A method for constructing a traffic safety knowledge graph of a continuous longitudinal slope road section of a highway

Abstract

The application belongs to the cross field of traffic engineering and information technology, and discloses a highway continuous longitudinal slope section traffic safety knowledge graph construction method.The method integrates section characteristics, traffic flow, historical accidents and environmental data, and constructs a multi-source data set through cleaning and standardization processing; defines section, accident, environment and vehicle entities, and constructs a multi-level knowledge graph containing a static topology layer, a dynamic attribute layer and a probability reasoning layer; uses natural language processing technology to extract knowledge triples, combines with a rule engine to dynamically fuse the correlation logic of slope, environment and accident data, realizes real-time risk assessment and early warning; uses a graph database to store the graph, supports semantic query and dynamic update.The application breaks through the limitations of traditional single data source analysis, solves the coupling risk assessment problem of dynamic environmental factors and static characteristics under complex terrain through multi-source fusion and probability reasoning mechanism, and significantly improves the traffic safety risk prediction accuracy.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of traffic engineering and information technology, and provides a highway continuous longitudinal slope section traffic safety knowledge graph construction method. BACKGROUND

[0002] With the rapid development of highway traffic network, continuous longitudinal slope sections become high-traffic-accident areas due to their special topographic conditions (such as large slope changes and small curve radii). Traditional traffic safety analysis methods are mostly based on a single data source (such as historical accident statistics or static road section parameters), and lack comprehensive utilization of multi-source heterogeneous data (such as dynamic traffic flow and real-time environmental parameters). For example, the existing technology often evaluates risk through the linear relationship between slope and slope length, but ignores the coupling influence of dynamic environmental factors such as weather and road friction coefficient on accident probability, resulting in insufficient prediction accuracy of the risk assessment model in actual scenarios.

[0003] In the field of knowledge graph technology, existing methods mostly focus on entity relationship modeling of general traffic networks (such as road topology and vehicle trajectory), and do not design a dedicated knowledge representation system for continuous longitudinal slope sections. For example, some studies use RDF models or property graphs to store static relationships, but lack real-time fusion capability for dynamic time-series data (such as weather state changes (t) and friction coefficient μ(t)), and the modeling of the association relationship between entities mostly relies on deterministic rules, without introducing a probabilistic reasoning mechanism to reflect the randomness of risk evolution.

[0004] Although recent studies have attempted to integrate multi-dimensional data to construct a risk graph (such as the patent CN 119399965 A of China Communications First Highway Survey & Design Institute Co., Ltd.), the model still has the following limitations:

[0005] 1) The association logic of static road section features and dynamic environmental parameters is not clearly distinguished, limiting the reasoning ability of the rule engine;

[0006] 2) A multi-level knowledge mode (such as the static topology layer, dynamic attribute layer, and probabilistic reasoning layer) is not established, making it difficult to achieve hierarchical mapping from data to knowledge;

[0007] 3) Real-time warning function relies on fixed threshold judgment, without dynamic adjustment of risk level based on Bayesian networks and other probabilistic models.

[0008] To address the above problems, the present patent proposes a traffic safety knowledge graph construction method for highway continuous longitudinal slope sections, which solves the deficiencies of traditional methods in data integration and real-time risk assessment through key technologies such as multi-source data fusion, hierarchical knowledge modeling, and dynamic probabilistic reasoning, and provides fine-grained decision support for traffic safety management under complex topographic conditions. SUMMARY

[0009] The purpose of the present application aims to solve the problems of insufficient integration of multi-source heterogeneous data, lack of coupling risk assessment of dynamic environmental factors and static road characteristics in traditional safety analysis method of highway continuous longitudinal slope section, and low prediction accuracy caused by real-time warning relying on fixed threshold, and to realize fine traffic safety management by constructing multi-level knowledge graph integrating static topology, dynamic attributes and probabilistic reasoning.

[0010] In order to achieve the above purpose, the present application adopts the following technical means:

[0011] The present application provides a highway continuous longitudinal slope section traffic safety knowledge graph construction method, comprising the following steps:

[0012] Step 1, data acquisition and preprocessing: integrate multi-source heterogeneous data, including road feature data, traffic flow data, historical accident data and environment data, and perform cleaning, deduplication and standardization processing on the data, and output standardized data set;

[0013] Step 2, knowledge representation and modeling: define the entity, attribute and relationship between entities of knowledge graph, construct a multi-level mode containing road entity, accident entity and environment entity, the mode includes static topology layer, dynamic attribute layer and probabilistic reasoning layer;

[0014] Step 3, knowledge extraction and construction: extract entity attribute value and entity relationship from structured data, perform named entity recognition, relationship extraction and event extraction on unstructured text data through natural language processing technology, fuse multi-source data and solve entity reference ambiguity and conflict, and generate knowledge triplets;

[0015] Step 4, knowledge reasoning and application: realize dynamic risk assessment based on preset rule engine, the rule engine integrates the correlation logic of slope characteristics, environmental parameters and accident data, and generates real-time risk warning signal;

[0016] Step 5, system deployment and update: store the knowledge graph in the graph database, and provide semantic query, risk warning and graph dynamic update functions through the visual interface.

[0017] Further, step 1 comprises the following steps:

[0018] Step 1.1, multi-source data acquisition, four types of heterogeneous data of highway continuous longitudinal slope section are collected:

[0019] Road feature data: slope , longitudinal slope continuous length , curve radius , represents the road ID, denoted as vector , wherein the uphill > 0, downhill <0;

[0020] Traffic flow data: time series traffic volume , vehicle average speed , constitute time series matrix ;

[0021] Accident data: accident type , timestamp , severity level {1,2,3}, constitute event set ;

[0022] Environmental data: weather state , road surface friction coefficient , visibility , constitute time series matrix ;

[0023] Step 1.2, clean, deduplicate and standardize the data, output the standardized dataset:

[0024] .

[0025] Further, the entities of the knowledge graph defined in step 2 specifically include the following steps:

[0026] Define four types of core entities and their attributes, and use vector space and probability model for formal description:

[0027] Let the entity set of the knowledge graph be:

[0028]

[0029] Define the attribute definition set of all entities = , defined as follows:

[0030] The attribute set of the road segment entity :

[0031]

[0032] Wherein: represents the road segment ID, is the slope, is the elevation difference, is the horizontal distance, is the length of longitudinal slope continuity, and R is the radius of curvature of the curve;

[0033] The attribute combination of the accident entity , representing the accident characteristics, is defined as a multivariate group:

[0034]

[0035] in: Code the accident type. This is the timestamp of the accident. The severity level of the accident is categorized as follows: 1: Minor, 2: Moderate, 3: Major.

[0036] Environmental entities The set of attributes:

[0037] ={ , , }

[0038] in: For weather conditions, For visibility, Let be the coefficient of friction of the road surface, which follows the law of friction. Conditional distribution:

[0039] =

[0040] vehicle entity The set of attributes representing vehicle traffic characteristics is defined as a set of dynamic parameters:

[0041]

[0042] in: Vehicle type , For real-time vehicle speed, Indicates vehicle acceleration. Indicates the braking performance coefficient, and Related.

[0043] The mapping relationship with Type is as follows:

[0044] .

[0045] Furthermore, step 2, specifically the relationships between attributes and entities, includes the following steps:

[0046] Define the set of relationships between entities as:

[0047]

[0048] Each relation is defined as a weighted directed edge, where the weight represents the strength or probability of the association, as follows:

[0049] Road section - causal relationship of the accident:

[0050] Modeling the causal relationship between road segment features and accidents using a conditional probability model:

[0051]

[0052] wherein denotes the road segment feature vector, , denotes the feature weight coefficient, corresponding to the weight of the slope , length of longitudinal slope , radius of curve , , , , and σ(·) denotes the Sigmoid function;

[0053] Environment-accident risk probability association:

[0054] Describing the dynamic influence of environmental factors on accident risk using a Bayesian network model:

[0055]

[0056] denotes the accident severity level, denotes the indicator function, taking 1 in bad weather and 0 otherwise, , , denotes the environmental factor weight;

[0057] Road segment-environment coupling risk:

[0058] Defining the interaction between slope and environmental factors using a coupling risk index:

[0059]

[0060] : weather-related road surface friction coefficient, is the coupling coefficient.

[0061] Road segment-environment-vehicle coupling risk:

[0062]

[0063] is the road surface friction coefficient provided by the environmental entity, where , , are all coupling coefficients.

[0064] Further, step 2 constructs a multi-level model containing road segment entities, accident entities, and environmental entities, which includes a static topology layer, a dynamic attribute layer, and a probability reasoning layer, specifically including the following steps:

[0065] Building a multi-level pattern

[0066] The knowledge graph pattern is represented using a hierarchical graph structure, defining three levels:

[0067] Static topology layer

[0068] Adjacency matrices are used to describe fixed relationships between entities. :

[0069]

[0070] This indicates that at least one exists;

[0071] Dynamic attribute layer

[0072] Using time tensors This indicates that entity attributes change over time. For the number of entities, For time step, For attribute dimensions;

[0073] Probabilistic Inference Layer

[0074] Integrate association rules and risk models to define the rule engine logic:

[0075] Risk assessment rules:

[0076]

[0077] Indicates the overall risk index The medium risk threshold, Indicates the overall risk index High risk threshold This represents a set of recent accident patterns, specifically those similar accidents that have occurred within a time window.

[0078] Furthermore, step 2 also includes schema extensibility steps:

[0079] Define the pattern extension protocol:

[0080] Entity compatibility: Adding new entities It must share at least one common attribute with existing entities. Specific attribute interface:

[0081]

[0082] Indicates the addition of a new entity. Represents the union of all existing entity attributes. This represents the intersection operation, which finds common attributes. denotes that the intersection is non-empty, i.e. there is at least one common attribute.

[0083] Relation inheritance: hierarchical expansion through relation generalization:

[0084]

[0085] related_to: newly added default generic relation, indicating that there is some unrefined association between entities, allowing gradual refinement.

[0086] Further, step 2 also includes a formal verification step:

[0087] To ensure the consistency of the schema logic, define the constraint conditions:

[0088] Causal closure: if then there must be a reverse association to ensure the bidirectional traceability of the knowledge graph, denotes the forward relationship, denotes the reverse relationship.

[0089] Further, step 4 includes the following steps:

[0090] Step 4.1, Rule Engine Initialization

[0091] Load preset association rules and risk models, reference conditional probability, Bayesian network and coupled risk index defined in section 2.2, build an extensible inference logic library:

[0092]

[0093] Step 4.2, Dynamic Risk Assessment

[0094] Input real-time data stream , perform multi-dimensional risk calculation:

[0095] Single-factor risk quantification

[0096] Slope accident probability:

[0097] denotes the slope-related risk probability of the kth accident under real-time data,

[0098] Environmental risk value:

[0099] Coupling risk coefficient:

[0100] Comprehensive risk model, define the weighted comprehensive risk index:

[0101] +

[0102] wherein , is a weight coefficient, satisfying , determined by historical accident data fitting;

[0103] Step 4.3, real-time risk early warning, generating a three-level early warning signal based on the risk threshold value:

[0104]

[0105] represents the medium risk threshold value of the comprehensive risk index , represents the high risk threshold value of the comprehensive risk index , represents a recent accident pattern set, specifically a same type of accident that has occurred within a time window.

[0106] Because the present application adopts the above technical means, it has the following beneficial effects:

[0107] 1. Through multi-source heterogeneous data fusion and standardization processing technology, the problem of data isolation and quality difference in traditional methods is solved, and a high-precision data foundation is constructed, as follows:

[0108] Technical means: integrating four types of heterogeneous data of road segment characteristics, traffic flow, accidents and environment, using cleaning, deduplication and standardization processing to generate a unified data set.

[0109] Technical problem: traditional methods rely on a single data source and cannot comprehensively consider dynamic environment and static road segment characteristics, resulting in one-sided risk assessment.

[0110] Effect: the data coverage completeness is improved to four dimensions (slope, traffic flow, accident, environment), and the standardization error is reduced, such as the road surface friction coefficient in environmental data is dynamically calibrated according to weather (rainy , sunny ), ensuring data reliability.

[0111] 2. Through multi-level knowledge graph modeling technology, the problem of static and dynamic data fragmentation is solved, and the spatio-temporal adaptability of risk reasoning is improved, as follows:

[0112] Technical means: constructing a static topology layer, a dynamic attribute layer and a probability reasoning layer, and defining the weighted relationship between entities (such as road segment-accident causal probability ).

[0113] Technical problem: traditional graphs lack dynamic temporal attributes and probability correlations, and cannot reflect real-time risk changes.

[0114] Effect: Dynamic attribute layer with time series tensor Support minute-level update, probabilistic inference layer integrates Bayesian network (such as ), and risk prediction accuracy is improved.

[0115] 3. Through the dynamic rule engine and the comprehensive risk model, the problem of static threshold early warning lag is solved, real-time risk grading control is realized, and the specific implementation is as follows:

[0116] Technical means: integrate slope conditional probability, environmental Bayesian network and coupled risk index , calculate weighted risk .

[0117] Technical problem: fixed threshold cannot adapt to dynamic changes such as weather and traffic flow, and false alarm rate is high.

[0118] Effect: early warning response time is shortened, and high risk identification rate is improved.

[0119] 4. Through the graph database and the visualization interface technology, the problems of low query efficiency and difficult update of the traditional system are solved, the operation and maintenance efficiency is improved, and the specific implementation is as follows:

[0120] Technical means: adopt Neo4j and other graph databases to store knowledge graph, provide semantic query interface and dynamic update function.

[0121] Technical problem: relational database cannot efficiently process complex entity association.

[0122] Effect: complex query response time <200ms, support thousands of concurrent updates per second, and graph version iteration period is reduced from weekly to hourly.

[0123] 6. Through the mode expansion protocol and formal verification, the problem of poor compatibility of knowledge graph is solved, the system scalability and logical consistency are ensured, and the specific implementation is as follows:

[0124] Technical means: define entity compatibility conditions and causal closure constraints.

[0125] Technical problem: adding entities easily causes data island and logical conflict.

[0126] Effect: support seamless access to new entities such as construction areas, and the consistency verification pass rate of the extended graph is 100%. BRIEF DESCRIPTION OF DRAWINGS

[0127] Figure 1 The flowchart of the present application. DETAILED DESCRIPTION

[0128] The embodiments of the present application will be described in detail below. Although the present application will be described and illustrated with respect to some specific embodiments, it is noted that the present application is not limited to only these embodiments. Rather, modifications and equivalents of the present application are intended to be covered by the scope of the claims of the present application.

[0129] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. Those skilled in the art will understand that the present application can also be implemented without these specific details.

[0130] The present application provides a highway continuous longitudinal slope road section traffic safety knowledge graph construction method, comprising the following steps:

[0131] Data collection and preprocessing: integrate multi-source heterogeneous data, including road section feature data, traffic flow data, historical accident data and environment data, and clean, remove duplicates and standardize the data;

[0132] Step 1.1, multi-source data collection

[0133] Four types of heterogeneous data of highway continuous longitudinal slope road section are collected:

[0134] Road section feature data: slope , longitudinal slope continuous length , curve radius , denoted as vector

[0135] (from GIS system);

[0136] Traffic flow data: time series vehicle flow , vehicle average speed , constituting a time series matrix (from traffic monitoring system);

[0137] Accident data: accident type (classification code), timestamp , severity level {1,2,3}, constituting an event set ;

[0138] Environment data: weather state , road surface friction coefficient (set according to rainy, snowy and sunny days), visibility , constituting a time series matrix .

[0139] Step 1.2, clean, remove duplicates and standardize the data, output standardized data set:

[0140] .

[0141] Knowledge representation and modeling: define the entities, attributes and relationships between entities of the traffic safety knowledge graph, and construct a multi-level schema containing road segment entities, accident entities and environment entities;

[0142] Step 2.1, define four types of core entities and their attributes, and use vector space and probability model for formal description:

[0143] Let the entity set of the knowledge graph be:

[0144]

[0145] Define the attribute set of all entities = , defined as follows:

[0146] The attribute set of the road segment entity :

[0147]

[0148] Among them: represents the road segment ID, is the slope, is the elevation difference, is the horizontal distance, is the length of the longitudinal slope, and R is the radius of curvature of the curve.

[0149] The attribute set of the accident entity represents the accident characteristics, defined as a multivariate set:

[0150]

[0151] Among them, is the accident type code (such as 1: rear-end, 2: rollover), is the accident timestamp, is the accident severity level (1: minor, 2: general, 3: major).

[0152] The attribute set of the environment entity :

[0153] ={ , , }

[0154] is the weather state, is the road surface friction coefficient, which is subject to conditional distribution:

[0155] =

[0156] Visibility (unit: m.

[0157] Vehicle entity Attribute set, representing vehicle passing characteristics, defined as a set of dynamic parameters:

[0158]

[0159] Where is the vehicle type, is the real-time vehicle speed;

[0160] is the acceleration, converted from speed-time

[0161] Step 2.2, define the association between entities, and build a probability association model

[0162] Define the set of association relationships between entities as:

[0163]

[0164] Each relationship is defined as a weighted directed edge, and the weight represents the association strength or probability, as follows:

[0165] 1. Road section-accident causal association (R) )

[0166] Model the causal relationship between road section characteristics and accident occurrence using a conditional probability model:

[0167]

[0168] Symbol interpretation:

[0169] : Road section feature vector (slope, length of longitudinal slope, radius of curve)

[0170] : Feature weight coefficient (e.g. indicates that an increase in slope will increase the probability of accidents)

[0171] : Accident type (e.g. rear-end collision, rollover, etc.)

[0172] : Activation function.

[0173] 2. Environment-accident risk probability association (environmental impact)

[0174] To describe the dynamic influence of environmental factors on accident risk, Bayesian networks are used to model:

[0175]

[0176] Symbol explanation:

[0177] : Accident severity level

[0178] : Indicator function (1 for bad weather, 0 otherwise)

[0179] : Environmental factor weight (needs to be fitted with historical data through existing conventional means);

[0180] 3. Road segment-environment coupling risk (initiation risk)

[0181] Define the interaction of slope and environmental factors using a coupling risk index, with a higher value indicating a higher risk:

[0182]

[0183] Symbol explanation:

[0184] : Weather-related road surface friction coefficient

[0185] : Coupling coefficient (determined by accident data regression analysis).

[0186] Road segment-environment-vehicle coupling risk:

[0187]

[0188] Provide the road surface friction coefficient for the environmental entity, where , , are coupling coefficients, and the formula is a dimensionless risk index.

[0189] Step 2.3, build multi-level schema

[0190] Use a hierarchical graph structure to represent the knowledge graph schema, defining three types of levels:

[0191] 1. Static topology layer

[0192] Describe the fixed relationship between entities using an adjacency matrix :

[0193]

[0194] This indicates that if at least one relation exists. Belongs to a set of relations And the relationship From the entity Pointing to entity If so, then the condition is met.

[0195] Example: This indicates that the road section is related to the accident.

[0196] 2. Dynamic Attribute Layer

[0197] Using time tensors Indicates how entity attributes change over time:

[0198] Entity count

[0199] Time step

[0200] Attribute dimensions (such as road segment entities) ,correspond )

[0201] 3. Probabilistic Inference Layer

[0202] Integrate association rules and risk models to define the rule engine logic:

[0203] Risk assessment rules:

[0204]

[0205] Indicates the overall risk index The medium risk threshold, Indicates the overall risk index High risk threshold This represents a set of recent accident patterns, specifically those similar accidents that have occurred within a time window.

[0206] Step 2.4, Schema Extensibility Design

[0207] To support future additions of entities (such as traffic signs and construction zones), a pattern extension protocol is defined:

[0208] Entity compatibility: Adding new entities It must share at least one common attribute with existing entities. Specific attribute interface:

[0209]

[0210] This formula is a compatibility rule for knowledge graph schema extension, requiring that the new entity must be associated with existing entities through at least one common attribute. This rule ensures data associability, logical consistency of reasoning, and overall maintainability of the schema.

[0211] : Set of attributes of the new entity (e.g., "Construction Area").

[0212] For example:

[0213] Represents the union of all existing entity attributes.

[0214] Suppose existing entities include "Road Section" and "Accident":

[0215]

[0216]

[0217] Then the union is: .

[0218] : Intersection operation, i.e., find common attributes.

[0219] If the new entity attribute is , the intersection of the union with existing attributes is .

[0220] : Non-empty intersection, i.e., at least one common attribute.

[0221] This constraint ensures data associability:

[0222] If the new entity has no relation to existing attributes (e.g., , it cannot be associated with existing entities (e.g., "Road Section") through common attributes (e.g., "Slope"), resulting in data islands in the knowledge graph.

[0223] Support knowledge reasoning:

[0224] Common attributes are the basis for establishing logical relationships between entities. For example, through the "Slope" attribute, we can reason about the impact of "Construction Area" on the safety of "Road Section".

[0225] Maintain schema consistency:

[0226] Avoid arbitrary attribute expansion leading to schema chaos, ensuring that new entities can be integrated into the existing framework.

[0227] Actual example

[0228] Legal expansion:

[0229] New entity "Traffic Sign" with attributes {Type, Location, Associated Road Segment}.

[0230] Common attribute: Associated Road Segment ("Road Segment ID" of Road Segment entity can be mapped to "Associated Road Segment").

[0231] Condition: Existence of common attribute "Associated Road Segment" associated with "Road Segment" entity.

[0232] Relation Inheritance: Hierarchical expansion through relation generalization:

[0233]

[0234] related_to: New default generic relation indicating some unspecified association between entities, allowing gradual refinement.

[0235] By adding the default relation related_to, we provide a "placeholder" association mechanism for new entities while preserving existing relations. This ensures system stability and leaves room for future detailed modeling.

[0236] Significance of relation inheritance:

[0237] Supporting gradual expansion:

[0238] When a new entity (such as "Construction Area") is added, it may not be immediately possible to define its specific relationship (such as blocks_traffic or increases_risk) with existing entities (such as "Road Segment"). At this time, use related_to to establish a preliminary association, and then gradually refine it later.

[0239] Avoiding rigid relation definition:

[0240] If all new entities are required to predefine specific relations, it may lead to overdesign or frequent pattern modifications. related_to provides a flexible intermediate state.

[0241] Maintaining knowledge graph consistency:

[0242] Even if the new entity has not been fully integrated into the business logic, related_to can ensure its basic association with other entities in the graph, avoiding data silos.

[0243] Specific role of related_to:

[0244] Scenario Application Example Significance Newly added entity initial association Construction area related_to road segment Indicate that the construction area is related to a certain road segment, but the specific impact type (such as closure, restricted driving) needs to be refined. Multi-entity generalized association Traffic sign related_to road segment, weather sensor related_to environment Unified management of various unclassified association relationships. Conflict resolution transition Newly added entity "temporary regulation" is associated with the accident entity through related_to, and is replaced with triggers later When the data conflict is not resolved, maintain the temporary association.

[0245] Synergy with original relationships

[0246] Original relationships (e.g., has_accident) are strong semantic, high-weight associations that directly support reasoning (e.g., cause-effect relationship between road segment and accident).

[0247] related_to is a weak semantic, low-weight association that only indicates a connection and needs further verification or refinement.

[0248] Example:

[0249] If a new entity "animal migration" is associated with "road segment" through related_to, subsequent data analysis may discover its potential relationship with accidents, and eventually upgrade related_to to "cause accidents."

[0250] Step 2.5, formal verification

[0251] To ensure the consistency of the pattern logic, define the constraint conditions:

[0252] Causal closure: if , then there must be a reverse association , ensuring the bidirectional traceability of the knowledge graph.

[0253] Forward relationship:

[0254] For example: "road segment has had an accident ", from the perspective of data, associate its historical accident records from the road segment.

[0255] Reverse relationship:

[0256] For example: "accident occurred on road segment ", further specifying that accident A (rollover) occurred on road segment X (5% slope, 2km length of vertical slope), from the perspective of data, locate the location of the accident from the accident.

[0257] Final output

[0258] The knowledge graph pattern can be represented as a four-tuple:

[0259]

[0260] The above constraint conditions ensure the consistency of data logic and physical consistency.

[0261] Step 3, knowledge extraction and construction: from structured data The entity attribute values and the relationship between entities are extracted, the unstructured text data is processed by natural language processing technology to recognize named entities, extract relationships and events, multi-source data is fused, entity reference ambiguity and conflicts are solved, and knowledge triples are generated;

[0262] Step 4, knowledge reasoning and application: based on the preset rule engine to realize dynamic risk assessment, the rule engine integrates the correlation logic of slope characteristics, environmental parameters and accident data;

[0263] Step 4.1, rule engine initialization

[0264] Load the preset correlation rules and risk models, refer to the conditional probability, Bayesian network and coupled risk index defined in section 2.2, and construct an extensible reasoning logic library:

[0265]

[0266] Wherein:

[0267] The conditional probability of slope characteristics and accidents;

[0268] The environmental coupling risk index;

[0269] The vehicle dynamics equation;

[0270] Step 4.2, dynamic risk assessment

[0271] Input real-time data stream , execute multi-dimensional risk calculation:

[0272] Wherein:

[0273] 1. Single-factor risk quantification

[0274] Slope accident probability:

[0275] The slope-related risk probability of the kth accident under real-time data is represented by σ(·), and σ(·) represents the Sigmoid function;

[0276] Environmental risk value:

[0277] Coupling risk coefficient:

[0278] Comprehensive risk model, define the weighted comprehensive risk index:

[0279] +

[0280] Wherein These are the weighting coefficients. ,satisfy Determined by fitting historical accident data;

[0281] Step 4.3, Real-time Risk Warning

[0282] A three-level early warning signal is generated based on the risk threshold:

[0283]

[0284] Indicates the overall risk index The medium risk threshold, Indicates the overall risk index High risk threshold, This represents a set of recent accident patterns, specifically those similar accidents that have occurred within a time window.

[0285] Triggering actions include:

[0286] High: Issue speed limit orders and activate variable message signs;

[0287] Medium: Generate maintenance inspection work orders;

[0288] Low: Update the risk status attribute of the knowledge graph;

[0289] System Deployment and Updates: The knowledge graph is stored using a graph database, and semantic query, risk warning, and dynamic graph update functions are provided through a visualization interface.

Claims

1. A method for constructing a traffic safety knowledge graph of a highway continuous longitudinal slope section, characterized in that, Comprising the following steps: Step 1, data collection and preprocessing: integrating multi-source heterogeneous data, including road segment feature data, traffic flow data, historical accident data and environmental data, and cleaning, deduplicating and standardizing the multi-source heterogeneous data to output a standardized dataset; Step 2, knowledge representation and modeling: defining entities, attributes and inter-entity relationship of the knowledge graph, and constructing a multi-level model containing road segment entities, accident entities, environmental entities and vehicle entities; the model includes: Static topology layer: used to describe the fixed relationship between entities, represented by an adjacency matrix; Dynamic attribute layer: used to represent the change of entity attributes over time, represented by a time series tensor; Probabilistic reasoning layer: used to integrate association rules and risk models, risk determination rules; Step 3, knowledge extraction and construction: extracting entity attribute values and inter-entity relationships from structured data, performing named entity recognition, relationship extraction and event extraction on unstructured text data through natural language processing technology, fusing multi-source data and solving entity reference ambiguity and conflicts, and generating knowledge triples; Step 4, knowledge reasoning and application: realizing dynamic risk assessment based on a pre-set rule engine that integrates the association logic of slope characteristics, environmental parameters and accident data, and generating real-time risk warning signals through the following sub-steps: Step 4.1, rule engine initialization: referencing conditional probability, Bayesian network and coupling risk index to construct an extensible reasoning logic library; Step 4.2, dynamic risk assessment: calculating slope accident probability, environmental risk value and coupling risk coefficient; Step 4.3, real-time risk warning: comparing the comprehensive risk index with the pre-set threshold to generate a three-level warning signal; Step 5, system deployment and update: storing the knowledge graph in a graph database and providing semantic query, risk warning and graph dynamic update functions through a visual interface. 2.The method of claim 1, wherein, Step 1 includes the following steps: Step 1.1 Multi-source data collection Collecting four types of heterogeneous data for continuous longitudinal road segments: Road segment feature data: slope , length of longitudinal slope continuity , radius of curve , represents the road segment ID, denoted as vector , where the uphill > 0, and the downhill < 0; Traffic flow data: time series vehicle flow , vehicle average speed , constituting a time series matrix ; Accident data: accident type , timestamp , severity level {1,2,3}, constituting an event set ; Environmental data: weather conditions , road friction coefficient , visibility , constituting a time series matrix ; Step 1.2 Cleaning, deduplicating and standardizing the data to output a standardized dataset: 。 3. The method according to claim 1, wherein the method is characterized by: In step 2, the entities of the knowledge graph are defined as follows: Define four types of core entities and their attributes, and use vector space and probability models for formal description: Let the entity set of the knowledge graph be: a set of property definitions defining all entities = , defined as follows: a collection of attributes of a road segment entity : wherein: represents a road segment ID, is a slope, is an elevation difference, is a horizontal distance, is a length of a longitudinal slope continuity, and R is a radius of curvature of a curve. Attributes of the accident entity representing the accident characteristics, defined as a tuple of attributes: wherein: is an accident type code, is an accident time stamp, is an accident severity level, 1: minor, 2: average, 3: major; environmental entity a set of attributes of the environmental entity: ={ ， ， } where: is the weather state, is the visibility, is the road friction coefficient, subject to a conditional distribution: = A vehicle entity a set of attributes representing vehicle passage characteristics, defined as a set of dynamic parameters: wherein: is a vehicle type , is a real-time vehicle speed, denotes a vehicle acceleration, denotes a braking performance coefficient, associated with a vehicle weight. The mapping relationship with Type is: 。 4. The method according to claim 3, wherein the method is characterized in that: In step 2, the attributes and inter-entity relationship are defined as follows: Define the inter-entity relationship set as: Each relationship is defined as a weighted directed edge, with the weight representing the association strength or probability, as follows: Road segment-accident causal association: Modeling the causal relationship between road segment characteristics and accidents using a conditional probability model: wherein denotes a road segment feature vector, , denotes a feature weight coefficient, respectively corresponding to the weight of the slope , the length of the longitudinal slope , the radius of the curve , , , , and σ(·) denotes a Sigmoid function. Environment-accident risk probability association: Describing the dynamic influence of environmental factors on accident risk using a Bayesian network model: represents the accident severity level, represents the indicator function, which takes the value 1 in bad weather and 0 otherwise, , , represents the environmental factor weight; Road segment-environment coupling risk: Defining the interaction between slope and environmental factors using a coupling risk index: : weather dependent road friction coefficient, is the coupling coefficient; Road segment-environment-vehicle coupling risk: a road surface friction coefficient provided for an environmental entity, wherein , , are coupling coefficients.

5. The method according to claim 3, wherein the method is characterized in that: In step 2, a multi-level model containing road segment entities, accident entities and environmental entities is constructed, including a static topology layer, a dynamic attribute layer and a probabilistic reasoning layer, as follows: Constructing a multi-level model Adopt hierarchical graph structure to represent schema of knowledge graph, define three types of hierarchy: Static topology layer Describing fixed relationships between entities using adjacency matrices : denotes at least one; Dynamic attribute layer Temporal tensor denotes the change of entity properties over time, for the number of entities, for the time steps, for the property dimension; Probabilistic reasoning layer Integrate association rules and risk models to define rule engine logic: Risk determination rules: a medium risk threshold of the composite risk index a medium risk threshold of the composite risk index a high risk threshold of the composite risk index a high risk threshold of the composite risk index a near-term accident pattern set, in particular a set of accidents of the same kind that have occurred within a time window.

6. The method according to claim 3, wherein the method is characterized in that: Step 2 also includes the extensibility step of Schema: Define schema extension protocol: Entity compatibility: new entity Must satisfy at least one common attribute with existing entity, specific attribute interface: represents the set of attributes of the new entity, represents the union of all existing entity attributes, represents the intersection operation, i.e. finding common attributes, represents that the intersection is non-empty, i.e. there is at least one common attribute; Relationship inheritance: achieve hierarchy extension through relationship generalization: related_to, a newly added default general relationship, represents the existence of some unspecified association between entities, allowing for gradual refinement.

7. The method according to claim 3, wherein the method is characterized by: Step 2 also includes the formal verification step: To ensure the logical consistency of the schema, define constraints: Causal closure: if then there must exist a reverse association ensuring bidirectional traceability of the knowledge graph, denotes a forward relationship, denotes a reverse relationship.

Images