Multi-scale traffic accident-prone road section cause analysis method

Abstract

The invention provides a multi-scale cause analysis method for a road section with multiple traffic accidents. The method comprises the following steps: S1, dividing urban roads in an analysis range into road sections as analysis units to identify road sections with frequent traffic accidents; step S2, from the perspective of urban planning design and operation management, constructing and calculating a multi-scale traffic accident influence factor index system meeting requirements on a macroscopic level, a middle level and a microscopic level; s3, training a road section traffic accident kernel density analysis model, and calculating an SHAP value of each influence factor by adopting an SHAP method; and S4, on each road section where traffic accidents occur frequently, comparing the influence degree of each influence factor on the traffic accidents on the road section through normalizing the SHAP absolute value, and identifying main influence factors. The method can solve the problems that the existing road safety analysis technology is insufficient in comprehensiveness and granularity and cannot meet the traffic safety management requirement under the current urban update background.

Description

Technical Field

[0001] The present invention relates to the field of urban planning, design and operation management, and in particular to a multi-scale cause analysis method for road sections prone to traffic accidents. Background Art

[0002] Road traffic safety is closely related to residents’ production and life and is an important part of urban governance.

[0003] Mainstream road traffic safety analysis technology mainly focuses on traffic influencing factors at the micro level. Starting from traffic accident data, it uses the behavior of traffic participants (such as traffic flow conflicts, driving habits, compliance with traffic regulations, etc.) to explain the occurrence of traffic accidents. However, many factors other than traffic participants also have a significant impact on traffic safety. For example, the socioeconomic level and population composition have a decisive role in the level of urban motorization, travel volume, and travel ratio, which in turn have an impact on traffic safety; land use type characteristics are also directly related to the risk of traffic accidents. For example, commercial land areas have a large number of motor vehicle trips and severe congestion, leading to a higher risk of accidents. Therefore, the comprehensiveness and comprehensiveness of road traffic safety analysis technology focusing on the micro level are insufficient, and it is impossible to incorporate traffic safety management into the systematic decision-making process of urban planning, design and management.

[0004] To this end, some studies have begun to attempt to quantitatively study the relationship between regional population economy, land use, traffic operation, road attributes and traffic safety from a macro perspective. This approach makes it possible to understand traffic safety in the context of urban planning, construction, operation and management. However, the analysis level of these studies is mainly focused on the traffic analysis zone (TAZ), which is relatively single in dimension and does not fully match the refined governance needs in the context of urban renewal at this stage.

[0005] Therefore, it is necessary to analyze the impact of non-traffic participant factors such as population distribution, land use type, public buildings, road infrastructure, etc. on the level of urban road traffic safety at multiple scales from the perspective of urban planning, design and operation management, and to determine the key influencing factors, which will not only help improve the level of traffic safety governance, but also provide a scientific basis for urban renewal and urban operation management. Summary of the invention

[0006] The present invention proposes a multi-scale cause analysis method for road sections with high traffic accident incidence, which can solve the problem that the existing road safety analysis technology is insufficient in comprehensiveness and granularity and cannot meet the current traffic safety management needs in the context of urban renewal.

[0007] The present invention adopts the following technical solutions.

[0008] A multi-scale traffic accident-prone road section cause analysis method, the analysis method comprising the following steps:

[0009] Step S1, dividing the urban roads within the analysis range into sections as analysis units to identify sections with high traffic accident incidence;

[0010] Step S2: From the perspective of urban planning, design and operation management, a multi-scale traffic accident influencing factor index system that meets the requirements is constructed and calculated at the macro level, the meso level and the micro level;

[0011] Step S3, training the road section traffic accident kernel density analysis model, and using the SHAP method to calculate the SHAP value of each influencing factor;

[0012] Step S4: On each road section with frequent traffic accidents, the influence of each influencing factor on the traffic accidents on the road section is compared by normalizing the SHAP absolute value, and the main influencing factor is identified.

[0013] The step S1 specifically includes the following steps:

[0014] Step S101: within the research scope, obtain the GIS data of the road network within the research scope, use the intersection as the segmentation point, and divide the road network into sections;

[0015] Step S102: Using the traffic accident history records, calculate the plane distribution of the traffic accident kernel density of the road section, specifically: using the geocoding tool provided by the Internet map service provider, convert the location description in the traffic accident history records into longitude and latitude, and then represent each accident record as a point on the map. Perform kernel density analysis on all accident points to obtain the traffic accident kernel density distribution of the study area; the geocoding tool includes Amap API and Baidu Map API;

[0016] Step S103: Calculate the traffic accident kernel density of the road section according to the traffic accident kernel density value at the location of the road section; specifically, divide each road section into equal parts according to a preset interval, and take the average value of the kernel density values ​​at all the equal division points as the traffic accident kernel density of the road section;

[0017] Step S104: Use the box plot method to count the traffic accident kernel density of all road sections, and use the upper quartile and interquartile range indicators to calculate the upper limit index; identify the road sections with traffic accident kernel density values ​​exceeding the upper limit as traffic accident-prone sections.

[0018] The preset distance in step S103 is 1 meter.

[0019] The step S2 specifically includes the following steps:

[0020] Step S201: Determine the factors affecting traffic accidents and their indicators at the macro, meso and micro levels respectively.

[0021] The macro level is the urban area, and the macro level scale influencing factors reflect the impact of the urban area where the road section is located on traffic accidents on the road section; the specific influencing factors include centrality and importance, the indicator related to centrality is the circle, and its construction principle is to reflect the distance relationship between the road section and the city center, and the indicator related to importance is betweenness centrality, and its construction principle is to reflect the importance of the road section in the entire road network;

[0022] The meso-level is the block. The meso-level scale influencing factors reflect the impact of the blocks around the road section on traffic accidents on the road section. The influencing factors include population and land use. The population-related indicator is population density, and its construction principle is to reflect the population size or density of the blocks around the road section. The land-related indicator is the proportion of various types of land areas, and the construction principle is to reflect the proportion of various types of major land in the blocks around the road section. The various types of major land include commercial service land, public management land, residential land, and industrial and mining land.

[0023] The micro level is the road, and the micro-level scale influencing factors reflect the impact of the attributes of each road section on traffic accidents; its influencing factors are traffic operation and road safety facilities; the indicator related to traffic operation is traffic saturation, and the construction principle is to be able to reflect the traffic flow characteristics of the road section, including flow, flow speed, and density; the indicators related to road safety facilities are whether to set up a separation between vehicles and non-motorized vehicles and whether to set up a central separation, and the construction principle is to be able to reflect the construction status of road safety facilities on the road section;

[0024] Step S202: using multi-source urban geographic information data, calculate the above indicators on each road section.

[0025] The step S3 specifically comprises the following steps:

[0026] Step S301: constructing a data set: assuming that the independent variable is the traffic accident influencing factor index, including: macro-scale index; meso-scale index; micro-scale index; the dependent variable is the traffic accident kernel density of the road section;

[0027] Macro-scale indicators include the circle and centrality; meso-scale indicators include population density and the proportion of various land areas; micro-scale indicators include traffic saturation, traffic flow, whether there is a separation between vehicles and non-motor vehicles, and whether there is a central separation;

[0028] Step S302: model training; on the data set, the XGBOOST algorithm is used to train a multi-scale road section traffic accident analysis model; specifically: all variables of the data set are standardized, and the training set and the validation set are divided according to a ratio of 5:1; using python, the model training function in the XGBOOST package is called, the dependent variables and independent variables in the training set and the validation set are input, and the hyperparameter optimization is performed using the GridSearchCV package. The result is the road section traffic accident kernel density analysis model, which is used to reflect the quantitative relationship between the traffic accident kernel density of each road section and the traffic accident influencing factor indicators;

[0029] Step S303: Model interpretation; Apply the SHAP method to interpret the model obtained by XGBOOST algorithm training, and obtain the SHAP value of each influencing factor index on each accident-prone road section; Specifically: Use the road section traffic accident kernel density analysis model, input the traffic accident influencing factor index of each road section, and obtain the model prediction value of the traffic accident kernel density of each road section; Use python to call the SHAP package, use the model interpretation function therein, input the model prediction value of the traffic accident kernel density of each road section and the traffic accident influencing factor index, and obtain the SHAP value of each traffic accident influencing factor index on each road section, which is used to reflect the degree of influence of each influencing factor on the traffic accident of the road section. The larger the SHAP absolute value, the greater the degree of influence.

[0030] The step S4 specifically includes the following steps:

[0031] Step S401: On each road section with frequent traffic accidents, the absolute values ​​of the SHAP values ​​of all influencing factor indicators are calculated and divided by the maximum value thereof to obtain the normalized SHAP absolute value of each indicator;

[0032] Step S402: Among all the indicators corresponding to each influencing factor, the largest normalized SHAP absolute value is taken as the influence degree of the influencing factor on the traffic accident on the road section;

[0033] Step S403: Divide all traffic accident influencing factors on each road section into three groups: main influencing factors, secondary influencing factors and other influencing factors in descending order of influence.

[0034] Step S403 specifically includes: traversing the influence degree characteristic values ​​of all influencing factors, taking the largest influence degree characteristic value as the reference value; dividing all influence degree characteristic values ​​by the reference value to obtain a ratio;

[0035] The influencing factors with a ratio greater than or equal to 0.67 were identified as the main influencing factors; the influencing factors with a ratio greater than or equal to 0.33 and less than 0.67 were identified as secondary influencing factors, and the influencing factors with a ratio less than 0.33 were defined as other influencing factors.

[0036] The beneficial effects of the present invention are: comprehensively considering the factors affecting traffic accidents at different scales, such as macro, meso and micro, and explaining the degree of their influence on traffic accidents, providing a basis for systematically solving traffic safety problems through urban planning, design and operation management; defining the basic unit of traffic safety analysis as the road section, and specifically calculating the main influencing factors of each road section with frequent traffic accidents, which can adapt to the needs of refined urban governance and support urban renewal actions. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0038] Attached Figure 1 It is a schematic diagram of the framework of the multi-scale urban road accident-prone section main influencing factors analysis method of the present invention;

[0039] Attached Figure 2 It is a schematic diagram of the process of road section division and accident-prone road section identification of the present invention;

[0040] Attached Figure 3 It is a schematic diagram of the construction and calculation process of the multi-scale traffic accident influencing factor index system of the present invention;

[0041] Attached Figure 4 It is a schematic diagram of the training of the road section traffic accident kernel density analysis model and the calculation process of the SHAP value of the influencing factors of the present invention;

[0042] Attached Figure 5 It is a schematic diagram of the flow of identifying influencing factors of a road section prone to traffic accidents according to the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0044] As shown in the figure, a multi-scale traffic accident-prone road section cause analysis method includes the following steps:

[0045] Step S1, dividing the urban roads within the analysis range into sections as analysis units to identify sections with high traffic accident incidence;

[0046] Step S2: From the perspective of urban planning, design and operation management, a multi-scale traffic accident influencing factor index system that meets the requirements is constructed and calculated at the macro level, the meso level and the micro level;

[0047] Step S3, training the road section traffic accident kernel density analysis model, and using the SHAP method to calculate the SHAP value of each influencing factor;

[0048] Step S4: On each road section with frequent traffic accidents, the influence of each influencing factor on the traffic accidents on the road section is compared by normalizing the SHAP absolute value, and the main influencing factor is identified.

[0049] The step S1 specifically includes the following steps:

[0050] Step S101: within the research scope, obtain the GIS data of the road network within the research scope, use the intersection as the segmentation point, and divide the road network into sections;

[0051] Step S102: Using the traffic accident history records, calculate the plane distribution of the traffic accident kernel density of the road section, specifically: using the geocoding tool provided by the Internet map service provider, convert the location description in the traffic accident history records into longitude and latitude, and then represent each accident record as a point on the map. Perform kernel density analysis on all accident points to obtain the traffic accident kernel density distribution of the study area; the geocoding tool includes Amap API and Baidu Map API;

[0052] Step S103: Calculate the traffic accident kernel density of the road section according to the traffic accident kernel density value at the location of the road section; specifically, divide each road section into equal parts according to a preset interval, and take the average value of the kernel density values ​​at all the equal division points as the traffic accident kernel density of the road section;

[0053] Step S104: Use the box plot method to count the traffic accident kernel density of all road sections, and use the upper quartile and interquartile range indicators to calculate the upper limit index; identify the road sections with traffic accident kernel density values ​​exceeding the upper limit as traffic accident-prone sections.

[0054] The preset distance in step S103 is 1 meter.

[0055] The step S2 specifically includes the following steps:

[0056] Step S201: Determine the factors affecting traffic accidents and their indicators at the macro, meso and micro levels respectively.

[0057] The macro level is the urban area, and the macro level scale influencing factors reflect the impact of the urban area where the road section is located on traffic accidents on the road section; the specific influencing factors include centrality and importance, the indicator related to centrality is the circle, and its construction principle is to reflect the distance relationship between the road section and the city center, and the indicator related to importance is betweenness centrality, and its construction principle is to reflect the importance of the road section in the entire road network;

[0058] The meso-level is the block. The meso-level scale influencing factors reflect the impact of the blocks around the road section on traffic accidents on the road section. The influencing factors include population and land use. The population-related indicator is population density, and its construction principle is to reflect the population size or density of the blocks around the road section. The land-related indicator is the proportion of various types of land areas, and the construction principle is to reflect the proportion of various types of major land in the blocks around the road section. The various types of major land include commercial service land, public management land, residential land, and industrial and mining land.

[0059] The micro level is the road, and the micro-level scale influencing factors reflect the impact of the attributes of each road section on traffic accidents; its influencing factors are traffic operation and road safety facilities; the indicator related to traffic operation is traffic saturation, and the construction principle is to be able to reflect the traffic flow characteristics of the road section, including flow, flow speed, and density; the indicators related to road safety facilities are whether to set up a separation between vehicles and non-motorized vehicles and whether to set up a central separation, and the construction principle is to be able to reflect the construction status of road safety facilities on the road section;

[0060] Step S202: using multi-source urban geographic information data, calculate the above indicators on each road section.

[0061] The step S3 specifically comprises the following steps:

[0062] Step S301: constructing a data set: assuming that the independent variable is the traffic accident influencing factor index, including: macro-scale index; meso-scale index; micro-scale index; the dependent variable is the traffic accident kernel density of the road section;

[0063] Macro-scale indicators include the circle and centrality; meso-scale indicators include population density and the proportion of various land areas; micro-scale indicators include traffic saturation, traffic flow, whether there is a separation between vehicles and non-motor vehicles, and whether there is a central separation;

[0064] Step S302: model training; on the data set, the XGBOOST algorithm is used to train a multi-scale road section traffic accident analysis model; specifically: all variables of the data set are standardized, and the training set and the validation set are divided according to a ratio of 5:1; using python, the model training function in the XGBOOST package is called, the dependent variables and independent variables in the training set and the validation set are input, and the hyperparameter optimization is performed using the GridSearchCV package. The result is the road section traffic accident kernel density analysis model, which is used to reflect the quantitative relationship between the traffic accident kernel density of each road section and the traffic accident influencing factor indicators;

[0065] Step S303: Model interpretation; Apply the SHAP method to interpret the model obtained by XGBOOST algorithm training, and obtain the SHAP value of each influencing factor index on each accident-prone road section; Specifically: Use the road section traffic accident kernel density analysis model, input the traffic accident influencing factor index of each road section, and obtain the model prediction value of the traffic accident kernel density of each road section; Use python to call the SHAP package, use the model interpretation function therein, input the model prediction value of the traffic accident kernel density of each road section and the traffic accident influencing factor index, and obtain the SHAP value of each traffic accident influencing factor index on each road section, which is used to reflect the degree of influence of each influencing factor on the traffic accident of the road section. The larger the SHAP absolute value, the greater the degree of influence.

[0066] The step S4 specifically includes the following steps:

[0067] Step S401: On each road section with frequent traffic accidents, the absolute values ​​of the SHAP values ​​of all influencing factor indicators are calculated and divided by the maximum value thereof to obtain the normalized SHAP absolute value of each indicator;

[0068] Step S402: Among all the indicators corresponding to each influencing factor, the largest normalized SHAP absolute value is taken as the influence degree of the influencing factor on the traffic accident on the road section;

[0069] Step S403: Divide all traffic accident influencing factors on each road section into three groups: main influencing factors, secondary influencing factors and other influencing factors in descending order of influence.

[0070] Step S403 specifically includes: traversing the influence degree characteristic values ​​of all influencing factors, taking the largest influence degree characteristic value as the reference value; dividing all influence degree characteristic values ​​by the reference value to obtain a ratio;

[0071] The influencing factors with a ratio greater than or equal to 0.67 were identified as the main influencing factors; the influencing factors with a ratio greater than or equal to 0.33 and less than 0.67 were identified as secondary influencing factors, and the influencing factors with a ratio less than 0.33 were defined as other influencing factors.

[0072] Example:

[0073] See also Figure 1 This embodiment provides a multi-scale method for analyzing the main influencing factors of accident-prone sections on urban roads, which specifically includes the following steps:

[0074] S1: Divide the urban roads within the study area into sections and identify sections with high traffic accident incidence.

[0075] S2: From the perspective of urban planning, design and operation management, a traffic accident influencing factor index system that meets the requirements is constructed and calculated at three different scales: macro, meso and micro.

[0076] S3: Train the road section traffic accident kernel density analysis model and use the SHAP method to calculate the SHAP value of each influencing factor.

[0077] S4: Further analyze the SHAP value on accident-prone sections to identify the main influencing factors.

[0078] Figure 2 It is a flow chart of road section division and accident-prone road section identification, which specifically includes the following steps:

[0079] S101: Obtain GIS data of the road network within the research scope, and divide the road network into sections by taking intersections as segmentation points.

[0080] S102: Use geocoding tools provided by Internet map service providers (e.g., Amap API, Baidu Map API) to convert the location descriptions in the traffic accident history records into longitude and latitude, and then represent each accident record as a point on the map. Perform kernel density analysis on all accident points to obtain the kernel density distribution of traffic accidents in the study area.

[0081] S103: Divide each road section into equal parts at intervals of 1 m, and take the average value of the kernel density values ​​at all the equally divided points as the traffic accident kernel density of the road section.

[0082] S104: Using the box plot method to count the traffic accident kernel density of all road sections, and using the upper quartile and interquartile range indicators to calculate the upper limit index. The road sections where the traffic accident kernel density value exceeds the upper limit are identified as road sections with frequent traffic accidents.

[0083] Figure 3 It is a flowchart for constructing and calculating the multi-scale traffic accident influencing factor index system, which specifically includes the following steps:

[0084] S201: Determine the factors affecting traffic accidents and their indicators at the macro, meso and micro scales. See Table 1 for details:

[0085] Table 1 Standard construction

[0086]

[0087] Macro-scale influencing factors reflect the impact of the urban area where the road section is located on traffic accidents on the road section, including two indicators: the circle and betweenness centrality.

[0088] Mesoscale traffic accident influencing factors reflect the impact of surrounding blocks on traffic accidents on the road section. It includes five indicators: population density of the grid, and the proportion of various types of land (commercial service land, public management land, residential land, and industrial and mining land).

[0089] Micro-scale traffic accident influencing factors reflect the impact of each road section's own attributes on traffic accidents, including traffic saturation, traffic flow, whether there is a separation between motor vehicles and non-motor vehicles, and whether there is a central separation.

[0090] S202: Calculate the above indicators on each road section using multi-source urban geographic information data. The calculation method of each indicator is shown in Table 2.

[0091] Table 2 Calculation method of traffic accident influencing factors

[0092]

[0093] Figure 4 This is a flowchart of the road section traffic accident kernel density analysis model training and the calculation of the SHAP value of the influencing factors, which specifically includes the following steps:

[0094] S301: Construct a data set. The independent variables are indicators of factors affecting traffic accidents, including: macro-scale indicators (circle, centrality); meso-scale indicators (population density, proportion of various land areas); micro-scale indicators (traffic saturation, traffic flow, whether to set up motor vehicle and non-motor vehicle separation, whether to set up central separation). The dependent variable is the traffic accident kernel density of the road section.

[0095] S302: Model training. Standardize all variables in the data set, and divide the training set and validation set into a ratio of 5:1. Use python to call the model training function in the XGBOOST package, input the dependent variables and independent variables in the training set and validation set, and use the GridSearchCV package to optimize the hyperparameters. The result is the road section traffic accident kernel density analysis model. It reflects the quantitative relationship between the traffic accident kernel density and the traffic accident influencing factor indicators of each road section. S303: Model interpretation. Use the road section traffic accident kernel density analysis model, input the traffic accident influencing factor indicators of each road section, and obtain the model prediction value of the traffic accident kernel density of each road section. Use python to call the SHAP package, use the model interpretation function therein, input the model prediction value of the traffic accident kernel density of each road section and the traffic accident influencing factor indicators, and obtain the SHAP value of each traffic accident influencing factor indicator on each road section. It reflects the degree of influence of each influencing factor on the traffic accidents on that road section. The larger the SHAP absolute value, the greater the degree of influence. [Explanation of "Using the SHAP method to explain the model"]

[0096] Figure 5This is a flow chart for identifying factors affecting accident-prone sections, which specifically includes the following steps:

[0097] S401: On each road section with a high incidence of traffic accidents, the absolute values ​​of the SHAP values ​​of all influencing factor indicators are calculated and divided by the maximum value thereof to obtain the normalized SHAP absolute value of each indicator.

[0098] S402: Among all the indicators corresponding to each influencing factor, take the largest normalized SHAP absolute value as the influence degree characteristic value of the influencing factor.

[0099] S403: Traverse the influence degree characteristic values ​​of all influencing factors, and take the largest influence degree characteristic value as the reference value. Divide all influence degree characteristic values ​​by the reference value to obtain a ratio. Identify the influencing factors with a ratio greater than or equal to 0.67 as the main influencing factors; identify the influencing factors with a ratio greater than or equal to 0.33 and less than 0.67 as the secondary influencing factors, and define the influencing factors with a ratio less than 0.33 as other influencing factors.

Claims

1. A multi-scale traffic accident-prone road section cause analysis method, characterized by: The analytical method The steps include: Step S1, dividing the urban roads within the analysis range into sections as analysis units to identify sections with high traffic accident incidence; Step S2: From the perspective of urban planning, design and operation management, a multi-scale traffic accident influencing factor index system that meets the requirements is constructed and calculated at the macro level, the meso level and the micro level; Step S3, training the road section traffic accident kernel density analysis model, and using the SHAP method to calculate the SHAP value of each influencing factor; Step S4: On each road section with frequent traffic accidents, the influence of each influencing factor on the traffic accidents on the road section is compared by normalizing the SHAP absolute value, and the main influencing factor is identified.

2. The multi-scale traffic accident-prone road section cause analysis method according to claim 1, characterized in that: The step S1 specifically includes the following steps: Step S101: within the research scope, obtain the GIS data of the road network within the research scope, use the intersection as the segmentation point, and divide the road network into sections; Step S102: using the traffic accident history records, calculating the plane distribution of the traffic accident kernel density of the road section, specifically: using the geocoding tool provided by the Internet map service provider to convert the location description in the traffic accident history records into longitude and latitude, and then representing each accident record as a point on the map; performing kernel density analysis on all accident points to obtain the traffic accident kernel density distribution of the study area; the geocoding tool includes Amap API and Baidu Map API; Step S103: Calculate the traffic accident kernel density of the road section according to the traffic accident kernel density value at the location of the road section; specifically, divide each road section into equal parts according to a preset interval, and take the average value of the kernel density values ​​at all the equal division points as the traffic accident kernel density of the road section; Step S104: Use the box plot method to count the traffic accident kernel density of all road sections, and use the upper quartile and interquartile range indicators to calculate the upper limit index; identify the road sections with traffic accident kernel density values ​​exceeding the upper limit as traffic accident-prone sections.

3. The multi-scale traffic accident-prone road section cause analysis method according to claim 2, characterized in that: The preset distance in step S103 is 1 meter.

4. The multi-scale traffic accident-prone road section cause analysis method according to claim 1, characterized in that: The step S2 specifically includes the following steps: Step S201: Determine the factors affecting traffic accidents and their indicators at the macro, meso and micro levels respectively. The macro level is the urban area, and the macro level scale influencing factors reflect the impact of the urban area where the road section is located on traffic accidents on the road section; the specific influencing factors include centrality and importance, the indicator related to centrality is the circle, and its construction principle is to reflect the distance relationship between the road section and the city center, and the indicator related to importance is betweenness centrality, and its construction principle is to reflect the importance of the road section in the entire road network; The meso-level is the block. The meso-level scale influencing factors reflect the impact of the blocks around the road section on traffic accidents on the road section. The influencing factors include population and land use. The population-related indicator is population density, and its construction principle is to reflect the population size or density of the blocks around the road section. The land-related indicator is the proportion of various types of land areas, and the construction principle is to reflect the proportion of various types of major land in the blocks around the road section. The various types of major land include commercial service land, public management land, residential land, and industrial and mining land. The micro level is the road, and the micro-level scale influencing factors reflect the impact of the attributes of each road section on traffic accidents; its influencing factors are traffic operation and road safety facilities; the indicator related to traffic operation is traffic saturation, and the construction principle is to be able to reflect the traffic flow characteristics of the road section, including flow, flow speed, and density; the indicators related to road safety facilities are whether to set up a separation between vehicles and non-motorized vehicles and whether to set up a central separation, and the construction principle is to be able to reflect the construction status of road safety facilities on the road section; Step S202: using multi-source urban geographic information data, calculate the above indicators on each road section.

5. The multi-scale traffic accident-prone road section cause analysis method according to claim 1, characterized in that: The step S3 specifically comprises the following steps: Step S301: constructing a data set: setting the independent variable as the traffic accident influencing factor index, including: macro-scale index; meso-scale index; micro-scale index; the dependent variable is the traffic accident kernel density of the road section; the macro-scale index includes the circle and centrality; the meso-scale index includes population density and the proportion of various types of land use; the micro-scale index includes traffic saturation, traffic flow, whether to set up a motor vehicle and non-motor vehicle separation, and whether to set up a central separation; Step S302: model training; on the data set, the XGBOOST algorithm is used to train a multi-scale road section traffic accident analysis model; specifically: all variables of the data set are standardized, and the training set and the validation set are divided according to a ratio of 5:1; using python, the model training function in the XGBOOST package is called, the dependent variables and independent variables in the training set and the validation set are input, and the hyperparameter optimization is performed using the GridSearchCV package. The result is the road section traffic accident kernel density analysis model, which is used to reflect the quantitative relationship between the traffic accident kernel density of each road section and the traffic accident influencing factor indicators; Step S303: Model interpretation: Apply the SHAP method to interpret the model obtained by XGBOOST training, and obtain the SHAP value of each influencing factor index on each accident-prone road section; specifically: use the road section traffic accident kernel density analysis model, input the traffic accident influencing factor index of each road section, and obtain the model prediction value of the traffic accident kernel density of each road section; use python to call the SHAP package, use the model interpretation function therein, input the model prediction value of the traffic accident kernel density of each road section and the traffic accident influencing factor index, and obtain the SHAP value of each traffic accident influencing factor index on each road section, which is used to reflect the degree of influence of each influencing factor on the traffic accident of the road section. The larger the SHAP absolute value, the greater the degree of influence.

6. The multi-scale traffic accident-prone road section cause analysis method according to claim 1, characterized in that: The step S4 specifically includes the following steps: Step S401: On each road section with frequent traffic accidents, the absolute values ​​of the SHAP values ​​of all influencing factor indicators are calculated and divided by the maximum value thereof to obtain the normalized SHAP absolute value of each indicator; Step S402: Among all the indicators corresponding to each influencing factor, the largest normalized SHAP absolute value is taken as the influence degree of the influencing factor on the traffic accident on the road section; Step S403: Divide all traffic accident influencing factors on each road section into three groups: main influencing factors, secondary influencing factors and other influencing factors in descending order of influence.

7. The multi-scale traffic accident-prone road section cause analysis method according to claim 6, characterized in that: Step S403 specifically includes: traversing the influence degree characteristic values ​​of all influencing factors, taking the largest influence degree characteristic value as the reference value; dividing all influence degree characteristic values ​​by the reference value to obtain a ratio; The influencing factors with a ratio greater than or equal to 0.67 were identified as the main influencing factors; the influencing factors with a ratio greater than or equal to 0.33 and less than 0.67 were identified as secondary influencing factors, and the influencing factors with a ratio less than 0.33 were defined as other influencing factors.

Images