Method for analyzing road sections with multiple road traffic incidents based on GIS fused spatio-temporal data

Abstract

According to the GIS fusion spatio-temporal data-based road traffic event-prone road section analysis method provided by the invention, the road traffic event information in the target area range is comprehensively acquired through different acquisition modes, and the comprehensiveness and diversity of the data are ensured. According to semantic description information for describing road traffic events based on natural languages, structured event occurrence time and position information are accurately extracted from gathered semantic description information by utilizing a semantic analysis technology and are converted into comprehensive information data fused with time information, so that the accuracy and efficiency of subsequent data processing are greatly improved. The converted space-time element fusion data of the road traffic event is associated and coupled with the vector electronic map road network data, so that deep fusion of the space-time data is realized; the fusion not only facilitates the unified management and efficient utilization of data, but also provides a solid foundation for the deepening application of the spatial position of the road traffic event.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent traffic control, and in particular to a method for analyzing road sections prone to road traffic incidents based on GIS-fused spatiotemporal data. Background Art

[0002] With the rapid development of motorization, the scale of road traffic infrastructure has expanded rapidly, and the number of motor vehicles, travel frequency and driving distance have continued to increase. As a result, abnormal road traffic incidents have occurred frequently, which has brought greater difficulties and more challenges to road traffic management.

[0003] In order to effectively deal with the difficulties in road traffic incident control, conducting spatiotemporal distribution analysis of road traffic incidents and identifying road objects with high incidence and multiple occurrences of different road traffic incidents are important means to grasp the laws of incident occurrence, analyze the causes of incidents, and then strengthen road traffic management in a targeted manner and curb the subsequent occurrence of road traffic incidents.

[0004] In the prior art, the spatiotemporal distribution analysis of traffic incidents, especially the spatial position analysis, is often carried out based on a fixed road or a designated road center position, and by setting a distance range as a condition. For example, the patent with publication number CN115880906A is a solution in the prior art, which discloses a method for identifying sections with frequent traffic accidents based on an adaptive aggregation algorithm. The method adaptively divides the section units based on each accident site as the starting point, and aggregates the section units that meet the judgment criteria to generate the identification results of sections with frequent traffic accidents. From the perspective of spatial analysis, the analysis method is relatively static analysis-based, so it is relatively extensive, resulting in low efficiency in the analysis and judgment stage of road traffic incidents, making it difficult to accurately reflect the sections with prominent traffic safety hazards in the designated area, and it is difficult to discover the deep-seated correlation between road traffic incidents and road objects, and thus it is difficult to "target" and propose effective governance countermeasures. Summary of the invention

[0005] In order to solve the problem that the analysis of sections with frequent traffic incidents in the prior art is mostly carried out based on a fixed road specified range, which makes it difficult to accurately reflect the sections with prominent traffic hazards, the present invention provides a method for analyzing sections with frequent traffic incidents based on GIS fusion of spatiotemporal data, which can accurately identify road sections with frequent traffic incidents in a specified area, improves the accuracy and reliability of analysis and judgment, and provides a scientific decision-making basis for traffic management departments.

[0006] The technical solution of the present invention is as follows: a method for analyzing road sections with frequent traffic incidents based on GIS fusion of spatiotemporal data, characterized in that it comprises the following steps:

[0007] S1: Construct a general structure of semantic parsing rules to describe road traffic events;

[0008] The general structure is: standard road traffic event address = administrative division name + region name + road traffic event type + occurrence time + road name + intersection name + road direction + absolute geographical location;

[0009] The absolute geographical location includes: road markers and the distance position relative to the markers on the road;

[0010] S2: Determine the target area and build a database of semantic elements of marker numbers corresponding to the target area;

[0011] The marker semantic element database includes: semantic descriptions and specific longitudes and latitudes of addresses corresponding to all markers in the target area;

[0012] S3: Obtaining road traffic incident information occurring on all roads within the target area through different collection methods;

[0013] S4: parsing structured information describing the time and location of the event from all the valid semantic description information describing the road traffic event based on natural language, and converting the structured information into comprehensive information data integrating the time information;

[0014] S5: associating and coupling the comprehensive information data recording effective road traffic events with the road network data of the vector electronic map, and performing spatial position matching and positioning of the road traffic events based on the vector electronic map;

[0015] S6: Based on the vector electronic map, sort and filter according to the set strategic conditions, aggregate the comprehensive information data of the road traffic incidents, and identify the road sections where traffic incidents are frequent;

[0016] S7: Visually fit and present the analyzed road sections with high traffic incident incidence on the vector electronic map.

[0017] It is further characterized by:

[0018] In step S3, the collection method includes some or all of the following methods:

[0019] The way traffic participants use different types of voice communication terminals to report event information, the way intelligent traffic monitoring equipment perceives and reports event information, the way server equipment installed in the computer room analyzes video information collected by intelligent traffic equipment to generate road traffic event information, the event information generated based on Internet road traffic data analysis, and the way road traffic managers discover event information through road inspections;

[0020] The road traffic incident information includes: road traffic incident type, occurrence time, and geographical name;

[0021] The traffic event types include: one or more of traffic congestion events, traffic accident events and abnormal traffic weather events;

[0022] Step S4 specifically includes the following steps:

[0023] a1: Aggregate all valid semantic description information based on natural language describing road traffic events provided by different sources;

[0024] The semantic description information includes: the type of road traffic event, the time of occurrence, and the name of the geographical location; the name of the geographical location includes: the name of the road, intersection, and landmark;

[0025] a2: parsing the structured information describing the road traffic incident from the semantic description information through natural language processing technology and machine learning algorithms;

[0026] The structured information includes: administrative division, region, road traffic event type, date, time, road type, road name, intersection name, direction and absolute geographical location; the absolute geographical location is: the distance position of the road marker relative to the marker on the road;

[0027] a3: converting the structured information into comprehensive information data integrating time information and geographic space location information;

[0028] Converting the geographical location included in the structured information described in text form into geographical location information in a geographical coordinate system, and obtaining the precise geographical location information corresponding to the road traffic event based on the geographical location information; and combining the time information corresponding to each road traffic event with the precise geographical location information to form comprehensive information data including time and space dimensions;

[0029] The geographical location information includes: administrative division code, area code, road code, intersection code, road section code, road direction code, lane code;

[0030] The precise geographic location information includes: earth longitude and earth latitude;

[0031] The comprehensive information data includes: longitude code, latitude code and time information code;

[0032] In step a2, the following contents are included:

[0033] a21: dividing the semantic description information into address elements according to the general structure of the semantic parsing rules;

[0034] The semantic description information is a character string including time and address, and the time and address elements contained in the semantic description information are separated and decomposed into different phrases;

[0035] a22: After the time and address are segmented, each keyword of the time and address is standardized:

[0036] The standardization processing includes: abbreviation and abbreviation standardization processing, alias and nickname standardization processing, and short or long word standardization processing;

[0037] a23: semantic element correction;

[0038] The operation includes: checking the segmented time and address elements one by one with time and address elements as the minimum granularity, optimizing the accuracy of time and address elements, and correcting common words and semantics;

[0039] a24: remove duplicate semantic elements;

[0040] The operations include: determining the duplication of the segmented time and address elements and removing duplicates;

[0041] a25: semantic element supplement;

[0042] The operations include: supplementing the missing intermediate levels in the segmented time and address elements to make the element semantic information of the time and address data more complete;

[0043] a26: Verification of correctness of semantic elements;

[0044] The operations include: verifying the accuracy and availability of time and address elements, and removing incorrect times and addresses without real geographical significance;

[0045] a27: Verification of semantic element integrity;

[0046] It is to conduct integrity check according to the semantic parsing rules of the set road traffic events;

[0047] Step S5 specifically includes the following steps:

[0048] b1: Model the association relationship of the vector road network; classify all roads in the target area;

[0049] Road types include: expressways, national and provincial roads, expressways, trunk roads, secondary roads and branch roads;

[0050] b2: Divide all types of roads in the target area into segments according to the smallest unit link;

[0051] When segmenting, follow these rules:

[0052] The uplink and downlink directions of the same section are set to different links;

[0053] Encode each link and ensure the uniqueness of the code;

[0054] Set a corresponding set of geospatial longitude and latitude coordinates for each link;

[0055] b3: construct all links in the target area into a link set;

[0056] Combining the link set and its corresponding longitude and latitude coordinate set to construct a unified road link-level spatial geographic database;

[0057] b4: Based on the comprehensive information data, the marker semantic element database is associated with the link-level spatial geographic database through the mapping of geographic space longitude and latitude, so as to achieve the coupling of the comprehensive information data of road traffic events with the road network data of the vector electronic map;

[0058] Accurately map each road traffic event to the corresponding vector electronic map road network, forming a close association between road traffic events and map coordinate elements;

[0059] b5: Based on the associated coupled vector electronic map road network data, perform spatial location matching and positioning of road traffic events, and locate the identified geographical spatial information of the location of the road traffic event on the corresponding unique road link information;

[0060] By comparing the geospatial location data corresponding to the road traffic event with the road vector elements in the vector electronic map, the precise location of the event is determined and accurately marked on the map; the road vector elements include: roads, intersections and points of interest POI;

[0061] In step S6, identifying road sections where traffic incidents frequently occur includes the following steps:

[0062] c1: construct strategy conditions;

[0063] The strategy conditions include: event type, time attribute and road space attribute;

[0064] c2: constructing event spatiotemporal element indexes for the comprehensive information data, supporting rapid retrieval in multiple dimensions;

[0065] The search dimensions include: event type, time interval, and road geographic location range;

[0066] c3: Based on the dynamic algorithm of spatiotemporal element clustering and aggregation, the indexed spatiotemporal elements are calculated, and similar events occurring at similar times and locations are merged and grouped to obtain an event data set;

[0067] c4: for the road sections within the target area, the number of different types of traffic events in the event data set is sorted from large to small, so as to generate a sequence of road sections where different types of traffic events frequently occur;

[0068] In step c3, the dynamic algorithm for clustering and aggregating spatiotemporal elements includes: a dynamic statistical method for time elements based on time periods, specifically including the following steps:

[0069] d1: Set the maximum length value WinMax of the road section where traffic incidents frequently occur;

[0070] d2: In the time dimension, in the target area, all the link segments are sorted in descending order according to the number of events that occurred in the link segments;

[0071] The sorted road segments link are denoted as: L1, L2, ..., L n , where n is the total number of links included in the target area;

[0072] d3: Initialize the sequence number i of the link to be processed: i=1;

[0073] Initialize the road sign number j=1;

[0074] d4: Get L i , recorded as pending link;

[0075] d5: Get the road where the link to be processed is located, recorded as: road to be confirmed;

[0076] Determine the total length of the road to be confirmed, denoted as: RLENGH;

[0077] Obtain the name of the road to be confirmed, recorded as: RName, and store RName into the road name array NameArray;

[0078] d6: Compare the total length of the road to be confirmed RLENGH with the maximum value WinMax of the road screening sliding window;

[0079] When RLENGH>WinMax, take the center point of the link to be processed as the base point and extend to both sides by a maximum of WinMax / 2, intercept this section, and record it as: the high-incidence section; execute step d7;

[0080] When 1km≥RLENGH≤WinMax, the entire road is selected as the L1 section, which is recorded as the high-incidence section; and step d7 is executed;

[0081] When RLENGH<1km, i=i+1; loop through steps d4 to d6;

[0082] d7: Update road marking number j: j=j+1;

[0083] d8: Get L j , get L j The road you are on, determine L j The name of the road, recorded as: TempName;

[0084] Search TempName in the road name array NameArray. If TempName exists in the road name array NameArray, then loop through steps d7 to d8.

[0085] Otherwise, if TempName does not exist in the road name array NameArray, assign the value of j to i, and execute steps d4 to d8 repeatedly until all road links are involved in the calculation; then execute step d9;

[0086] d9: Acquire all the road sections with frequent traffic incidents, that is, the road sections with frequent traffic incidents included in the target area;

[0087] In step c3, the spatiotemporal element clustering dynamic algorithm includes: a spatial distance element sliding window dynamic statistical method based on a length range, specifically including the following steps:

[0088] e1: Initialize variables;

[0089] max_sum: stores the maximum value of the sum of the number of events that meet the conditions, initialized to 0;

[0090] current_a_sum: stores the sum of the number of events S in the current window, initialized to 0;

[0091] start: the starting index of the current window, initialized to 1;

[0092] best_start: the starting index of all links that meet the conditions and have the sum of event numbers, initialized to 0;

[0093] e2: Set the length of the road length screening sliding window to WinLengh;

[0094] e4: Acquire all roads included in the target area, assuming that the target area includes N roads;

[0095] e5: Obtain any road, denoted as: the road to be confirmed;

[0096] Obtain all the road segments link included in the road to be confirmed, sort all the road segments ling in the order of adjacent first and last, denoted as: Link1, Link2,......Link m , where m is the total number of road segments link included in the road to be confirmed;

[0097] e6: Starting from Link start , sequentially add the subsequent road segments link on the road to be confirmed to the road length screening sliding window in the order of adjacent first and last;

[0098] e7: Confirm whether the road segment link with the largest serial number included in the road length screening sliding window is completely included in the window;

[0099] If so, execute step e8;

[0100] If not, remove the road segment link with the largest serial number from the window, and then execute step e8;

[0101] e8: Calculate the sum current_s_sum of the number of all road traffic events that occurred between the specified time period [t min , t max among all the road segments link included in the road length screening sliding window;

[0102] e9: Compare current_s_sum and max_sum;

[0103] If current_s_sum > max_sum, replace the value in max_sum with the value of current_s_sum, and store the serial numbers of all the road segments link included in the road length screening sliding window in best_start;

[0104] Otherwise, keep the current value in max_sum;

[0105] e10: Compare start and m;

[0106] If start = m, execute step e11;

[0107] Otherwise, when start < m, update the starting index start of the previous window: start = start + 1; loop and execute steps e6 - e10;

[0108] e11: Get the value stored in max_sum and the starting index of all link segments stored in best_start.

[0109] All the road segments link stored in best_start are connected end to end to obtain the road segment with frequent road events corresponding to the road to be confirmed; the value stored in max_sum is the number of events in the road segment with frequent road events;

[0110] e12: Obtain the road sections with high incidence of road events corresponding to the N roads in the area, and obtain the N road sections with high incidence of road events corresponding to the target area.

[0111] The method for analyzing road traffic incident-prone sections based on GIS fusion spatiotemporal data provided by the present application comprehensively obtains road traffic incident information within the target area through different collection methods, ensuring the comprehensiveness and diversity of the data. For the semantic description information of road traffic incidents based on natural language, semantic parsing technology is used to accurately extract structured event occurrence time and location information from the aggregated semantic description information, and convert it into comprehensive information data with fusion time information, greatly improving the accuracy and efficiency of subsequent data processing. By associating and coupling the converted spatiotemporal element fusion data of road traffic incidents with the vector electronic map road network data, the deep fusion of spatiotemporal data is achieved; such fusion not only contributes to the unified management and efficient use of data, but also provides a solid foundation for the in-depth application of the spatial location of road traffic incidents. Based on the vector electronic map, by aggregating the spatiotemporal element information of road traffic incidents, the road sections with frequent traffic incidents can be accurately identified, which improves the accuracy and reliability of analysis and judgment, and provides a scientific decision-making basis for the traffic management department. By visually fitting and presenting the analyzed road sections with frequent traffic incidents on the vector electronic map, the analysis results are more intuitive and easy to understand, which not only helps the traffic management department to quickly grasp the road traffic conditions, but also improves the decision-making efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0112] Figure 1 It is a flowchart of a method for analyzing a road section with frequent traffic incidents;

[0113] Figure 2 An example of the source method for obtaining all road traffic incident information within the target area;

[0114] Figure 3 An example of a vector electronic map with visual markers. DETAILED DESCRIPTION

[0115] In this method, after the information data of various types of road traffic incidents in the target area are sorted out, including those reported in natural language, and combined with the Geographic Information System (GIS), they can be retrieved according to actual needs to obtain analysis results based on electronic maps that can accurately reflect the sections with prominent traffic safety hazards in the specified area.

[0116] like Figure 1 As shown, the present invention includes a method for analyzing road sections with high incidence of road traffic incidents based on GIS fusion of spatiotemporal data, which includes the following steps.

[0117] S1: Construct a general structure of semantic parsing rules to describe road traffic events;

[0118] The general structure is: standard road traffic incident address = administrative division name + region name + road traffic incident type + occurrence time + road name + intersection name + road direction + absolute geographical location. The absolute geographical location includes: road markers and the distance position relative to the markers on the road.

[0119] Common road markers include: electric poles numbered 01 (using a 2-digit code to indicate the administrative division) + 037 (using a 2-digit code to indicate the pole serial number) + 9 (using a 2-digit code to indicate the type or attribute identification), street lights numbered 03 (using a 2-digit code to indicate the administrative division) + 10201 (using a 5-digit code to indicate the road number) + 015 (using a 3-digit code to indicate the pole number), intersections, buildings with the house number XX Road XX, highway mileage signs and hundred-meter signs, etc.

[0120] S2: Determine the target area and build a database of semantic elements of markers corresponding to the target area according to the actual situation in the target area;

[0121] The marker semantic element database includes: semantic descriptions and specific longitudes and latitudes of addresses corresponding to all markers in the target area.

[0122] During specific implementation, a database of semantic elements of marker data can be collected and constructed based on historical data of road traffic incidents in the target area.

[0123] S3: Obtain the road traffic incident information on all roads within the target area through different collection methods. For example, in this embodiment, based on this method, at least two main road sections with a length of ≤5km and ≥1km in the main urban area of ​​the city where the most road traffic incidents occur are identified. In order to obtain the road traffic incident information, it is necessary to clarify the target area to which the traffic incident belongs, that is, to determine all the main roads within the main urban area of ​​the city as the target roads for analysis.

[0124] like Figure 2As shown, the collection method includes some or all of the following methods:

[0125] 1) The way traffic participants use different types of voice communication terminals to report event information; such as different types of voice terminals such as telephones and mobile phones;

[0126] 2) The way in which intelligent traffic monitoring equipment perceives and reports event information; intelligent traffic monitoring equipment, such as: traffic video surveillance equipment, laser radar, integrated radar and millimeter wave radar;

[0127] 3) A method in which a server device installed in a computer room analyzes video information collected by intelligent traffic equipment to generate traffic event information;

[0128] 4) Event information generated based on analysis of Internet road traffic data

[0129] 5) The way road traffic managers discover incident information through road inspections.

[0130] The traffic event information collected in this method includes: road traffic event type, occurrence time, and geographical name; the traffic event type includes: one or more of traffic congestion events, traffic accident events, and abnormal traffic weather events.

[0131] In specific implementation, according to the needs of data analysis, the collection time period can be a specified period of time in the past, including historical data; it can also be specified from a short period of time now to certain time periods in the future; it can also be real-time collection and analysis, and the analysis results can be dynamically updated at any time. The specific collection time period is set according to actual needs. Compared with the existing analysis methods for sections with frequent traffic incidents, which are mostly based on data collected by traffic monitoring equipment on the road surface, the data sources in this method are more diverse, and the data collection efficiency is higher and more comprehensive, ensuring that the subsequent analysis results are more accurate.

[0132] S4: From all the effective semantic description information based on natural language describing road traffic events, parse out the structured information describing the time and location information of the event, and at the same time convert the structured information into comprehensive information data integrating time information.

[0133] Step S4 specifically includes the following steps.

[0134] a1: Aggregate all valid semantic description information based on natural language describing road traffic events provided by different sources.

[0135] The semantic description information includes: the type of road traffic event, the time of occurrence, and the name of the geographical location; the name of the geographical location includes: the name of the road, intersection, and landmark.

[0136] a2: From the semantic description information, through natural language processing technology and machine learning algorithms, parse out the structured information describing the road traffic incident.

[0137] In order to facilitate data storage, retrieval and processing, all data describing traffic events in this application are converted into structured information. Structured information includes: administrative division, region, road traffic event type, date, time (hour, minute), road type, road name, intersection name, direction and absolute geographical location. Among them, the absolute geographical location is: the distance position of the road marker relative to the marker on the road. The specific data form can be expressed in the form of kilometers + meters or the distance position relative to markers such as street lamp poles.

[0138] a3: Convert structured information into comprehensive information data that integrates time information and geographic spatial location information;

[0139] The geographical location included in the structured information described in text form is converted into geographical location information in a geographical coordinate system; based on the geographical location information, the precise geographical location information corresponding to the road traffic event is obtained; and the time information corresponding to each road traffic event is combined with the precise geographical location information to form comprehensive information data including time and space dimensions;

[0140] Geographic location information includes: administrative division code, area code, road code, intersection code, section code, road direction code, lane code;

[0141] The geospatial precise location information includes: Earth longitude and Earth latitude;

[0142] The comprehensive information data includes: longitude code, latitude code and time information code.

[0143] The geographical location information includes the administrative division code + region code: a 6-digit code is used, wherein the 1st to 4th digits represent the administrative division, and the 5th to 6th digits represent the region;

[0144] Road code: 13-digit code is used, in which the first digit indicates the road category, the second to fifth digits indicate the road code, the sixth to ninth digits indicate the highway mileage pile number or the urban road intersection section number, and the tenth to thirteenth digits indicate the specific number of meters;

[0145] Road direction code: 2-bit code, indicating the east, south, west, north, southeast, northeast, southwest, northwest directions of the intersection and the up and down directions of the road section;

[0146] Lane numbering: Numbered from left to right along the direction of vehicle travel, starting with 1.

[0147] Time information code: 19-bit code, accurate to seconds yyyy-MM-dd hh:mm:ss set

[0148] Road traffic event type: 1-bit code is used, such as character a to represent a road traffic congestion event. i represents the i-th road traffic congestion event.

[0149] The structured geographical location information of road traffic incidents described in text form can be expressed as ai(xxxxxxxxxxxxxxxxxxxxxxxxx (geographic location information code), xxxxxxxxxxxxxxxxxxxx (time information code)).

[0150] Longitude information code: 10-bit code, unit is degree (°). The first bit is the sign bit, east longitude is "+" (optional), west longitude is "-", the second to fourth bits are the longitude value, the fifth bit is the decimal point, the sixth to tenth bits are the longitude minutes and seconds, converted to the decimal place of degree, retaining 5 decimal places.

[0151] Longitude information code: 10-bit code, unit is degree (°). The first bit is the sign bit, the north latitude is "+" (optional); the south latitude is "-", the second to fourth bits are the longitude value, the fifth bit is the decimal point, the sixth to tenth bits are the longitude minutes and seconds, converted to the decimal place of degree, retaining 5 decimal places.

[0152] Constituent comprehensive information data: a i (xxxxxxxxxx (longitude code), xxxxxxxxxx (latitude code), xxxxxxxxxxxxxxxxxxxx (time information code)).

[0153] The conversion relationship is:

[0154] (1)a i (Semantically represented road traffic incident address information) is converted to: a i (Structured geolocation information);

[0155] (2)a i (Structured geographic location information)+a i (time information) = a i (Comprehensive information data).

[0156] In this method, road traffic incident information described in natural language is converted into structured data based on existing natural language processing technology and machine learning algorithms, ensuring that this method can be more widely based on data from various different information sources and is more practical.

[0157] Wherein, step a2 includes the following operations.

[0158] a21: segment the semantic description information into address elements according to the general structure of the semantic parsing rules;

[0159] The description information is a string including time and address. The time and address elements contained in the description information are separated and decomposed into different phrases.

[0160] According to the time and address character string, the time and address dictionary, and the address model, the time and address elements contained in the time and address are separated and decomposed into different phrases;

[0161] For example: Now, there is a traffic jam in front of the hardware and building materials store on Wanhangdu Road, Jing'an District, Shanghai.

[0162] According to the general structure of the semantic parsing rules of road traffic events, the standard road traffic event address is decomposed into:

[0163] Administrative divisions—Shanghai;

[0164] Area Name—Jing'an District;

[0165] Type of road traffic incident – ​​traffic congestion;

[0166] Time of occurrence - now (e.g., 9:30 am, November 25, 2024 (Monday));

[0167] Road name—Wanhangdu Road;

[0168] Road direction - south to north (determined by absolute geographical location);

[0169] Absolute geographical location—the landmark is the hardware and building materials store, and the distance relative to the landmark on the road is the door.

[0170] a22: After the time and address are segmented, the keywords of time and address are standardized:

[0171] Standardization includes: abbreviation and abbreviation standardization, alias and nickname standardization, and short or long word standardization;

[0172] The time dictionary is composed of general semantic words that describe time, such as now, 5 minutes ago, yesterday, etc., which express fuzzy time. By comparing the time dictionary, the fuzzy time can be converted into precise year / month / day / hour / minute.

[0173] The address dictionary is composed of semantic words that describe the landmarks of the address, such as the North Gate of Xihui Park, Wuxi Zoo and other descriptive words that describe the landmarks on the road. By referring to the address dictionary, the described landmarks on the road can be converted into the road name, intersection name and road direction where the landmarks are located.

[0174] Standardization of abbreviations and abbreviations: "Nanchang No. 1 Middle School" is standardized to "Nanchang No. 1 Middle School", "Suzhou Housing and Urban-Rural Development Bureau" is standardized to "Suzhou Housing and Urban-Rural Development Bureau", etc.

[0175] Standardization of aliases and nicknames: "Zhe" is standardized to "Zhejiang", "Yu" is standardized to "Chongqing", etc.

[0176] Standardization of few or many words: "Nanjing East Inner Ring", "East Inner Ring of Yuhuatai District, Nanjing" is standardized to "Nanjing Yuhuatai District Inner Ring East Road", etc.

[0177] a23: semantic element correction;

[0178] Mainly check the segmented time and address elements one by one with time and address elements as the smallest granularity, optimize the accuracy of time and address elements, and correct common words and semantics;

[0179] For example, from 9:00 on October 27, 2024 to now (9:30), the east entrance of the intersection of Yingyang Road and Xinxin South Road in Yu Shui District, Xinyu City has been congested for 40 minutes.

[0180] After decomposing the general structure according to the semantic parsing rules of road traffic events, the corrected congestion start time should be 8:50, and the corrected intersection name of Yingyang Road-Xinxin South Road should be Yingyang Road-Xinxin South Avenue Intersection.

[0181] a24: remove duplicate semantic elements;

[0182] Mainly judge the duplication of the segmented time and address elements and remove duplicates;

[0183] For example, at 9:30 am on November 25, 2024, a traffic jam occurred in front of a hardware and building materials store on Wanhangdu Road, Jing'an District, Shanghai. Since the time 9:30 itself is morning time, after deduplication of semantic elements, it is converted into "At 9:30 am on November 25, 2024, a traffic jam occurred in front of a hardware and building materials store on Wanhangdu Road, Jing'an District, Shanghai". It is then further parsed according to the general structure of the semantic parsing rules for road traffic events.

[0184] a25: semantic element supplement;

[0185] It mainly supplements the missing intermediate levels in the segmented time and address elements to make the element semantic information of time and address data more complete;

[0186] For example: There is a traffic jam in front of the hardware and building materials store on Wanhangdu Road, Jing'an District, Shanghai.

[0187] According to the analyzed absolute geographical location landmark being the hardware and building materials store on Wanhangdu Road, the road direction information can be supplemented as "south to north".

[0188] a26: Verification of correctness of semantic elements;

[0189] The main purpose is to verify the accuracy and availability of time and address elements, and remove incorrect times and addresses that have no real geographical significance. In specific implementation, the time and address elements can be verified based on preset rules, and the verification method is implemented based on existing technologies.

[0190] a27: Verification of semantic element integrity;

[0191] The main purpose is to perform integrity checks according to the set semantic parsing rules for road traffic events.

[0192] According to the general structure of the semantic parsing rules of road traffic events, the parsed road traffic event address elements are automatically checked with the eight elements of the standard road traffic event address to ensure that the information structure is complete and comprehensive.

[0193] If it is determined after automatic verification that it does not meet the integrity requirements, it can be further reviewed manually and the relevant semantic elements can be manually completed. If it cannot be completed, the data is considered invalid.

[0194] S5: Associating and coupling the precise geographic spatial location information of the effective road traffic events with the road network data of the vector electronic map, and performing spatial location matching and positioning of the road traffic events based on the vector electronic map; specifically including the following steps.

[0195] b1: Model the association relationship of the vector road network; classify all roads in the target area;

[0196] Road types include: expressways, national and provincial roads, expressways, main roads, secondary roads and branch roads.

[0197] b2: Divide all types of roads in the target area into segments according to the smallest unit link;

[0198] When segmenting, follow these rules:

[0199] The uplink and downlink directions of the same section are set to different links;

[0200] Encode each link and ensure the uniqueness of the code;

[0201] Set a corresponding set of geospatial longitude and latitude coordinates for each link.

[0202] The road segment link in this application follows the definition of term 3.2 "road segment link" in the national standard GB / T 33171-2016: a directional road segment defined by two endpoints and containing an intersection or an entrance and exit.

[0203] Let the geographic longitude variable be E, the geographic dimension variable be N, and a certain road link be k i , i is a positive integer, corresponding to this section link k i The geographical longitude and latitude coordinate set of k is denoted as Q, then: Q = {k i =k(E,N)|e min ≤E≤e max , n min ≤N≤n max};

[0204] Among them, e min is the minimum value of the geographical longitude of the link to be processed, e max The maximum value of the geographical longitude of the link to be processed, n min is the minimum value of the geographical latitude of the link to be processed, n max The maximum value of the geographical latitude of the link to be processed.

[0205] b3: Build all links in the target area into a link set;

[0206] Combine the link set and its corresponding longitude and latitude coordinate set to build a unified link-level spatial geographic database;

[0207] Assume that the geographic longitude variable is recorded as E, the geographic dimension variable is recorded as N, the road link is recorded as k, and the set of all road links in the administrative area is R, then: R = {k i =k(E,N)|E min ≤E≤E max , N min ≤N≤N max , i is a positive integer}

[0208] Among them, E min is the minimum value of the geographical longitude of the administrative area to be processed, E max The maximum value of the geographical longitude of the administrative area to be processed, N min is the minimum value of the geographical latitude of the administrative area to be processed, N max It is the maximum value of the geographical latitude of the administrative area to be processed.

[0209] b4: Based on the comprehensive information data, through the mapping of geographic space longitude and latitude, the marker semantic element database is associated with the link-level spatial geographic database of the road section, realizing the coupling of the comprehensive information data of road traffic events with the road network data of the vector electronic map; each road traffic event is accurately mapped to the corresponding vector electronic map road network, forming a close association between road traffic events and map coordinate elements.

[0210] The data stored in the marker semantic element database are: the semantic description and specific longitude and latitude of the addresses corresponding to all markers in the target area. In summary, the process of establishing an association between the marker semantic element database and the link-level spatial geographic database in this application is as follows:

[0211] After the data in the marker semantic element database is structured, it is converted into geographic location information to form a geographic location information database; the geographic location information includes: administrative division code, area code, road code, intersection code, section code, road direction code, and lane code.

[0212] The geographic location information is then converted into precise geographic location information, and the time information is combined with the precise geographic location information to obtain comprehensive information data, and a geographic longitude and latitude information database is constructed based on the comprehensive information data. The precise geographic location information includes: earth longitude and earth latitude; the comprehensive information data includes longitude code, latitude code and time information code.

[0213] The geographic longitude and latitude information database is associated with the link-level spatial geographic database of the road segment in the target area. Finally, the marker semantic element database is associated with the link-level spatial geographic database of the road segment.

[0214] b5: Based on the associated coupled vector electronic map road network data, perform spatial location matching and positioning of road traffic events, and locate the identified geographical spatial information of the location of the road traffic event on the corresponding unique road link information;

[0215] By comparing the geospatial location data corresponding to the road traffic incident with the road vector elements in the vector electronic map, the precise location of the incident is determined and accurately marked on the map; road vector elements include: roads, intersections and points of interest (POIs).

[0216] The identified geographical spatial information of the location of the road traffic incident is located on the corresponding unique road link information. i (comprehensive information data) = a i (longitude code, longitude code, time code), converted to a i (link code of road section, time code);

[0217] The link code of a road section is denoted as k and is encoded using 8 digits.

[0218] S6: Based on the vector electronic map, sort and filter according to the set strategic conditions, aggregate the comprehensive information data of road traffic incidents, and identify road sections where traffic incidents are frequent; specifically, the following steps are included.

[0219] c1: construct strategy conditions;

[0220] The strategic conditions include: event type, time attributes (date, hour, minute, etc.) and road space attributes (road name, type, length, etc.). As shown in Table 1, in this method, the strategic conditions are constructed according to the three-dimensional element combination setting of event type, time attributes (date, hour, minute, etc.) and road space attributes (road name, type, length, etc.), so as to determine the prerequisites for clustering and aggregation calculation of spatiotemporal element information of road traffic events, and ensure that the screened event data can reflect the real and important road traffic conditions.

[0221] Table 1: Example of policy conditions

[0222]

[0223] In the specific implementation, each dimension element can be selected in any combination. In the embodiment of the present application, the specific strategy is: the event type is all, the time attribute is 1 month, and the road quantity and length attributes are no less than 2 main road sections with a length of ≤5km and ≥1km.

[0224] c2: Build event spatiotemporal element indexes for comprehensive information data, supporting fast retrieval in multiple dimensions;

[0225] The search dimensions include: event type, time interval and road geographic location range.

[0226] c3: Based on the clustering and aggregation dynamic algorithm, the indexed spatiotemporal elements are calculated to merge and group similar events that occurred at similar times and locations to form an event dataset containing rich spatiotemporal details;

[0227] Common clustering aggregation dynamic calculation methods mainly include two commonly used algorithms: K-means algorithm (K-means) and ISODATA algorithm. The clustering aggregation dynamic algorithm in this application can be implemented based on existing technologies, or the algorithm can be constructed by itself according to actual needs. The spatiotemporal element clustering aggregation dynamic algorithm in this embodiment includes: a time element dynamic statistical algorithm based on time period intervals and a spatial distance element dynamic programming or sliding window dynamic statistical algorithm based on length ranges.

[0228] In specific implementation, according to the set time attribute, the number of road traffic event information that meets the strategy requirements is accumulated with each trunk road as a unit to obtain the absolute number of road traffic events occurring on each trunk road in the time dimension. After obtaining the absolute number of road traffic events occurring on each trunk road in the time dimension, the dynamic programming or sliding window dynamic statistical algorithm based on the spatial distance element of the length range is applied. Specifically, in an embodiment of the present application, the window of the sliding window dynamic statistical algorithm based on the spatial distance element of the length range is the trunk road section with the most road traffic events occurring with a length of ≤5km and ≥1km, and the step length is the trunk road section link. The section link refers to the basic unit in the road model of the high-precision vector electronic map, which is the smallest directional road section.

[0229] In step c3, the dynamic algorithm for clustering and aggregating spatiotemporal elements includes: a dynamic statistical method for time elements based on time periods, specifically including the following steps:

[0230] d1: Set the maximum length value WinMax of the road section where traffic incidents frequently occur; WinMax is set according to actual needs. In this embodiment, WinMax is set to 5km;

[0231] d2: In the time dimension, in the target area, all the link segments are sorted in descending order according to the number of events that occurred in the link segments;

[0232] The sorted road links are recorded as: L1, L2, ..., L n , where n is the total number of links included in the target area;

[0233] d3: Initialize the sequence number i of the link to be processed: i=1;

[0234] Initialize the road sign number j=1;

[0235] d4: Get L i , recorded as pending link;

[0236] d5: Get the road where the link to be processed is located, recorded as: road to be confirmed;

[0237] Determine the total length of the road to be confirmed, denoted as: RLENGH;

[0238] Get the name of the road to be confirmed, recorded as: RName, and store RName into the road name array NameArray;

[0239] d6: Compare the total length of the road to be confirmed RLENGH with the maximum value WinMax of the road screening sliding window;

[0240] When RLENGH>WinMax, take the center point of the link to be processed as the base point and extend to both sides by a maximum of WinMax / 2, intercept this section, and record it as: the high-incidence section; execute step d7;

[0241] When 1km≥RLENGH≤WinMax, the entire road is selected as the L1 section, which is recorded as the high-incidence section; and step d7 is executed;

[0242] When RLENGH<1km, i=i+1; loop through steps d4 to d6;

[0243] d7: Update road marking number j: j=j+1;

[0244] d8: Get L j , get L j The road you are on, determine L j The name of the road, recorded as: TempName;

[0245] Search TempName in the road name array NameArray. If TempName exists in the road name array NameArray, then loop through steps d7 to d8.

[0246] Otherwise, if TempName does not exist in the road name array NameArray, assign the value of j to i, and execute steps d4 to d8 repeatedly until all road links are involved in the calculation; then execute step d9;

[0247] d9: Get all the high-incident road sections, that is, the high-incident road sections within the target area.

[0248] In step c3, the dynamic algorithm for clustering and aggregation of spatiotemporal elements includes: a dynamic statistical method of sliding window of spatial distance elements based on a length range, which specifically includes the following steps.

[0249] Before applying the dynamic statistical method of sliding window of spatial distance elements based on length range, it is necessary to first clarify the road traffic event type conditions, time attribute conditions, and road attribute conditions.

[0250] Assume: the road traffic event type variable is a, the time variable is t, and the road attribute condition is l;

[0251] Let: a = A (A = traffic congestion event), l = L (L = main road), t min is the lower limit of the set time, t max The time limit is set.

[0252] Note: At is the number of A events occurring at time t, t∈(t min ,tmax ), L p is the pth trunk road within the main urban area, k q is the qth trunk road link, the geographic longitude variable is recorded as E, and the geographic latitude variable is recorded as N, where variables p and q are positive integers.

[0253] S represents the number of events A occurring on link k of the trunk road under the set strategy conditions, then k q The number of occurrences of event A under the set policy conditions is:

[0254]

[0255] in, Indicates k q The number of occurrences of event A under the set policy conditions, represents the geographical longitude of kq, represents the geographical dimension of kq; based on the above formula, the number of occurrences of A events for each trunk road link k under the set strategy conditions can be calculated.

[0256] The method to determine the geographic space length of each link is as follows: Assume that a link k q The starting and ending latitude and longitude (E start , N start , E end , N end) , you can use the Haversine formula to calculate the distance between two geographic coordinates:

[0257]

[0258] in:

[0259] φ is latitude, λ is longitude, the starting point coordinates are (λ1, φ1), and the end point coordinates are (λ2, φ2); Δφ=φ2-φ1, Δλ=λ2-λ1, and r is the radius of the earth (usually 6371km). For the link kq, its length can be calculated According to the above formula, the geographic space length of link k on each trunk road can be calculated.

[0260] The specific method for screening road sections with frequent traffic incidents includes the following steps:

[0261] e1: Initialize variables;

[0262] max_sum: stores the maximum value of the sum of the number of road traffic events that meet the conditions, initialized to 0;

[0263] current_a_sum: stores the sum of the number of road traffic events S in the current window, initialized to 0;

[0264] start: the starting index of the current window, initialized to 1;

[0265] best_start: the starting index of all links that meet the conditions and have the sum of event numbers, initialized to 0;

[0266] e2: Set the length of the road length screening sliding window to WinLengh; WinLengh is set according to actual needs. In this embodiment, WinLengh is set to 5km;

[0267] e4: Get all roads included in the target area. Assume that the target area includes N roads.

[0268] e5: Get any road, recorded as: road to be confirmed;

[0269] Get all the road segments link included in the road to be confirmed, and sort all the road segments ling in the order of the first adjacent first, and record them as: Link1, Link2, ... Link m , where m is the total number of links included in the road to be confirmed;

[0270] e6: Link start As the starting position, the subsequent road segments link on the road to be confirmed are added to the road length screening sliding window in the order of the first and the last adjacent;

[0271] e7: confirm whether the road segment link with the largest sequence number included in the road length screening sliding window is completely included in the window;

[0272] If yes, execute step e8;

[0273] If not, remove the link with the largest sequence number from the window, and then execute step e8;

[0274] e8: Calculate the road length of all road segments included in the sliding window link in the specified time period [t min ,t max ]The sum of all road traffic incidents occurring between current_s_sum;

[0275] e9: compare current_s_sum and max_sum;

[0276] If current_s_sum > max_sum, replace the value in max_sum with the value of current_s_sum, and store the sequence numbers of all sections link included in the road length screening sliding window into best_start;

[0277] Otherwise, retain the current value in max_sum;

[0278] e10: Compare start and m;

[0279] If start = m, execute step e11;

[0280] Otherwise, when start < m, update the starting index start of the previous window: start = start + 1; loop to execute steps e6 - e10;

[0281] e11: Obtain the value stored in max_sum and the starting indices of all sections link stored in best_start,

[0282] Connect all sections link stored in best_start end to end, then the accident - prone sections corresponding to the road to be confirmed are obtained; the value stored in max_sum is the number of accident values of the accident - prone sections of the road;

[0283] e12: Obtain the accident - prone sections corresponding to N roads in the area respectively, then N accident - prone sections corresponding to the target area are obtained. Specifically, how many accident - prone sections are selected for each road can be selected according to actual needs, and it is controlled based on the number of the sum of road traffic accident numbers stored in max_sum. In this embodiment, one accident - prone section is selected for each road.

[0284] In this application, the accident - prone sections of road traffic in the target area are calculated in two ways: the dynamic statistical method of the spatial distance element sliding window based on the length range and the dynamic statistical method of the time element based on the time period interval. The results obtained by the two methods are the same. Among them, the calculation efficiency of the dynamic statistical method of the time element based on the time period interval is higher, while the accuracy of the dynamic statistical method of the spatial distance element sliding window based on the length range is higher. In practical applications, to ensure the accuracy of the calculation results, the two methods can be verified with each other. If they are inconsistent, the problem points can be checked. Generally speaking, it is preferably to select the dynamic statistical method of the spatial distance element sliding window based on the length range for calculation.

[0285] c4: For the selected road sections within the target area, the number of different types of traffic events in the event data set is sorted from large to small, so as to generate the order of road sections where different types of traffic events are frequent.

[0286] In this embodiment, two urban trunk roads are finally selected. The two roads have different names and are not continuous. The length of each sampled road is ≥1km and ≤5km.

[0287] S7: Visually fitting and presenting the analyzed road sections with frequent traffic incidents on the vector electronic map, which specifically includes the following steps.

[0288] d1: Based on the types of traffic events and the number of each type of traffic event, the sorting and screening results are displayed on a high-precision vector electronic map to visualize the distribution map of accident-prone sections.

[0289] d2: Set different visual markers for different road sections; present these markers in a dynamic or static manner on the vector electronic map to intuitively display the road sections where traffic incidents are frequent. Visual markers include: color, icon or line thickness.

[0290] In this embodiment, the two main road sections with the most road traffic incidents that are ≤5km and ≥1km in length that are sorted and screened out are used to draw a visual distribution map on a high-precision vector electronic map, and different visual marks such as different colors, icons, and line thicknesses are set for different sections.

[0291] d3: Generate a list of accident-prone road sections.

[0292] Reference Figure 3 , the left side of the figure is a list of accident-prone sections in the target area; the right side of the figure is a vector electronic map with visual markings corresponding to one of the accident-prone sections.

[0293] After using the technical solution of the present invention, the ability to monitor and analyze road traffic incidents can be improved, which helps traffic management departments to more accurately grasp road traffic conditions, take effective measures in a timely manner, ensure the safety and smoothness of road traffic, and improve the ability to control road traffic incidents. In addition, the application of this method also helps to promote the development of intelligent transportation technology and systems and improve the intelligent level of road traffic management.

Claims

1. A road traffic incident-prone section analysis method based on GIS fusion of spatiotemporal data, characterized in that: It includes the following steps: S1: Construct a general structure of semantic parsing rules to describe road traffic events; The general structure is: standard road traffic incident address = administrative division name + region name + road traffic incident type + occurrence time + road name + intersection name + road direction + absolute geographical location; The absolute geographical location includes: road markers and the distance position relative to the markers on the road; S2: Determine the target area and build a database of semantic elements of marker numbers corresponding to the target area; The marker semantic element database includes: semantic descriptions and specific longitudes and latitudes of addresses corresponding to all markers in the target area; S3: Obtaining road traffic incident information occurring on all roads within the target area through different collection methods; S4: parsing structured information describing the time and location of the event from all the valid semantic description information describing the road traffic event based on natural language, and converting the structured information into comprehensive information data integrating the time information; S5: associating and coupling the comprehensive information data recording effective road traffic events with the road network data of the vector electronic map, and performing spatial position matching and positioning of the road traffic events based on the vector electronic map; S6: Based on the vector electronic map, sort and filter according to the set strategic conditions, aggregate the comprehensive information data of the road traffic incidents, and identify the road sections where traffic incidents are frequent; S7: Visually fit and present the analyzed road sections with high traffic incident incidence on the vector electronic map.

2. The method for analyzing road sections with high incidence of road traffic incidents based on GIS fusion of spatiotemporal data according to claim 1 is characterized by: In step S3, the collection method includes some or all of the following methods: The method in which traffic participants use different types of voice communication terminals to report event information, the method in which intelligent traffic monitoring equipment perceives and reports event information, the method in which server equipment installed in the computer room analyzes video information collected by intelligent traffic equipment to generate road traffic event information, the event information generated based on Internet road traffic data analysis, and the method in which road traffic managers discover event information through road inspections.

3. The method for analyzing road sections with high traffic incident incidence based on GIS fusion of spatiotemporal data according to claim 1 is characterized by: The road traffic incident information includes: road traffic incident type, occurrence time, and geographical name; The traffic event types include: one or more of traffic congestion events, traffic accident events and abnormal traffic weather events.

4. The method for analyzing road sections with high traffic incident incidence based on GIS fusion of spatiotemporal data according to claim 1 is characterized by: Step S4 specifically includes the following steps: a1: Aggregate all valid semantic description information based on natural language describing road traffic events provided by different sources; The semantic description information includes: the type of road traffic event, the time of occurrence, and the name of the geographical location; the name of the geographical location includes: the name of the road, intersection, and landmark; a2: parsing the structured information describing the road traffic incident from the semantic description information through natural language processing technology and machine learning algorithms; The structured information includes: administrative division, region, road traffic event type, date, time, road type, road name, intersection name, direction and absolute geographical location; the absolute geographical location is: the distance position of the road marker relative to the marker on the road; a3: converting the structured information into comprehensive information data integrating time information and geographic space location information; Converting the geographical location included in the structured information described in text form into geographical location information in a geographical coordinate system, and obtaining the precise geographical location information corresponding to the road traffic event based on the geographical location information; and combining the time information corresponding to each road traffic event with the precise geographical location information to form comprehensive information data including time and space dimensions; The geographical location information includes: administrative division code, area code, road code, intersection code, road section code, road direction code, lane code; The precise geographic location information includes: earth longitude and earth latitude; The comprehensive information data includes: longitude code, latitude code and time information code.

5. The method for analyzing road sections with high traffic incident incidence based on GIS fusion of spatiotemporal data according to claim 4 is characterized by: In step a2, the following contents are included: a21: dividing the semantic description information into address elements according to the general structure of the semantic parsing rules; The semantic description information is a character string including time and address, and the time and address elements contained in the semantic description information are separated and decomposed into different phrases; a22: After the time and address are segmented, each keyword of the time and address is standardized: The standardization processing includes: abbreviation and abbreviation standardization processing, alias and nickname standardization processing, and short or long word standardization processing; a23: semantic element correction; The operation includes: checking the segmented time and address elements one by one with time and address elements as the minimum granularity, optimizing the accuracy of time and address elements, and correcting common words and semantics; a24: remove duplicate semantic elements; The operations include: determining the duplication of the segmented time and address elements and removing duplicates; a25: semantic element supplement; The operations include: supplementing the missing intermediate levels in the segmented time and address elements to make the element semantic information of the time and address data more complete; a26: Verification of correctness of semantic elements; The operations include: verifying the accuracy and availability of time and address elements, and removing incorrect times and addresses without real geographical significance; a27: semantic element integrity verification; The integrity check is performed according to the set semantic parsing rules of road traffic events.

6. The method for analyzing road sections with high incidence of road traffic incidents based on GIS fusion of spatiotemporal data according to claim 1 is characterized by: Step S5 specifically includes the following steps: b1: Model the association relationship of the vector road network; classify all roads in the target area; Road types include: expressways, national and provincial roads, expressways, trunk roads, secondary roads and branch roads; b2: Divide all types of roads in the target area into segments according to the smallest unit link; When segmenting, follow these rules: The uplink and downlink directions of the same section are set to different links; Encode each link and ensure the uniqueness of the code; Set a corresponding set of geospatial longitude and latitude coordinates for each link; b3: construct all links in the target area into a link set; Combining the link set and its corresponding longitude and latitude coordinate set to construct a unified road link-level spatial geographic database; b4: Based on the comprehensive information data, the marker semantic element database is associated with the link-level spatial geographic database through the mapping of geographic space longitude and latitude, so as to achieve the coupling of the comprehensive information data of road traffic events with the road network data of the vector electronic map; Accurately map each road traffic event to the corresponding vector electronic map road network, forming a close association between road traffic events and map coordinate elements; b5: Based on the associated coupled vector electronic map road network data, perform spatial location matching and positioning of road traffic events, and locate the identified geographical spatial information of the location of the road traffic event on the corresponding unique road link information; By comparing the geographic spatial location data corresponding to the road traffic event with the road vector elements in the vector electronic map, the precise location of the event is determined and accurately marked on the map; the road vector elements include: roads, intersections and points of interest POI.

7. The method for analyzing road sections with high traffic incident incidence based on GIS fusion of spatiotemporal data according to claim 1 is characterized by: In step S6, identifying road sections where traffic incidents frequently occur includes the following steps: c1: construct strategy conditions; The strategy conditions include: event type, time attribute and road space attribute; c2: constructing event spatiotemporal element indexes for the comprehensive information data, supporting rapid retrieval in multiple dimensions; The search dimensions include: event type, time interval, and road geographic location range; c3: Based on the dynamic algorithm of spatiotemporal element clustering and aggregation, the indexed spatiotemporal elements are calculated, and similar events occurring at similar times and locations are merged and grouped to obtain an event data set; c4: For the road sections within the target area, the number of different types of traffic events in the event data set is sorted from large to small, so as to generate a sequence of road sections where different types of traffic events are frequent.

8. The method for analyzing road sections with high incidence of road traffic incidents based on GIS fusion of spatiotemporal data according to claim 7 is characterized by: In step c3, the dynamic algorithm for clustering and aggregating spatiotemporal elements includes: a dynamic statistical method for time elements based on time periods, specifically including the following steps: d1: Set the maximum length value WinMax of the road section where traffic incidents frequently occur; d2: In the time dimension, in the target area, all the link segments are sorted in descending order according to the number of events that occurred in the link segments; The sorted road segments link are denoted as: L1, L2, ..., L n , where n is the total number of links included in the target area; d3: Initialize the sequence number i of the link to be processed: i=1; Initialize the road sign number j=1; d4: Get L i , recorded as pending link; d5: Get the road where the link to be processed is located, recorded as: road to be confirmed; Determine the total length of the road to be confirmed, denoted as: RLENGH; Obtain the name of the road to be confirmed, recorded as: RName, and store RName into the road name array NameArray; d6: Compare the total length of the road to be confirmed RLENGH with the maximum value WinMax of the road screening sliding window; When RLENGH>WinMax, take the center point of the link to be processed as the base point and extend to both sides by a maximum of WinMax / 2, intercept this section, and record it as: the high-incidence section; execute step d7; When 1km ≥ RLENGH ≤ WinMax, select the entire road as section L1, denoted as: accident-prone section; execute step d7; When RLENGH < 1km, i = i + 1; loop and execute steps d4 - d6; d7: Update the road marking serial number j: j = j + 1; d8: Get L j , get L j The road you are on, determine L j The name of the road, recorded as: TempName; Query TempName in the road name array NameArray. If TempName exists in the road name array NameArray, loop and execute steps d7 - d8; Otherwise, if TempName does not exist in the road name array NameArray, assign the value of j to i, loop and execute steps d4 - d8 until all sections link are involved in the calculation; execute step d9; d9: Obtain all the accident-prone sections, which are the accident-prone sections of road traffic events included in the target area.

9. The method for analyzing road sections with high incidence of road traffic incidents based on GIS fusion of spatiotemporal data according to claim 7 is characterized by: In step c3, the spatio-temporal element clustering and aggregation dynamic algorithm includes: a dynamic statistical method for spatial distance elements sliding window based on length range, specifically including the following steps: e1: Initialize variables; max_sum: Store the maximum value of the sum of the number of events meeting the conditions, initialized to 0; current_a_sum: Store the sum of the number of events S in the current window, initialized to 0; start: The starting index of the current window, initialized to 1; best_start: The starting index of all sections link that meet the conditions and have the sum of the number of events, initialized to 0; e2: Set the length of the road length screening sliding window to WinLengh; e4: Obtain all the roads included in the target area. Assume there are N roads in the target area; e5: Obtain any one road, denoted as: road to be confirmed; Get all the road segments link included in the road to be confirmed, and sort all the road segments ling in the order of the first adjacent first, and record them as: Link1, Link2, ... Link m , where m is the total number of links included in the road to be confirmed; e6: Link start As the starting position, the subsequent road segments link on the road to be confirmed are added to the road length screening sliding window in the order of the first and the last adjacent; e7: Confirm whether the section link with the largest serial number included in the road length screening sliding window is completely included in the window; If so, execute step e8; If not, remove the section link with the largest serial number from the window, and then execute step e8; e8: Calculate the road length of all road segments link included in the sliding window in the specified time period [t min ,t max ]The sum of all road traffic incidents occurring between current_s_sum; e9: Compare current_s_sum and max_sum; If current_s_sum > max_sum, replace the value in max_sum with the value of current_s_sum, and store the serial numbers of all section links included in the road length screening sliding window in best_start; Otherwise, retain the current value in max_sum; e10: Compare start and m; If start = m, execute step e11; Otherwise, when start < m, update the starting index start of the previous window: start = start + 1; loop and execute steps e6 - e10; e11: Obtain the value stored in max_sum and the starting index of all section links stored in best_start, All the road segments link stored in best_start are connected end to end to obtain the road segment with frequent road events corresponding to the road to be confirmed; the value stored in max_sum is the number of events in the road segment with frequent road events; e12: Obtain the road sections with high incidence of road events corresponding to the N roads in the area, and obtain the N road sections with high incidence of road events corresponding to the target area.

Images