Urban road traffic vehicle track visual analysis system

Abstract

The invention relates to an urban road traffic vehicle trajectory visual analysis system, which comprises a data processing and storage module, which is used for carrying out comprehensive data cleaning and single trajectory aggregation on the obtained urban road traffic vehicle trajectory, and then clustering group trajectories. The global control module A is used for carrying out time range selection, road selection and track search; the road flow and congestion condition analysis module is used for displaying and analyzing congestion and traffic flow data of each road; the track abnormal behavior analysis module D is used for identifying vehicle track abnormal behaviors; the individual track analysis module is used for displaying and analyzing detailed information of a single track; and the group track analysis module F is used for analyzing the characteristics of the vehicle group track. Through multi-view design, four types of analysis modules, namely 12 visual views, are constructed, a'macroscopic-microscopic 'panoramic analysis chain is formed, interactive screening of global control modules is supported, an analysis process of'what you see is what you get' is realized, and the analysis efficiency is improved. The problems that a traditional analysis tool needs multi-step operation, results are scattered, and it is difficult for a single chart to present multi-dimensional association features of track data are solved.

Description

Technical Field

[0001] The present invention relates to a method for visual analysis of traffic data, and particularly to a visual analysis system for vehicle trajectories in urban road traffic. Background Art

[0002] In recent years, the rapid improvement of the economic level and the advancement of the urbanization process have led to an increasing demand for urban transportation. The urban transportation system has become a core component of the operation of modern cities. However, the rapid growth of the population and the increase in the number of motor vehicles have also brought unprecedented pressure to the urban transportation system. Urban transportation problems are becoming increasingly serious, and traffic problems such as traffic congestion and traffic accidents occur frequently. How to effectively solve traffic problems and ensure the travel safety and efficiency of people has become the top priority in the field of traffic research. A large part of the reasons for traffic problems such as traffic congestion and traffic accidents lies in abnormal driving behaviors of vehicles, such as speeding and reverse driving. These behaviors will affect the normal operation and traffic efficiency of road traffic. Therefore, analyzing and studying vehicle behaviors helps to discover and improve urban traffic problems.

[0003] With the rapid development of global positioning systems, geographic information systems, the Internet of Things, sensor technology, and positioning technology, a large number of data collection devices have been widely used in the field of urban transportation, and a vast amount of traffic data has been collected, including traffic camera data, on-vehicle sensor data, roadside sensor data, mobile phone positioning data, GPS trajectory data, etc. These data contain rich traffic information and provide new possibilities for traffic management.

[0004] Traffic trajectory data contains rich traffic spatio-temporal characteristics and vehicle behavior information, providing a data basis for studying traffic problems. However, characteristics such as large volume, high dimension, and strong spatio-temporal coupling of trajectory data make it difficult to intuitively understand. Traditional analysis methods are difficult to directly extract valuable information from trajectory data. When dealing with high-dimensional and non-linear trajectory data, traditional methods often face problems such as high computational complexity and low detection accuracy. The complexity of traffic data is not only reflected in its huge scale and diverse dimensions, but also in its unique spatio-temporal attributes. Vehicle trajectory data not only contains the position information of vehicles, but also contains multi-dimensional information such as time, speed, and direction, reflecting rich spatio-temporal characteristic behavior patterns. These patterns have important reference values for understanding urban traffic flow, optimizing traffic resource allocation, preventing traffic accidents, etc. However, with the sharp increase in the amount of data, how to efficiently mine vehicle behavior patterns from massive trajectory data and conduct effective analysis has become a hot and difficult point in current research. Summary of the Invention

[0005] In view of the above problems existing in the prior art, the technical problem to be solved by the present invention is: The present invention provides a visual analysis system for urban road traffic vehicle trajectories, which solves the problems of non-intuitive and inaccurate existing analysis.

[0006] To solve the above technical problems, the present invention adopts the following technical solutions: A visual analysis system for urban road traffic vehicle trajectories, comprising:

[0007] A data processing and storage module: acquiring urban road traffic vehicle trajectories and performing comprehensive data cleaning; performing single-trajectory aggregation to merge trajectory points belonging to the same vehicle into a complete trajectory; performing trajectory expansion to add attribute information of acceleration, road, and lane to obtain a high-quality trajectory data set; extracting implicit features in urban road traffic vehicle trajectories through an encoder-decoder model based on a BiLSTM-Transformer hybrid architecture, and then using the k-means clustering algorithm to cluster urban road traffic vehicle trajectories in the feature space to complete the clustering of group trajectories.

[0008] A global control module A: used for time range selection, road selection, and trajectory search;

[0009] Time range selection: setting the time range, with the time divided at 5-minute intervals, for selecting the specific time period to be analyzed.

[0010] Road selection: selecting a specific road, and the traffic flow and average speed trend graph, trajectory abnormal behavior table, and trajectory abnormal behavior statistical graph will be updated with the information of the selected road.

[0011] Trajectory search: through the function of searching for trajectories by vehicle ID, after inputting the vehicle ID for search, the individual trajectory analysis module will display the detailed information of the trajectory, and at the same time, the trajectory map will be updated to the time range where the trajectory is located and the trajectory will be highlighted.

[0012] An analysis module for road traffic flow and congestion conditions, used for analyzing the laws of road traffic flow and congestion conditions, discovering the spatio-temporal distribution of traffic congestion, and the spatio-temporal distribution of road traffic flow and average speed conditions;

[0013] A trajectory abnormal behavior analysis module D, used for identifying vehicle trajectory abnormal behaviors, and vehicle trajectory abnormal behaviors include three categories: reverse driving behavior, speeding behavior, and sudden acceleration and deceleration behavior;

[0014] An individual trajectory analysis module, used for displaying and analyzing the detailed information of a single trajectory;

[0015] A group trajectory analysis module F, used for analyzing the characteristics of vehicle group trajectories.

[0016] Further, the analysis module for road traffic flow and congestion includes four visualization views to display the congestion data and traffic flow data of each road, so as to analyze the road traffic flow and congestion situation. The four visualization views are the road congestion situation distribution map B1, the road traffic flow heat map B2, the road traffic flow and average speed statistical chart B3, and the traffic flow and average speed trend chart B4.

[0017] The road congestion situation distribution map is used to display the spatio-temporal distribution of the road congestion state. The horizontal axis is the time axis, which divides one hour into twelve five-minute time periods to show the evolution of the congestion situation within one hour. The specific time period to be displayed is selected through the global control module A. The vertical axis represents the spatial dimension, and all roads in the area are marked according to two driving directions, showing the congestion distribution on different roads. The size and color of the data points represent the congestion index.

[0018] The road traffic flow heat map uses a table heat map to display the traffic flow of each road per hour.

[0019] The road traffic flow and average speed statistical chart uses a combination of a bar chart and a line chart to comprehensively display the traffic flow and average speed of the road. Among them, the bar chart is used to represent the road traffic flow, and the height of the bar is proportional to the traffic flow. The line chart is used to represent the average traffic speed, and its change reflects the operating speed characteristics of different roads under different traffic flow conditions.

[0020] The traffic flow and average speed trend chart shows the changes in traffic flow and average speed over time. The horizontal axis represents time, and the vertical axis is divided into two parts, which are used to represent traffic flow and average speed respectively. Among them, the change trend of traffic flow is presented through a stacked bar chart. This stacked chart shows the traffic flow of small vehicles, trucks, and buses respectively, and stacks the three to form the total traffic flow. The change trend of the average speed is shown through a line chart, which records the average speed changes of all vehicles, small vehicles, trucks, and buses per minute.

[0021] Further, the trajectory abnormal behavior analysis module D includes three views, and the three views include the trajectory abnormal behavior table, the trajectory abnormal behavior distribution map, and the trajectory abnormal behavior statistical chart.

[0022] The trajectory abnormal behavior table lists the detailed information of all detected vehicle abnormal behaviors in tabular form.

[0023] The trajectory abnormal behavior distribution map uses a scatter plot to display the geographical location distribution of all abnormal behaviors. Each data point represents a trajectory where an abnormal behavior has occurred.

[0024] The trajectory anomaly behavior statistical chart presents the anomaly behavior data by hour in the form of a stacked bar chart. Different types of anomaly behaviors are stacked in the bar chart in the same three different colors as the trajectory anomaly behavior distribution chart, showing the quantity and proportion of various anomaly behaviors in different time periods.

[0025] Further, the individual trajectory analysis module includes three views, and the three views include a trajectory map C, an individual trajectory detailed information map E1, and an individual trajectory characteristic radar chart E2.

[0026] The trajectory map uses road map data and is presented using the Leaflet map rendering technology, showing the structure and layout of the road network in the research area and the historical trajectory data of vehicles. On the trajectory map, vehicle trajectories are presented in the form of lines, showing all trajectories within the selected time period.

[0027] The individual trajectory detailed information map shows the information of a single trajectory. Above the individual trajectory detailed information map, the basic information of the trajectory is shown through four labels, including vehicle ID, vehicle type, trajectory start time, and trajectory end time. Below the map, in the form of a line chart, it details the changes in multiple key attributes such as the speed, acceleration, direction, road, and lane of the trajectory over time.

[0028] The individual trajectory characteristic radar chart comprehensively displays the characteristic values of the trajectory from multiple dimensions. This radar chart covers multiple key characteristic values such as maximum speed, average speed, maximum acceleration, minimum acceleration, number of rapid accelerations and decelerations, number of lane changes, maximum steering change, and duration. Each characteristic value is represented by an independent axis on the radar chart, and the length of the axis represents the value range of the characteristic value. By connecting the performance of the trajectory on each characteristic value in the form of a graph, a closed polygon is formed to present the relative advantages and disadvantages of the trajectory in different characteristic dimensions.

[0029] Further, the group trajectory analysis module F includes two views: a trajectory spatio-temporal feature clustering chart and a trajectory parallel coordinate chart;

[0030] The trajectory spatio-temporal feature clustering chart shows the characteristic clustering scatter plot of all vehicle trajectories within the selected time period.

[0031] The trajectory parallel coordinate chart shows the specific values of the multi-dimensional features of all trajectories within the selected time period. This trajectory parallel coordinate chart includes multiple key characteristic dimensions such as vehicle ID, maximum speed, average speed, maximum acceleration, minimum acceleration, average acceleration, parking ratio, total distance, and clustering category. In the parallel coordinate chart, each polyline represents a vehicle trajectory, and the values of the polyline on different coordinate axes correspond to the performance of the trajectory in each characteristic dimension.

[0032] Further method, the calculation method of the congestion index is as follows:

[0033]

[0034] where v pred is the average speed when the road is unobstructed, and is the average driving speed of vehicles on this road within a period of time.

[0035] Further method, the judgment method of the abnormal behavior of vehicle trajectories is as follows:

[0036] Reverse driving behavior: In terms of space, it satisfies that the deviation degree of the vehicle driving direction Δθ(t) > 90°; in terms of time, the continuous abnormal duration T ≥ 2s; when these two conditions are met simultaneously, it is determined that the vehicle has a reverse driving behavior.

[0037] Speeding behavior: Extract the road speed limit value as the threshold v limit for speeding detection. When the driving speed of the vehicle is greater than the road speed limit value v limit , it can be determined that the vehicle trajectory has a speeding behavior.

[0038] Sudden acceleration and deceleration behavior:

[0039] The acceleration threshold is set as |a| > 3m / s 2 , and when the absolute value of the acceleration exceeds the threshold, it is determined that there is a sudden acceleration and deceleration behavior.

[0040] Compared with the prior art, the present invention has at least the following advantages:

[0041] 1. The present invention first processes the data to obtain a vehicle trajectory data set. Then, vehicle behavior pattern mining is carried out from two aspects: vehicle abnormal behavior detection and trajectory clustering analysis. In order to visually display traffic laws and characteristics, visualization is carried out from both macroscopic and microscopic perspectives. The macroscopic perspective shows the overall situation of urban traffic, such as traffic flow, congested areas, etc.; the microscopic perspective analyzes the behavior characteristics of individual vehicles, such as driving paths, trajectory characteristics, etc. Finally, through integrating data processing, vehicle behavior pattern mining and visualization, a vehicle trajectory visual analysis system that supports the overall perception of traffic situation and the fine analysis of microscopic behavior is developed to help users understand traffic laws, explore vehicle trajectory behaviors and spatio-temporal characteristics, and provide support for traffic management and optimization governance.

[0042] 2. The present invention uses a feature extraction method based on deep learning to replace manual feature engineering, combined with k-means clustering, which can more reasonably handle the problems of high dimensionality and spatio-temporal coupling of trajectory data, fully extract trajectory features, divide the trajectories into groups with similar behavior patterns, and provide reliable trajectory clustering results for subsequent analysis.

[0043] 3. Through multi-view design, the present invention constructs 12 visualization views in 4 types of analysis modules, forming a "macro-micro" panoramic analysis chain, supporting interactive screening of the global control module, realizing an "what you see is what you get" analysis process, and solving the problems of traditional analysis tools that require multi-step operations, scattered results, and the difficulty of presenting the multi-dimensional correlation characteristics of trajectory data with a single chart. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic structural diagram of an urban road traffic vehicle trajectory visual analysis system.

[0045] Figure 2 It is a trajectory map.

[0046] Figure 3 is a distribution map of road congestion, where (a) shows the road congestion situation and (b) shows the floating display of congestion information.

[0047] Figure 4 is the visualization of road traffic flow information, where (a) is a road traffic flow heat map and (b) is a statistical chart of road traffic flow and average speed.

[0048] Figure 5 It is a trend chart of traffic flow and average speed.

[0049] Figure 6 It is a table of trajectory abnormal behaviors.

[0050] Figure 7 It is a distribution map of trajectory abnormal behaviors.

[0051] Figure 8 It is a statistical chart of trajectory abnormal behaviors.

[0052] Figure 9 It is a detailed information map of an individual trajectory.

[0053] Figure 10 It is a radar chart of individual trajectory characteristics.

[0054] Figure 11 It is a clustering chart of trajectory spatio-temporal characteristics.

[0055] Figure 12 It is a parallel coordinate chart of trajectories.

[0056] Figure 13 It is a schematic diagram of the present invention.

[0057] Figure 14 It is the traffic flow trend during the morning rush hour.

[0058] Figure 15 It is the traffic flow situation during the morning rush hour in the road traffic flow heat map.

[0059] Figure 16 It is the road traffic flow and average speed during the 8 am time period.

[0060] Figure 17 It is the distribution of road congestion during the morning rush hour.

[0061] Figure 18 It is the trend chart of traffic flow and average speed of Road 1.

[0062] Figure 19 It is the trend chart of traffic flow and average speed of Road 13.

[0063] Figure 20 is the statistical chart of trajectory abnormal behaviors of Road 13 and all roads, where (a) is the statistical chart of trajectory abnormal behaviors of Road 13, and (b) is the statistical chart of trajectory abnormal behaviors of all roads.

[0064] Figure 21 It is the trajectory map of Road 13.

[0065] Figure 22 Spatial distribution characteristics of trajectory abnormal behaviors.

[0066] Figure 23 Proportion of motor vehicle speeding behaviors.

[0067] Figure 24 shows the selected speeding trajectory information, where (a) are the multi-dimensional characteristics of the selected speeding trajectories, and (b) are the detailed information of the selected speeding trajectories.

[0068] Figure 25 Trajectory map of the selected speeding trajectories.

[0069] Figure 26 Statistical chart of trajectory abnormal behaviors of Road 9.

[0070] Figure 27 Trajectory clustering during the morning rush hour from 8:55 to 9:00.

[0071] Figure 28 shows the parallel coordinate plot of the selected clustering clusters, where (a) selects the clustering cluster cluster3, and (b) selects the clustering cluster cluster2.

[0072] Figure 29 shows the selected cluster2 trajectory information, where (a) are the multi-dimensional characteristics of the selected cluster2 trajectories, and (b) are the detailed information of the selected cluster2 trajectories.

[0073] Figure 30 Trajectory map of the selected cluster2 trajectories. Specific implementation manners

[0074] The present invention will be further described in detail below.

[0075] 1. Trajectory data, obtaining the trajectories of existing urban road traffic vehicles.

[0076] 2. Preprocessing of trajectory data, cleaning and aggregation of trajectory data

[0077] First, according to the type attribute in the trajectory data, all motor vehicle data are filtered out, including small vehicles, passenger cars and truck types. To ensure the validity of the data, all records with type value of -1 and seq value of -1 are excluded. These data usually represent unrecognized abnormal objects.

[0078] A trajectory aggregation algorithm is adopted to aggregate the trajectory data according to the vehicle ID, and a CSV trajectory data file is created for each vehicle. Each file represents the single trajectory data of the vehicle. Through the trajectory aggregation process, a total of 43,889 independent vehicle trajectory data files are generated. Then, short trajectories are filtered. The screening threshold is set to 12, and trajectories with less than 12 trajectory points are filtered out, that is, to ensure that each trajectory exists for at least 6 seconds. After the operation of filtering short trajectories, a total of 35,685 trajectories are obtained. This operation ensures the high quality and validity of the trajectory data, laying a solid foundation for subsequent traffic situation analysis and micro-trajectory behavior pattern mining.

[0079] 3. Expansion of trajectory data

[0080] Combining the road lane data with the trajectory data, the shortest distance algorithm is used to calculate the road section ID where each trajectory data is located, realizing the accurate matching of the trajectory and the section, and expanding the road information into the trajectory data. The acceleration of the road is calculated as supplementary information.

[0081] 4. To scientifically quantify the road operation state, with a 5-minute dynamic analysis window, by fusing the vehicle speed, travel time and flow characteristics of the section, the refined monitoring of road traffic congestion and the visualization expression are realized. The calculation method of the congestion index is as follows:

[0082] where v pred is the average speed when the road is unobstructed, is the average driving speed of vehicles on this road within a period of time. This formula characterizes the congestion index through the relative change rate of speed, effectively eliminating the influence of road section length differences on the calculation results.

[0083] 5. Detection of abnormal behaviors in vehicle trajectories:

[0084] Reverse driving behavior: By calculating the difference between the driving direction of the vehicle and the specified driving direction of the road, and judging according to the set threshold combined with the spatio-temporal double detection strategy, the reverse driving behavior in the vehicle trajectory data can be detected.

[0085] When detecting reverse driving behavior, two conditions of space and time are considered simultaneously: spatially, the deviation degree of the vehicle driving direction Δθ(t) > 90°; temporally, the continuous abnormal duration T ≥ 2s; when both conditions are met simultaneously, it is determined that the vehicle has reverse driving behavior.

[0086] Speeding behavior: Usually detected based on the comparison result of the vehicle driving speed and the road speed limit value. The road speed limit value can generally be extracted from road data. If there is no road speed limit value in the data, a default threshold can be set according to the road grade (such as highway, urban arterial road). Extract the road speed limit value as the threshold v for speeding detection limit When the driving speed of the vehicle is greater than the road speed limit value v limit it can be determined that the vehicle trajectory has a speeding behavior.

[0087] Rapid acceleration and deceleration behavior: After obtaining the acceleration data, it is possible to judge whether the vehicle has rapid acceleration or deceleration behavior based on the value of the acceleration. This judgment process usually depends on a preset acceleration threshold. If the acceleration value exceeds the set positive acceleration threshold, it is determined as rapid acceleration; if it is lower than the set negative acceleration threshold, it is determined as rapid deceleration.

[0088] The acceleration threshold is set to |a| > 3m / s 2 When the absolute value of the acceleration exceeds the threshold, it is determined that there is rapid acceleration and deceleration behavior. Specifically, the judgment threshold for rapid acceleration is 3m / s 2 and the judgment threshold for rapid deceleration is -3m / s 2 .

[0089] 6. Extract the implicit features in the urban road traffic vehicle trajectory through an encoder-decoder model based on the BiLSTM-Transformer hybrid architecture, and then use the k-means clustering algorithm to cluster the urban road traffic vehicle trajectory in the feature space to complete the clustering of the group trajectory.

[0090] 7. Vehicle trajectory visual analysis system

[0091] The design goal of the urban road traffic vehicle trajectory visual analysis system is to support the full-process analysis needs of users from macroscopic traffic situation perception to microscopic vehicle behavior analysis through multi-dimensional and multi-granularity visualization and visual analysis methods:

[0092] G1: Macroscopic traffic situation perception: Realize the visual analysis of traffic flow and congestion situation

[0093] Support the display of the traffic flow distribution and congestion change rules of the road from multi-dimensional perspectives such as time and space, comprehensively master the macroscopic traffic operation situation, and identify congested time periods and sections.

[0094] G2: Micro-individual Behavior Analysis: Provide the display and analysis functions of individual vehicle trajectories

[0095] Through visualization means such as trajectory maps, trajectory attribute-time curves, and multi-dimensional feature radar charts, it supports the detailed analysis of individual trajectories and realizes the micro-perspective analysis of individual vehicle trajectories.

[0096] G3: Abnormal Behavior Detection and Analysis: Analyze trajectory abnormal behaviors

[0097] Combined with abnormal behavior detection algorithms and interactive visualization technologies, it provides exploratory analysis of abnormal behavior trajectories and identifies and studies abnormal behavior patterns in traffic.

[0098] G4: Group Behavior Analysis: Reveal the spatio-temporal laws and pattern characteristics of vehicle group behaviors

[0099] Through methods such as trajectory clustering, it realizes the group analysis of trajectory behaviors, mines the common behaviors of vehicle groups on specific road sections or during specific time periods, analyzes their spatio-temporal distribution characteristics, and reveals the group behavior characteristics and laws in traffic flow.

[0100] G5: Interactive Exploration and Linkage Analysis: Integrate multi-level interaction and collaborative analysis capabilities

[0101] Through the multi-view collaboration and linkage mechanism, it supports the exploration needs of traffic trajectory data from macro to micro. From the macro traffic road flow and congestion analysis to the multi-level visual analysis views of micro-individual trajectories, it supports the linkage filtering and collaborative reasoning between views and completes the complete analysis link from overall situation perception to specific problem tracing.

[0102] The urban road traffic vehicle trajectory visual analysis system designed by the present invention. The system adopts a modular design and supports the full-process requirements from macro traffic situation perception to micro-individual behavior analysis through multi-view linkage and interactive exploration. The overview diagram of the system is as Figure 1 shown, and it includes a total of five functional modules and more than a dozen interactive views.

[0103] The global control module A provides global control functions, including time range selection, road selection, and trajectory search. The analysis module of road flow and congestion includes four views: road congestion distribution map B1, road flow heat map B2, road flow and average speed statistics map B3, and traffic flow and average speed trend map B4. It focuses on the analysis of road flow and congestion, and is an important window for grasping the macro situation of traffic, providing an overall perception of traffic congestion, flow and average speed. The trajectory abnormal behavior analysis module D includes three views: trajectory abnormal behavior table, trajectory abnormal behavior distribution map, and trajectory abnormal behavior statistics map. It is used for trajectory abnormal behavior analysis and provides a display of trajectory abnormal behavior and analysis of spatiotemporal distribution characteristics. The individual trajectory analysis module includes three views: trajectory map C, individual trajectory detailed information map E1, and individual trajectory feature radar map E2. It is dedicated to individual trajectory analysis, deeply explores the characteristics of a single trajectory from a micro level, and can provide detailed feature information of a single trajectory. The group trajectory analysis module F includes two views: trajectory spatiotemporal feature clustering map and trajectory parallel coordinate map. By clustering the trajectory, the group characteristics of the trajectory are mined.

[0104] 8. Track map

[0105] In the urban road traffic vehicle trajectory visual analysis system, the intuitive display of trajectories is the basic and key part, providing the most direct and easy-to-understand way to recognize vehicle trajectories. The trajectory map transforms the originally abstract and complex trajectory data into intuitive and easy-to-understand visual elements through graphical expression. This transformation lowers the threshold for data interpretation and improves the readability and comprehensibility of the data. In order to display vehicle trajectory information in detail and intuitively, the trajectory map is Figure 2 shown.

[0106] The trajectory map is based on road map data and is presented in detail using Leaflet map rendering technology. It can show the structure and layout of the road network in the study area, as well as the historical trajectory data of vehicles. On the trajectory map, vehicle trajectories are presented in the form of lines, showing all trajectories within the selected five-minute time period. Users can view traffic conditions in a specific area by zooming, panning, and other operations. After selecting a single trajectory, the trajectory will be highlighted in red on the map. Clicking the play button can restore the motion animation of the trajectory, allowing users to intuitively feel the vehicle's driving trajectory and speed changes. At the same time, the map will also display detailed information diagrams of individual trajectories, including key feature information such as vehicle ID and speed curve, to facilitate users to deeply analyze micro-traffic behavior.

[0107] 9. Road congestion distribution map

[0108] The road congestion distribution map is used to show the spatiotemporal distribution of road congestion status, as shown in Figure 3 (a). This view is based on the optimized scatter bubble chart. The horizontal axis is set as the time axis, and one hour is divided into twelve five-minute time periods, thus showing the evolution of congestion conditions within this hour. The vertical axis represents the spatial dimension, and all roads in the area are marked according to two driving directions, showing the distribution of congestion on different roads. The size and color of the data points in the figure reflect the congestion status, which is specifically reflected by the congestion index, a quantitative indicator. According to the high and low congestion index, the road congestion status is divided into four levels, and progressive color coding is used to realize the visualization of road congestion status. The corresponding relationship between traffic congestion degree, congestion index and color mapping used for visualization is shown in Table 1. When the mouse hovers over a data point, the detailed congestion information of the point will be displayed, including congestion index, road, congestion degree, and time, as shown in Figure 3 (b). When the user clicks on any data point in the figure, other views will react in a linked manner and automatically adjust to the selected time period and road, providing support for the user's subsequent detailed analysis.

[0109] Table 1 Road congestion index and color mapping

[0110]

[0111] 10. Visualization of road traffic information

[0112] Two visualization charts of different dimensions, road flow heat map and road flow and average speed statistics chart, were constructed to provide an effective situation overview for analyzing road traffic flow.

[0113] The road traffic heat map adopts the form of a tabular heat map to realize the spatiotemporal visualization mapping of the traffic flow conditions of each road every hour, as shown in (a) in Figure 4. In this heat map, the color depth is used as an intuitive representation of the size of the traffic flow, and a mapping relationship between traffic flow and color depth is constructed. Specifically, the larger the traffic volume, the darker the color of the corresponding road cell. This color gradient change effectively reflects the degree of difference in traffic flow. At the same time, the heat map further enhances the readability and accuracy of the data by marking the specific traffic volume values, so that the observer can accurately obtain the traffic information of each road at different times. From the perspective of spatiotemporal analysis, the road traffic heat map can clearly show the dynamic change characteristics of traffic flow. For example, by comparing the heat maps of different time periods, the changing patterns of traffic flow can be discovered; observing the heat map performance of different roads in the same time period helps to identify traffic bottleneck roads.

[0114] The statistical chart of road traffic flow and average speed combines a bar chart and a line chart to comprehensively display the traffic flow and average speed of roads, aiming to reveal the internal relationship between traffic flow and operation efficiency, as shown in (b) of Figure 4. Among them, the bar chart is used to represent the road traffic flow, and the height of the bar is proportional to the traffic flow size, visually showing the traffic load degree of each road. The line chart is used to represent the average passing speed of the road, and its change reflects the operation speed characteristics of different roads under different traffic flow conditions. This statistical chart sorts the roads according to the traffic flow size, which is convenient for users to directly locate the roads with the largest and smallest traffic flows, and also helps to explore the relationship between traffic flow size and speed. This view shows the road traffic flow and average speed within one hour selected by the user. Clicking on different hours of the heat map can switch to the statistical chart corresponding to that hour, forming a supplement to the traffic flow heat map.

[0115] 11. Statistical Chart of Traffic Flow and Average Speed Trend

[0116] The traffic flow and average speed trend (as Figure 5 ) shows the trends of traffic flow and average speed changing over time within the research area. By comprehensively analyzing traffic flow and speed, an overall overview of the traffic situation can be obtained, discovering the fluctuation rules of traffic flow, the characteristics of peak hours, and the changing trends of speed.

[0117] Figure 5 The horizontal axis in it represents time, with a 1-minute time interval, covering the continuous time period from 7:00 am to 4:00 pm. Users can freely adjust and scale the time range of concern as needed by dragging the range selector below the horizontal axis to deeply analyze the traffic conditions in a specific time period. The vertical axis is divided into two parts, respectively used to represent traffic flow and average speed. Among them, the changing trend of traffic flow is presented by a stacked bar chart. This stacked chart shows the traffic flow situations of small vehicles, trucks, and passenger cars respectively, and stacks the three to form the total traffic flow. This helps users understand the contributions of different vehicle types to traffic flow, as well as the composition and distribution characteristics of traffic flow. The changing trend of average speed is shown by a line chart, which respectively records the average speed changes of all vehicles, small vehicles, trucks, and passenger cars per minute, enabling users to intuitively feel the speed fluctuations and at the same time analyze the relationship between speed changes and traffic flow.

[0118] This view can show the overall traffic flow and average speed situations of all roads, and can also show the situations of each road separately. When the mouse hovers over the chart, the detailed flow and speed situations of specific types of vehicles at the selected minute will be displayed. In addition, by clicking on the legend, the display of this type of data can be cancelled, so as to only analyze the traffic flow or speed of the vehicle types that need to be concerned.

[0119] 12. Trajectory abnormal behavior analysis module

[0120] The trajectory abnormal behavior analysis module integrates the trajectory abnormal behavior table, trajectory abnormal behavior distribution diagram and trajectory abnormal behavior statistics diagram to display and analyze vehicle abnormal behavior.

[0121] The trajectory abnormal behavior table lists the details of all detected vehicle abnormal behaviors in a table, such as Figure 6 As shown. The table contains multiple key fields: the serial number is the unique identifier of each data; the vehicle ID identifies the trajectory where the abnormal behavior occurs; the abnormal time records the specific time when the abnormal behavior occurs; the road to which it belongs clarifies the specific road where the abnormal behavior occurs; the abnormal event describes the specific category of abnormal behavior, which is divided into wrong-way driving, speeding, sudden acceleration, and sudden deceleration. When the user clicks on a single piece of information in the table, the trajectory map will switch to the time period when the abnormal behavior occurred and highlight the movement route of the selected trajectory. At the same time, the individual trajectory analysis module will display the detailed information of the selected trajectory.

[0122] Trajectory abnormal behavior distribution diagram (such as Figure 7 ) uses a scatter plot to show the geographical distribution of all abnormal behaviors. Each data point represents a trajectory where an abnormal behavior occurred. In order to distinguish different types of abnormal behaviors, the figure uses three different colors to represent the three types of abnormal behaviors: driving against the flow, speeding, and sudden acceleration and deceleration. Through this color coding, users can quickly identify the main types of abnormal behaviors in different road areas and understand the spatial distribution of abnormal behaviors. Clicking a single data point can link other views to view detailed information on the trajectory.

[0123] Trajectory abnormal behavior statistics (such as Figure 8 ) presents hourly abnormal behavior data in the form of a stacked bar chart. Each bar represents a one-hour time period, and the height of the bar represents the total number of abnormal behaviors in that hour. Different types of abnormal behaviors are stacked in the bar chart in the same three different colors as the trajectory abnormal behavior distribution chart, intuitively showing the number and proportion of various types of abnormal behaviors in different time periods. On the right side of the bar chart, a semi-circular pie chart shows the total amount of abnormal behaviors in all time periods, and uses colors to distinguish different abnormal behaviors, showing their number and proportion. After clicking on different abnormal behavior types or hours in the statistical chart, the trajectory abnormal behavior table will be updated accordingly, showing all abnormal behavior information for the selected hour and the selected abnormal behavior type. The trajectory abnormal behavior distribution chart will respond according to time and show the distribution of all abnormal behaviors in the selected hour, so that users can focus on abnormal behaviors in a specific time period and analyze their spatiotemporal distribution characteristics.

[0124] 13. Individual trajectory analysis module

[0125] The individual trajectory analysis module is used to analyze the detailed information of a single trajectory. It mainly consists of two parts: the individual trajectory detailed information graph and the individual trajectory feature radar graph, providing users with a comprehensive and detailed perspective on trajectory analysis, and helping users gain insights into the characteristics and laws behind the trajectory from multiple dimensions.

[0126] The individual trajectory detailed information graph (such as Figure 9 ) presents the rich information of a single trajectory in an intuitive and comprehensive way. At the top of the graph, the basic information of the trajectory is displayed through four labels, including vehicle ID, vehicle type, trajectory start time, and trajectory end time. This information facilitates users to quickly understand the basic attributes of the trajectory. At the bottom of the graph, a line chart is used to detail the changes in multiple key attributes such as speed, acceleration, direction, road, and lane over time. The horizontal axis is finely divided at intervals of 0.5 seconds, highly restoring the dynamic information during the actual movement of the trajectory, enabling precise grasp of the state changes of the trajectory at each moment. The vertical axis corresponds to different attributes. Hovering the mouse over the line can view the specific values of each attribute at that moment. The individual trajectory detailed information graph enhances users' understanding of the multi-dimensional characteristics of the trajectory over time. For example, by observing the line of speed change over time, users can analyze the driving speed characteristics of the vehicle in different sections and determine whether there are abnormal situations such as speeding or low-speed driving; the line of acceleration change helps to detect the vehicle's sudden acceleration and deceleration behaviors, further evaluating the smoothness and safety of driving.

[0127] The individual trajectory feature radar graph (such as Figure 10 ) comprehensively displays the characteristic values of the trajectory from multiple dimensions. This radar graph covers multiple key characteristic values such as maximum speed, average speed, maximum acceleration, minimum acceleration, number of sudden acceleration and deceleration times, number of lane changes, maximum steering change, and duration. These characteristic values comprehensively cover the performance of the trajectory in multiple important aspects such as speed, acceleration, steering, and driving duration, providing a rich data basis for comprehensively evaluating the trajectory. Each characteristic value is represented by an independent axis on the radar graph, and the length of the axis represents the value range of the characteristic value. By connecting the performances of the trajectory on each characteristic value in the form of a graph to form a closed polygon, it intuitively presents the relative advantages and disadvantages of the trajectory in different characteristic dimensions. For example, if the radar graph extends longer on the maximum speed and average speed axes, it indicates that the vehicle corresponding to this trajectory has a higher driving speed; if it is longer on the number of sudden acceleration and deceleration times axis, it shows that there are more sudden acceleration and deceleration behaviors during the vehicle's driving, and the driving style is relatively aggressive, possibly posing a safety risk.

[0128] 14. Group Trajectory Analysis Module

[0129] The group trajectory analysis module integrates the trajectory spatio-temporal feature clustering map and the trajectory parallel coordinate map, and can comprehensively and deeply analyze the trajectory behavior patterns and characteristics of group vehicles from both macroscopic and microscopic levels.

[0130] The trajectory spatio-temporal feature clustering map (such as Figure 11 ) focuses on showing the feature clustering scatter plot of all vehicle trajectories within the selected five-minute time period. Its core purpose is to measure the similarity of trajectory behavior patterns. During the trajectory clustering process, an encoder-decoder model with a BiLSTM-Transformer hybrid architecture is used for feature extraction, extracting the most representative and discriminative features from the trajectory data. These features cover multiple dimensions such as the speed, acceleration, and direction change of vehicle travel, comprehensively reflecting the motion state of the vehicle at different times and in different spaces. After feature extraction, a clustering algorithm is used to cluster the trajectories. The clustering results are visually presented in the form of a scatter plot. Each scatter point represents a vehicle trajectory, and the distribution position of the scatter points in the graph reflects the similarity degree between the trajectories. Trajectories with higher similarity will be clustered together to form different clustering clusters. By observing the distribution of the clustering clusters, different types of trajectory behavior patterns can be identified.

[0131] The trajectory parallel coordinate map (such as Figure 12 ) shows the specific values of the multi-dimensional features of all trajectories within the selected five-minute time period. This graph includes multiple key feature dimensions such as vehicle ID, maximum speed, average speed, maximum acceleration, minimum acceleration, average acceleration, parking ratio, total distance, and clustering. In the parallel coordinate map, each polyline represents a vehicle trajectory, and the values taken by the polyline on different coordinate axes correspond to the performance of the trajectory in each feature dimension. By observing the trend and distribution of the polyline. For example, on the maximum speed and average speed coordinate axes, trajectories with higher polyline positions indicate faster vehicle travel speeds; while on the parking ratio coordinate axis, trajectories with higher polyline positions indicate more parking times or longer parking durations for the vehicle.

[0132] Embodiment 1: An urban road traffic vehicle trajectory visual analysis system, including:

[0133] The dataset used is sourced from the Tianyi Transportation Intelligent Transportation Holographic Road Dataset from 7:00 to 16:00 on April 13, 2023, provided by the ChinaVis2023 Data Visualization Competition. This dataset detects and records the driving trajectory information of traffic participants through roadside sensors and consists of two parts: the road trajectory dataset and the supporting map dataset. The road trajectory dataset covers the trajectory data of all traffic participants on the roads within the collected area. The collection time interval for each piece of data is 0.5 seconds, with a time range of 9 hours from 7:00 to 16:00, and it is stored in JSON file format. Each trajectory data contains detailed information such as ID (id), position (position), shape (shape), motion status (is_moving), motion speed (velocity), motion direction (heading), type (type), timestamp (time_meas), etc. The original trajectory dataset includes various traffic participants such as small vehicles, passenger cars, trucks, pedestrians, and non-motor vehicles, with a total of more than 17 million pieces of data. The supporting map dataset provides static road feature information such as road boundaries, road lanes, crosswalks, traffic lights, etc. in the collected research area. Among them, the road lane data includes specific attributes such as lane line ID (fid), road section ID (road_sec_id), position (geometry), speed limit (limit_speed), lane type (category), etc., providing detailed road information for the analysis of trajectory data. The map area covers a continuous road with 8 intersections and a length of approximately 10 kilometers.

[0134] 1. Identification and Analysis of Traffic Peak Periods

[0135] In actual traffic scenarios, the identification of traffic peak periods is crucial for traffic management and optimization. Through macroscopic analysis of traffic trajectory data, it is possible to identify a significant increase in traffic flow during specific time periods, thereby determining the specific time slots of the peak periods.

[0136] View the traffic flow data in the traffic flow and average speed trend chart and observe that the traffic flow from 8:00 am to 9:30 am is continuously higher than other time periods, showing obvious peak characteristics, as Figure 14 shown. Combining with the road traffic heat map, it is found that (as Figure 15 ) the road traffic flow at 8:00 and 9:00 is significantly higher than other time periods, and the road traffic flow at 8:00 is the highest. Combining with the calendar information, April 13, 2023 is Thursday, a working day, and it is inferred that this time period is the morning rush hour for commuting.

[0137] To gain a deeper understanding of the traffic conditions during the morning rush hour, a detailed analysis of the road traffic flow and average speed within the 8:00 time period was continued (as Figure 16) The results show that the traffic flows of Road 14, Road 1, and Road 4 rank among the top three, indicating that these three road segments bear relatively high traffic pressure during the morning rush hour. Further analysis of the average road speed reveals that as the road traffic flow decreases, the average road speed generally shows an upward trend, which is consistent with the operating characteristics of traffic flow. However, there are also some exceptions. The average speeds of Road 1, Road 5, Road 15, and Road 6 show abnormal decreases. Analyzing the regional map, it is found that these road segments are all located at the edge of the region. The relatively low average speed may be due to insufficient recorded data and the influence of traffic lights at intersections, resulting in a relatively low average speed of vehicles on these roads. Conducting a specific analysis of the roads with the top three traffic flows, the average speeds of Road 1 and Road 4 are both lower than that of Road 14, which ranks first in terms of traffic flow. This indicates that there may be traffic congestion or bottleneck problems on Road 1 and Road 4 during the morning rush hour, leading to a decrease in vehicle driving speed and further exacerbating the traffic pressure.

[0138] From the overall view of the road traffic flow heat map (as Figure 15 ), the traffic flows of Road 14, Road 4, Road 1, and Road 13 are relatively high throughout the day, indicating that these road segments are high-frequency roads worthy of attention and have been bearing relatively high traffic pressure for a long time.

[0139] 2. Road Congestion Identification and Spatiotemporal Correlation Analysis

[0140] After selecting 8 o'clock, the distribution of road congestion during the morning rush hour can be observed, as Figure 17 shown. It can be seen from the figure that during the 8 o'clock period, congestion occurred in multiple road segments. Among them, the congestion on the right side of Road 1 and the downward direction of Road 13 are the most significant and last for a long time.

[0141] For Road 1, this road has a large traffic flow and a low passing speed during the morning rush hour, and it is a traffic bottleneck road. Further analyzing the traffic flow and average speed trend chart of Road 1, it is found that the traffic volume of Road 1 from 8 am to 9 am is significantly higher than other time periods, and the average speed is relatively low, as Figure 18 shown. It can be inferred that traffic congestion occurred on the right side of Road 1 during the morning rush hour due to the large traffic volume.

[0142] For Road 13, observing the statistical chart of traffic flow and average speed ( Figure 16 ), it is found that its traffic volume ranks fourth during the morning rush hour. Further analyzing the traffic flow and average speed trend chart, the traffic volume of Road 13 during the morning rush hour is higher than other time periods, but its average passing speed is not only relatively low during the morning rush hour but also shows a relatively low level overall, as Figure 19As shown. This indicates that the congestion on Road 13 may not be solely caused by high traffic flow. To delve deeper into the causes of congestion, analyze the statistical chart of abnormal trajectory behavior (as shown in Figure 20), and it is found that there are relatively many reverse driving behaviors on Road 13, reaching a total of 59 times, accounting for 59% of all reverse driving behaviors on the roads. Through the trajectory map (such as Figure 21 ), it can be found that Road 13 is significantly narrower than other roads, with a single lane in both directions left and right, and reverse driving behaviors are mostly concentrated here. This shows that due to its narrowness, this section of the road can no longer meet the traffic demand, and the behavior of occupying the oncoming lane occurs frequently. Therefore, it can be inferred that the traffic congestion on Road 13 is likely related to its road conditions, that is, the narrow road is the main reason for the congestion on Road 13.

[0143] 3. Identification of Abnormal Trajectory Behavior and Analysis of Spatiotemporal Characteristics

[0144] By observing the distribution map of abnormal trajectory behavior (such as Figure 22 ), the spatial distribution characteristics of abnormal trajectory behavior can be analyzed. Specifically, two types of abnormal behaviors, namely motor vehicle speeding and sudden acceleration and deceleration, mainly occur concentratedly on the two long roads up and down. This may be related to the large traffic flow and high vehicle speed on these two long roads. Reverse driving behaviors mostly occur on Road 12 and Road 13, which may be related to factors such as road design, traffic flow conditions, or driving habits.

[0145] Further observing the statistical chart of abnormal trajectory behavior ( Figure 23 ), the time distribution law and composition types of abnormal trajectory behavior can be revealed. The statistical results show that the speeding behavior accounts for the highest proportion in the total abnormal behaviors, reaching 72%. During the four time periods of 7 am, 8 am, 9 am, and 11 am, the occurrence frequency of abnormal behaviors is significantly higher than other time periods. Among them, the speeding and sudden acceleration and deceleration behaviors are particularly prominent. Combining with the road flow and traffic flow analysis views, it can be found that the traffic flow during the time periods of 8 am, 9 am, and 11 am is relatively large. These time periods are mostly working hours, and drivers may be more likely to have speeding and sudden acceleration and deceleration behaviors due to being in a hurry. This may be one of the reasons for the increase in abnormal behaviors during these time periods. However, it is worth noting that although the traffic flow during the 7 am time period is not significantly more than other time periods, the amount of abnormal behaviors ranks first, and it is mostly dominated by speeding and sudden acceleration and deceleration. This may be because the traffic flow on the road is less during the 7 am time period, the vehicle driving is relatively smooth, and some drivers may therefore relax their vigilance, resulting in an increase in speeding and sudden acceleration and deceleration behaviors.

[0146] To explore the specific characteristics of abnormal behavior in more depth, a trajectory with a speeding behavior occurring at 7 o'clock is selected for detailed analysis, as shown in Figure 24. The vehicle ID of this trajectory is 196561123, the trajectory duration is 47 seconds, the average speed reaches 16.78 m / s, and the maximum speed even reaches 18.272 m / s. The driving speed exceeds the speed limit of 60 km / h (i.e., 16.67 m / s). As shown in Figure 24(b), from the detailed information map of the individual trajectory, it can be seen that during the entire driving period, this trajectory has been driving at a relatively high speed, the driving state is very stable, and the acceleration change is not significant. Combining with the trajectory map, it can be seen that ( Figure 25 ) during the whole process of this trajectory driving from Road 6 to Road 10, it is a straight driving without obvious turning or lane-changing behavior. Observing the road traffic flow heat map, it can be found that during this time period, the road traffic flow of all roads passed by this trajectory is very small, which indicates that the driving environment of the vehicle is relatively loose during this time period, and the driver may be more likely to generate speeding behavior. Through the trajectory abnormal behavior statistical chart, it can be found that ( Figure 26 ) there are a large number of speeding behaviors on Road 9 at 7 o'clock in the morning, which further confirms the analysis of the reasons for the vehicle's speeding.

[0147] 4. Trajectory Behavior Pattern Recognition and Analysis

[0148] The trajectory spatio-temporal feature clustering map can distinguish different trajectory categories, forming an intuitive understanding of the trajectory. Clicking on the trajectory data point can display the detailed information of the trajectory, which is convenient for operating and exploring the trajectory. Taking the time period from 8:55 to 9:00 during the morning rush hour as an example, it can be observed from the trajectory clustering map that ( Figure 27 ) most of the trajectories are highly aggregated into two categories, and only a small number of trajectories are assigned to cluster2 and cluster3, indicating that these trajectories assigned to cluster2 and cluster3 are very likely to have abnormal characteristics and are worthy of further in-depth study. In the trajectory parallel coordinate map, by brushing and selecting the clustering clusters cluster3 and cluster2, it can be found that the trajectories in these two clustering clusters have some remarkable characteristics, as shown in Figure 28. The trajectories in cluster3 are characterized by relatively fast speeds. Their maximum speeds all exceed 10 m / s, and the absolute values of the maximum acceleration and the minimum acceleration are relatively large, indicating that these trajectories have experienced relatively drastic speed changes during driving. At the same time, the total driving distances of these trajectories are also relatively long. The trajectories in cluster2 are similar to those in cluster3, also having relatively high speeds, but their driving distances are longer than those of the trajectories in cluster3, and the parking ratio is relatively low. These characteristics reveal the driving rules and driving behavior characteristics of the trajectories in these two clustering clusters.

[0149] To further explore these trajectory behavior patterns, a trajectory in cluster2 was selected for in-depth analysis, as shown in Figure 29. The trajectory ID is 211755516, the trajectory duration is up to 433 seconds, the maximum speed reaches 13.788 m / s, the absolute values of both the maximum acceleration and the minimum acceleration exceed 3, the number of lane changes is 45, and the maximum steering change reaches 6. As shown in (b) of Figure 29, it can be seen from the detailed trajectory information diagram that the speed and acceleration of this trajectory fluctuate frequently and the direction changes violently. Combining with the trajectory map, it can be found that ( Figure 30 ), this trajectory made a U-turn on Road 11, traveled a long distance, and stopped many times during the whole driving process, which indicates that the vehicle encountered many red lights or other traffic obstacles during the driving process. Through in-depth analysis of this trajectory, it can be inferred that the driver of this vehicle may have a relatively aggressive driving style, tend to drive at high speeds and change lanes frequently, but is also restricted by traffic signals and other traffic conditions.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An urban road traffic vehicle trajectory visual analysis system, characterized in that, Including: Data processing and storage module: Obtain the trajectories of urban road traffic vehicles and perform comprehensive data cleaning; Perform single-trajectory aggregation, merge the trajectory points belonging to the same vehicle into a complete trajectory; perform trajectory expansion, add attribute information such as acceleration, road, and lane to obtain a high-quality trajectory dataset; extract the implicit features in the trajectories of urban road traffic vehicles through an encoder-decoder model based on the BiLSTM-Transformer hybrid architecture, and then use the k-means clustering algorithm to cluster the trajectories of urban road traffic vehicles in the feature space to complete the clustering of group trajectories; Global control module A: Used for time range selection, road selection, and trajectory search; Time range selection: Set the time range, which is divided at 5-minute intervals, for selecting the specific time period to be analyzed; Road selection: Select a specific road, and the traffic flow and average speed trend chart, trajectory abnormal behavior table, and trajectory abnormal behavior statistical chart will be updated with the information of the selected road; Trajectory search: Through the function of searching for trajectories by vehicle ID, after inputting the vehicle ID for search, the individual trajectory analysis module will display the detailed information of the trajectory, and at the same time the trajectory map will be updated to the time range where the trajectory is located and the trajectory will be highlighted; Analysis module for road traffic flow and congestion conditions, used to analyze the laws of road traffic flow and congestion conditions, and discover the spatio-temporal distribution of traffic congestion, the spatio-temporal distribution of road traffic flow and average speed conditions; Trajectory abnormal behavior analysis module D, used to identify abnormal vehicle trajectory behaviors, and the abnormal vehicle trajectory behaviors include three categories: reverse driving behavior, speeding behavior, and sudden acceleration and deceleration behavior; Individual trajectory analysis module, used to display and analyze the detailed information of a single trajectory; Group trajectory analysis module F, used to analyze the characteristics of vehicle group trajectories.

2. The urban road traffic vehicle trajectory visual analysis system according to claim 1, wherein: The analysis module for road traffic flow and congestion conditions includes four visualization views to display the congestion data and traffic flow data of each road to realize the analysis of road traffic flow and congestion conditions. The four visualization views are the road congestion situation distribution map B1, the road traffic flow heat map B2, the road traffic flow and average speed statistical chart B3, and the traffic flow and average speed trend chart B4; The road congestion situation distribution map is used to display the spatio-temporal distribution of road congestion status; the horizontal axis is the time axis, which divides one hour into twelve five-minute time periods to show the evolution of congestion status within one hour, and the specific time period to be displayed is selected through the global control module A; the vertical axis represents the spatial dimension, and all roads in the area are marked according to two driving directions to show the congestion distribution on different roads; the size and color of the data points represent the congestion index; The road traffic flow heat map uses a table heat map to display the traffic flow passing conditions of each road per hour; The road traffic flow and average speed statistical chart uses a combination of a bar chart and a line chart to comprehensively display the traffic flow and average speed of the road; among them, the bar chart is used to represent the road traffic flow, and the height of the bar is proportional to the traffic flow size; the line chart is used to represent the average road traffic speed, and its change reflects the operating speed characteristics of different roads under different traffic flow conditions; The traffic flow and average speed trend chart shows the changes in traffic flow and average speed over time; the horizontal axis represents time, and the vertical axis is divided into two parts, respectively used to represent traffic flow and average speed; among them, the change trend of traffic flow is presented by a stacked bar chart; this stacked chart shows the traffic flow of small vehicles, trucks, and passenger cars respectively, and stacks the three to form the total traffic flow; the change trend of average speed is shown by a line chart, which records the average speed changes of all vehicles, small vehicles, trucks, and passenger cars per minute.

3. The urban road traffic vehicle trajectory visual analysis system according to claim 2, wherein: The trajectory abnormal behavior analysis module D includes three views, and the three views include a trajectory abnormal behavior table, a trajectory abnormal behavior distribution map, and a trajectory abnormal behavior statistical chart; The trajectory abnormal behavior table lists the detailed information of all detected vehicle abnormal behaviors in tabular form; The trajectory abnormal behavior distribution map uses a scatter plot to show the geographical location distribution of all abnormal behaviors, and each data point represents a trajectory where an abnormal behavior occurred; The trajectory abnormal behavior statistical chart presents the abnormal behavior data counted by hour in the form of a stacked bar chart; Different types of abnormal behaviors are stacked in the bar chart in the same three different colors as the trajectory abnormal behavior distribution map, showing the quantity and proportion of various abnormal behaviors in different time periods.

4. The urban road traffic vehicle trajectory visual analysis system according to claim 3, wherein: The individual trajectory analysis module includes three views, and the three views include a trajectory map C, an individual trajectory detailed information map E1, and an individual trajectory characteristic radar chart E2; The trajectory map uses road map data and is presented using the Leaflet map rendering technology, showing the structure and layout of the road network in the research area, as well as the historical trajectory data of vehicles; On the trajectory map, vehicle trajectories are presented in the form of lines, showing all trajectories within the selected time period; The individual trajectory detailed information map shows the information of a single trajectory; above the individual trajectory detailed information map, the basic information of the trajectory is shown through four labels, including vehicle ID, vehicle type, trajectory start time, and trajectory end time; below the map, in the form of a line chart, it details the changes in multiple key attributes such as the speed, acceleration, direction, road, and lane of the trajectory over time; The individual trajectory characteristic radar chart comprehensively shows the characteristic values of the trajectory from multiple dimensions; this radar chart covers multiple key characteristic values such as maximum speed, average speed, maximum acceleration, minimum acceleration, number of sudden accelerations and decelerations, number of lane changes, maximum steering change, and duration; each characteristic value is represented by an independent axis on the radar chart, and the length of the axis represents the value range of the characteristic value; by connecting the performance of the trajectory on each characteristic value in graphical form, a closed polygon is formed to present the relative advantages and disadvantages of the trajectory in different characteristic dimensions.

5. The urban road traffic vehicle trajectory visual analysis system according to claim 4, wherein: The group trajectory analysis module F includes two views, namely a trajectory spatio-temporal feature clustering chart and a trajectory parallel coordinate chart; The trajectory spatio-temporal feature clustering chart shows the characteristic clustering scatter plot of all vehicle trajectories within the selected time period; The trajectory parallel coordinate diagram shows the specific values of the multi-dimensional features of all trajectories within the selected time period; the trajectory parallel coordinate diagram includes multiple key feature dimensions such as vehicle ID, maximum speed, average speed, maximum acceleration, minimum acceleration, average acceleration, parking ratio, total distance, and clustering category; In the parallel coordinate diagram, each polyline represents a vehicle trajectory, and the values of the polyline on different coordinate axes correspond to the performance of the trajectory in each feature dimension.

6. The urban road traffic vehicle trajectory visual analysis system according to claim 5, characterized in that: The calculation method of the congestion index is as follows: Among them, v pred is the average speed when the road is unobstructed, and is the average driving speed of vehicles on this road over a period of time.

7. The urban road traffic vehicle trajectory visual analysis system according to claim 6, wherein: The judgment method of the abnormal behavior of the vehicle trajectory is as follows: Reverse driving behavior: In terms of space, it satisfies that the vehicle driving direction deviation Δθ(t)>90°; in terms of time, the continuous abnormal duration T≥2s; when these two conditions are met simultaneously, it is determined that the vehicle has a reverse driving behavior; Speeding behavior: Extract the road speed limit value as the threshold v for speeding detection limit , when the vehicle speed is greater than the road speed limit value v limit , it can be determined that the vehicle trajectory has a speeding behavior; Sudden acceleration and deceleration behavior: The acceleration threshold is set as |a| > 3 m / s 2 , and it is determined that there is a rapid acceleration or deceleration behavior when the absolute value of the acceleration exceeds the threshold.

Images