A pre-and-post-accident situation evolution analysis method based on floating car trajectory data

Abstract

The application relates to a pre-and-post-accident situation evolution analysis method based on floating car trajectory data. The method first constructs a road network model, matches road network nodes in the situation points in the situation record, and sorts the up-and-down directions. Then, based on the accident point position and direction, the reachability subgraph upstream of the accident point is extracted, and the situation record is filtered in combination with the accident time, record time and position. The core step is to grid the reachability subgraph according to the distance and time dimensions, and construct a path-level dynamic situation grid model. Finally, based on the model, the congestion stages (such as generation, diffusion and dissipation) are divided, and the accident influence level is calculated. Compared with the prior art, the application can solve the problems of inaccurate spatial matching, difficult definition of time and space influence range and difficult quantification of congestion propagation law in the prior art.

Description

Technical Field

[0001] This invention relates to the field of road traffic safety analysis, and in particular to a method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data. Background Technology

[0002] In recent years, the expressway network has been expanding and traffic flow has been steadily increasing, leading to frequent road traffic accidents. Major accidents can significantly alter the operational status of road sections in a short period, generating complex phenomena such as queue diffusion, sudden speed drops, and upstream propagation of congestion waves. The ability to quickly and objectively assess the spatiotemporal impact and degree of influence on the road system after an accident is a crucial prerequisite for traffic management departments to implement tiered responses, develop diversion and detour plans, and restore traffic order.

[0003] Existing accident impact assessment methods mainly rely on fixed detectors (such as inductive loop detectors and toll station cross-section flow detectors) or video checkpoint data, which often only obtain speed and flow information on a few monitoring sections, making it difficult to accurately depict the congestion evolution process along the continuous space of the road. Some studies are based on floating car data or congestion situation data for analysis, but they mostly remain at the "point-to-segment" level of statistical comparison, lacking unified high-precision road network support, and the spatial matching error between accident location and congested road section is relatively large.

[0004] Furthermore, traditional studies often use time series or single-section indicators to describe changes in operation before and after an accident, making it difficult to construct a two-dimensional spatiotemporal model that takes into account both "distance along the road" and "time evolution." This makes it impossible to intuitively depict key characteristics such as the location of the congestion wave front, propagation speed, maximum impact distance, and recovery process. Moreover, there is a lack of a systematic method that combines the duration of congestion with the propagation distance to define the impact level of an accident.

[0005] For example, Chinese patent CN114093168B, "A Method for Evaluating Urban Road Traffic Operation Status Based on a Resilience Perspective," constructs an urban road network topology model, integrates macroscopic, mesoscopic, and microscopic multi-level indicators to generate key performance indicators, and utilizes seepage theory to identify the critical threshold for road network congestion diffusion, thus achieving a macroscopic evaluation of traffic operation status from the perspective of system resilience. However, this existing technology still has the following problems in practical applications, especially in dealing with the typical scenario of sudden traffic accidents: 1. The analysis granularity is too macroscopic and lacks specificity for specific accidents: The existing technology focuses on evaluating the evolution of the overall road network resilience, with the entire road network or large connected groups as the evaluation unit. This macroscopic perspective makes it difficult to capture the local and rapidly evolving traffic congestion situation triggered by a single accident, and it cannot accurately depict the spatiotemporal evolution details of traffic flow upstream of the accident point. Therefore, it is difficult to provide direct and precise decision support for rapid response and traffic management at the accident site.

[0006] 2. While the evaluation indicators are highly comprehensive, their real-time dynamic tracking capabilities are insufficient: The key performance indicator P(t) used in this technology is a comprehensive indicator that integrates multi-dimensional information such as topology, attributes, and performance. Although it has strong systematicity, its calculation is complex, and its output congestion index (CI) reflects more the systemic resilience "loss" than the specific propagation distance, propagation speed, and congestion dissipation process of the accident's impact. This makes it less intuitive and agile in dynamically and in real-time tracking the spatiotemporal propagation process of the accident's impact.

[0007] 3. Focuses on state identification, but is weak in process evolution analysis: The existing technology identifies the "critical threshold" of congestion through seepage theory, which is a state judgment. It fails to automatically divide and quantify the entire process of congestion after an accident, from "generation → diffusion → peak → dissipation," which is exactly the information most needed in accident emergency management to assess the severity of the accident, predict the duration of the impact, and formulate phased control strategies.

[0008] Therefore, it is necessary to propose an accident impact analysis method that combines high-precision road network construction, local multi-candidate point map matching, upstream and downstream accessibility screening based on graph structure, and path-level dynamic situation grid modeling, in order to solve the problems of inaccurate spatial matching, difficulty in defining the spatiotemporal impact range, and difficulty in quantifying the congestion propagation law in existing technologies. Summary of the Invention

[0009] The purpose of this invention is to provide a method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data in order to overcome the shortcomings of the existing technology.

[0010] The objective of this invention can be achieved through the following technical solutions: A method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data includes: Step S1: Construct a road network model, wherein the road network model includes multiple road network nodes; Step S2: Read the situation records, match one or more nearest neighbor road network nodes for at least two situation points in the situation records, and sort the nearest neighbor road network nodes according to the order of the situation points to obtain the uplink and downlink directions of each situation record. Step S3: Extract the upstream reachability subgraph of the accident point from the road network model based on the location and direction of the accident point; Step S4: Based on the upstream reachability subgraph of the accident point and the time of the accident point, the situation records are filtered by combining the recording time and up / down direction of each situation record and the position of the situation point in the situation record. Step S5: Rasterize the accessibility sub-map of the accident point according to the distance and time dimensions, and map the selected situation records to each grid to obtain a path-level dynamic situation grid model; Step S6: Generate congestion stages based on the path-level dynamic situation grid model and calculate the accident impact level.

[0011] In step S1, some of the road network nodes are obtained from the original process, while the other part is obtained by interpolation between the road network nodes obtained from the original process.

[0012] The processing of a single situation record in step S2 includes: Step S2-1: Match multiple road network nodes for each situation point in the situation record according to the nearest neighbor matching principle; Step S2-2: Based on the connection direction of the road network nodes matched by each situation point and the trajectory direction of the corresponding situation point, select one of the road network nodes matched by each situation point as the first road node, and cover the situation point sequence in the situation record with the first road node sequence obtained by finding the union of the first road nodes of all situation points. Step S2-3: Determine the up and down directions of the situation record based on the connection direction of the first road node.

[0013] Step S2-2 specifically includes: Step S2-2-1: Initialize the voting value of all situation points to 0, and combine all situation points in the situation record in pairs to obtain multiple situation point combinations; Step S2-2-2: Select an untraversed combination of situation points as the current situation point combination, and construct the first set of road network points by matching the road network nodes of the previous situation point in the current situation point combination, and construct the second set of road network points by matching the road network nodes of the next situation point. Step S2-2-3: Select one road network node from each of the first and second road network node sets to form a road network node combination, and traverse all road network nodes in the first and second road network node sets to obtain all road network node combinations; Step S2-2-4: For each combination of road network nodes, calculate the direction vector from the road network node in the first set of road network nodes to the road network node in the second set of road network nodes as the road network node direction, and compare it with the direction vector from the previous situation point to the next situation point in the current situation point combination. Add the voting value once to the two road network nodes corresponding to the closest road network node directions. Step S2-2-5: Determine whether the traversal of all situation point combinations has been completed. If yes, proceed to step S2-2-6; otherwise, return to step S2-2-2. Step S2-2-6: Take the road network node with the largest vote value among the road network nodes matched by each situation point as the first road node of each situation point, and cover the situation point sequence in the situation record with the first road node sequence obtained by taking the union of the first road nodes of all situation points.

[0014] In step S4, the following situation records are filtered: Situational records not located on the reachability submap of the accident site; Situation records located outside the accident analysis time interval, wherein the accident analysis time interval is determined based on the time of the accident point, specifically: tracing back a first set time period based on the accident point, and tracing back a second set time period based on the accident point.

[0015] Step S5 includes: Step S5-1: In the accessibility sub-map of the accident point, divide the upstream road segment into sub-segments according to the first set step size; Step S5-2: Select multiple discrete time points for the accident analysis time interval according to the first time step; Step S5-3: Pair discrete time points and sub-segments to obtain multiple two-dimensional grid points, and initialize all two-dimensional grid points; Step S5-4: Based on the recording time of the situation record, calculate the absolute value of the difference between each of the remaining discrete time points as the first time difference, and take the discrete time point with the smallest first time difference as the matching time of the situation record. Calculate the road distances between each first road node in the situation record and the accident point, and use the maximum and minimum road distances of all first road nodes as the boundary values ​​of the distance interval of the situation record. All two-dimensional grid points whose time dimension is the matching time and whose distance dimension is within the distance interval are taken as the first matching point of the corresponding situation record; Step S5-5: Record the average speed and congestion status of each situation to the corresponding first matching point; Steps S5-6: Determine the average speed and congestion status of each two-dimensional grid point, where: If a two-dimensional grid point only records an average speed and congestion state, then this recorded average speed and congestion state are directly used as the average speed and congestion state of the two-dimensional grid point. If a two-dimensional grid point records multiple average speeds and congestion states, then the average of all recorded average speeds is taken as the average speed of the two-dimensional grid point, and the most severe congestion state among all recorded congestion states is taken as the congestion state of the two-dimensional grid point. If the two-dimensional grid points do not record average speed and congestion status, then the average speed of the two-dimensional grid points is set to the road speed limit, and all recorded congestion statuses are set to non-congested.

[0016] Step S6 includes: Step S6-1: For each discrete time point, determine the farthest upstream distance point: in: Discrete time points t m Corresponding to the farthest upstream point, For the sub-section i Distance along the road from the location of the accident. For the sake of and and t m The average velocity of the two-dimensional grid points, This is the maximum speed limit for congestion. Step S6-2: Calculate the maximum impact distance caused by the accident: in: denoted as the maximum impact distance caused by the accident, where m is the index of the discrete time point; Step S6-3: Calculate the congestion propagation speed: in: Discrete time points t m The corresponding congestion propagation speed, where Δt is the first time step. Discrete time points t m+1 For the farthest upstream point, if the congestion propagation speed is less than 0, it means that the congestion is propagating upstream; otherwise, it means that the congestion is beginning to dissipate. Step S6-4: Calculate the interval average speed and the duration of the congestion interval at each discrete time point: in: Discrete time points t m The corresponding interval average velocity, The number of sub-segments, Let k be the duration of congestion in sub-segment k. Discrete time points t m The congestion characteristic value of the corresponding sub-segment k, For the set of sub-segment numbers; Step S6-5: Calculate the first-order difference for each discrete time point based on the interval average velocity at each discrete time point: in: Discrete time points t m+1 The corresponding first-order difference, Discrete time points t m+1 The corresponding interval average velocity; The congestion stage is determined by combining the furthest upstream distance, where: , Growth is the stage where congestion is generated. , If the congestion continues to increase, it enters the spreading stage. and If the level remains near the maximum value, then congestion is approaching its peak. , The gradual reduction indicates the dissipation phase of congestion; Step S6-6: Calculate the impact level of the accident: in: Let α be the accident impact level of sub-segment k, α be the weight of congestion duration, and β be the weight of maximum impact distance. Steps 6-7: Calculate the time when the maximum congestion distance occurs: in: This represents the time when the maximum congestion distance corresponding to sub-segment k occurs.

[0017] In step S6-1, if the average speed of all two-dimensional grid points corresponding to a discrete time point is greater than the upper limit of congestion speed, then the farthest upstream distance point of that discrete time point is set to 0.

[0018] An accident pre- and post-accident situation evolution analysis device based on floating car trajectory data includes a memory, a processor, and a program stored in the memory. When the processor executes the program, it implements the method described above.

[0019] A storage medium having a program stored thereon, which, when executed, implements the method described above.

[0020] Compared with the prior art, the present invention has the following beneficial effects: 1. By constructing a high-precision road topology map, a situation matching strategy based on local multi-candidate points, an accessibility constraint screening mechanism, and a path-level dynamic situation evolution model, we can achieve accurate identification of the scope of accident impact and quantitative analysis of the congestion propagation pattern after the accident.

[0021] 2. Interpolation can add intermediate nodes between the original nodes, making the road network model denser and more continuous, reducing the spatial gap when matching situation points, and improving the accuracy and coverage of the road network model.

[0022] 3. By matching multiple road network nodes for each situation point in a single situation record and determining the uplink and downlink directions, we can avoid the problem of not being able to include road network nodes in the same uplink and downlink directions in the matching results because the nearest neighbor points are located in opposite directions. On the other hand, by determining the final road network nodes through subsequent directional connections, we can improve the accuracy, robustness and directional accuracy of the situation record and road network matching.

[0023] 4. After initializing the voting values, all combinations of situation points are traversed, and the directions of road network points and trajectory points are compared. Voting values ​​are accumulated for the nodes corresponding to the closest directions, and the node with the largest voting value is finally selected as the first road node. This mechanism, based on the majority consensus principle, ensures directional consistency among multiple situation points and reduces local errors. This enhances the overall consistency of situation records, thereby solving the problem of inaccurate spatial matching and making the matching results more reliable.

[0024] 5. By filtering out situation records that are not on the reachability submap of the incident point or outside the incident analysis time interval, data quality and analysis efficiency are improved, and interference from irrelevant data is reduced.

[0025] 6. Divide the upstream road segment into sub-segments by step size (distance dimension), and discretize the time interval by step size (time dimension) to form two-dimensional grid points. Map the situation records onto the grid points. Rasterization converts continuous spatiotemporal data into discrete grids, allowing average speed and congestion status to be statistically analyzed and aggregated at the grid unit level. This solves the problem of "difficulty in quantifying the laws of congestion propagation" because the grid model provides a unified framework, making indicators such as speed and congestion easy to calculate and visualize.

[0026] 7. Calculate indicators such as the farthest upstream point, maximum impact distance, congestion propagation speed, average speed within the region, and congestion duration, and divide the process into stages based on first-order differences and distance changes. For example, the formula defines the congestion propagation speed and impact level, enabling automatic identification of the congestion process from the data. This solves the problem of the difficulty in quantifying congestion propagation patterns because the algorithm transforms subjective descriptions into calculable indicators, providing a scientific basis for evaluation.

[0027] 8. By setting the farthest upstream distance point to 0 when the average speed of all two-dimensional grid points is greater than the congestion speed limit, the boundary case is handled, ensuring the rationality and stability of the model. Attached Figure Description

[0028] Figure 1 This is a schematic diagram of the main steps of the method of the present invention; Figure 2 This is a schematic diagram of the trajectory matching effect. Detailed Implementation

[0029] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0030] A method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data, such as Figure 1 As shown, it includes: Step S1: Construct a road network model, which includes multiple road network nodes. Some of the road network nodes are obtained from the original process, while the other part is obtained by interpolation between the road network nodes obtained from the original process.

[0031] First, all road features within the target road area are extracted from the open-source map (OpenStreetMap) to obtain raw vector data including road geometry, node coordinates, and road direction information. Because the raw road data suffers from uneven node distribution and large segment length differences, which are detrimental to subsequent trajectory matching and accessibility analysis, each road geometric curve is resampled at a preset 10-meter interval to construct high-precision road network data. Specifically, based on the original road line features, new sampling point coordinates are obtained by interpolating along the curve at a fixed step size of 10 meters. These sampled points are then incorporated into the original road line features, forming a high-density road network. This equidistant sampling significantly improves the spatial resolution of the road network and greatly reduces the matching error between subsequent accident points and situation points. To achieve subsequent matching of accident points and situation points, a KDTree spatial index is constructed using all sampled coordinates.

[0032] Step S2: Read the situation records, match one or more nearest neighbor road network nodes for at least two situation points in the situation records, and sort the nearest neighbor road network nodes according to the order of the situation points to obtain the uplink and downlink directions of each situation record. Situational records typically include recording time, average speed, distance, congestion status, and a series of situational points.

[0033] The processing procedure for a single situation record includes: Step S2-1: Match multiple road network nodes for each situation point in the situation record according to the nearest neighbor matching principle; Step S2-2: Based on the connection direction of the road network nodes matched to each situation point and the trajectory direction of the corresponding situation point, select one of the road network nodes matched to each situation point as the first road node, and cover the situation point sequence in the situation record with the first road node sequence obtained by taking the union of the first road nodes of all situation points. Specifically, this includes: Step S2-2-1: Initialize the voting value of all situation points to 0, and combine all situation points in the situation record in pairs to obtain multiple situation point combinations; Step S2-2-2: Select an untraversed combination of situation points as the current situation point combination, and construct the first set of road network points by matching the road network nodes of the previous situation point in the current situation point combination, and construct the second set of road network points by matching the road network nodes of the next situation point. Step S2-2-3: Select one road network node from each of the first and second road network node sets to form a road network node combination, and traverse all road network nodes in the first and second road network node sets to obtain all road network node combinations; Step S2-2-4: For each combination of road network nodes, calculate the direction vector from the road network node in the first set of road network nodes to the road network node in the second set of road network nodes as the road network node direction, and compare it with the direction vector from the previous situation point to the next situation point in the current situation point combination. Add the voting value once to the two road network nodes corresponding to the closest road network node directions. Step S2-2-5: Determine whether the traversal of all situation point combinations has been completed. If yes, proceed to step S2-2-6; otherwise, return to step S2-2-2. Step S2-2-6: Take the road network node with the largest vote value among the road network nodes matched by each situation point as the first road node of each situation point, and cover the situation point sequence in the situation record with the first road node sequence obtained by finding the union of the first road nodes of all situation points. In this process, by finding the union, the problem of introducing new errors caused by adjacent situation points being matched with the same road network node can be solved.

[0034] Step S2-3: Determine the up and down directions of the situation record based on the connection direction of the first road node.

[0035] This method reduces the complexity of the algorithm and eliminates candidate points with inconsistent directions while eliminating uplink and downlink matching errors in highway networks.

[0036] By using the above method, all situation records are traversed, thus obtaining the uplink and downlink directions of all situation records. Furthermore, the situation points in the situation records are converted into road network nodes, specifically the first road nodes. The effect is as follows: Figure 2 As shown.

[0037] Step S3: Extract the upstream reachability subgraph of the accident point from the road network model based on the location and direction of the accident point; The constructed accident point accessibility subgraph includes ramps and main roads. Based on this, a partial upstream direction graph of the accident point is constructed according to the road network graph of the accident point. This facilitates the shortest path search to be performed with the accident point as the source in the later stage. Compared with the complexity of full graph search, this method takes the accident point as the unit and only analyzes the situational influence near the accident point, which greatly improves efficiency. Step S4: Based on the upstream reachability subgraph of the accident point and the time of the accident point, the situation records are filtered by combining the recording time and up / down direction of each situation record and the position of the situation point in the situation record. Specifically, the following situation records are filtered: Situational records not located on the reachability submap of the accident site; Situation records located outside the accident analysis time interval, wherein the accident analysis time interval is determined based on the time of the accident point, specifically: tracing back a first set time period based on the accident point, and tracing back a second set time period based on the accident point. Under normal circumstances, the second set time period should be significantly longer than the first set time period.

[0038] Step S5: Rasterize the accessibility sub-map of the incident point according to the distance and time dimensions, and map the selected situation records to each grid to obtain a path-level dynamic situation grid model, including: Step S5-1: In the accessibility sub-map of the accident point, divide the upstream road segment into sub-segments according to the first set step length. The first set step length should be the distance along the road.

[0039] Step S5-2: Select multiple discrete time points for the accident analysis time interval according to the first time step; Step S5-3: Pair discrete time points and sub-segments to obtain multiple two-dimensional grid points, and initialize all two-dimensional grid points; Step S5-4: Based on the recording time of the situation record, calculate the absolute value of the difference between each of the remaining discrete time points as the first time difference, and take the discrete time point with the smallest first time difference as the matching time of the situation record. Calculate the road distances between each first road node in the situation record and the accident point, and use the maximum and minimum road distances of all first road nodes as the boundary values ​​of the distance interval of the situation record. All two-dimensional grid points whose time dimension is the matching time and whose distance dimension is within the distance interval are taken as the first matching point of the corresponding situation record; Step S5-5: Record the average speed and congestion status of each situation to the corresponding first matching point; Steps S5-6: Determine the average speed and congestion status of each two-dimensional grid point, where: If a two-dimensional grid point only records an average speed and congestion state, then this recorded average speed and congestion state are directly used as the average speed and congestion state of the two-dimensional grid point. If a two-dimensional grid point records multiple average speeds and congestion states, then the average of all recorded average speeds is taken as the average speed of the two-dimensional grid point, and the most severe congestion state among all recorded congestion states is taken as the congestion state of the two-dimensional grid point. If the two-dimensional grid points do not record average speed and congestion status, then the average speed of the two-dimensional grid points is set to the road speed limit, and all recorded congestion statuses are set to non-congested.

[0040] Step S6: Generate congestion stages and calculate accident impact levels based on the path-level dynamic situation grid model, including: Step S6-1: For each discrete time point, determine the farthest upstream distance point: in: Discrete time points t m Corresponding to the farthest upstream point, For the sub-section i Distance along the road from the location of the accident. For the sake of and and t m The average velocity of the two-dimensional grid points, This is the maximum speed limit for congestion. Step S6-2: Calculate the maximum impact distance caused by the accident: in: denoted as the maximum impact distance caused by the accident, where m is the index of the discrete time point; Step S6-3: Calculate the congestion propagation speed: in: Discrete time points t m The corresponding congestion propagation speed, where Δt is the first time step. Discrete time points t m+1For the farthest upstream point, if the congestion propagation speed is less than 0, it means that the congestion is propagating upstream; otherwise, it means that the congestion is beginning to dissipate. Step S6-4: Calculate the interval average speed and the duration of the congestion interval at each discrete time point: in: Discrete time points t m The corresponding interval average velocity, The number of sub-segments, Let k be the duration of congestion in sub-segment k. Discrete time points t m The congestion characteristic value of the corresponding sub-segment k, For the set of sub-segment numbers; Step S6-5: Calculate the first-order difference for each discrete time point based on the interval average velocity at each discrete time point: in: Discrete time points t m+1 The corresponding first-order difference, Discrete time points t m+1 The corresponding interval average velocity; The congestion stage is determined by combining the furthest upstream distance, where: , Growth is the stage where congestion is generated. , If the congestion continues to increase, it enters the spreading stage. and If the level remains near the maximum value, then congestion is approaching its peak. , The gradual reduction indicates the dissipation phase of congestion; Step S6-6: Calculate the impact level of the accident: in: Let α be the accident impact level of sub-segment k, α be the weight of congestion duration, and β be the weight of maximum impact distance. Steps 6-7: Calculate the time when the maximum congestion distance occurs: in: This represents the time when the maximum congestion distance corresponding to sub-segment k occurs.

[0041] In step S6-1, if the average speed of all two-dimensional grid points corresponding to a discrete time point is greater than the upper limit of congestion speed, then the farthest upstream distance point of that discrete time point is set to 0.

[0042] The overall research was completed through the above steps. The study concluded that, based on the establishment of a unified high-precision road network, accident point data and road condition data were spatiotemporally matched to construct a two-dimensional distance-time dynamic situation grid model, thereby studying the congestion impact level and congestion situation evolution mechanism after an accident.

[0043] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method for analyzing the pre- and post-accident situational evolution based on floating car trajectory data, characterized in that, include: Step S1: Construct a road network model, wherein the road network model includes multiple road network nodes; Step S2: Read the situation records, match one or more nearest neighbor road network nodes for at least two situation points in the situation records, and sort the nearest neighbor road network nodes according to the order of the situation points to obtain the uplink and downlink directions of each situation record. Step S3: Extract the upstream reachability subgraph of the accident point from the road network model based on the location and direction of the accident point; Step S4: Based on the upstream reachability subgraph of the accident point and the time of the accident point, the situation records are filtered by combining the recording time and up / down direction of each situation record and the position of the situation point in the situation record. Step S5: Rasterize the accessibility sub-map of the accident point according to the distance and time dimensions, and map the selected situation records to each grid to obtain a path-level dynamic situation grid model; Step S6: Generate congestion stages based on the path-level dynamic situation grid model and calculate the accident impact level.

2. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 1, characterized in that, In step S1, some of the road network nodes are obtained from the original process, while the other part is obtained by interpolation between the road network nodes obtained from the original process.

3. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 1, characterized in that, The processing of a single situation record in step S2 includes: Step S2-1: Match multiple road network nodes for each situation point in the situation record according to the nearest neighbor matching principle; Step S2-2: Based on the connection direction of the road network nodes matched by each situation point and the trajectory direction of the corresponding situation point, select one of the road network nodes matched by each situation point as the first road node, and cover the situation point sequence in the situation record with the first road node sequence obtained by finding the union of the first road nodes of all situation points. Step S2-3: Determine the up and down directions of the situation record based on the connection direction of the first road node.

4. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 3, characterized in that, Step S2-2 specifically includes: Step S2-2-1: Initialize the voting value of all situation points to 0, and combine all situation points in the situation record in pairs to obtain multiple situation point combinations; Step S2-2-2: Select an untraversed combination of situation points as the current situation point combination, and construct the first set of road network points by matching the road network nodes of the previous situation point in the current situation point combination, and construct the second set of road network points by matching the road network nodes of the next situation point. Step S2-2-3: Select one road network node from each of the first and second road network node sets to form a road network node combination, and traverse all road network nodes in the first and second road network node sets to obtain all road network node combinations; Step S2-2-4: For each combination of road network nodes, calculate the direction vector from the road network node in the first set of road network nodes to the road network node in the second set of road network nodes as the road network node direction, and compare it with the direction vector from the previous situation point to the next situation point in the current situation point combination. Add the voting value once to the two road network nodes corresponding to the closest road network node directions. Step S2-2-5: Determine whether the traversal of all situation point combinations has been completed. If yes, proceed to step S2-2-6; otherwise, return to step S2-2-2. Step S2-2-6: Take the road network node with the largest vote value among the road network nodes matched by each situation point as the first road node of each situation point, and cover the situation point sequence in the situation record with the first road node sequence obtained by taking the union of the first road nodes of all situation points.

5. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 1, characterized in that, In step S4, the following situation records are filtered: Situational records not located on the reachability submap of the accident site; Situation records located outside the accident analysis time interval, wherein the accident analysis time interval is determined based on the time of the accident point, specifically: tracing back a first set time period based on the accident point, and tracing back a second set time period based on the accident point.

6. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 5, characterized in that, Step S5 includes: Step S5-1: In the accessibility sub-map of the accident point, divide the upstream road segment into sub-segments according to the first set step size; Step S5-2: Select multiple discrete time points for the accident analysis time interval according to the first time step; Step S5-3: Pair discrete time points and sub-segments to obtain multiple two-dimensional grid points, and initialize all two-dimensional grid points; Step S5-4: Based on the recording time of the situation record, calculate the absolute value of the difference between each of the remaining discrete time points as the first time difference, and take the discrete time point with the smallest first time difference as the matching time of the situation record. Calculate the road distances between each first road node in the situation record and the accident point, and use the maximum and minimum road distances of all first road nodes as the boundary values ​​of the distance interval of the situation record. All two-dimensional grid points whose time dimension is the matching time and whose distance dimension is within the distance interval are taken as the first matching point of the corresponding situation record; Step S5-5: Record the average speed and congestion status of each situation to the corresponding first matching point; Steps S5-6: Determine the average speed and congestion status of each two-dimensional grid point, where: If a two-dimensional grid point only records an average speed and congestion state, then this recorded average speed and congestion state are directly used as the average speed and congestion state of the two-dimensional grid point. If a two-dimensional grid point records multiple average speeds and congestion states, then the average of all recorded average speeds is taken as the average speed of the two-dimensional grid point, and the most severe congestion state among all recorded congestion states is taken as the congestion state of the two-dimensional grid point. If the two-dimensional grid points do not record average speed and congestion status, then the average speed of the two-dimensional grid points is set to the road speed limit, and all recorded congestion statuses are set to non-congested.

7. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 6, characterized in that, Step S6 includes: Step S6-1: For each discrete time point, determine the farthest upstream distance point: in: Discrete time points t m Corresponding to the farthest upstream point, For the sub-section i Distance along the road from the location of the accident. For the sake of and and t m The average velocity of the two-dimensional grid points, This is the maximum speed limit for congestion. Step S6-2: Calculate the maximum impact distance caused by the accident: in: denoted as the maximum impact distance caused by the accident, where m is the index of the discrete time point; Step S6-3: Calculate the congestion propagation speed: in: Discrete time points t m The corresponding congestion propagation speed, where Δt is the first time step. Discrete time points t m+1 For the farthest upstream point, if the congestion propagation speed is less than 0, it means that the congestion is propagating upstream; otherwise, it means that the congestion is beginning to dissipate. Step S6-4: Calculate the interval average speed and the duration of the congestion interval at each discrete time point: in: Discrete time points t m The corresponding interval average velocity, For the number of sub-segments, Let K be the duration of congestion in sub-segment k. Discrete time points t m The congestion characteristic value of the corresponding sub-segment k, For the set of sub-segment numbers; Step S6-5: Calculate the first-order difference for each discrete time point based on the interval average velocity at each discrete time point: in: Discrete time points t m+1 The corresponding first-order difference, Discrete time points t m+1 The corresponding interval average velocity; The congestion stage is determined by combining the furthest upstream distance, where: , Growth is the stage where congestion is generated. , If the congestion continues to increase, it enters the spreading stage. and If the level remains near the maximum value, then congestion is approaching its peak. , The gradual reduction indicates the dissipation phase of congestion; Step S6-6: Calculate the impact level of the accident: in: Let α be the accident impact level of sub-segment k, α be the weight of congestion duration, and β be the weight of maximum impact distance. Steps 6-7: Calculate the time when the maximum congestion distance occurs: in: This represents the time when the maximum congestion distance corresponding to sub-segment k occurs.

8. The method for analyzing the pre- and post-accident situation evolution based on floating car trajectory data according to claim 5, characterized in that, In step S6-1, if the average speed of all two-dimensional grid points corresponding to a discrete time point is greater than the upper limit of congestion speed, then the farthest upstream distance point of that discrete time point is set to 0.

9. A device for analyzing the pre- and post-accident situational evolution based on floating car trajectory data, comprising a memory, a processor, and a program stored in the memory, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-8.

10. A storage medium having a program stored thereon, characterized in that, When the program is executed, it implements the method as described in any one of claims 1-8.

Images