Traffic anomaly multi-index attribution analysis method and system based on intra-group and inter-group distance

Abstract

The invention belongs to the field of traffic control, and provides a traffic anomaly multi-index attribution analysis method and system based on intra-group and inter-group distances in order to solve the problem that suggestions fitting the reality cannot be provided for traffic control and emergency command in the prior art. The traffic anomaly multi-index attribution analysis method based on the intra-group and inter-group distance comprises the following steps: calculating a comprehensive safety index according to a quantized traffic safety index and a corresponding weight value thereof; when monitoring that the comprehensive safety index in the current period abnormally fluctuates, comparing the traffic safety index in the current period with the traffic safety index in the previous normal period, and calculating the difference of the traffic safety indexes in the groups and the dispersion degree of the traffic safety indexes in the groups; the contribution degree of each traffic safety index to the change of the comprehensive safety index is quantified, leading factors are accurately identified, and finally an interpretable attribution report is generated, so that multi-index attribution analysis can be performed on traffic abnormity in a quantified and interpretable manner, and instructive suggestions are provided for traffic control and emergency command.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of traffic control, and particularly relates to a traffic anomaly multi-index attribution analysis method and system based on intra-group and inter-group distance. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute prior art.

[0003] By using Internet of Things sensors, cameras and other devices, multi-dimensional traffic flow data (such as average vehicle speed, traffic volume, time occupancy, vehicle spacing, lane change frequency, etc.) of a road section can be continuously collected. When a traffic accident, abnormal congestion, severe weather and other sudden conditions occur, how to quickly and accurately locate the key dominant factors leading to the occurrence of the conditions from the massive monitoring data is a core challenge to realize precise traffic control and emergency command.

[0004] At present, the analysis methods for traffic anomalies include threshold alarm method, statistical analysis method and machine learning model. Among them, the threshold alarm method sets a fixed threshold for a single index (such as vehicle speed below 30 km / h). This method is simple but extremely rough, cannot handle complex situations with multiple index coupling, cannot quantify the contribution of each index, has a high false alarm rate, and cannot answer the question of "which index is the main cause". Statistical analysis (such as correlation analysis) finds abnormal indexes by calculating the deviation (such as Z-score) of each index in the abnormal period from the normal state, but can only reflect the change amplitude and cannot reveal the causal relationship. A large change in an index may only be "effect" but not "cause", and the contributions of multiple indexes cannot be effectively sorted and quantified. Machine learning models (such as classification models) judge traffic states by training models (such as SVM, random forest). These models are "black box" or "gray box" and lack interpretability. Even if it can judge "abnormality", it is difficult to clearly and quantitatively explain which features (indexes) and their respective importance caused the abnormality, i.e. it does not have "attribution" ability, making it difficult for traffic management personnel to understand and trust, and thus unable to guide specific control measures.

[0005] In summary, the prior art lacks an effective means for quantitatively and interpretively analyzing the multi-index attribution of traffic anomalies, and cannot provide practical suggestions for traffic control and emergency command. SUMMARY

[0006] To solve the above technical problems, the present application provides a traffic anomaly multi-index attribution analysis method and system based on intra-group and inter-group distance, which can quantitatively and interpretively analyze the multi-index attribution of traffic anomalies and provide guidance suggestions for traffic control and emergency command.

[0007] In order to achieve the above object, the present application adopts the following technical solutions: The first aspect of the present application provides a traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance.

[0008] In one or more embodiments, a traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance is provided, comprising: According to the point cloud data of the set road section, the vehicle trajectory across frames is established, the continuous motion information of each vehicle is obtained, and then the multi-dimensional monitoring index is calculated in real time; the multi-dimensional monitoring index includes average vehicle speed, traffic flow, time occupancy rate, average vehicle head time interval and lane speed difference; According to the multi-dimensional monitoring index, the traffic safety index is constructed and quantified, and the comprehensive safety index is calculated according to the quantified traffic safety index and the corresponding weight value; the traffic safety index includes following risk, lane changing risk and speed risk; When the comprehensive safety index in the current period is monitored to appear abnormal fluctuation, the traffic safety index of the current period and the last normal period is compared, the difference of each traffic safety index between groups and the dispersion degree of each traffic safety index in the group is calculated, the contribution degree of each traffic safety index to the change of the comprehensive safety index is quantified, and then the dominant factor is identified and the explainable attribution report is generated.

[0009] As an implementation mode, the process of quantifying the contribution degree of each traffic safety index to the change of the comprehensive safety index is: For each traffic safety index, the attribution score is calculated according to the ratio of the inter-group difference to the intra-group dispersion degree; The attribution score of each traffic safety index is normalized to obtain its contribution degree and is sorted according to the contribution degree.

[0010] As an implementation mode, the contribution degree of each traffic safety index to the change of the comprehensive safety index is as follows: ; Wherein, , are the mean and variance of the ith traffic safety index in the abnormal group respectively; , are the mean and variance of the ith traffic safety index in the normal group respectively.

[0011] As an implementation mode, the corresponding weight of each traffic safety index is determined by calculating the information entropy of each traffic safety index.

[0012] As an implementation mode, the following risk is represented by the vehicle head time interval THW and the collision time TTC; THW=D / V_ego; When V_ego>V_front, TTC=D / (V_ego-V_front) is established; Wherein, D is the distance between the ego vehicle and the front vehicle, V_ego is the ego vehicle speed, and V_front is the front vehicle speed.

[0013] As an embodiment, the calculation formula of lane changing risk is: Lane changing risk = alpha x f(TTC_front) + beta x f(TTC_rear); When the current vehicle speed is greater than the front vehicle, TTC_front=d_front / (v_ego-v_front); When the rear vehicle speed is greater than the ego vehicle, TTC_rear=d_rear / (v_rear-v_ego); f(TTC)=1 / TTC; Wherein, TTC is the collision time; TTC_front and TTC_rear are the collision times of the front and rear vehicles of the target lane respectively; f() is the inverse function of the collision time; d_front and d_rear are the distances of the front and rear vehicles of the target lane respectively; V_ego is the ego vehicle speed, V_front is the front vehicle speed; alpha and beta are the weight values corresponding to the inverse function of the collision time; alpha+beta =1.

[0014] As an embodiment, the speed risk is represented by the speed fluctuation rate of the road section , and the expression is: ; ; Wherein, M is the total number of effective vehicles passing through the road section in a time period; Indicates the average acceleration and deceleration intensity of the vehicle, and the greater the value, the more unstable the driving behavior; is the instantaneous speed of the vehicle at the t time; is the instantaneous speed of the vehicle at the t-1 time; N is the effective sampling point number of the vehicle in the window T.

[0015] The second aspect of the application provides a traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance.

[0016] In one or more embodiments, a traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance comprises: A multi-dimensional monitoring index calculation module is configured to establish a vehicle trajectory across frames according to point cloud data of a set road section, obtain continuous motion information of each vehicle, and further calculate multi-dimensional monitoring indexes in real time; the multi-dimensional monitoring indexes include average vehicle speed, vehicle flow, time occupancy rate, average vehicle headway, and lane-to-lane speed difference. a comprehensive safety index calculation module configured to construct and quantify traffic safety indexes according to multi-dimensional monitoring indexes, and calculate a comprehensive safety index according to the quantified traffic safety indexes and corresponding weight values; the traffic safety indexes include car-following risk, lane-changing risk and speed risk; an index attribution analysis module configured to compare traffic safety indexes of a current period with those of a last normal period when an abnormal fluctuation of the comprehensive safety index in the current period is monitored, quantify a contribution of each traffic safety index to a change in the comprehensive safety index by calculating a difference between groups and a dispersion within a group of each traffic safety index, and further identify a dominant factor and generate an interpretable attribution report.

[0017] A third aspect of the present application provides a computer-readable storage medium.

[0018] A computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps in the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance as described above.

[0019] A fourth aspect of the present application provides an electronic device.

[0020] An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the steps in the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance as described above when executing the program.

[0021] Compared with the prior art, the present application has the following advantages: The present application constructs and quantifies traffic safety indexes according to multi-dimensional monitoring indexes, and further calculates a comprehensive safety index in combination with corresponding weight values, compares traffic safety indexes of a current period with those of a last normal period when an abnormal fluctuation of the comprehensive safety index in the current period is monitored, quantifies a contribution of each traffic safety index to a change in the comprehensive safety index by calculating a difference between groups and a dispersion within a group of each traffic safety index, and further identifies a dominant factor and generates an interpretable attribution report. The present application uses normal data of the last period as a baseline, effectively eliminates the influence of external factors such as season, week, and weather, and ensures that the attribution result focuses on the real abnormal fluctuation. Moreover, the intra-group dispersion is introduced as a denominator to automatically punish indexes that are naturally volatile and unstable, avoid misjudgment, and make the attribution result more robust and reliable. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The embodiments of these drawings illustrate the preferred embodiments of the present application and, together with the description, serve to explain the principles of the present application.

[0023] Figure 1 is a traffic anomaly multi-index attribution analysis method flowchart based on inter-group distance within groups of embodiments of the present application; Figure 2 is a traffic anomaly multi-index attribution analysis system structure schematic diagram based on inter-group distance within groups of embodiments of the present application; Figure 3 is a schematic diagram of an electronic device of an embodiment of the present application. DETAILED DESCRIPTION

[0024] The present application will be further described below with reference to the accompanying drawings and embodiments.

[0025] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present application. Unless otherwise indicated, all technical and scientific terms used herein have the same meaning as would be understood by one of ordinary skill in the art to which the present application pertains.

[0026] It should be noted that the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit exemplary embodiments according to the present application. As used herein, the singular form is intended to include the plural form unless the context clearly indicates otherwise, and it should also be understood that when the terms "comprise" and / or "include" are used in the specification, they refer to the presence of a feature, step, operation, device, component, and / or combinations thereof.

[0027] Term explanation: Anomalous event: refers to an abnormal event that suddenly occurs and deviates from the normal state in highway traffic operation. Its manifestations include but are not limited to: severe congestion, traffic accidents, vehicle breakdown, adverse weather influence (such as fog, heavy rain and snow), cargo spillage and other situations that cause traffic flow state to deteriorate sharply.

[0028] Multi-dimensional monitoring metrics: refers to multi-dimensional data collected by sensor networks deployed along the highway for depicting traffic flow state. Its core indicators include: Average speed: the average value of the speed of all vehicles passing through the road section within a specified time period.

[0029] Volume: the total number of vehicles passing through the road section within a specified time period.

[0030] Time occupancy: the percentage of time that the sensor is occupied by vehicles within a specified time period, which is a direct indicator of traffic density.

[0031] Average Time Gap / Distance Gap: The average value of time or space gap between adjacent vehicles.

[0032] Speed Difference between Lanes: The difference of average vehicle speed in different lanes, used to evaluate the stability and conflict risk of traffic flow.

[0033] Attribution Analysis: In this invention, a computational process used to quantify the contribution degree of each monitoring indicator to a specific incident, and to rank their importance, aiming to identify the dominant and secondary factors leading to the incident.

[0034] Within-group / Between-group Distance: A statistical term applied in this field.

[0035] Between-group Variability: The difference between the mean value of a monitoring indicator in the “normal state” data set and the mean value in the “abnormal state” data set. The larger this value, the greater the deviation of the indicator when the state changes.

[0036] Within-group Variability: The dispersion (usually measured by variance) of a monitoring indicator within its own data set, either “normal state” or “abnormal state”. The larger this value, the greater the natural fluctuation of the indicator itself.

[0037] Contribution Degree: A quantitative output result defined, which is calculated by the within-group / between-group distance model, and used to represent the impact size of a single monitoring indicator on the current incident. This value can be further normalized into a percentage form for comparison and ranking.

[0038] Historical Normal State Dataset: The baseline data used for comparison with the current “abnormal state”. This data set is not arbitrary historical data, but a collection of traffic flow data identified as “normal state” in a historical period with similar conditions (such as the same type of week, the same time period, similar weather) as the current abnormal occurrence time.

[0039] Comprehensive State Index: A composite index (e.g., congestion index) used to comprehensively evaluate the current traffic state. This index can be formed by the weighted fusion of multiple basic monitoring indicators, or composed of the anomaly scores output by the machine learning model, and is the basis for triggering the attribution analysis.

[0040] DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a density-based spatial clustering algorithm. It can identify clusters of arbitrary shape and effectively handle noisy data.

[0041] Kalman filtering is an algorithm for predicting and estimating the optimal state of a dynamic system from a series of observation data containing noise.

[0042] Figure 1 The principle diagram of the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance of the embodiment of the application is given. According to Figure 1 , the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance of the embodiment can include the following steps S101~ S103.

[0043] The specific implementation process of steps S101~ S103 is as follows: Step S101: Establish vehicle trajectories across frames according to point cloud data of the set road section, obtain continuous motion information of each vehicle, and then calculate multi-dimensional monitoring indicators in real time; the multi-dimensional monitoring indicators include average vehicle speed, vehicle flow, time occupancy rate, average vehicle headway, and lane-to-lane speed difference.

[0044] In one or more embodiments, a multi-line beam lidar (such as 16 lines, 32 lines, or higher) deployed on a highway roadside column or gantry is used as the core perception device to achieve all-weather, high-precision, three-dimensional perception of traffic flow. The specific implementation steps are as follows: Step S1011: Raw point cloud data acquisition; The lidar continuously emits laser beams at a set frequency (such as 10 Hz) and receives reflected signals from the road surface, vehicles, and other objects, generating a raw point cloud data stream containing three-dimensional coordinates (X, Y, Z), reflection intensity, and time stamp.

[0045] Step S1012: Point cloud preprocessing; Filter and denoise the raw point cloud (remove interference points such as rain, snow, and birds) and ground segmentation to separate the dynamic target point cloud clusters above the road surface.

[0046] Step S1013: Vehicle target detection and tracking; The dynamic point cloud clusters are segmented using a clustering algorithm (such as DBSCAN) to identify individual vehicle targets. Subsequently, through Kalman filtering or a multi-target tracking algorithm (such as the point cloud adapted version of SORT, DeepSORT), vehicle trajectories across frames are established to obtain continuous motion information for each vehicle.

[0047] Step S1014: Multi-dimensional monitoring index calculation; Based on the vehicle trajectory data obtained through tracking, the following core monitoring indicators are calculated in real time within a set road section analysis unit (such as a 500-meter road section) and time window (such as 1 minute): Average vehicle speed : the arithmetic mean of the instantaneous speeds of all tracked vehicles within the statistical window; ； is the instantaneous speed of the ith vehicle; N is the total number of effective vehicles passing through the road section within the time window; Vehicle flow (Volume): the total number of vehicles passing through the downstream section of the road within a unit of time.

[0048] Volume = N; N is the number of vehicles that successfully pass through the downstream detection section within a unit of time (such as 1 minute).

[0049] Time occupancy (Time Occupancy): calculate the proportion of space occupied by vehicles within the laser radar detection range over time, and integrate to obtain the time occupancy.

[0050] ; is the time when the ith vehicle completely occupies a certain space position; T is the total observation time (such as 60 seconds).

[0051] Average time headway (Average Time Headway, THW) / distance gap (Average Distance Gap, ADG): the average value of the time interval or spatial distance between adjacent vehicles in the same lane when passing through a certain reference section.

[0052] ; is the time when the front vehicle passes through the section; is the time when the rear vehicle passes through the same section; M is the number of effective paired following samples.

[0053] ; Space distance between the jth pair of front and rear vehicles (unit: meter).

[0054] Speed difference between lanes: Calculate the average speed of vehicles in each lane, and then take the difference between the maximum and minimum values.

[0055] ; The average speed of all vehicles in the kth lane; k = 1, 2,...; K is the total number of lanes (e.g. K = 3 or 4).

[0056] Step S102: According to the multi-dimensional monitoring indicators, traffic safety indicators are constructed and quantified, and according to the quantified traffic safety indicators and their corresponding weight values, a comprehensive safety indicator is calculated; the traffic safety indicators include car-following risk, lane-changing risk and speed risk. Using laser radar vehicle tracking trajectory data, high-risk driving behaviors are identified, and their risk values, i.e. traffic safety indicators, are quantified, including car-following risk (Car-Following Risk), lane-changing risk (Lane-Changing Risk) and speed risk (Speed Risk).

[0057] Specifically, the car-following risk is characterized by the headway THW and the time to collision TTC; THW = D / V_ego; When V_ego > V_front, TTC = D / (V_ego - V_front) holds; Where D is the distance between the ego vehicle and the front vehicle, V_ego is the ego vehicle speed, and V_front is the front vehicle speed.

[0058] Physical meaning and risk explanation: THW is too small, indicating that the driver's reaction time is insufficient. TTC is a direct measure of the degree of impending collision, and the smaller the value, the higher the risk. TTC < 3s is usually considered to be extremely high risk.

[0059] Specifically, the calculation formula of the lane-changing risk is: Lane-changing risk = a x f(TTC_front) + b x f(TTC_rear); When the current vehicle speed is greater than the front vehicle, TTC_front = d_front / (v_ego - v_front); When the rear vehicle speed is greater than the ego vehicle, TTC_rear = d_rear / (v_rear - v_ego); f(TTC) = 1 / TTC; Where TTC is the time to collision; TTC_front and TTC_rear are the time to collision with the front and rear vehicles in the target lane respectively; f() is the inverse function of time to collision; d_front and d_rear are the distance to the front and rear vehicles in the target lane respectively; V ego is the ego vehicle speed, V front is the front vehicle speed; a and b are the weight values corresponding to the inverse function of time to collision; a + b = 1. a and b can be adjusted according to the scene (e.g., focusing on the front vehicle at high speed and focusing on the rear vehicle in urban areas).

[0060] Specifically, the speed risk is represented by the speed fluctuation rate of the road segment , and its expression is: ; ; Where M is the total number of effective vehicles passing through the road segment within the time period; represents the average acceleration and deceleration intensity of the vehicle, and the larger the value, the more unstable the driving behavior; is the instantaneous speed of the vehicle at time t; is the instantaneous speed of the vehicle at time t-1; N is the number of effective sampling points of the vehicle within the window T.

[0061] The micro-individual risk is aggregated within a specific time window (e.g., 1 minute) and spatial range (e.g., 500-meter road segment) to form an index representing the overall safety level of the road segment.

[0062] The number of events with TTC < 3s is counted for all vehicles within the time window; the number of events with THW < 1s is counted; the number of aggressive lane changes exceeding a certain threshold is counted; to avoid bias under different traffic, the above numbers can be divided by the total number of vehicles in the time period to obtain the standardized risk event rate (e.g., the number of high-risk following events per 100 vehicles).

[0063] The comprehensive safety index of the road segment is calculated by entropy weight method, and the process is as follows: Step a: Construct the evaluation matrix; each row is a time window, and each column is a traffic safety index (following risk, lane changing risk and speed risk); Step b: Data standardization; eliminate the dimension influence of different indicators; calculate the information entropy: calculate the information entropy of each indicator. The greater the difference in indicator value (i.e., the greater the fluctuation, the more information provided), the smaller the entropy; determine the weight, the smaller the entropy, the greater the weight of the indicator.

[0064] Step c: Calculate the comprehensive score Safety_Score, i.e., the comprehensive safety index; Safety_Score = w1*I1 + w2*I2 + w3*I3 +...; wherein: I1, I2, I3 are the normalized values of each risk indicator; w1, w2, w3 are the corresponding weights calculated by entropy weight method; Safety state discrimination and visualization: Divide the safety level, according to the long-term historical data, the dynamic threshold (threshold 1, threshold 2 and threshold 3) can be set, and the safety indicators are divided into different levels: Safety (green representation can be used): Safety_Score < threshold 1. Traffic flow is stable, and risk events are rare.

[0065] Attention (yellow representation can be used): threshold 1 ≤ Safety_Score < threshold 2. Risk events begin to appear and need attention.

[0066] Warning (orange representation can be used): threshold 2 ≤ Safety_Score < threshold 3. High-risk events occur frequently, and the probability of accidents increases significantly.

[0067] Danger (red representation can be used): Safety_Score ≥ threshold 3. In an extremely dangerous state, traffic accidents are likely to occur.

[0068] Step S103: When the abnormal fluctuation of the comprehensive safety indicator in the current period is monitored (for example, the comprehensive safety indicator drops sharply), the traffic safety indicators of the current period and the last normal period are compared, the difference between each traffic safety indicator in the group and the dispersion degree in the group are calculated, the contribution of each traffic safety indicator to the change of the comprehensive safety indicator is quantified, and the dominant factor is identified and an interpretable attribution report is generated.

[0069] When the comprehensive safety indicator drops sharply, the above step S103 can be performed to quickly locate the specific traffic safety indicator that causes the comprehensive safety indicator to drop sharply (such as: following risk surge), and analyze the specific reasons, such as large vehicles mixing, weather reasons or traffic accidents themselves.

[0070] Analysis dimension: month dimension (M). The same can be applied to day dimension (D).

[0071] Data grouping: Anomaly group (G_anomaly): defined as the current period in which the indicator fluctuates, i.e. all data in the current month.

[0072] Normal group (G_normal): defined as the most comparable historical period as the current period, i.e. all data in the last month.

[0073] Risk factor (i): a set of sub-indicators that constitute the comprehensive safety indicator Safety_Score, for example, i = 1 for following risk; i = 2 for lane-changing conflict rate; i = 3 for speed risk.

[0074] Wherein, the process of quantifying the contribution degree of each traffic safety indicator to the change of the comprehensive safety indicator is: For each traffic safety indicator, calculate its attribution score according to the ratio of inter-group difference to intra-group dispersion; normalize the attribution score of each traffic safety indicator to obtain its contribution degree and sort it according to the contribution degree.

[0075] The contribution degree (i.e. attribution score AttributionScore) of each traffic safety indicator to the change of the comprehensive safety indicator is: The formula is: ; Wherein, , are the mean and variance of the i-th traffic safety indicator in the abnormal group, respectively; , are the mean and variance of the i-th traffic safety indicator in the normal group, respectively.

[0076] Numerator (inter-group difference): measures the average shift amplitude of the traffic safety indicator between this month and last month. The larger the value, the more significant the change.

[0077] Denominator (intra-group dispersion): measures the inherent fluctuation level of the traffic safety indicator in the respective group. The larger the value, the less stable the traffic safety indicator itself, and the less significant the change.

[0078] Normalize the attribution score of each traffic safety indicator to obtain its contribution degree (ContributionDegree, i), and sort it according to the contribution degree.

[0079] ; Wherein n is the total number of traffic safety indicators. Directly reflects the contribution proportion of the i-th risk factor to the fluctuation of the comprehensive safety indicator this time, the larger the value, the greater the contribution.

[0080] The application will be further described in conjunction with the embodiments.

[0081] Background: this month's safety indicator Safety_Score_M1 is 72 (dangerous), last month's Safety_Score_M0 is 60 (caution), and the indicator has deteriorated ΔS = +12.

[0082] Table 1 Risk factors and their rankings;​

[0083] According to the analysis of Table 1, the main reason for the deterioration of traffic safety this month (AS = +12) is the sharp increase in following risk, which contributes up to 58.1%. Specifically, the frequency of high-risk following events with TTC < 3 seconds is significantly higher than last month. The secondary reason is the difference in speed risk, i.e., the increase in speed difference (contribution 25.2%), indicating an increase in behaviors such as sudden acceleration and sudden deceleration in traffic flow. It is recommended to focus on checking whether there are bottleneck points (such as newly appearing road occupation construction, failed traffic lights, etc.) that cause frequent start-stop of traffic flow this month, and to carry out special governance of close following on high-risk road sections.

[0084] The output of the present application is not only data, but also a clear guide directly pointing to governance actions (for example: following behavior should be focused on control), forming a complete closed loop of "monitoring-diagnosis-governance". The present application organically combines micro perception, macro index construction and advanced attribution model, forming a complete, closed-loop, intelligent traffic safety diagnosis and evaluation system, which embodies significant technical integration innovation.

[0085] As shown in Figure 2 The traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance provided by the embodiment of the present application can be realized in a software manner, and the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance includes the following software modules: a multi-dimensional monitoring index calculation module 201, a comprehensive safety index calculation module 202, and an index attribution analysis module 203.

[0086] The functions of each software module in the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance will be introduced as follows: The multi-dimensional monitoring index calculation module 201 is used to establish vehicle trajectories across frames according to point cloud data of a set road section, obtain continuous motion information of each vehicle, and then calculate multi-dimensional monitoring indexes in real time; the multi-dimensional monitoring indexes include average vehicle speed, traffic volume, time occupancy, average headway, and lane-to-lane speed difference; The comprehensive safety index calculation module 202 is used to construct traffic safety indexes and quantify them according to the multi-dimensional monitoring indexes, and calculate a comprehensive safety index according to the quantified traffic safety indexes and corresponding weight values; the traffic safety indexes include following risk, lane-changing risk, and speed risk; The index attribution analysis module 203 is used to compare traffic safety indexes of the current period with those of the last normal period when abnormal fluctuations of the comprehensive safety index in the current period are monitored, quantify the contribution of each traffic safety index to the change of the comprehensive safety index by calculating the difference between the traffic safety indexes in the inter-group and the dispersion degree in the intra-group, and then identify the dominant factor and generate an interpretable attribution report.

[0087] It should be noted that each module in the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance in the embodiment of the application corresponds to each step in the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance in the above embodiment one by one, and the specific implementation process is the same, which will not be repeated here.

[0088] The structure of the electronic device in the embodiment of the application will be described in detail below. Figure 3 The schematic diagram of the component structure of the electronic device provided in the embodiment of the application can be understood as follows. Figure 3 Only exemplary structures of the electronic device are shown, not all structures, and part or all of the structures shown can be implemented as needed.

[0089] The electronic device provided in the embodiment of the application includes at least one processor 301, a memory 302, a user interface 303, and at least one network interface 304. Each component in the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance is coupled together through a bus system 305. It can be understood that the bus system 305 is used to realize the connection and communication between the components. In addition to the data bus, the bus system 305 also includes a power bus, a control bus, and a status signal bus. However, for the purpose of clear illustration, all kinds of buses are marked as the bus system 305 in the Figure 3

[0090] The user interface 303 can include a display, a keyboard, a mouse, a trackball, a click wheel, a key, a button, a touchpad, or a touch screen, etc.

[0091] It can be understood that the memory 302 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. The memory 302 in the embodiment of the application can store data to support the operation of the terminal. Examples of these data include any computer programs for operating on the terminal, such as an operating system and an application program. The operating system contains various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application program can include various application programs.

[0092] ​In some embodiments, the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance provided by the embodiments of the present application can be implemented in a combination of software and hardware. For example, the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance provided by the embodiments of the present application can be a processor in the form of a hardware decoding processor programmed to execute the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance provided by the embodiments of the present application. For example, the processor in the form of a hardware decoding processor can be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic elements.

[0093] For example, the processor 301 can be an integrated circuit chip with a processing capability of signals, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor.

[0094] As an example of the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance implemented in hardware provided by the embodiments of the present application, the apparatus provided by the embodiments of the present application can directly use the processor 301 in the form of a hardware decoding processor to complete the execution, for example, by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic elements to execute the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance provided by the embodiments of the present application.

[0095] The memory 302 in the embodiments of the present application is used to store various types of data to support the operation of the traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance, or to store the instructions for executing the traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance provided by the embodiments of the present application. Figure 1program code of the method. Examples of such data include any executable code such as an executable instruction for operating on a traffic anomaly multi-index attribution analysis system based on inter / intra-group distance, and a program implementing the traffic anomaly multi-index attribution analysis method based on inter / intra-group distance according to the embodiments of the present application can be included in the executable instruction.

[0096] In particular, according to the embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments of the present application include a computer program product comprising a computer program which is carried on a computer readable medium, and the computer program comprises instructions for executing the various functions defined in the apparatus of the present application. Figure 1 program code of the method. In such embodiments, the computer program can be downloaded and installed from a network by the communication section, and / or installed from a detachable medium. When the computer program is executed by the central processing unit, various functions defined in the apparatus of the present application are executed.

[0097] The present application is described with reference to the flowcharts and / or block diagrams of the methods, apparatus (system) and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one flow or multiple flows and / or blocks Figure 1 Figure 1 an apparatus that implements the functions specified in one block or multiple blocks.

[0098] The above merely describes the preferred embodiments of the present application, and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A traffic anomaly multi-index attribution analysis method based on intra-group and inter-group distance, characterized in that, Comprise: According to the point cloud data of the set section, the vehicle trajectory across frames is established, the continuous motion information of each vehicle is obtained, and then the multi-dimensional monitoring index is calculated in real time; The multi-dimensional monitoring index includes average vehicle speed, traffic flow, time occupancy, average headway and lane speed difference; According to the multi-dimensional monitoring index, the traffic safety index is constructed and quantified, and the comprehensive safety index is calculated according to the quantified traffic safety index and the corresponding weight value; The traffic safety index includes following risk, lane changing risk and speed risk; When the comprehensive safety index in the current period is monitored to appear abnormal fluctuation, the traffic safety index of the current period and the last normal period is compared, the difference of each traffic safety index between groups and the dispersion degree in the group is calculated, the contribution degree of each traffic safety index to the change of the comprehensive safety index is quantified, and then the dominant factor is identified and the explainable attribution report is generated. 2.The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, The process of quantifying the contribution degree of each traffic safety index to the change of the comprehensive safety index is: For each traffic safety index, the attribution score is calculated according to the ratio of the difference between groups and the dispersion degree in the group; The attribution score of each traffic safety index is normalized to obtain its contribution degree and sorted according to the contribution degree. 3.The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, Contribution of each traffic safety index to change of comprehensive safety index The formula is: ； wherein, , respectively the mean and variance of the i-th traffic safety indicator in the abnormal group; , respectively the mean and variance of the i-th traffic safety indicator in the normal group. 4.The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, The corresponding weight of each traffic safety index is determined by calculating the information entropy of each traffic safety index. 5.The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, The following risk is represented by the headway THW and the collision time TTC; THW=D / V_ego; When V_ego>V_front, TTC=D / (V_ego-V_front) is established; Where D is the distance between the ego vehicle and the front vehicle, V_ego is the ego vehicle speed, and V_front is the front vehicle speed. 6.The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, The calculation formula of lane changing risk is: Lane changing risk=α×f(TTC_front)+β×f(TTC_rear); When the current speed is greater than the front vehicle, TTC_front=d_front / (v_ego-v_front); When the rear vehicle speed is greater than the vehicle, TTC_rear=d_rear / (v_rear-v_ego); f(TTC)=1 / TTC; Where TTC is the collision time; TTC_front and TTC_rear are the collision times of the front and rear vehicles of the target lane respectively; f() is the inverse function of the collision time; d_front and d_rear are the distances of the front and rear vehicles of the target lane respectively; V_ego is the ego vehicle speed, V_front is the front vehicle speed; α and β are the weight values corresponding to the inverse function of the collision time; α+β =1.

7. The traffic anomaly multi-index attribution analysis method based on inter-group distance in groups according to claim 1, wherein, Speed risk is characterized by speed fluctuation rate at link level which is expressed as: ； ； Wherein, M is the total number of effective vehicles passing through the road section within the time period; represents the average acceleration and deceleration intensity of the vehicle, and the larger the value is, the more unstable the driving behavior is; is the instantaneous speed of the vehicle at the tth moment; is the instantaneous speed of the vehicle at the (t-1)th moment; and N is the number of effective sampling points of the vehicle within the window T.

8. A traffic anomaly multi-index attribution analysis system based on intra-group and inter-group distance, characterized in that, Based on the traffic anomaly multi-index attribution analysis method based on the distance between groups and groups in any one of claims 1-7, comprising: A multi-dimensional monitoring index calculation module is used to establish the vehicle trajectory across frames according to the point cloud data of the set section, obtain the continuous motion information of each vehicle, and then calculate the multi-dimensional monitoring index in real time; The multi-dimensional monitoring index includes average vehicle speed, traffic flow, time occupancy, average headway and lane speed difference; The comprehensive safety index calculation module is configured to construct and quantify traffic safety indexes according to multi-dimensional monitoring indexes, and to calculate a comprehensive safety index according to the quantified traffic safety indexes and corresponding weight values; the traffic safety indexes include following indexes: car-following risk, lane-changing risk and speed risk; The index attribution analysis module is configured to compare traffic safety indexes of the current period with traffic safety indexes of the last normal period when it is monitored that the comprehensive safety index of the current period abnormally fluctuates, to quantify the contribution of each traffic safety index to the change of the comprehensive safety index by calculating the difference between groups and the dispersion within groups of each traffic safety index, and to identify the dominant factor and generate an interpretable attribution report.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The program, when executed by the processor, implements the steps of the traffic anomaly multi-index attribution analysis method based on group-in-group distance according to any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the steps of the traffic anomaly multi-index attribution analysis method based on group-in-group distance according to any one of claims 1-7 when executing the program.

Images