A multi-class traffic environment anomaly diagnosis method and system

Abstract

This invention provides a method and system for diagnosing multiple types of traffic environment anomalies, belonging to the field of intelligent transportation technology. The method includes: real-time collection of driving status data from multiple vehicle-mounted terminals, extraction of driving behavior features, association with spatiotemporal information, and uploading to a data analysis platform; dividing the target road segment into several road segment units; mapping the driving behavior features of multiple vehicles to corresponding road segment units and lanes using the correlation between spatiotemporal information and driving behavior features; performing cluster analysis on the driving behavior features of multiple vehicles within a preset spatiotemporal window, where the driving behavior features include lateral and longitudinal features; obtaining the group driving behavior features corresponding to the multiple vehicles based on the cluster analysis; and diagnosing multiple types of traffic environment anomalies based on the consistency of the group driving behavior features. This invention relies on the group driving behavior of multiple vehicles to achieve accurate identification of multiple types of traffic environment anomalies, improving the reliability and real-time performance of anomaly diagnosis, and providing effective support for road safety management.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically to a method and system for diagnosing various types of traffic environment anomalies. Background Technology

[0002] With the rapid growth of global traffic flow, the road traffic environment is becoming increasingly complex, and various sudden or hidden abnormal events such as road debris, temporary construction, blind spots, and road geometric defects occur frequently. These traffic environment anomalies are often difficult for drivers to detect in time, which can easily lead to traffic accidents such as emergency braking and rear-end collisions, seriously threatening driving safety. Therefore, the ability to identify such anomalies in advance and provide timely warnings to following vehicles has become a key requirement for improving road safety and traffic efficiency.

[0003] Traditional methods for identifying traffic anomalies primarily rely on fixed infrastructure such as video surveillance and geomagnetic coils, or manual inspections. Limited by deployment density and maintenance costs, their coverage is limited, and they exhibit low sensitivity to transient, unstructured anomalies such as nighttime debris and low-lying obstacles. In recent years, some studies have attempted to utilize single-vehicle sensors such as millimeter-wave radar and cameras for local environmental perception, and to achieve individual risk avoidance through systems like Forward Collision Warning (FCW). However, these solutions only use the perception results for the vehicle's own decision-making, failing to effectively aggregate multi-vehicle data to form collective cognition. A large amount of valuable driving behavior data remains unutilized, failing to reveal common environmental anomalies behind individual behaviors, thus limiting overall early warning capabilities.

[0004] The two most prominent problems with existing technologies are: first, they fail to effectively aggregate data from multiple vehicles to form a collective understanding; and second, their coverage is limited and their early warnings are delayed. Summary of the Invention

[0005] In view of this, it is necessary to provide a multi-type traffic environment anomaly diagnosis method and system to solve the technical problems of existing technologies, such as failure to effectively aggregate multi-vehicle data to form a collective understanding, limited coverage, and delayed early warning.

[0006] To address the aforementioned technical problems, in a first aspect, the present invention provides a method for diagnosing multiple types of traffic environment anomalies, characterized by comprising: Real-time driving status data is collected through vehicle-mounted terminals in multiple vehicles. Driving behavior features are extracted based on the driving status data, and the driving behavior features are correlated with spatiotemporal information and then uploaded to the data analysis platform. The target road segment is divided into several road segment units through the data analysis platform; the driving behavior characteristics of multiple vehicles are mapped to the corresponding road segment units and lanes by using the spatiotemporal information and the correlation between the driving behavior characteristics; The data analysis platform performs cluster analysis on the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window. The driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to the multiple vehicles are obtained. Based on the consistency of the group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and multiple types of traffic environment anomaly diagnosis results are obtained.

[0007] In one possible implementation, the multi-type traffic environment anomaly diagnosis method also includes: The data analysis platform is used to match and associate the diagnostic results of the various traffic environment anomalies with the high-precision map to the corresponding road segment units and lanes, thereby obtaining the abnormal road segment units and abnormal lanes. When a target vehicle enters the abnormal road segment or abnormal lane within a preset time, the target vehicle's navigation software sends a graphic and voice warning message containing the type of traffic environment anomaly and driving suggestions to the target vehicle.

[0008] In one possible implementation, the multi-type traffic environment anomaly diagnosis results are associated with high-precision map matching to the corresponding road segment units and lanes, which is achieved through a map matching algorithm.

[0009] In one possible implementation, the driving state data includes lane number, collision time, longitudinal deceleration, lateral acceleration, current lateral position within the lane, and latitude and longitude coordinates; the driving behavior features include emergency braking, excessively short collision time, lane changing, and lateral position deviation within the lane; the extraction of driving behavior features based on the driving state data includes: When the absolute value of the longitudinal deceleration is greater than a preset braking deceleration threshold, emergency braking behavior features are extracted. When the collision time is less than a preset collision time threshold, extract the behavior feature of excessively low collision time. When the lane number changes, and the lateral position and latitude and longitude coordinates of the current lane are synchronously matched with the lane switching trajectory, and the lateral acceleration shows a symmetrical temporal pattern of first increasing and then decreasing, the lane changing behavior features are extracted. When the current lateral position in the lane exceeds the preset normal lateral position range in the lane and exceeds the preset offset time, the lateral position offset behavior feature in the lane is extracted.

[0010] In one possible implementation, the behavior characteristic of excessively low collision time is used to characterize the potential risk of a rear-end collision between the vehicle and the vehicle in front. The preset collision time threshold is adjusted in conjunction with the target vehicle's current speed; the higher the current speed, the larger the preset collision time threshold; the lower the current speed, the smaller the preset collision time threshold.

[0011] In one possible implementation, dividing the target road segment into several road segment units includes: Based on the equal-spacing grid division method, the target road segment is divided into several road segment units according to the preset road segment length.

[0012] In one possible implementation, the spatiotemporal information includes a timestamp, latitude and longitude coordinates, and lane number; the step of mapping the driving behavior characteristics of multiple vehicles to corresponding road segment units and lanes using the spatiotemporal information and the association relationship between the driving behavior characteristics includes: Establish a correlation between the characteristics of multi-vehicle driving behavior and corresponding spatiotemporal information; The timestamp, latitude and longitude coordinates, and lane number are matched with the road segment units and lanes of the target road segment using a map matching algorithm. The matched multi-vehicle driving behavior features are mapped to the corresponding road segment units and lanes within the target road segment.

[0013] In one possible implementation, the clustering analysis of the driving behavior features of multiple vehicles within a preset spatiotemporal window includes: K-means clustering was performed on the horizontal and vertical features of driving behavior characteristics to obtain the cluster categories and isolated samples of the horizontal features, and the cluster categories and isolated samples of the vertical features. Isolated samples of the horizontal features and isolated samples of the vertical features are removed. Based on the horizontal and vertical feature clustering categories after removing isolated samples, different categories of group driving behavior are obtained.

[0014] In one possible implementation, the step of diagnosing multiple types of traffic environment anomalies and obtaining diagnostic results for multiple types of traffic environment anomalies based on the consistency of group driving behavior characteristics includes: The presence of an invisible obstacle was determined based on the concentrated lane-changing behavior of multiple unrelated vehicles within a similar time and space range; The lane is temporarily closed or under construction based on a sudden drop in vehicle density in a certain lane, a significant increase in density in adjacent lanes, and concentrated lane-changing behavior upstream. The vehicle density is the number of vehicles per unit space for the corresponding lane of the target road segment unit within a preset time and space window, based on the location data, lane number, and timestamp uploaded by multiple vehicle terminals. The systematic offset of the lateral position distribution of multiple vehicles in the same lane without changing lanes is determined to be a road geometric anomaly or a road surface anomaly. A road segment is identified as a dynamically high-risk section if the proportion of emergency braking or collisions with excessively short time exceeds a preset risk ratio threshold within a specific time and space range.

[0015] Secondly, the present invention also provides a multi-type traffic environment anomaly diagnosis system, including: The driving data acquisition and association unit is used to collect driving status data in real time through the vehicle terminals of multiple vehicles, extract driving behavior features based on the driving status data, and upload the driving behavior features to the data analysis platform after associating them with spatiotemporal information. The road segment division and behavior mapping unit is used to divide the target road segment into several road segment units through the data analysis platform; and to map the driving behavior characteristics of multiple vehicles to the corresponding road segment units and lanes by using the spatiotemporal information and the correlation between the driving behavior characteristics. The group behavior clustering and anomaly diagnosis unit is used to perform cluster analysis on the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window through the data analysis platform. The driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to multiple vehicles are obtained. Based on the consistency of the group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and multiple types of traffic environment anomaly diagnosis results are obtained.

[0016] The beneficial effects of this invention are as follows: The multi-type traffic environment anomaly diagnosis method provided by this invention firstly collects driving status data in real time through the on-board terminals of multiple vehicles, extracts driving behavior features based on the driving status data, and uploads the driving behavior features to a data analysis platform after associating them with spatiotemporal information. This achieves real-time collection, feature extraction, and effective uploading of multi-vehicle driving data, providing comprehensive and representative basic data for subsequent traffic environment anomaly diagnosis and avoiding the impact of the randomness of single vehicle data on diagnostic accuracy. Secondly, the target road segment is divided into several road segment units through the data analysis platform, and the driving behavior features of multiple vehicles are mapped to the corresponding road segments using the association relationship between the spatiotemporal information and the driving behavior features. The system identifies units and lanes to precisely bind driving behavior characteristics with specific spatial locations, clearly defining the road segments and lanes corresponding to abnormal behaviors, thus providing spatial positioning support for accurate anomaly diagnosis. Finally, the data analysis platform performs cluster analysis on the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window. These driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the system obtains the group driving behavior characteristics corresponding to multiple vehicles. By diagnosing the consistency of the group driving behavior characteristics, the system obtains diagnostic results for various types of traffic environment anomalies. This approach can uncover patterns in the driving behavior of multiple vehicle groups, eliminate invalid interference, improve the reliability and comprehensiveness of anomaly diagnosis, and achieve accurate identification of various types of traffic environment anomalies. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic flowchart of an embodiment of the multiple traffic environment anomaly diagnosis method provided by the present invention; Figure 2 A schematic flowchart of an embodiment of the abnormal road segment matching and navigation warning provided by the present invention; Figure 3 This is a schematic flowchart of another embodiment of the multi-type traffic environment anomaly diagnosis method provided by the present invention; Figure 4 This is a schematic diagram of an embodiment of the present invention for extracting driving behavior features based on driving state data; Figure 5 This is a schematic diagram of an embodiment of the present invention that maps the driving behavior characteristics of multiple vehicles to corresponding road segment units and lanes; Figure 6 This is a schematic flowchart of an embodiment of the present invention for clustering analysis of driving behavior characteristics of multiple vehicles within a preset spatiotemporal window; Figure 7 A schematic flowchart illustrating an embodiment of the present invention for diagnosing multiple types of traffic environment anomalies and obtaining diagnostic results for multiple types of traffic environment anomalies; Figure 8 This is a schematic diagram of an embodiment of the multi-type traffic environment anomaly diagnosis system provided by the present invention. Detailed Implementation

[0019] 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 a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0020] In the description of the embodiments of the present invention, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0021] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0022] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0023] Before demonstrating the embodiments, the following terms will be explained.

[0024] Vehicle-mounted terminal: A device on a vehicle used to collect and transmit driving-related data. In this invention, it is used to collect driving status data of multiple vehicles in real time and upload it to a data analysis platform.

[0025] Driving status data: Various status parameters generated during vehicle operation, including lane number, collision time, longitudinal deceleration, etc., are the basis for extracting driving behavior features.

[0026] Driving behavior characteristics: Based on driving state data, these are characteristics that reflect the vehicle's driving operation and state. In this invention, these include emergency braking, lane changing, and lateral position deviation.

[0027] Spatiotemporal information: Information used to characterize the vehicle's travel time and spatial location. In this invention, it includes timestamps, latitude and longitude coordinates, and lane numbers, which are used to associate driving behavior characteristics with specific locations.

[0028] Data analysis platform: A platform used to receive, process, and analyze data uploaded by multiple vehicles. In this invention, it undertakes core functions such as road segmentation, cluster analysis, and anomaly diagnosis.

[0029] Road segment unit: The smallest spatial unit after dividing the target road segment according to preset rules. In this invention, it is divided based on the equal-spacing grid division method, which is used to accurately locate abnormal locations.

[0030] Preset spatiotemporal window: A combination of a pre-defined fixed time range and a fixed spatial range, used in this invention to limit the analysis range of multi-vehicle driving behavior characteristics and ensure the effectiveness of the analysis.

[0031] Lateral characteristics: These are features that reflect the lateral movement of a vehicle. In this invention, they are related to the vehicle's lateral acceleration and lateral position deviation within the lane.

[0032] Longitudinal characteristics: characteristics that reflect the longitudinal driving state of a vehicle, which in this invention are related to the vehicle's longitudinal deceleration, collision time, etc.

[0033] Group driving behavior characteristics: The driving behavior characteristics obtained by clustering the driving behavior characteristics of multiple vehicles can reflect the commonalities of a certain group. In this invention, these characteristics are used to diagnose traffic environment anomalies.

[0034] Consistency: In this invention, it specifically refers to the degree of similarity in the driving behavior characteristics of a multi-vehicle group. Specifically, it is reflected in the common trends in the lateral and longitudinal characteristics of multiple vehicles within the same group after cluster analysis. This is the core judgment criterion for this invention to diagnose various traffic environment anomalies.

[0035] This invention provides a method and system for diagnosing various types of traffic environment anomalies, which will be described below.

[0036] Figure 1 This invention provides a method for diagnosing various types of traffic environment anomalies, such as... Figure 1 As shown, various methods for diagnosing traffic environment anomalies include: S101. Real-time collection of driving status data through vehicle-mounted terminals of multiple vehicles, extraction of driving behavior features based on driving status data, and uploading of driving behavior features to a data analysis platform after association with spatiotemporal information. S102. Divide the target road segment into several road segment units through a data analysis platform; map the driving behavior characteristics of multiple vehicles to the corresponding road segment units and lanes by using the correlation between spatiotemporal information and driving behavior characteristics; S103. Using a data analysis platform, cluster analysis is performed on the driving behavior characteristics of multiple vehicles within a preset time and space window. The driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to multiple vehicles are obtained. Based on the consistency of the group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and the diagnostic results of multiple types of traffic environment anomalies are obtained.

[0037] It is important to note that the core design of this invention is the use of multi-vehicle data for traffic environment anomaly diagnosis, which is closely related to the extraction of group driving behavior characteristics and the analysis of behavioral consistency. Driving data from a single vehicle is easily affected by accidental factors such as individual driver habits and sudden, unexpected actions, making it impossible to objectively and accurately reflect the actual traffic conditions of a road segment. Relying solely on single-vehicle data makes it difficult to extract representative group driving behavior characteristics, let alone achieve accurate anomaly diagnosis based on behavioral consistency.

[0038] Collecting driving status data from multiple vehicles can aggregate driving information from different drivers and vehicles within the same time and space, effectively avoiding the random bias of single vehicle data and providing comprehensive, objective, and reliable basic data for subsequent cluster analysis and extraction of group driving behavior characteristics.

[0039] Meanwhile, the core of group driving behavior characteristics lies in "group commonality". Only based on multi-vehicle data can isolated samples be eliminated through cluster analysis, and the common driving patterns of most vehicles be extracted. Then, by analyzing the consistency of group driving behavior characteristics, it can be determined whether the common pattern is caused by traffic environment anomalies, and finally the accurate identification of multiple types of traffic environment anomalies can be achieved. This is also the key difference between this invention and the existing single-vehicle diagnostic scheme.

[0040] In summary, the multi-type traffic environment anomaly diagnosis method provided by this invention firstly collects driving status data in real time through the on-board terminals of multiple vehicles, extracts driving behavior features based on the driving status data, and uploads the driving behavior features to a data analysis platform after associating them with spatiotemporal information. This achieves real-time collection, feature extraction, and effective uploading of multi-vehicle driving data, providing comprehensive and representative basic data for subsequent traffic environment anomaly diagnosis and avoiding the impact of the randomness of single vehicle data on diagnostic accuracy. Secondly, the target road segment is divided into several road segment units through the data analysis platform, and the driving behavior features of multiple vehicles are mapped to corresponding road segment units using the association relationship between the spatiotemporal information and the driving behavior features. The system identifies lanes and road segments to precisely bind driving behavior characteristics to specific spatial locations, clearly defining the road segments and lanes corresponding to abnormal behaviors, thus providing spatial positioning support for accurate anomaly diagnosis. Finally, the data analysis platform performs cluster analysis on the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window. These driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the system obtains the group driving behavior characteristics corresponding to multiple vehicles. By diagnosing the consistency of the group driving behavior characteristics, the system obtains diagnostic results for various types of traffic environment anomalies. This approach can uncover patterns in the driving behavior of multiple vehicle groups, eliminate invalid interference, improve the reliability and comprehensiveness of anomaly diagnosis, and achieve accurate identification of various types of traffic environment anomalies.

[0041] In some embodiments of the present invention, such as Figure 2 As shown, various methods for diagnosing traffic environment anomalies also include: S201. Through the data analysis platform, the diagnostic results of multiple types of traffic environment anomalies are matched and associated with high-precision maps to the corresponding road segment units and lanes to obtain abnormal road segment units and abnormal lanes. S202. When the target vehicle enters an abnormal road segment or abnormal lane within a preset time, the target vehicle's navigation software sends graphic and voice warning information containing the type of traffic environment abnormality and driving suggestions to the target vehicle.

[0042] It should be noted that the above two steps are follow-up early warning operations after the diagnosis of multiple types of traffic environment anomalies, and their effectiveness is closely related to the traffic environment anomaly diagnosis methods described above. The previous analysis of the consistency of driving behavior characteristics of multi-vehicle groups enabled the accurate identification of multiple types of traffic environment anomalies, clarifying the road segment units and lanes corresponding to the anomaly type and location. This provided a reliable prerequisite for the accurate delivery and timely reminder of subsequent early warning information. Only by achieving accurate diagnosis of traffic environment anomalies can the pertinence and timeliness of early warning information be ensured, thereby providing scientific and timely safety guidance for target vehicle drivers and effectively avoiding driving risks caused by traffic environment anomalies.

[0043] In some embodiments of the present invention, a data analysis platform is used to match and associate the diagnostic results of various traffic environment anomalies with high-precision maps to corresponding road segment units and lanes, thereby obtaining abnormal road segment units and abnormal lanes. Through a high-precision map matching algorithm and based on geographic coordinates, the data analysis platform maps each type of traffic environment anomaly result to the corresponding road segment, road station, and lane on the digital map, generating structured abnormal electronic map data. The data includes anomaly type, road segment interval, road station, and lane number (if any). The sent warning messages can be set according to different types of traffic environment anomalies. For example, when the diagnosis result is the presence of an invisible obstacle, the warning message can be set as "An invisible obstacle exists ahead. Please slow down and proceed with caution." When the diagnosis result is temporary lane closure or construction, the warning message can be set as "The corresponding lane ahead is temporarily closed / under construction. Please change lanes in advance and slow down." When the diagnosis result is road geometry anomaly or road surface anomaly, the warning message can be set as "Road geometry / road surface anomaly exists ahead. Please control your speed and drive cautiously." When the diagnosis result is a dynamic high-risk road section, the warning message can be set as "Ahead is a dynamic high-risk road section with a risk of rear-end collision. Please maintain a safe distance and slow down." The above warning messages not only clarify the type of anomaly but also provide targeted driving advice. They can be presented simultaneously through navigation software in the form of graphic pop-ups and voice broadcasts to ensure that drivers are clearly aware and can respond in a timely manner.

[0044] Figure 3 This is a flowchart illustrating another embodiment of the multi-type traffic environment anomaly diagnosis method provided by the present invention. The multi-type traffic environment anomaly diagnosis system of the present invention consists of three core modules: an in-vehicle terminal, a data analysis platform, and in-vehicle navigation software. These three modules work together to form a closed-loop working chain of "data acquisition - core diagnosis - precise early warning." The specific collaborative relationship and working principle are as follows: (1) Vehicle-mounted terminal: Data acquisition and preprocessing module As the system's data input, the vehicle-mounted terminal is responsible for collecting real-time driving status data from multiple vehicles, including but not limited to lane number, collision time, longitudinal deceleration, lateral acceleration, current lateral position within the lane, and latitude and longitude coordinates. Based on the collected driving status data, the vehicle-mounted terminal further extracts four types of driving behavior features: emergency braking, excessively short collision time, lane changing, and lateral position deviation within the lane. These extracted driving behavior features are then correlated with spatiotemporal information such as timestamps and latitude and longitude coordinates, and uploaded to the data analysis platform in structured data format. This provides multi-dimensional and multi-source foundational data support for subsequent group behavior analysis and anomaly diagnosis.

[0045] (2) Data analysis platform: core diagnostic and mapping processing module As the core processing unit of the system, the data analysis platform undertakes the crucial function of anomaly diagnosis. First, the platform divides the target road segment into several road segment units using a preset equidistant grid method. Utilizing the correlation between spatiotemporal information and driving behavior characteristics, it maps the driving behavior characteristics of multiple vehicles to corresponding road segment units and lanes, achieving precise binding of behavioral data with spatial location. Second, within a preset spatiotemporal window, the platform performs K-means clustering analysis on the lateral and longitudinal driving behavior characteristics of multiple vehicles, removing isolated samples to obtain group driving behavior characteristics. Based on the consistency of these group driving behavior characteristics, it diagnoses various types of traffic environment anomalies, including the presence of invisible obstacles, temporary lane closures or construction, road geometric or pavement anomalies, and dynamically high-risk road sections. Finally, the platform matches the anomaly diagnosis results with a high-precision map to generate a structured anomaly map, clearly identifying the anomaly type, the road segment unit corresponding to the anomaly location, and the lane, providing accurate anomaly information support for the warning operations of the in-vehicle navigation software.

[0046] (3) In-vehicle navigation software: warning output and driving guidance module As the system's early warning output, the in-vehicle navigation software is responsible for providing accurate and timely safety guidance to the target vehicle. The software collects real-time driving status data from the target vehicle and, combined with a structured anomaly map generated by a data analysis platform, predicts when the target vehicle will enter an abnormal road segment or lane. When the target vehicle enters the corresponding abnormal area within a preset time, the in-vehicle navigation software obtains the anomaly type and targeted driving suggestions from the data analysis platform. It then distributes real-time warning information to the target vehicle through graphical pop-ups and voice announcements, such as "There is an invisible obstacle ahead; please slow down and proceed with caution" or "The corresponding lane ahead is temporarily closed / under construction; please change lanes in advance and slow down before proceeding," ensuring that the driver is clearly aware of the abnormal situation and can take timely countermeasures.

[0047] Overall, the vehicle-mounted terminal, data analysis platform, and vehicle navigation software work closely together to form a complete collaborative system: the vehicle-mounted terminal provides reliable data input, the data analysis platform completes accurate anomaly diagnosis and mapping, and the vehicle navigation software provides timely early warning output, thereby ensuring the accuracy, comprehensiveness, and timeliness of anomaly diagnosis in various traffic environments and providing strong support for road safety management.

[0048] In some embodiments of the present invention, multiple traffic environment anomaly diagnosis results are associated with high-precision map matching to corresponding road segment units and lanes, which is achieved through map matching algorithms.

[0049] In some embodiments of the present invention, the map matching algorithms applicable to the present invention include, but are not limited to, the following preferred types: (1) Geometry-based map matching algorithms, such as point-to-curve matching algorithms and curve-to-curve matching algorithms. The advantage of this type of algorithm is that it has low computational complexity and fast response speed, which can meet the application scenarios with high real-time requirements; the disadvantage is that it is more sensitive to positioning noise data, and the matching accuracy is easily affected in complex road conditions or when the positioning accuracy is low.

[0050] (2) Map matching algorithms based on probability statistics, such as Hidden Markov Model (HMM) map matching algorithms. The advantage of this type of algorithm is that it can comprehensively consider the probability distribution of positioning data and road topology information, has strong robustness to noisy data, and has higher matching accuracy; the disadvantage is that the computational complexity is relatively high and it requires a certain amount of computing power from the system.

[0051] (3) Machine learning-based map matching algorithms, such as deep learning-based map matching models. The advantage of this type of algorithm is that it can automatically learn the complex mapping relationship between location data and road network, and performs better in traffic scenarios with dense traffic and complex road topology; the disadvantage is that it requires a large amount of labeled data for model training, and the model deployment and maintenance costs are high.

[0052] It should be noted that the selection of map matching algorithm in this invention can be flexibly adjusted according to the real-time requirements, data accuracy and computing power conditions of the actual application scenario. The above algorithm is only a preferred embodiment, and this invention does not limit its scope of protection. Any map matching algorithm that can achieve accurate matching and association between abnormal diagnosis results and high-precision maps can be applied to this invention.

[0053] In some embodiments of the present invention, driving state data includes lane number, collision time, longitudinal deceleration, lateral acceleration, current lateral position within the lane, and latitude and longitude coordinates; driving behavior features include emergency braking, excessively short collision time, lane changing, and lateral position deviation within the lane; driving behavior features are extracted based on the driving state data, such as... Figure 4 As shown, it includes: S401. When the absolute value of the longitudinal deceleration is greater than the preset braking deceleration threshold, extract the emergency braking behavior features. S402. When the collision time is less than the preset collision time threshold, extract the behavior feature of excessively low collision time. S403. When the lane number changes, and the lateral position and latitude and longitude coordinates of the current lane are synchronously matched with the lane switching trajectory, and the lateral acceleration shows a symmetrical temporal pattern of first increasing and then decreasing, the lane changing behavior features are extracted. S404. When the current lateral position in the lane exceeds the preset normal lateral position range in the lane and continues to exceed the preset offset time, extract the lateral position offset behavior feature in the lane.

[0054] In some embodiments of the present invention, the behavior characteristic of excessively low collision time is used to characterize the potential rear-end collision risk between the vehicle and the vehicle in front. The preset collision time threshold is adjusted based on the target vehicle's current speed; the higher the current speed, the larger the preset collision time threshold; the lower the current speed, the smaller the preset collision time threshold.

[0055] In some embodiments of the present invention, four types of driving behavior features are extracted based on driving state data: (1) Emergency braking behavior: The vehicle terminal collects longitudinal deceleration data of the vehicle within the initial set period, and uses the first 1 percentile value of the deceleration distribution within the period as the emergency braking behavior discrimination threshold; after the set period ends, the longitudinal deceleration is monitored in real time. When the deceleration exceeds the discrimination threshold and the vehicle speed drops sharply within the corresponding time period, an emergency braking behavior is determined to have occurred, and its timestamp and geographical coordinates are recorded; at the same time, the vehicle terminal will continuously collect historical deceleration data and dynamically adjust the discrimination threshold so that the discrimination threshold is adaptively optimized as the individual driving style evolves. (2) Collision time too low behavior: The vehicle terminal monitors the collision time between the vehicle and the vehicle in front in real time, and dynamically sets a judgment threshold in combination with the current vehicle speed. The higher the vehicle speed, the larger the judgment threshold. When the collision time is lower than the dynamic judgment threshold, it is determined that a collision time too low behavior has occurred, and its timestamp and geographical coordinates are recorded. (3) Lane changing behavior: Lane changing operation is identified based on lateral acceleration timing signal. When the lateral acceleration is detected to show a symmetrical change pattern of first unidirectional enhancement and then reverse callback, it is determined that the vehicle has changed lanes. The lane changing direction is determined according to the initial change direction of the lateral acceleration: if the initial change is positive, it is determined to be a right lane changing behavior; if it is negative, it is determined to be a left lane changing behavior. At the same time, it is recorded based on the timestamp, geographical coordinates, the lane before lane changing and the lane after lane changing.

[0056] (4) Lane lateral position deviation behavior: When a vehicle continuously deviates from the lane centerline in the current lane beyond the preset lateral deviation threshold without changing lanes, and the duration exceeds the preset time window, it is determined that a lane lateral position deviation behavior has occurred, and the lane, geographical coordinates, timestamp, duration and average deviation are recorded.

[0057] Based on the extracted driving behavior features and the recorded geographical coordinates, timestamps, lane information, and other data, the vehicle terminal uploads the structured data to the data analysis platform.

[0058] In some embodiments of the present invention, the most standard urban roads and highway lanes in China are preferred, with a conventional lane width of 3.5m to 3.75m; Longitudinal deceleration can be acquired by an onboard inertial measurement unit (IMU) or wheel speed sensors. The preset braking deceleration threshold can be set to 3 m / s² depending on the scenario. 2 Up to 5m / s 2 The preferred range, for example, when a vehicle is driving on urban roads, if the absolute value of the longitudinal deceleration exceeds 4 m / s². 2 This can be used to determine emergency braking, and this feature can be used to reflect sudden situations ahead or rear-end collision risks. Time to Collision (TC) refers to the relative distance between the vehicle and the vehicle in front divided by their relative speeds. The preset collision time threshold can be dynamically adjusted based on vehicle speed. For example, the threshold is set to 2 seconds when the vehicle speed is 60 km / h and 3 seconds when the vehicle speed is 100 km / h. When the TTC is lower than the corresponding threshold, it indicates that the following distance is too close and there is a risk of rear-end collision. This feature can be used to identify dynamic high-risk road sections. Lane number changes can be provided by lane line recognition cameras or high-precision maps. The lateral position and latitude and longitude must match the continuous trajectory from the original lane to the target lane. For example, when moving from the left lane to the right lane, the lateral position gradually shifts from -1.8m (with the lane center as the origin) to +1.8m, while the lateral acceleration first increases to 0.5m / s². 2 Then, the value is reset to 0 to form a symmetrical pattern, thereby avoiding misjudging small deviations within the lane as lane changes and improving the accuracy of feature extraction. The preset normal lateral position range can be set to ±0.5m from the center of the lane. When the vehicle continuously deviates to more than ±0.8m, it indicates that there may be driver distraction, uneven road surface, or road geometry abnormalities. This feature can be used for subsequent diagnosis of road geometry abnormalities or road surface abnormalities.

[0059] It should be noted that the above thresholds, ranges and modes are only preferred embodiments and can be flexibly adjusted according to actual application scenarios (such as urban roads, highways, and different vehicle models). This invention does not limit its scope of protection. Any parameter settings that can accurately extract the corresponding driving behavior characteristics can be applied to this invention.

[0060] In some embodiments of the present invention, dividing the target road segment into several road segment units includes: Based on the equal-spacing grid division method, the target road segment is divided into several road segment units according to the preset road segment length.

[0061] It should be noted that the equal-spacing grid division method is a commonly used method for road space division in existing technologies. Its core principle is to divide the continuous target road segment equally according to a preset fixed length, forming several independent road segment units with consistent range and clear boundaries. The length of each road segment unit is uniform, they do not overlap, and they completely cover the entire target road segment, which facilitates standardized and normalized zoning analysis of the road segment.

[0062] In the context of the application scenario of this invention, the purpose of this partitioning method is to break down a large target road segment into smaller road segment units, thereby achieving precise binding between driving behavior characteristics and spatial location. The driving behavior characteristics of multiple vehicles can be mapped to specific road segment units and lanes through spatiotemporal information, avoiding the ambiguity of abnormal location due to the large road segment range. This provides a regularized spatial basis for subsequent cluster analysis, extraction of group driving behavior characteristics, and diagnosis of traffic environment anomalies within a preset spatiotemporal window.

[0063] Specifically, the preset road segment length can be flexibly adjusted according to the actual application scenario, with an optimal range of 50 meters to 100 meters. For example, in urban road scenarios, due to the dense intersections and complex traffic scenarios, the preset road segment length can be set to 50 meters to achieve more precise anomaly localization. In highway scenarios, due to the flat road sections and stable traffic flow, the preset road segment length can be set to 100 meters to balance analysis efficiency and positioning accuracy.

[0064] It should be noted that the above-mentioned preset road segment length is only a preferred embodiment. The specific implementation parameters of the equal-spacing grid division method can be flexibly adjusted according to the actual length of the target road segment and the complexity of the traffic scenario. This invention does not limit its scope of protection. Any implementation method that divides the target road segment into several road segment units based on the principle of equal spacing is within the scope of protection of this invention.

[0065] In some embodiments of the present invention, the spatiotemporal information includes timestamps, latitude and longitude coordinates, and lane numbers; the spatiotemporal information is used to map the driving behavior characteristics of multiple vehicles to corresponding road segment units and lanes, such as... Figure 5 As shown, it includes: S501. Establish the correlation between multi-vehicle driving behavior characteristics and corresponding spatiotemporal information; S502. Using a map matching algorithm, the timestamp, latitude and longitude coordinates, and lane number are matched with the road segment units and lanes of the target road segment; S503: Map the matched multi-vehicle driving behavior features to the corresponding road segment units and lanes within the target road segment.

[0066] In some embodiments of the present invention, a correlation between multi-vehicle driving behavior characteristics and spatiotemporal information is established. The conventional method of multi-source data association in the prior art is adopted, that is, each driving behavior characteristic of each vehicle (such as emergency braking, lane changing, etc.) is assigned a unique combination of timestamp, latitude and longitude coordinates, and lane number to ensure that all kinds of behavior characteristics can be accurately matched to specific time, location and lane. For example, the "lateral position offset" feature of a vehicle is bound to the timestamp 16:35:20, latitude and longitude coordinates (39.9142°, 116.4174°), and lane number 3, which lays the foundation for subsequent matching work.

[0067] During the matching process, a map matching algorithm suitable for the scenario of this invention can be selected from the existing technology, preferably a point-to-curve matching algorithm or a hidden Markov model (HMM) map matching algorithm. For example, when using the point-to-curve matching algorithm, the latitude and longitude coordinates are first mapped to the road curves on the high-precision map, and the specific lane is locked in combination with the lane number. Then, the corresponding road segment unit is matched according to the road range where the latitude and longitude coordinates are located. When using the hidden Markov model algorithm, the positioning probability of latitude and longitude coordinates, road topology information and lane number can be comprehensively considered to reduce the impact of positioning noise on matching accuracy and adapt to complex traffic scenarios.

[0068] After mapping, the driving behavior features of multiple vehicles will be grouped and standardized according to road segment units and lanes. For example, if multiple vehicles are matched to the 12th road segment unit and lane number 2 within a similar time frame and with similar latitude and longitude coordinates, the extracted features such as "low collision time" and "emergency braking" will be uniformly grouped under the corresponding lane of that road segment unit. This standardized mapping provides a clear data foundation for subsequent cluster analysis and extraction of group driving behavior features of multiple vehicles in the same road segment unit and lane within a preset spatiotemporal window, ensuring accurate discovery of the consistency of group behavior and thus achieving accurate diagnosis of traffic environment anomalies. It should be noted that the above association methods, map matching algorithm selection, and mapping examples are all preferred embodiments based on existing technologies. This invention is not limited to these; any implementation method that can achieve accurate mapping of multi-vehicle driving behavior features to road segment units and lanes can be applied to this invention.

[0069] By associating and mapping the characteristics of multi-vehicle driving behavior with spatiotemporal information to corresponding road segment units and lanes, it is possible to achieve precise spatial positioning and regular collection of driving behavior data, providing reliable data support for subsequent extraction of group driving behavior characteristics and consistency analysis, and effectively improving the positioning accuracy and precision of traffic environment anomaly diagnosis.

[0070] In some embodiments of the present invention, cluster analysis is performed on the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window, including: S601. Perform K-means clustering on the horizontal and vertical features of driving behavior characteristics to obtain the cluster categories and isolated samples of the horizontal features, and the cluster categories and isolated samples of the vertical features. S602. Remove isolated samples with horizontal features and isolated samples with vertical features. S603. Based on the horizontal and vertical feature clustering categories after removing isolated samples, different group driving behavior categories are obtained.

[0071] In some embodiments of this invention, K-means clustering is a commonly used unsupervised clustering algorithm, suitable for the multi-vehicle behavior feature classification requirements of this invention. For example, lateral features can be selected as lateral acceleration and lateral position offset within the lane. During clustering, 3-5 categories are set, grouping features with similar offset trends and accelerations into one category, while those deviating too far from the category center (e.g., exceeding 2 standard deviations) are considered lateral isolated samples. Vertical features can be selected as longitudinal deceleration and collision time. Similarly, samples with abnormally large deceleration and abnormally small collision times are classified as vertical isolated samples. After removing these isolated samples caused by accidental factors such as single vehicle misoperation, the group driving behavior categories such as "stable lateral offset + gentle longitudinal deceleration" and "concentrated lane changes + symmetrical lateral acceleration" can be identified by combining the lateral and vertical clustering categories.

[0072] It should be noted that the clustering analysis of the present invention is not limited to K-means clustering. Other clustering algorithms adapted to the classification of multi-vehicle behavior features in the prior art can also be used, such as the density clustering DBSCAN algorithm that can adaptively identify feature distribution, or the hierarchical clustering algorithm that completes classification through layer-by-layer aggregation. All of the above algorithms can achieve horizontal and vertical feature classification and isolated sample identification, and are applicable to the present invention.

[0073] This embodiment uses horizontal and vertical features to cluster and remove isolated samples, which can filter out the interference of occasional abnormal behavior of a single vehicle, accurately extract the common driving patterns of multiple vehicles, and classify reliable group driving behavior categories. This provides core data support for subsequent diagnosis of traffic environment anomalies based on the consistency of group behavior, and improves the accuracy and reliability of anomaly diagnosis.

[0074] In some embodiments of the present invention, based on the consistency of group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and multiple types of traffic environment anomaly diagnosis results are obtained, including: S701. Based on the concentrated lane-changing behavior of multiple unrelated vehicles within a similar time and space range, it is determined that there is an invisible obstacle. S702. Based on a sudden drop in vehicle density in a lane, a significant increase in density in adjacent lanes, and concentrated lane-changing behavior upstream, it is determined that the lane is temporarily closed or under construction. Vehicle density is the number of vehicles per unit space for the corresponding lane of the target road segment unit within a preset time and space window, based on the location data, lane number, and timestamp uploaded by multiple vehicle terminals. S703. Systematic offset of lateral position distribution when multiple vehicles in the same lane have not changed lanes is determined to be road geometric anomaly or road surface anomaly. S704. Road sections that are identified as dynamically high-risk sections are those where the proportion of emergency braking behavior or collision time that is too short exceeds a preset risk ratio threshold within a specific time and space range.

[0075] In some embodiments of the present invention, the following traffic environment anomaly diagnosis is performed based on the consistency of driving behavior in both the lateral and longitudinal dimensions: (1) Invisible obstacle diagnosis: Based on timestamps and geographic coordinates, spatiotemporal clustering analysis is performed on vehicle lane-changing behavior. If, under the condition of no navigation guidance, multiple unrelated vehicles change lanes within the time window Δt and spatial range Δs of the same road segment, and the lanes they were in before changing lanes are highly concentrated, then it is determined that there are invisible obstacles such as nighttime debris, low obstacles, accident debris, etc. in front of the lane. (2) Lane temporary closure or construction zone diagnosis: Based on timestamps and geographic coordinates, spatiotemporal clustering analysis is performed on vehicle lane changing behavior. If the upstream area of ​​a certain road segment is detected, most vehicles complete lane changing before entering the road segment, and the vehicle density of the lane before lane changing drops sharply to near zero in the road segment, while the density of adjacent lanes increases significantly, then the target lane is determined to be in a temporary closure or construction state. (3) Diagnosis of road geometry or pavement anomalies: Based on the geographical coordinates and lane number, spatial cluster analysis is performed on the lateral position data of the lane. If multiple vehicles that have not changed lanes show a systematic shift in lateral position distribution in the same lane, specifically, the mean of lateral position deviates from the lane centerline by more than a preset threshold and the variance is low, then it is determined that the road section has potholes, unclear markings or abnormal line design. (4) Dynamic high-risk road segment diagnosis: Based on timestamps and geographic coordinates, spatiotemporal clustering analysis is performed on emergency braking behavior and collision time too low behavior. If the proportion of vehicles that trigger emergency braking or collision time too low behavior exceeds the preset proportion threshold in a specific road segment and at a specific time, the road segment is determined to be a dynamic high-risk road segment.

[0076] The above diagnostic results are combined with geographical coordinates, the lane (if any), and the road segment section to form a structured and saved traffic environment anomaly results.

[0077] This embodiment relies on the consistency of driving behavior among multiple vehicles to diagnose anomalies, which can eliminate the accidental interference from individual vehicle operations, achieve accurate and reliable judgment of various traffic environment anomalies, and improve the authenticity and practicality of diagnostic results.

[0078] To better implement the multi-type traffic environment anomaly diagnosis method in the embodiments of the present invention, based on the multi-type traffic environment anomaly diagnosis method, correspondingly, such as Figure 8 As shown, this embodiment of the invention also provides a multi-type traffic environment anomaly diagnosis system. The multi-type traffic environment anomaly diagnosis system 800 includes: The driving data acquisition and association unit 801 is used to collect driving status data in real time through the vehicle terminals of multiple vehicles, extract driving behavior features based on the driving status data, and upload the driving behavior features to the data analysis platform after associating them with spatiotemporal information. The road segment division and behavior mapping unit 802 is used to divide the target road segment into several road segment units through the data analysis platform; and to map the driving behavior characteristics of multiple vehicles to the corresponding road segment units and lanes by using the correlation between spatiotemporal information and driving behavior characteristics. The group behavior clustering and anomaly diagnosis unit 803 is used to perform cluster analysis on the driving behavior characteristics of multiple vehicles within a preset time and space window through a data analysis platform. The driving behavior characteristics include lateral and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to multiple vehicles are obtained. Based on the consistency of the group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and multiple types of traffic environment anomaly diagnosis results are obtained.

[0079] The multi-type traffic environment anomaly diagnosis system 800 provided in the above embodiments can realize the technical solutions described in the above multi-type traffic environment anomaly diagnosis method embodiments. The specific implementation principles of each module or unit can be found in the corresponding content in the above multi-type traffic environment anomaly diagnosis method embodiments, and will not be repeated here.

[0080] This implementation constructs a complete "data-driven, vehicle-cloud collaborative" intelligent early warning system. The system uses a massive number of connected vehicles as real-time, dynamic sensing nodes, extracting and lightweighting high-confidence behavioral features at the vehicle end, ensuring the effectiveness and reliability of the data source. On the data analysis platform, big data clustering and spatiotemporal pattern recognition technologies are used to aggregate discrete micro-behaviors into a group-consistent signal reflecting the road environment state, thereby achieving early detection and accurate diagnosis of various hidden and sudden anomalies. Finally, through a low-latency communication network, structured early warning information is combined with high-precision maps and real-time vehicle trajectories to achieve personalized and scenario-based proactive safety services. This system not only significantly improves the perception and decision-making capabilities of individual vehicles but also forms a self-evolving and continuously optimizing "herd immunity" mechanism for road safety, providing core capability support for the safety protection of future intelligent transportation systems. Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0081] The above provides a detailed description of the various traffic environment anomaly diagnosis methods and system media provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this invention. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A method for diagnosing multiple types of traffic environment anomalies, characterized in that, include: Real-time driving status data is collected through vehicle-mounted terminals in multiple vehicles. Driving behavior features are extracted based on the driving status data, and the driving behavior features are correlated with spatiotemporal information and then uploaded to the data analysis platform. The target road segment is divided into several road segment units using the data analysis platform. By utilizing the spatiotemporal information and the correlation between the driving behavior characteristics, the driving behavior characteristics of multiple vehicles are mapped to corresponding road segment units and lanes; The data analysis platform performs cluster analysis on the driving behavior characteristics of multiple vehicles within a preset time and space window. The driving behavior characteristics include lateral features and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to multiple vehicles are obtained; Based on the consistency of group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and diagnostic results for multiple types of traffic environment anomalies are obtained.

2. The method according to claim 1, characterized in that, The method further includes: The data analysis platform is used to match and associate the diagnostic results of the various traffic environment anomalies with the high-precision map to the corresponding road segment units and lanes, thereby obtaining the abnormal road segment units and abnormal lanes. When a target vehicle enters the abnormal road segment or abnormal lane within a preset time, the target vehicle's navigation software sends a graphic and voice warning message containing the type of traffic environment anomaly and driving suggestions to the target vehicle.

3. The method according to claim 2, characterized in that, The diagnostic results of various traffic environment anomalies are matched with high-precision maps and associated with corresponding road segment units and lanes, which is achieved through map matching algorithms.

4. The method according to claim 1, characterized in that, The driving status data includes lane number, collision time, longitudinal deceleration, lateral acceleration, current lateral position within the lane, and latitude and longitude coordinates; the driving behavior characteristics include emergency braking, excessively short collision time, lane changing, and lateral position deviation within the lane. The extraction of driving behavior features based on the driving state data includes: When the absolute value of the longitudinal deceleration is greater than a preset braking deceleration threshold, emergency braking behavior features are extracted. When the collision time is less than a preset collision time threshold, extract the behavior feature of excessively low collision time. When the lane number changes, and the lateral position and latitude and longitude coordinates of the current lane are synchronously matched with the lane switching trajectory, and the lateral acceleration shows a symmetrical temporal pattern of first increasing and then decreasing, the lane changing behavior features are extracted. When the lateral position of the current lane exceeds the preset normal lateral position range within the lane and continues to exceed the preset offset time, the lateral position offset behavior feature within the lane is extracted.

5. The method according to claim 4, characterized in that, The behavior characteristic of excessively low collision time is used to characterize the potential rear-end collision risk between the vehicle and the vehicle in front. The preset collision time threshold is adjusted in conjunction with the target vehicle's current speed; the higher the current speed, the larger the preset collision time threshold; the lower the current speed, the smaller the preset collision time threshold.

6. The method according to claim 1, characterized in that, The process of dividing the target road segment into several road segment units includes: Based on the equal-spacing grid division method, the target road segment is divided into several road segment units according to the preset road segment length.

7. The method according to claim 1, characterized in that, The spatiotemporal information includes timestamps, latitude and longitude coordinates, and lane numbers; the step of mapping the driving behavior characteristics of multiple vehicles to corresponding road segment units and lanes using the spatiotemporal information and the association relationship between the driving behavior characteristics includes: Establish a correlation between the characteristics of multi-vehicle driving behavior and corresponding spatiotemporal information; The timestamp, latitude and longitude coordinates, and lane number are matched with the road segment units and lanes of the target road segment using a map matching algorithm. The matched multi-vehicle driving behavior features are mapped to the corresponding road segment units and lanes within the target road segment.

8. The method according to claim 1, characterized in that, The clustering analysis of the driving behavior characteristics of multiple vehicles within a preset spatiotemporal window includes: K-means clustering was performed on the horizontal and vertical features of driving behavior characteristics to obtain the cluster categories and isolated samples of the horizontal features, and the cluster categories and isolated samples of the vertical features. Isolated samples of the horizontal features and isolated samples of the vertical features are removed. Based on the horizontal and vertical feature clustering categories after removing isolated samples, different categories of group driving behavior are obtained.

9. The method according to claim 1, characterized in that, Based on the consistency of group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and diagnostic results for these anomalies are obtained, including: The presence of an invisible obstacle was determined based on the concentrated lane-changing behavior of multiple unrelated vehicles within a similar time and space range; The lane is temporarily closed or under construction based on a sudden drop in vehicle density in a certain lane, a significant increase in density in adjacent lanes, and concentrated lane-changing behavior upstream. The vehicle density is the number of vehicles per unit space for the corresponding lane of the target road segment unit within a preset time and space window, based on the location data, lane number, and timestamp uploaded by multiple vehicle terminals. The systematic offset of the lateral position distribution of multiple vehicles in the same lane without changing lanes is determined to be a road geometric anomaly or a road surface anomaly. A road segment is identified as a dynamically high-risk section if the proportion of emergency braking or collisions with excessively short time exceeds a preset risk ratio threshold within a specific time and space range.

10. A multi-type traffic environment anomaly diagnosis system, characterized in that, include: The driving data acquisition and association unit is used to collect driving status data in real time through the vehicle terminals of multiple vehicles, extract driving behavior features based on the driving status data, and upload the driving behavior features to the data analysis platform after associating them with spatiotemporal information. The road segment division and behavior mapping unit is used to divide the target road segment into several road segment units through the data analysis platform; By utilizing the spatiotemporal information and the correlation between the driving behavior characteristics, the driving behavior characteristics of multiple vehicles are mapped to corresponding road segment units and lanes; The group behavior clustering and anomaly diagnosis unit is used to perform clustering analysis on the driving behavior characteristics of multiple vehicles within a preset time and space window through the data analysis platform. The driving behavior characteristics include lateral features and longitudinal features. Based on the cluster analysis, the group driving behavior characteristics corresponding to multiple vehicles are obtained; Based on the consistency of group driving behavior characteristics, multiple types of traffic environment anomalies are diagnosed and diagnostic results for multiple types of traffic environment anomalies are obtained.

Images