528 results about "Traffic system" patented technology

The invention belongs to the technical field of traffic management and control, and particularly relates to a congestion feed-forward intervention method based on traffic flow phase change critical point recognition, which comprises the following steps: collecting and preprocessing multi-source heterogeneous traffic data; carrying out multi-scale traffic flow feature engineering; identifying a traffic flow phase change critical point based on a space-time dynamic graph neural network and critical moderation effect analysis; generating a multi-objective optimization congestion feedforward intervention strategy; and performing intervention, evaluating the effect and performing adaptive learning. According to the technical scheme, accurate prevention and early intervention can be performed before congestion occurs, and the operation efficiency and reliability of an urban traffic system are remarkably improved.

The invention discloses a congestion management system and method based on vehicle-road cooperation and dynamic group path optimization, and relates to the technical field of intelligent traffic systems. The method comprises the following steps: S1, collecting original traffic data through a multi-modal sensor array to carry out space-time alignment, and constructing a microscopic traffic flow data set; s2, bidirectional information interaction is carried out through a vehicle-road cooperative communication network; s3, calculating a multi-modal fusion confidence coefficient parameter based on the data quality index, and generating traffic state information and road section real-time saturation; s4, fusing historical and static road network data, constructing a dynamic digital twinborn model and a road network state feature map, and generating a road network load balancing index; and S5, making a decision by adopting a hierarchical-distributed architecture, and generating a group path induction strategy through multi-objective optimization. The efficient collaborative decision is realized through the graph convolutional network and the attention mechanism, and the road network traffic efficiency is comprehensively improved through stable and reliable multi-objective optimization on the premise of guaranteeing the user fairness.

The invention relates to three-dimensional traffic navigation, in particular to a three-dimensional traffic navigation system based on space-time grid coding, which comprises a low-altitude path planning module for planning an optimal flight path for an aircraft in a low-altitude airspace based on a four-dimensional space-time grid coding model and adjusting the path in real time in combination with dynamic environmental factors; the airspace conflict detection module is used for carrying out conflict detection on aircrafts in a low-altitude airspace, judging whether conflicts exist or not and finding potential conflicts in time; the path collaborative planning module is used for carrying out collaborative path planning on the multiple aircrafts with conflicts in the low-altitude airspace and solving the problem of conflicts among the multiple aircrafts; the cross-modal traffic coordination module is used for integrating ground and low-altitude traffic systems, realizing seamless connection and efficient coordination of an optimal ground path and an optimal flight path, and providing an integrated three-dimensional traffic navigation scheme; the method can overcome the defects that in the prior art, path planning efficiency is low, airspace conflicts are difficult to accurately detect, and a cross-modal traffic cooperation mechanism is lacked.

The invention relates to the field of intelligent transportation systems, discloses an intelligent highway maintenance system and method, and aims to solve the fundamental defects that in the prior art, the data acquisition dimension is single, the maintenance decision depends on artificial experience, and prospective prediction is lacked. The method comprises the following steps: acquiring multi-modal dynamic sensing data covering a whole road domain; constructing and updating a four-dimensional space-time digital twinborn model in real time; driving the causal graph neural network model to perform structure health state prediction and diagnosis; generating an active maintenance instruction through a multi-objective optimization engine; and the autonomous maintenance execution unit is dispatched to complete unmanned closed-loop operation. According to the method and the system, fundamental transformation of road maintenance from passive response to active prevention, from artificial experience to intelligent decision making and from discrete operation to closed-loop automation is realized, and the maintenance efficiency, the safety and the health level of the whole life cycle of the road are remarkably improved.

The present disclosure is related to Intelligent Transport Systems (ITS), and in particular, to service dissemination basic services (SDBS) and / or collective perception service (CPS) of an ITS Station (ITS-S). Implementations of how the SDBS and / or CPS is arranged within the facilities layer of an ITS-S, different conditions for service dissemination messages (SDMs) and / or collective perception message (CPM) dissemination, and format and coding rules of the SDM / CPS generation are provided.

The invention relates to a traffic simulation agent system construction method based on a large model, and the method comprises the steps: constructing a simulation tool library, and achieving the precise evaluation and continuous optimization of a simulation result through the fusion of multi-source heterogeneous traffic data, the construction of standardized input, and the establishment of a quantitative evaluation system. A large language model is utilized to understand a natural language instruction of a user, tasks are intelligently disassembled, an execution process is planned, dependence management and parallel scheduling are carried out in combination with a directed acyclic graph, and professional tools are driven to automatically execute. And performing evaluation, problem diagnosis and adaptive re-planning on an execution result through a large language model reflection mechanism to form an understanding-planning-execution-reflection closed loop. According to the method, the problems of how to assist a user to interact with a traffic system by utilizing an agent technology driven by a large language model, reducing the technical threshold of traffic simulation software use and saving time cost and labor cost are solved, the traffic simulation use threshold is reduced, the automation and intelligence level is improved, and efficient and accurate traffic system interaction and optimization are realized.

The invention discloses a mixed traffic flow-oriented automatic driving taxi dynamic scheduling system and a collaborative decision-making method, belongs to the field of intelligent traffic systems, and solves the problem that in the coexistence environment of manual driving vehicles and automatic driving taxies, the automatic driving taxies cannot be automatically scheduled. The technical problem of how to efficiently and cooperatively dispatch vehicles, accurately predict demands, optimize energy management and improve the overall operation efficiency of the system is solved. The system comprises a scheduling server which is connected with a road side sensing unit, a vehicle-mounted control unit and a charging station management platform. The scheduling server comprises a traffic flow analysis module; a demand prediction module; a dynamic scheduling module; and an energy collaboration module. The system is mainly used for realizing real-time, dynamic and intelligent scheduling and energy management of the automatic driving taxis in the mixed traffic flow, improving the operation efficiency, relieving the traffic jam and optimizing the charging resource utilization.

The invention discloses an electric vehicle man-machine cooperative scheduling strategy for multiple scenes of an electric power traffic coupling network, and aims to solve the problem of electric vehicle charging optimization scheduling caused by deep coupling of an electric power system and a traffic system in different scenes. The method comprises the steps that firstly, topological information and behavior characteristics are fused through a graph generative adversarial network, a graph structured model is constructed, and an electric power traffic coupling operation scene is generated; secondly, establishing a multi-objective optimization mechanism by utilizing hierarchical reinforcement learning, constructing an electric vehicle charging optimization scheduling finite Markov decision model in a conventional scene and a fault scene, and designing an algorithm based on knowledge distillation to solve a scheduling strategy; and finally, realizing strategy migration of charging redistribution and path emergency adjustment in a fault scene by combining a man-machine cooperative regulation and control technology and fusing a user instruction. Experimental results show that the strategy can effectively improve the toughness of the power grid, relieve traffic congestion, reduce charging queuing time and increase user satisfaction.

The invention relates to the technical field of computer vision and target detection, and particularly discloses a lightweight real-time two-wheeled vehicle helmet detection method. According to the method, firstly, a video stream is collected and preprocessed through a traffic monitoring camera, then feature extraction and fusion are carried out through a lightweight backbone network, an encoder and a neck network in sequence, finally, a detection result is output through a decoder, a StripCGLU module is introduced to reduce the calculation complexity, a Pola Former module is adopted to enhance the feature interaction capability, and finally, the detection result is output through a decoder. And a GLBiFPN network is designed to optimize multi-scale feature fusion. According to the method, the calculation complexity and parameter quantity of the model are remarkably reduced, the requirement of edge calculation equipment for efficiency is met while high detection precision is guaranteed, and an effective technical means is provided for safety monitoring in an intelligent traffic system.

The invention belongs to the technical field of traffic management evaluation, and particularly discloses and provides an urban traffic toughness quantitative evaluation method based on multi-mode traffic collaboration, and the method comprises the steps: carrying out the time-space standardization processing of multi-source data of motor vehicles, public traffic, slow traffic and external disturbance events, and generating a standardized time-space data set; based on the data set, respectively calculating a comprehensive transportation efficiency index, an alternative path redundancy index and a connection connectivity index, and constructing a toughness evaluation index set; based on the historical sequence of each index, generating a dynamic combination weight through a subjective and objective combination weighting method; and based on the index set and the dynamic weight at the current moment, calculating a comprehensive toughness index and generating a time sequence thereof to output an evaluation result. According to the method, the defects that in the prior art, an isolated analysis single mode depends on a static model are effectively overcome, and accurate multi-dimensional quantitative evaluation of the real toughness level of the complex traffic system under disturbance is achieved.

A transportation system for generating transportation recommendations may (1) receive a transportation request associated with a user; (2) identify, using the transportation request, a first location and a second location associated with the transportation request; (3) determine, using the first location and the second location, routes between the first location and the second location; (4) receive travel data associated with each of the plurality of the routes, the travel data including information indicating a current or predicted future travel condition along each of the routes; (5) generate, using historic transportation characteristics associated with the user and the travel data, a safety or other score for each route, the score indicating an estimated level of safety of traveling along the route; and / or (6) generate a user interface providing indicators associated with the scores of the routes.

The invention discloses a point cloud intelligent cutting and dynamic optimization method based on edge computing, and belongs to the technical field of intelligent traffic systems and edge computing, and the method comprises the steps: collecting three-dimensional point cloud data through a multi-line laser radar disposed on an RSU; generating a three-dimensional bounding box through a three-dimensional target detection network in an RSU edge calculation unit, generating an ROI mask, expanding a buffer region, and performing adaptive region cutting on the point cloud to form an ROI point set and a background point set; a perception-driven cutting decision is realized, and ROI extraction is converted into a self-learning process based on semantic features from fixed threshold judgment; voxelizing the non-uniform resolution point cloud, generating a unique hash index for each voxel unit, comparing a hash set of a current frame with a hash set of a previous frame, extracting newly added and disappeared point sets, and constructing a differential data packet according to a change ratio; while high timeliness is kept, data redundancy is reduced, and the point cloud updating rate and the incremental transmission efficiency are improved; and dynamic optimization of communication-calculation cooperation is realized.

The invention relates to the technical field of intelligent traffic systems, in particular to a highway network global traffic flow prediction method, device and equipment, and the method comprises the steps: constructing an initial feature sequence based on the multi-source data of an ETC portal system; inputting the data into a multi-head attention model driven by a query mechanism, and outputting a spatial dependency relationship between ETC door frame nodes; inputting the spatial dependency relationship into an extraction fusion model fusing a causal convolutional network and a selective state space network, extracting short-term time sequence features through the causal convolutional network, modeling long-term time sequence features through the selective state space network, and fusing the long-term time sequence features into fused time sequence features; and a global traffic flow prediction result is generated based on the fusion time sequence features, so that the problems of high traffic flow prediction cost, data coverage missing and the like in related technologies are solved.

The invention provides a ramp traffic signal adaptive control method and system based on unmanned aerial vehicle cooperation, and belongs to the technical field of intelligent traffic control, and the method comprises the steps: collecting traffic data in real time based on an unmanned aerial vehicle cluster dynamically deployed over a ramp, constructing a global traffic state matrix, and carrying out the initial processing through edge calculation; carrying out edge calculation on the primarily processed data by utilizing an edge node to generate a local optimal signal timing suggestion in real time; the local optimal signal timing suggestions uploaded by the multiple unmanned aerial vehicles are integrated at the cloud, and a global optimal signal timing strategy is generated through unified training; signal lamp parameters are adjusted in real time on the basis of a global optimal signal timing strategy, the direct intervention capability on site traffic is enhanced through an air guiding function, and a high-reliability solution is provided for an intelligent traffic system.

The invention relates to the field of intelligent traffic systems, in particular to an intelligent intersection multi-vehicle cooperative control method. The method comprises the following steps: S1, sensing a vehicle state and sharing vehicle node information by adopting a dual-mode communication mechanism; s2, fusing state data at edge computing nodes; establishing a dynamic traffic model according to the fused state data; s3, acquiring node data of the edge calculation node, calculating a vehicle passing priority index according to the node data, and performing vehicle priority decision according to the passing priority index; s4, performing grading processing on the vehicles with the overlapped tracks; s5, the RSU device broadcasts the control instruction, and if the vehicle-mounted unit receives the control instruction within the set time, an execution report is transmitted back to the RSU device; and if the vehicle-mounted unit does not receive the control instruction within the set time, a redundancy execution mechanism is started to detect and control the running state of the vehicle. According to the invention, millisecond-level synchronous transmission and high-precision perception of information are realized, and the real-time performance and accuracy of traffic environment modeling are improved.

The invention belongs to the technical field of intelligent traffic system and artificial intelligence crossing, and discloses an intelligent questioning and answering system and method for highway tunnel traffic events. The method comprises the steps of receiving a natural language query input by a user; guiding the context understanding to output a standardized event label in preset five types of events through a prompt project; calling a preset Cypher template, dynamically filling an event name, generating a structured query statement, executing query on a pre-constructed tunnel management and control knowledge graph Neo4j, and obtaining a structured sub-graph comprising a trigger condition, an influence region and response equipment; constructing a structured knowledge dictionary; guiding the large language model to generate a natural language answer through the strong constraint cue word; and if the identification result is a non-five-class event, directly performing answer rejection. According to the method, an interpretable, verifiable and landing technical path is provided for intelligent operation and maintenance of the expressway tunnel, and the method has important engineering application value and popularization prospect.

The embodiment of the invention provides a multi-level federal multi-mode large model privacy protection enhancement method and electronic equipment, and belongs to the technical field of intelligent traffic systems. According to the method, a'end-edge-cloud 'three-level federated learning architecture is constructed, firstly, local differential privacy disturbance is applied to local multi-mode traffic data at a traffic terminal node, and an end-side local model is trained; then aggregating a plurality of end side models on an edge server, and generating a personalized small model reflecting regional characteristics; then cooperatively training a plurality of personalized small models in a cloud center server to generate a global multi-modal large model; finally, knowledge of the global large model is fed back to a lower-level model through knowledge migration, and a continuously evolved iterative closed loop is formed. According to the method, privacy protection is carried out at a data source, so that original sensitive data is ensured not to go out of the local, the problems of data islands and privacy leakage in traffic large model collaborative training are effectively solved, and efficient and credible collaborative modeling is realized on the premise of ensuring data sovereignty.

The invention relates to the field of automatic driving system test and evaluation, in particular to an automatic driving automobile intelligence degree evaluation method based on subjective and objective mapping of a large language model, and can overcome the defects of a traditional evaluation method in the aspects of objectivity, efficiency and nonlinear fitting ability. The method comprises the following steps: converting automatic driving interaction data into natural language description; constructing an automatic driving field knowledge base, and performing knowledge enhancement on the large language model to improve the understanding of the large language model on professional rules; performing quantitative evaluation on interaction data in multiple dimensions (safety, comfort, efficiency, social interactivity and influence on a traffic system) by using the enhanced large language model; a nonlinear mapping model from interaction data to intelligent scoring is established by adopting a multi-layer perceptron, and efficient and objective comprehensive evaluation is realized. The method can effectively reduce the dependence on artificial experts, improves the evaluation consistency and efficiency, and is suitable for the intelligent level comprehensive evaluation of different levels of automatic driving systems.

The invention relates to a different intelligence traffic subject interaction information model construction method based on a three-dimensional digital model. The method comprises the following steps: S1, constructing a multi-level scene based on spatial levels of a road traffic system and designing functions of the multi-level scene; s2, defining autonomous levels of the traffic system, and describing the perception capability, decision logic and execution precision of each terminal device of the vehicle road cloud of each level; s3, according to different attributes of traffic entities, designing a perception-transmission-decision-control four-stage traffic information circulation process and realizing unambiguous collaboration among different intelligence subjects; and S4, converting the standardized and defined information model into a plurality of instantiation units, and deploying the instantiation units into computing entities of various end-edge-cloud traffic subjects to support actual operation of a traffic service scene. According to the method, the problems of difficult cross-domain collaboration, non-uniform information models and insufficient global optimization caused by non-uniform intelligent level of traffic subjects in the prior art are effectively solved, and methodological support is provided for design, verification and implementation of a vehicle-road cloud integrated system.

The invention discloses a distributed sensor abnormal event identification method for intelligent traffic, and particularly relates to the technical field of traffic information perception and identification, and the method comprises the following steps: generating an abnormal information initial confidence value through deploying a sensor node with a confidence value dynamic adjustment function; during fusion processing, independent identification priorities of single node anomalies are reserved; after sequence reconstruction of a unified time reference, judging whether a response condition is met or not in combination with trajectory evolution; if yes, an abnormal response process is triggered, a space-time compensation set is constructed based on historical data of the sensing blind area for verification, and finally an abnormal intervention instruction is output and a traffic scheduling system is linked; according to the method, the sensing sensitivity, the fusion accuracy and the response timeliness of the abnormal information in a complex traffic environment are improved, the problem of an identification blind area that single-point abnormity is covered is avoided, state reconstruction and supplementary verification of the sensing blind area are realized, and the rapid intervention capability of an intelligent traffic system on emergencies is enhanced.

The invention discloses an expressway complex scene multi-target sensing method based on cross-modal alignment self-adaptive regulation and control, and the method comprises the steps: forming uniform feature representation which is high in discrimination and is consistent through constructing a deep cross-modal feature alignment and self-adaptive fusion mechanism; a Transform-based set prediction decoder is adopted to convert target detection into a set prediction task of a fixed number of query vectors so as to enhance the multi-scale target sensing ability; a lightweight student network and an efficient knowledge distillation mechanism are introduced, so that the student network can effectively learn feature distribution and query representation of a complete teacher model, and the high-precision reasoning ability is kept while the parameter quantity and the calculation overhead are remarkably reduced. According to the method, more accurate sensing capability is provided under the conditions of complex traffic flow, environment change and emergencies of the expressway, so that the overall performance and real-time response capability of the system are improved, stable and reliable multi-target sensing support is provided for an intelligent traffic system, and road safety and traffic flow optimization are promoted.

The invention discloses a road berth charging and control system for realizing multi-level fault tolerance and a method thereof, and belongs to the technical field of intelligent traffic systems and Internet of Things. The system adopts an edge computing and distributed sensing collaborative architecture, and is composed of an edge controller (ECU) and a sensing and control unit (SCU) deployed in a berth. High availability of the system is ensured through fault-tolerant design of three core dimensions: firstly, perception layer fault tolerance dynamically adjusts data fusion weights of geomagnetism, millimeter wave radar and visual AI according to real-time environment data such as rainfall and electromagnetic interference by integrating an environment perception sub-module; secondly, network layer fault tolerance utilizes an immutable transaction log and a Saga distributed transaction compensation mechanism to realize local charging and digital RMB double offline payment when cloud connection is interrupted, and data consistency is ensured after network recovery; and finally, a hardware layer establishes a neighborhood cooperation protocol based on a signature agent command through fault tolerance, and a healthy node is allowed to act as an agent fault node through safety verification to execute an unlocking instruction. According to the invention, the problems of environmental interference, network interruption, single-point hardware failure and the like in unattended parking management are effectively solved, and the robustness and financial security of the system are remarkably improved.

Disclosed in the present invention is a traffic all-domain safety early warning system based on real-time safety information fusion technology, comprising a traffic all-domain big data cloud center, an onboard intelligent terminal OBU, an intelligent terminal having a client APP installed thereon, and a roadside unit RSU; the onboard intelligent terminal OBU is used for acquiring vehicle self-collected information serving as a first-layer information source; the roadside unit RSU is used for acquiring RSU information serving as a second-layer information source; the intelligent terminal having a client APP installed thereon is used for acquiring all-domain traffic information serving as a third-layer information source; and the traffic all-domain big data cloud center corrects the accuracy of dynamic traffic data comprising the vehicle self-collected information, an RSU information source and the all-domain traffic information, and fuses the corrected data into a unified traffic electronic map. The present invention ensures that all types of individuals entering the public traffic space can be detected in advance, thereby improving the information transparency and operation efficiency of the traffic system, and travel safety.

The invention discloses a subway interval interruption scene-oriented planned operation bus route fine adjustment method, which mainly comprises the following steps of: constructing a candidate bus route set and a rail station set capable of being conveniently transferred, and generating a multi-point connection path; on the premise of restraining the total demand of the detained passengers, the number of passengers getting on and off each bus station, the dynamic passenger capacity in the bus, the bus route arrival and departure time and the detour connection time are calculated, and the conventional operation service level is guaranteed; and constructing an optimization model aiming at minimizing the total delay of all system passengers, and solving the optimization model. By scientifically deploying existing operation bus resources, a multi-station dispersed transportation strategy is implemented, and travel requirements of passengers are accurately matched. The method effectively avoids the problem of congestion caused by centralized transportation of a single destination, and the constructed optimization model gives consideration to the operation efficiency of the public transportation system and the evacuation demands of detained passengers in a subway, remarkably reduces the overall delay cost, greatly improves the emergency response speed and the toughness of the public transportation system, and powerfully supports the safe and stable operation of an urban traffic system.

Systems, methods, apparatuses, and computer program products for discontinuation rest for predictable traffic. One method may include receiving, by a user equipment, a discontinuous rest configuration, and starting, by the user equipment, a discontinuous rest timer indicated by the discontinuous rest configuration upon an end of an active time and if a number of packets transmitted and a number of packets received before the expiration of the active time is smaller than a corresponding threshold indicated in the discontinuous rest configuration. The discontinuous rest timer is associated with a discontinuous reception group.

The invention relates to the technical field of smart cities and intelligent traffic systems, in particular to a smart city road traffic potential safety hazard road section screening system and method. Comprising the following steps: S1, receiving a real-time multi-modal data stream, and calculating a dynamic entropy value by a real-time entropy evaluation module; s2, if the dynamic entropy value exceeds a preset dynamic triggering threshold value, starting an abnormal event screening process; and S3, if the dynamic entropy value does not exceed the preset dynamic trigger threshold value, calling a preset conventional monitoring model to carry out basic monitoring on the multi-modal data stream. According to the method, the contradiction that a large amount of computing resources are consumed ineffectively for a long time in order to deal with small-probability emergencies is solved, and the rigid restriction relationship between the risk discrimination precision and the system response timeliness is broken through.

Examples relate to a concept for an entry-exit matching system, and in particular to an evaluation device, a method and a computer program for person re-identification for entry-exit matching in a transportation system. The evaluation device comprises processing circuitry configured to obtain a plurality of re-identification codes. Each re-identification code represents a person being recorded by at least one camera when entering or exiting at least a section of the transportation system. The processing circuitry is configured to match the plurality of re-identification codes using a global matching scheme to obtain a plurality of matched pairs of re-identification codes, such that each matched pair of re-identification codes comprises a re-identification code of a person entering and a re-identification code of a person exiting. The global matching scheme is based on reducing an overall distance between the re-identification codes of the matched pairs of re-identification codes over the plurality of matched pairs of re-identification codes. The processing circuitry is configured to determine points of entry and exit for the plurality of matched pairs of re-identification codes.

The invention belongs to the field of electric vehicles and intelligent traffic systems. An electric vehicle charging demand prediction method based on a vehicle-pile-network interactive coupling model is characterized by comprising the following steps: step 1, taking a data layer as a system input end, integrating the electric vehicle track data set, the actual traffic network data obtained from the OSM open platform, the number of electric vehicles at the charging station in real time and the parameter data of the power distribution network; 2, the model layer constructs a cross-domain collaborative model based on the output of the data layer; 3, urban-level travel OD matrix distribution is extracted from the trajectory data, a traffic flow space-time thermodynamic diagram is generated, and macroscopic demand input is provided for path planning; and 4, performing multi-level joint simulation on the data layer, the model layer and the algorithm layer, establishing a three-layer closed-loop simulation framework, and finally simulating and predicting the charging demand load of the electric vehicle. According to the method, the charging demand is predicted in real time.

The invention relates to the technical field of intelligent traffic systems and data processing, and discloses a highway toll station traffic flow space-time prediction method fusing multi-source information, and the method comprises the following steps: constructing a node flow feature tensor and a basic edge feature tensor containing a time delay attribute based on toll station running water and vehicle origin-destination data; calculating a dynamic impedance factor correction basic edge feature by using real-time meteorological data, and generating a meteorological enhanced dynamic edge feature tensor; constructing a space-time diagram attention network, explicitly introducing dynamic edge features to calculate an attention coefficient, and predicting short-time traffic; constructing a negative correlation weight and an efficiency weighted loss function based on the toll station operation efficiency score, and optimizing model parameters; and generating a control instruction and feeding back actual traffic data to update the operation efficiency score, thereby realizing closed-loop application. According to the method, high-precision closed-loop prediction for key congestion nodes in a complex environment is realized through weather perception dynamic enhanced edge features and an efficiency feedback weighted loss function.

The invention provides a multi-scene driving risk assessment method based on transfer learning. The method is suitable for real-time assessment and short-time early warning of driving risks in tunnel and non-tunnel environments. According to the method, a multi-source data set collected by multiple sensors is constructed based on a natural driving test, static and dynamic characteristics are extracted by using a sliding time window technology, and the driving risk is evaluated through a multi-scene driving risk evaluation model. The scene driving risk assessment model integrates a static information encoder, a variable selection network and an interpretable multi-head self-attention mechanism, has strong feature selection and time sequence modeling capabilities, and improves the adaptability and generalization capability to scarce tunnel data through a transfer learning strategy. Test results show that the method is superior to the existing mainstream model in the aspects of alarm rate, accuracy rate and false alarm rate, and is suitable for a risk assessment and early warning module in an intelligent traffic system.