow detection all-in-one machine (responsible for collecting lane traffic flow and vehicle speed data), the TRV8-2 event evidence collection all-in-one machine (responsible for collecting intersection emergencies and video data), the EC211 intersection intelligent all-in-one machine (integrating edge computing capabilities and preprocessing some data), a 2-megapixel high-definition camera (supporting the GB28181 protocol and collecting real-time video), and a traffic light detector (collecting traffic light phase and green ratio data). All devices are installed 3.5m above the entrance and exit lanes of the intersection to ensure that the collection view covers the entire lane and the core area of ​​the intersection.

[0148] In terms of server configuration, the message middleware server uses an 8-core 16G cloud server (running CentOS 7.9 system), the streaming media server uses a 16-core 32G cloud server (supporting hardware transcoding to ensure video processing efficiency), and the platform server uses a cluster of 3 8-core 16G cloud servers to achieve high availability and load sharing.

[0149] When setting up the message middleware, install EMQX on the CentOS 7.9 server using the command `yum install emqx`. After starting it, check the default port (1883) using `emqx_ctllisteners`. Log in to the EMQX Dashboard, create a data push topic `traffic / data`, and configure the edge devices to push data to this topic using the MQTT protocol (QoS 1).

[0150] When setting up a streaming media server, first download and compile the zlmediakit source code, and modify rtmp.port=1935 in the config.ini file; then pull the wvp-pro source code, modify media.rtmp.port=1935 in the application.yml file to ensure port consistency, package wvp-pro into a JAR file, and run it in the server background using the command nohup java -jar wvp-pro.jar&; during testing, the front end requests the video playback URL, the WVP service first performs a security check, and after passing the check, sends a SIP signaling to the camera to request streaming. The camera pushes the RTMP stream to zlmediakit, zlmediakit transcodes it and pushes the video stream to the front end to implement the playback function, and triggers a stop streaming command when the page is closed.

[0151] During platform construction, the data persistence module deployed MySQL 8.0 (stores structured data such as users and permissions), TDengine 3.0 (stores time-series data such as vehicle trajectories), Redis 6.0 (caches hot data), and a 10TB NAS storage system (stores videos and images from the past 30 days). The basic component module deployed Nacos 2.2.0 (a 3-node cluster for service registration and configuration), Skywalking 9.2.0 (collects service call data for tracing), and snail-job 1.0 (distributed task scheduling for periodically calculating historical traffic indicators). Sub-services included a data access service (receiving EMQX data and writing it to TDengine), a data analysis service (periodically calculating six key indicators and triggering alarms when thresholds are exceeded), a digital dashboard support service (providing 3D modeling and data projection interfaces), and an authentication service (using JWT for user login verification). Simultaneously, a load-balanced cluster was built using Nginx 1.20.1 + Keepalived 2.0.24, configured with a virtual IP 192.168.1.100 to distribute frontend requests, and Sentinel... 1.8.6 Configure the QPS threshold of the data analysis service to 1000. When the threshold is exceeded, rate limiting will be triggered and a "System busy" message will be returned.

[0152] Taking a city intersection (center latitude and longitude 116.39°, 39.91°) as an example, when modeling, enter the system scene management page to create a new intersection. First, enter the center latitude and longitude of the intersection, then draw the waiting area (area 20㎡, located in front of the left-turn lane), zebra crossing (width 4m, length 20m), and median strip (width 1.5m, length 50m); then enter the lane information: the approach lane has 3 lanes (left turn, straight, right turn, each lane is 3.5m wide, 50m long, and has a slope). 0.5%), the exit lane has 2 lanes (3.5m wide and 40m long); when drawing lane lines, double yellow lines (15cm wide) are used between the left turn lane and the straight lane, and solid white lines (10cm wide) are used between the straight lane and the right turn lane; finally, the stop line (20cm wide, coordinates 116.3901°, 39.9101°) and lane arrow signs (left turn arrow latitude and longitude 116.3902°, 39.9102°, direction northwest) are entered.

[0153] After modeling is completed, the edge device pushes data every 5 seconds, and the data analysis service calculates traffic indicators every minute: During the morning peak hours (7:30-8:30), the queue length of the left-turn lane at the entrance reaches 80m, the lane length is 100m, the overflow degree is 0.8 (slight overflow), the traffic flow anomaly index is -0.2 (normal), and the traffic imbalance index is 0.3 (slight imbalance); when the queue length of the left-turn lane increases to 95m at 8:00, the overflow degree reaches 0.95 (severe overflow), and the system immediately pushes an alarm message to the administrator's mobile phone, storing the time of the anomaly as "05-20 08:00:00".

[0154] After the administrator triggered the rollback function, the system reverted to 07:59:45. The left side of the holographic screen displayed the real-time trend of the overflow level increasing from 0.8 to 0.95, the middle area projected the trajectory of the queue of vehicles in the left-turn lane, and the right side played real-time video from 07:59:45 to 08:00:00. Through multi-dimensional information, it was confirmed that the cause of the anomaly was insufficient left-turn green light duration (originally 30 seconds). Subsequently, the left-turn green light duration was adjusted to 45 seconds. When monitored again, the queue length of the left-turn lane decreased to 40m, and the overflow level returned to 0.4 (normal). The problem was effectively resolved.

[0155] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0156] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A traffic situation analysis system based on holographic intersections, characterized in that: include: Edge devices are used to collect traffic data at intersections. The traffic data includes at least cross-sectional flow data, cross-sectional speed data, cross-sectional queue data, vehicle trajectory data, and traffic signal control data. The data access module communicates with the edge device and is used to receive traffic data collected by the edge device. The intersection modeling unit is used to construct a 3D digital model of the intersection. The 3D digital model includes at least the spatial information of the lanes, stop lines, and zebra crossing areas. The data analysis module is connected to the data access module and the intersection modeling unit respectively. It is used to calculate at least one traffic situation indicator based on traffic data and 3D digital model. The traffic situation indicator is selected from intersection overflow degree, traffic flow anomaly index, traffic imbalance index, intersection traffic flow imbalance index, intersection empty space degree, and congestion degree. The data storage module, which communicates with the data analysis module, is used to store traffic data, 3D digital model data, and abnormal records generated when traffic situation indicators exceed thresholds. The anomaly backtracking module is connected to the data storage module and the intersection modeling unit, respectively, and is used to trigger the state backtracking before the anomaly occurred based on the time information of the anomaly record; The backtracking steps of the exception backtracking module include: Obtain the time point T of the abnormal event recorded in the data storage module, determine the backtracking start time Ts = T - 15 seconds and the playback time length Tlength = 15 seconds; based on the preset playback redundancy time deviation Tm and the number of frames per second f, determine the query time range of the trajectory data as (Ts - Tm, T), and calculate the total number of playback frames Ftotal = Tlength × f and the time span of each frame Tf = (Tlength × 1000) / Ftotal based on Tlength and f; The system extracts vehicle trajectory datasets, intersection video stream data, and traffic operation data within the query time range from the data storage module, and arranges the vehicle trajectory data in reverse chronological order. Based on the calculated total number of frames and the time span of each frame, it initializes a frame list of length Ftotal, iterates through the vehicle trajectory data, calculates the corresponding frame index based on the difference between the timestamps Ttemp and Ts of each data entry, inserts the data into the positive frame list or the negative offset frame list respectively, and finally flips the negative offset frame list and merges it with the positive frame list in chronological order to form a complete trajectory frame sequence. The control streaming media module calls the video playback interface to locate the real-time video of the intersection to the retrospective start time Ts and start playback; at the same time, the trajectory frame sequence is sent to the intersection modeling unit to drive the 3D digital model to dynamically project the vehicle trajectory frame by frame, and the data analysis module recalculates the traffic situation indicators from the abnormal event time point T to the retrospective start time Ts based on the extracted traffic data, and displays them synchronously on the 3D digital model.

2. The traffic situation analysis system based on holographic intersections according to claim 1, characterized in that: Edge devices include at least two or more of the following: radar, camera, traffic flow detection integrated machine, event evidence collection integrated machine, and intersection intelligent integrated machine.

3. The traffic situation analysis system based on holographic intersections according to claim 1, characterized in that: The 3D digital model constructed by the intersection modeling unit also includes spatial information on the intersection center area, waiting area, infrastructure area, median strip area, road area, lane lines, lane arrow indicators, and lane channelization methods; The spatial information of a lane includes at least the lane number, lane entrance / exit direction, lane width, lane gradient, and lane length. The spatial information of lane markings includes at least the lane marking pattern, lane marking width, and lane marking color. The spatial information of lane arrow indicators should include at least latitude and longitude, arrow direction, and arrow type; the spatial information of lane channelization methods should include at least channelization type and channelization road locations.

4. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The steps in the data analysis module to calculate the overflow level at the intersection include: Step 1: Obtain the lane length from the 3D digital model of the intersection modeling unit; Step 2: Extract the vehicle queue length for each lane from the traffic data; Step 3: Calculate the ratio of vehicle queue length to lane length for each lane, and take the maximum value among all ratios as the degree of intersection overflow. Step 4: Classify the overflow status according to the maximum value: maximum value ≤ 0.7 is normal status, 0.7 < maximum value ≤ 0.8 is mild overflow status, 0.8 < maximum value ≤ 0.9 is moderate overflow status, and maximum value > 0.9 is severe overflow status.

5. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The steps in the data analysis module to calculate the traffic flow anomaly index include: Select the current time period as the statistical period, and calculate the historical traffic data S1, S2, ..., within the same period of one month from the data storage module. Traffic S during the current period C ; The historical average flow rate is calculated as follows: in, This represents the average historical traffic volume for the same period within the past month, where n is the number of historical traffic data samples. The root mean square error of historical flow rates is calculated as follows: in, is the mean squared error of historical traffic; n is the number of historical traffic data samples. Calculate the flow rate S for the current time period. C Then calculate the current average traffic, as shown in the following formula: in, This represents the current average traffic, and n+1 is the total number of data items. The mean square error of the current flow rate is calculated as follows: in, This represents the mean squared error of the total traffic volume, including the traffic volume for the current time period. The traffic flow anomaly index is calculated as follows: Traffic flow anomaly index If the index is less than 0, the traffic is normal; if the index is greater than or equal to 0, the traffic is abnormal.

6. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The steps in the data analysis module to calculate the traffic imbalance index include: First, extract the phase, phase sequence, cycle duration, and green ratio of the intersection from the traffic data; The saturation of each direction of rotation is calculated as shown in the following formula: Among them, X i The saturation of directional turning is represented by Qᵢ, which is the directional turning traffic volume, and CAPᵢ is the directional turning traffic capacity. The imbalance index of the same-direction import lane is calculated as follows: in, The same direction import imbalance index, This represents the maximum saturation value for different steering angles at the entrance. This represents the minimum saturation value for different steering directions at the entrance; The imbalance index of the opposite inlet is calculated as follows: in, Let j be the imbalance index of the j-th opposing import lane. Let i be the saturation level from inlet channel i to inlet channel j. The saturation level from inlet channel j to inlet channel i; Indicates taking and The maximum value in the range is used for normalization; Data is collected m times at 5-minute intervals to form an m×n saturation matrix, where n is the number of phases. The matrix is ​​then dimensionless, as shown in the following formula: in, This is the original value of the index in the i-th row and j-th column. For all The minimum value in; Calculate the standard deviation ω of each column of the matrix. J J=1,2,…,n, for ω J Normalization is performed to obtain ω J *; The traffic imbalance index is calculated as follows: in, The traffic imbalance index, Let J be the dimensionless mean saturation value of the J-th column, where J = 1, 2, ..., n. This is the normalized result of the standard deviation of saturation in column J.

7. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The steps in the data analysis module to calculate the intersection traffic flow imbalance index include: Obtain lane type information of the approach lane from the intersection modeling unit to determine whether the approach lane contains both a straight lane and a left-turn lane. If included, calculate the straight lane saturation SL. 直 and left-turn lane saturation SL 左 ; The intersection traffic flow imbalance index is calculated using the following formula: γ=|SL 左 -SL 直 | / max(SL 左 ,SL 直 ) Wherein, γ represents the intersection traffic flow imbalance index.

8. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The data analysis module calculates the degree of vacancy at intersections using the following steps: Spatial information of lanes is obtained from intersection modeling units, and the vehicle passage dataset of lane p within the statistical period is extracted from traffic data. Extract the set of vehicle headway times for lane p. When the time distance of the xth train head is greater than a preset threshold, the time distance of that train head is used as the dividing point. Divided into multiple periodic vehicle sets; Extract the first vehicle passing time T of the periodic vehicle set in lane p during period j. fqk And the first vehicle passage time T of the next release phase in the corresponding direction of that lane. fpj ; The green light duration for lane p is calculated using the following formula: Where T wpj n represents the duration of the green light being left idle. p d represents the number of vehicles passing through lane p during this period. sat This is the saturation headway. The green light utilization rate of lane p is calculated as follows: Where σ represents the green light utilization rate, the average value of σ over multiple cycles within the statistical period is taken as the green light utilization rate of lane p; the maximum value of the green light utilization rate of all lanes is taken as the green light utilization rate of the intersection. The maximum value > the preset threshold indicates a low degree of idle light, and the maximum value ≤ the preset threshold indicates a high degree of idle light.

9. A traffic situation analysis system based on a holographic intersection according to claim 1, characterized in that: The data analysis module calculates congestion levels using the following steps: First, extract the average vehicle speed V of the road segment from the traffic data; Divide the set time into equal intervals of ≤15 minutes, calculate the average vehicle speed for each interval, and extract the interval vehicle speed data within a historical period. Sort the interval vehicle speed data within the historical time period from largest to smallest, and take the average of the first 1 / 10 of the data as the free-flow speed V. f ; Compare V and V f The larger the difference, the higher the congestion level; the smaller the difference, the lower the congestion level.

Images