Traffic risk processing method and system based on space intelligent scene

Abstract

The invention provides a traffic risk processing method and system based on a spatial intelligent scene, and relates to the technical field of traffic management. The method comprises the following steps: a cloud sends an inspection strategy to inspection equipment, the inspection equipment acquires multi-source heterogeneous data of a to-be-detected area according to the inspection strategy, constructs a digital twin initial model of the to-be-detected area according to the multi-source heterogeneous data, and uploads the digital twin initial model and the multi-source heterogeneous data to the cloud; and the cloud constructs a digital twin target model of the to-be-detected area according to the digital twin initial model and the multi-source heterogeneous data, determines a traffic risk level of the to-be-detected area according to the digital twin target model of the to-be-detected area, and issues a risk processing strategy matched with the traffic risk level to the inspection equipment and the risk processing equipment. The invention aims to improve the problems that the traffic condition monitoring is not comprehensive, the risk early warning is lagged, the emergency disposal efficiency is low, and the overall traffic safety level is influenced in the existing method.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of traffic management technology, and in particular to a traffic risk management method and system based on spatial intelligent scenarios. Background Art

[0002] With the acceleration of urbanization, the number of traffic participants—motor vehicles, non-motor vehicles, and pedestrians—continues to rise, increasing the operational pressure on urban road traffic systems. The combined effects of high-density mixed traffic, multi-hour traffic fluctuations, and complex road network structures are leading to an increasing frequency of traffic violations, accidents, and public safety emergencies, becoming key and challenging issues that urgently need to be addressed in urban management.

[0003] The traditional on-site law enforcement model, primarily based on manual patrols and fixed-point monitoring, has exposed numerous shortcomings in practical application. For one thing, manual inspections are limited by the scope of manpower deployment and perception accuracy, making it difficult to achieve comprehensive coverage and real-time monitoring of large, dynamically changing areas. Furthermore, in key areas such as traffic violation evidence collection, accident scene responsibility determination, and emergency response, the traditional model suffers from incomplete data collection, untimely information upload, and inefficient response and dispatch, severely hindering the improvement of urban traffic management capabilities. Summary of the Invention

[0004] The present application provides a method for handling traffic risks based on a spatial intelligence scenario. The present application adopts the following technical solutions: In a first aspect, an embodiment of the present application provides a method for handling traffic risks based on a spatial intelligence scenario, the method comprising: The cloud sends inspection policies to inspection devices, which then acquire multi-source heterogeneous data from the area to be inspected based on the policies. The policies indicate the initial inspection routes and data collection rules for the inspection devices. The inspection equipment builds an initial digital twin model of the inspection area based on multi-source heterogeneous data and uploads the initial digital twin model and multi-source heterogeneous data to the cloud. The cloud constructs a digital twin target model of the area to be inspected based on the digital twin initial model and multi-source heterogeneous data; The cloud determines the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, and sends risk handling strategies that match the traffic risk level to the inspection equipment and risk handling equipment.

[0005] In an optional embodiment, the inspection device includes a first type of inspection device and a second type of inspection device. The cloud sends an inspection policy to the inspection device, and the inspection device obtains multi-source heterogeneous data of the area to be inspected according to the inspection policy, including: The cloud sends a first inspection strategy to the first type of inspection device. The first type of inspection device obtains multi-source macro traffic situation data of the area to be inspected according to the first inspection strategy. The first inspection strategy is used to indicate the initial inspection route and data collection rules of the first type of inspection device. The cloud sends a second inspection strategy to the second type of inspection device. The second type of inspection device obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy. The second inspection strategy is used to indicate the initial inspection route and data collection rules of the second type of inspection device.

[0006] In an optional embodiment, the inspection equipment constructs an initial digital twin model of the area to be inspected based on multi-source heterogeneous data, including: The first type of inspection equipment pre-processes multi-source macro traffic situation data to obtain key spatial structure characteristics and dynamic traffic distribution characteristics of the area to be inspected; Construct a three-dimensional scene structure model of the area to be inspected based on key spatial structure characteristics and dynamic traffic distribution characteristics; The second type of inspection equipment pre-processes multi-source microscopic traffic situation data to obtain the spatiotemporal behavior characteristics of traffic participants; A traffic behavior semantic model of the area to be detected is constructed based on the spatiotemporal behavior characteristics of traffic participants.

[0007] In an optional embodiment, uploading the digital twin initial model and multi-source heterogeneous data to the cloud includes: The first type of inspection equipment determines the spatial traffic risk level of the area to be inspected based on the 3D scene structure model. If the spatial traffic risk level is less than the threshold, the 3D scene structure model and multi-source macro traffic situation data are uploaded to the cloud. The second type of inspection equipment determines the behavioral traffic risk level of the area to be inspected based on the traffic behavior semantic model. When the behavioral traffic risk level is less than the threshold, the semantic model and multi-source micro-traffic situation data are uploaded to the cloud.

[0008] In an optional embodiment, the method further includes: When the spatial traffic risk level is greater than or equal to the threshold, the first type inspection device sends a risk handling strategy that matches the spatial traffic risk level to the risk handling device and the second type inspection device; When the behavior traffic risk level is greater than or equal to the threshold, the second-type patrol device sends a risk processing strategy that matches the behavior traffic risk level to the risk processing device and the first-type patrol device.

[0009] In an optional embodiment, the method further includes: The first type of inspection device adjusts the first inspection strategy based on the risk management strategy issued by the second type of inspection device or the cloud; The second type of inspection device adjusts the second inspection strategy according to the risk management strategy issued by the first type of inspection device or the cloud.

[0010] In an optional embodiment, the cloud constructs a digital twin target model of the area to be inspected based on the digital twin initial model and multi-source heterogeneous data, including: Perform spatial coordinate alignment and timestamp synchronization on the three-dimensional scene structure model uploaded by the first type of inspection equipment and the traffic behavior semantic model uploaded by the second type of inspection equipment; Based on the aligned 3D scene structure model and traffic behavior semantic model, the relationship between the behavior trajectory, state changes and environmental factors of traffic participants in specific spatial locations is determined to generate a structure-behavior semantic mapping dataset. The structure-behavior semantic mapping data is jointly modeled with multi-source macro traffic situation data and multi-source micro traffic situation data to construct a digital twin target model of the area to be inspected.

[0011] In an optional embodiment, the cloud determines the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, including: Based on the digital twin target model of the area to be inspected, the set of traffic element change factors of the next inspection cycle of the area to be inspected is predicted. The set of traffic element change factors includes at least spatial structure characteristics, traffic density characteristics, and traffic participant behavior characteristics; Input the traffic element change factor set of the next inspection cycle of the inspection area into the preset traffic risk time series prediction model to calculate the comprehensive traffic risk score; The traffic risk level of the area to be inspected is determined based on the matching results of the comprehensive traffic risk score and the preset risk level classification standards.

[0012] In a second aspect, an embodiment of the present application provides a traffic risk management system based on a spatial intelligence scenario, the system comprising: The first policy delivery module is used to send inspection policies from the cloud to inspection devices. The inspection devices acquire multi-source heterogeneous data from the inspection area based on the inspection policies. The inspection policies are used to indicate the initial inspection routes and data collection rules for the inspection devices. The data acquisition module is used by the inspection equipment to build a digital twin initial model of the inspection area based on multi-source heterogeneous data, and upload the digital twin initial model and multi-source heterogeneous data to the cloud; The model construction module is used in the cloud to build a digital twin target model of the area to be inspected based on the digital twin initial model and multi-source heterogeneous data; The risk processing module is used in the cloud to determine the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, and to issue risk processing strategies that match the traffic risk level to the inspection equipment and risk processing equipment.

[0013] In an optional implementation, the first policy issuing module includes: A first strategy delivery submodule is configured to transmit a first inspection strategy from the cloud to a first type of inspection device. The first type of inspection device obtains multi-source macro traffic situation data of the inspection area according to the first inspection strategy. The first inspection strategy is configured to indicate an initial inspection route and data collection rules for the first type of inspection device. The second strategy sending sub-module is used to send the second inspection strategy to the second type of inspection equipment in the cloud. The second type of inspection equipment obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy. The second inspection strategy is used to indicate the initial inspection route and data collection rules of the second type of inspection equipment.

[0014] In an optional embodiment, the data acquisition module includes: The first data processing submodule is used for the first type of inspection equipment to pre-process the multi-source macro traffic situation data to obtain the key spatial structure characteristics and dynamic traffic distribution characteristics of the area to be inspected; The first model building submodule is used to build a three-dimensional scene structure model of the area to be detected based on key spatial structure characteristics and dynamic traffic distribution characteristics; A second data processing submodule is used for the second type of patrol equipment to pre-process the multi-source microscopic traffic situation data to obtain the spatiotemporal behavior characteristics of traffic participants; The second model building submodule is used to build a traffic behavior semantic model of the area to be detected based on the spatiotemporal behavior characteristics of traffic participants.

[0015] In an optional embodiment, the data acquisition module further includes: A first data uploading submodule is used for the first type of inspection equipment to determine the spatial traffic risk level of the inspection area based on the three-dimensional scene structure model, and upload the three-dimensional scene structure model and multi-source macro traffic situation data to the cloud when the spatial traffic risk level is less than a threshold; The second data uploading submodule is used for the second type of inspection equipment to determine the behavioral traffic risk level of the area to be inspected based on the traffic behavior semantic model. When the behavioral traffic risk level is less than the threshold, the semantic model and multi-source micro-traffic situation data are uploaded to the cloud.

[0016] In an optional implementation, the system further includes a second policy issuing module, which includes: The third strategy issuing submodule is configured to, when the spatial traffic risk level is greater than or equal to the threshold, cause the first type inspection device to issue a risk handling strategy that matches the spatial traffic risk level to the risk handling device and the second type inspection device; The fourth strategy issuing submodule is used for, when the behavior traffic risk level is greater than or equal to the threshold, the second type inspection device to issue a risk handling strategy matching the behavior traffic risk level to the risk handling device and the first type inspection device.

[0017] In an optional embodiment, the system further includes an inspection strategy adjustment module, and the inspection strategy adjustment includes: A first policy issuing submodule is configured for the first type of inspection device to adjust the first inspection policy according to the risk handling policy issued by the second type of inspection device or the cloud; The second strategy issuing submodule is used for the second type of inspection device to adjust the second inspection strategy according to the risk handling strategy issued by the first type of inspection device or the cloud.

[0018] In an optional embodiment, the model building module includes: A data processing submodule, configured to align spatial coordinates and synchronize timestamps of the three-dimensional scene structure model uploaded by the first type of inspection equipment and the traffic behavior semantic model uploaded by the second type of inspection equipment; The implicit relationship processing submodule is used to determine the behavioral trajectories, state changes, and correlations between environmental factors of traffic participants in specific spatial locations based on the aligned 3D scene structure model and traffic behavior semantic model, and generate a structure-behavior semantic mapping dataset. The fusion submodule is used to jointly model the structure-behavior semantic mapping data with multi-source macro traffic situation data and multi-source micro traffic situation data to construct a digital twin target model of the area to be inspected.

[0019] In an optional embodiment, the risk processing module includes: The prediction submodule is used to predict the set of traffic element change factors of the next detection cycle in the detection area based on the digital twin target model of the detection area. The set of traffic element change factors includes at least spatial structure characteristics, traffic density characteristics and traffic participant behavior characteristics; The evaluation submodule is used to input the set of traffic element change factors of the next detection cycle in the detection area into the preset traffic risk time series prediction model to calculate the comprehensive traffic risk score; The risk determination submodule is used to determine the traffic risk level of the area to be inspected based on the matching results of the comprehensive traffic risk score and the preset risk level classification standards.

[0020] In a third aspect, the present application also provides an electronic device, comprising: a memory and one or more processors, the memory being coupled to the processor; wherein computer program code is stored in the memory, the computer program code comprising computer instructions, and when the computer instructions are executed by the processor, the electronic device executes the method in any possible design mode of the above-mentioned first aspect.

[0021] In a fourth aspect, the present application provides a computer-readable storage medium comprising computer instructions; when the computer instructions are executed on an electronic device, the electronic device executes the method in the first aspect and any possible design thereof.

[0022] In a fifth aspect, the present application provides a computer program product, which, when executed on an electronic device, enables the electronic device to execute the method in the first aspect and any possible design thereof.

[0023] This application provides a traffic risk management method based on spatial intelligence scenarios. By using the actual traffic monitoring needs of the inspection area as a basis, it enables targeted task assignment, avoiding inspection blind spots and resource waste. Secondly, the collection and integration of multi-source heterogeneous data at the macro and micro levels addresses the shortcomings of existing methods in fine-grained behavior recognition and global situation characterization. Thirdly, inspection equipment constructs a local digital twin initial model, enabling rapid initial risk assessment, reducing cloud load and improving overall response efficiency. Furthermore, the cloud constructs a high-precision digital twin target model through heterogeneous model alignment and data fusion, providing a more realistic and detailed representation of traffic conditions. Based on this model, the system can predict the evolution trend of traffic elements in the next inspection cycle and quantitatively assess future risk levels through a risk time series model, enhancing the foresight and accuracy of early warnings. Finally, by distributing the matched risk management strategies to inspection and risk management devices, a closed-loop risk management mechanism is established, centered on data-driven, model-supported, and policy-linked approaches. This fundamentally overcomes the problems of incomplete traffic condition monitoring, delayed risk warnings, and inefficient emergency response in traditional methods, effectively improving overall traffic safety assurance capabilities.

[0024] Among them, the technical effects of the second to fifth aspects refer to the technical effects of the first aspect and any of its embodiments, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A system architecture diagram provided for an embodiment of the present application; Figure 2 A flowchart of a traffic risk management method based on a spatial intelligence scenario provided in an embodiment of the present application; Figure 3A schematic structural diagram of a traffic risk management system based on a spatial intelligent scenario provided in an embodiment of the present application. DETAILED DESCRIPTION

[0026] The terms used in the following embodiments are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification and claims of the present application, the singular expressions "a", "a", "above", "the" and "this" are intended to also include expressions such as "one or more", unless there is a clear contrary indication in the context. It should also be understood that in the following embodiments of the present application, "at least one", "one or more" refer to one or more (including two). The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0027] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0028] In the following, the terms "first," "second," etc., are used for descriptive convenience only and should not be understood as indicating or implying relative importance or implicitly specifying the number of the technical features indicated. Thus, a feature defined as "first," "second," etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more. For example, "plurality of processing units" refers to two or more processing units.

[0029] Furthermore, in the embodiments of the present application, "upper," "lower," "left," and "right" are not limited to being defined relative to the orientation of the components schematically shown in the drawings. It should be understood that these directional terms can be relative concepts. They are used for relative description and clarification, and may change accordingly based on changes in the orientation of the components in the drawings. In the drawings, the thickness of layers and regions is exaggerated for clarity, and the dimensional ratios between the components in the drawings do not reflect the actual dimensional ratios.

[0030] In the embodiments of this application, unless otherwise specified or limited, the term "connection" should be understood in a broad sense. For example, "connection" can mean fixed connection, detachable connection, or integration; it can mean direct connection or indirect connection through an intermediate medium. In addition, the term "electrical connection" can mean direct electrical connection or indirect electrical connection through an intermediate medium.

[0031] In the embodiments of the present application, the term "module" generally refers to a functional structure divided according to logic. The "module" can be implemented by pure hardware or a combination of hardware and software. In the embodiments of the present application, "and / or" describes the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, B exists alone, and A and B exist at the same time.

[0032] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0033] With the acceleration of urbanization, the number of traffic participants—motor vehicles, non-motor vehicles, and pedestrians—continues to rise, increasing the operational pressure on urban road traffic systems. Driven by the high density of mixed traffic, multi-hour traffic fluctuations, and complex road network structures, traffic violations, accidents, and public safety emergencies are becoming increasingly frequent, becoming key and difficult issues that urgently need to be addressed in urban management. Especially during special scenarios such as rush hour, large-scale events, and inclement weather, the uncertainty and risk of traffic operations increase significantly. Traffic congestion, vehicle conflicts, and pedestrians cutting in are commonplace, seriously impacting urban operational efficiency and public safety.

[0034] The traditional on-site law enforcement model, which primarily relies on manual patrols and fixed-point monitoring, has exposed multiple shortcomings in practical application. On the one hand, manual inspections are limited by the scope of manpower deployment, operating time, and perception accuracy, making it difficult to achieve comprehensive coverage and real-time monitoring of large, dynamically changing areas, and prone to "blind spots" or "dead corners" in supervision. On the other hand, in key links such as collecting evidence of traffic violations, determining responsibility at accident scenes, and rapidly responding to emergencies, the traditional model often relies on the subjective judgment and experience of on-site law enforcement personnel. This leads to incomplete data collection, untimely information feedback, and inefficient dispatch and coordination, making it difficult to meet the needs of refined management and intelligent response in modern cities.

[0035] Furthermore, current traffic management systems are plagued by widespread data silos. Data from multiple sources, including cameras, sensors, vehicles, and mobile devices, lacks effective integration and real-time sharing. This results in limited urban traffic perception and forecasting capabilities, and management decisions often lag behind the actual evolution of traffic conditions. Traditional systems struggle to achieve semantic understanding and trend prediction of complex traffic behaviors, limiting the ability to proactively prevent and control traffic and precisely intervene. Therefore, there is an urgent need to introduce new technology systems with spatial perception, semantic analysis, and intelligent reasoning capabilities to establish an intelligent, systematic traffic management platform covering key urban roads and key areas, thereby achieving comprehensive perception, intelligent assessment, and efficient handling of urban traffic risks.

[0036] Based on this, the inventive concept of this application is proposed: through information sharing and functional collaboration among multiple distributed jammers, a multi-level interference area that combines main and auxiliary and is dynamically reconstructed is constructed, which can flexibly adjust the interference strategy and beam parameters according to the spatial position, behavioral characteristics and signal response of the target receiver, to achieve high-precision and high-efficiency adaptive collaborative interference, and effectively improve the response speed, interference effect and system robustness of the jamming system in complex environments.

[0037] See Figure 1 First, the application scenarios and system architecture involved in this application are explained. This application provides a spatial intelligent scene traffic risk management system for complex traffic environments, which mainly includes a cloud 101, a first type of inspection equipment 102, a second type of inspection equipment 103 and a risk management equipment 104.

[0038] Cloud 101 can be a traffic police command center, responsible for building, maintaining, and updating a digital twin target model of the traffic area in real time. The first type of inspection device 102 can be a drone, and the second type of inspection device 103 can be a traffic service robot. These devices integrate AI visual recognition modules, voice interaction units, and autonomous navigation chassis. They can automatically identify illegal parking, driving against traffic, and illegally occupying lanes while on the ground, and provide contactless persuasion to vehicle owners via voice or display. They also collect evidence and upload it to Cloud 101 in real time. The risk management device 104 can be a quadruped robot (i.e., a robot dog) equipped with flexible motion control and a variety of sensors, including but not limited to infrared sensors, ultrasonic rangefinders, lidar, and high-definition vision modules. They can perform autonomous navigation, environmental mapping, anomaly detection, and assisted intervention in environments such as confined spaces, complex terrain, and accident scenes. Using integrated simultaneous localization and mapping (SLAM) technology, they achieve high-precision positioning and real-time mapping in unknown or dynamic environments, adapting to complex operational scenarios such as narrow alleys, collapsed debris piles, and uneven terrain. Furthermore, the robot dog's high-degree-of-freedom quadruped structure and adaptive motion planning algorithm allow it to flexibly adjust its posture and gait based on 360-degree spatial constraints. This provides strong stability and reliability in obstacle avoidance, obstacle crossing, and posture maintenance, significantly improving operational capabilities and response efficiency in high-risk environments. Three types of inspection equipment and Cloud 101 form a collaborative working system. UAVs and robot dogs form an air-ground interconnected perception network, while traffic service robots are responsible for near-ground law enforcement and human-machine interaction. Cloud 101 centrally dispatches various devices to conduct risk assessments, level assessments, and policy issuance, thus establishing an urban traffic risk management platform with integrated air-ground, virtual-real, and intelligent linkage capabilities.

[0039] Reference Figure 2 The embodiment of the present application provides a traffic risk management method based on a spatial intelligence scenario, which is applied to the above architecture and may specifically include the following steps: S201: The cloud sends an inspection strategy to the inspection device, and the inspection device obtains multi-source heterogeneous data of the area to be inspected according to the inspection strategy.

[0040] In this embodiment, the inspection zone refers to specific areas or sections within the transportation system that require inspection and monitoring, such as roads, bridges, tunnels, intersections, or other critical areas of transportation facilities. Based on traffic management needs and real-time conditions, the cloud sends inspection strategies to inspection equipment. These strategies define specific inspection objectives, key inspection items, the types and methods of data to be collected, and the time and frequency of inspections. Following these strategies, inspection equipment utilizes a variety of sensors and collection methods within the inspection zone to acquire heterogeneous data from multiple sources, including road surface images, traffic flow data, environmental parameters (such as temperature and humidity), vehicle identification information, and structural vibration data. This provides a comprehensive and accurate picture of transportation facility and road conditions.

[0041] Depending on the type of inspection equipment, it may include: S2011: The cloud sends a first inspection strategy to the first type of inspection device, and the first type of inspection device obtains multi-source macro traffic situation data of the area to be inspected according to the first inspection strategy.

[0042] In this embodiment, the first type of patrol equipment is used to monitor the traffic conditions in the inspection area at a macro level. The first type of patrol equipment can be equipped with a variety of sensors such as visible light cameras, infrared thermal imagers, lidars, millimeter-wave radars, etc. Under the guidance of the initial patrol routes and data collection rules issued by the cloud, it performs tasks such as high-altitude cruising and image transmission, perceives the overall operation status of traffic in real time, and ensures comprehensive and continuous monitoring of traffic conditions over a large area.

[0043] Multi-source macro traffic situation data refers to comprehensive information collected by first-type inspection equipment through multiple types of sensors that reflects the overall traffic operation status. These data usually cover multiple dimensions such as traffic flow, speed distribution, vehicle density, road congestion, traffic events (such as accidents, construction), weather and environmental conditions. By integrating heterogeneous perception data collected by lidar, thermal imagers, image acquisition modules and communication modules, first-type inspection equipment can dynamically construct traffic situation maps within a large spatial range, comprehensively reflecting the operating status and changing trends of the traffic network, and providing accurate data support for traffic management and risk warning analysis in the cloud. S2012: The cloud sends a second inspection strategy to the second type of inspection device, and the second type of inspection device obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy.

[0044] In this embodiment, the second type of inspection equipment can be a traffic service robot, used to conduct refined ground-level traffic inspections within the inspection area. Equipped with perception and interaction modules such as an AI visual recognition module, a voice interaction unit, and an autonomous navigation chassis, this device automatically identifies and records behavioral information of ground traffic participants, such as illegal parking, driving against traffic, non-motorized vehicles occupying the road, and pedestrians jaywalking, during the inspection process, following inspection routes and collection rules issued by the cloud. This allows for close-range, dynamic perception of traffic conditions in key areas.

[0045] Multi-source micro-traffic situation data refers to fine-grained traffic information collected through the integration of multiple near-field sensors integrated into Type II inspection equipment. This data typically includes real-time locations of pedestrians and vehicles, their movement trajectories, behavioral categories, interactions between traffic participants, road resource usage, and the distribution of temporary obstacles. Data sources include high-resolution cameras, lidar, infrared sensors, voice acquisition units, millimeter-wave radar, and other sources.

[0046] S202: The inspection equipment constructs a digital twin initial model of the area to be inspected based on the multi-source heterogeneous data, and uploads the digital twin initial model and the multi-source heterogeneous data to the cloud.

[0047] After collecting multi-source heterogeneous data, inspection equipment can locally build an initial digital twin model of the area to be inspected based on this data. This initial digital twin model can be a lightweight model that reflects the basic traffic structure and dynamic situation of the area to be inspected under current spatiotemporal conditions. It can also perform functions such as spatial scene reconstruction, traffic element annotation, and preliminary analysis of behavioral information. This initial digital twin model can be a lightweight model with low computational overhead, fast loading speed, and suitability for edge device operation. It can achieve local mapping and preliminary modeling of the traffic environment. The initial digital twin model can then be used to make preliminary risk assessments for the area to be inspected, and different operational steps can be executed based on the results of the risk assessments.

[0048] Specifically, the steps for building the initial digital twin model may include: S2021: Type 1 inspection equipment pre-processes multi-source macro traffic situation data to obtain key spatial structure characteristics and dynamic traffic distribution characteristics of the area to be inspected; S2022: Construct a three-dimensional scene structure model of the area to be inspected based on key spatial structure characteristics and dynamic traffic distribution characteristics; S2023: The second type of inspection equipment pre-processes multi-source microscopic traffic situation data to obtain the spatiotemporal behavior characteristics of traffic participants; S2024: Construct a traffic behavior semantic model of the area to be detected based on the spatiotemporal behavior characteristics of traffic participants.

[0049] In the implementation schemes S2021 to S2024, the first type of inspection equipment fuses and preprocesses collected large-scale, multimodal sensory data (such as images, laser point clouds, radar reflections, and thermal imaging data), extracting the spatial topological relationships of static spatial elements such as road network structures, bridges, tunnels, and intersections. It also identifies dynamic traffic indicators such as traffic density, average speed, congestion distribution, and abnormal events, forming a core feature set that characterizes the macroscopic traffic patterns of the area to be inspected. After extracting key spatial structural features and traffic distribution data, the system uses modeling techniques such as point cloud reconstruction, image stitching, and SLAM (Simultaneous Localization and Mapping) to restore the inspection area to a three-dimensional spatial structure model and spatially annotate different traffic functional areas (such as lane divisions, traffic light locations, speed limit zones, etc.). Traffic flow information can be dynamically mapped into the model, achieving a preliminary coupled expression of the three-dimensional traffic environment and dynamic traffic situation, and constructing a twin skeleton model with both spatial and macroscopic dynamic dimensions.

[0050] The second type of inspection equipment analyzes and processes microscopic sensory information collected, including close-range high-precision images, lidar data, and voice signals. It extracts spatiotemporal behavioral characteristics, including the trajectories, motion states (such as stopping, accelerating, turning), and behavior types (such as illegal parking and driving against traffic) of traffic participants, including pedestrians, non-motorized vehicles, and vehicles, at different times and spatial locations. These characteristics reflect the operational details and behavioral complexity of the local traffic system, providing behavioral layer information for subsequent traffic semantic modeling. Based on the extracted spatiotemporal behavioral characteristics, combined with the behavior recognition model and semantic labeling system, a model is constructed that can express the semantic relationships between various types of traffic behaviors, such as semantic units such as "vehicle illegally changes lanes at an intersection" and "pedestrian crossing the road causes a traffic conflict." The traffic behavior semantic model will be integrated with the three-dimensional scene structure model to form the semantic enhancement layer of the initial digital twin model.

[0051] The steps for uploading the digital twin initial model and multi-source heterogeneous data to the cloud can include: S2025: The first type inspection device determines the spatial traffic risk level of the inspection area based on the three-dimensional scene structure model. If the spatial traffic risk level is less than a threshold, the three-dimensional scene structure model and multi-source macro traffic situation data are uploaded to the cloud. S2026: The second type of inspection equipment determines the behavioral traffic risk level of the area to be inspected based on the traffic behavior semantic model. When the behavioral traffic risk level is less than the threshold, the semantic model and multi-source micro-traffic situation data are uploaded to the cloud.

[0052] In the implementation of S2025 to S2026, the spatial traffic risk level is assessed by the first type of inspection equipment based on a constructed three-dimensional scene structure model. This assessment score can be derived using indicators such as spatial structural complexity (such as lane layout, intersection density, and limited sight distance), dynamic traffic pressure (such as traffic saturation, speed fluctuations, and congestion index), and environmental interference factors (such as severe weather and construction impacts) through a preset weight model or lightweight neural network. The behavioral traffic risk level is determined by the second type of inspection equipment in conjunction with a traffic behavior semantic model. The behavioral traffic risk level can be derived by extracting risk features and scoring them based on the frequency of abnormal behaviors of traffic participants (such as illegal parking and driving against traffic), behavioral aggregation characteristics, duration, and impact range, through AI visual recognition and spatiotemporal behavior sequence analysis.

[0053] To reduce computing pressure on the cloud and improve overall risk identification and response speed, the initial digital twin models constructed by inspection equipment (such as the 3D scene structure model and traffic behavior semantic model) adopt a lightweight design, with relatively limited accuracy and integrity. They are mainly used for local rapid initial screening and risk perception. Therefore, to avoid uploading large amounts of data from low-risk or no-risk areas to the cloud, which would cause redundant transmission and waste of resources, the corresponding model is only uploaded to the cloud along with the original multi-source heterogeneous data when the traffic risk level assessed locally by the equipment is below the preset threshold, that is, when the current model cannot confirm the existence of a substantial risk or only captures the boundary situation. The cloud will then use a high-precision fusion model to further judge and process it.

[0054] When the risk level identified by the local model is higher than the threshold, it means that the risk has been significantly detected even under the condition of limited model accuracy. The device can directly trigger the local response mechanism, which can include the following steps: When the spatial traffic risk level is greater than or equal to the threshold, the first type inspection device sends a risk handling strategy that matches the spatial traffic risk level to the risk handling device and the second type inspection device; When the behavior traffic risk level is greater than or equal to the threshold, the second-type patrol device sends a risk processing strategy that matches the behavior traffic risk level to the risk processing device and the first-type patrol device.

[0055] In this embodiment, both the first-type inspection device and the second-type inspection device have preliminary risk analysis and response capabilities, and can directly issue linkage instructions to other devices after detecting a high-risk event, thereby establishing a local rapid autonomous linkage processing system. Specifically, when the spatial traffic risk level is greater than or equal to a threshold, the first-type inspection device will generate a spatial risk processing strategy that matches the spatial risk source identified in the locally constructed three-dimensional scene structure model, combined with the preset risk level and scene matching rules. The strategy is then sent to the risk processing device and the second-type inspection device via the local network or 5G communication link to guide them in performing subsequent inspection reinforcement, on-site intervention, or abnormal handling operations. When the behavioral traffic risk level is greater than or equal to the threshold, the second-type inspection device analyzes the high-risk behavior characteristics identified in the micro-traffic behavior semantic model (such as concentrated illegal parking, wrong-way driving, illegal road occupation, etc.), matches the corresponding behavioral intervention strategy based on the behavioral risk level, and similarly sends the strategy to the risk processing device and the first-type inspection device to organize them to assist in completing tasks such as rapid perception supplementation, air-ground linkage response, or further image recording in high-risk areas. This processing mechanism relies on a pre-defined risk level-strategy mapping table and inter-device communication protocol, enabling the system to achieve multi-device collaboration, autonomous and efficient local risk processing capabilities without waiting for cloud responses.

[0056] In a feasible implementation, when the cloud receives the digital twin initial model and multi-source heterogeneous data uploaded by the first type of inspection device or the second type of inspection device but cannot process them in time due to insufficient computing resources, scheduling delays or other abnormal conditions, the cloud can transmit the digital twin initial model and multi-source heterogeneous data to the other type of inspection device closest to the area to be inspected based on the spatial distance matching relationship between the devices to collaboratively complete the processing. Specifically, if the three-dimensional scene structure model and macro traffic situation data uploaded by the first type of inspection device are received, the cloud will determine the second type of inspection device that is closer to the area to be inspected and send the relevant data to it. The second type of inspection device will further identify and model key targets or abnormal behaviors based on the ground perspective and semantic analysis capabilities. The second type of inspection device will send the risk treatment strategy to the first type of inspection device and the risk treatment device based on the risk identification results. Conversely, if the cloud receives traffic behavior semantic models and micro-traffic situation data uploaded by Type II patrol devices, it distributes this data to idle Type I patrol devices closer to the area, enabling them to perform supplementary modeling and analysis of behavior trajectories, spatial location distribution, and other information from an aerial perspective. Based on the risk identification results, the Type I patrol devices then transmit risk management strategies to Type II patrol devices and risk management devices. This distance-prioritized matching relationship is calculated based on parameters such as real-time positioning information, historical response delay records, device movement speed, and current task load, ensuring that data processing tasks are quickly handled by the optimally located device. This enables edge collaboration, efficient, and reliable traffic risk perception and response capabilities, even when cloud processing is limited.

[0057] In a feasible implementation manner, the method further includes: The first type of inspection device adjusts the first inspection strategy based on the risk management strategy issued by the second type of inspection device or the cloud; The second type of inspection device adjusts the second inspection strategy according to the risk management strategy issued by the first type of inspection device or the cloud.

[0058] In this embodiment, specifically, the above-mentioned adjustment process embodies a cross-device information sharing and task linkage mechanism, which aims to improve the overall system's response efficiency to sudden traffic risks and the flexibility of resource scheduling. As an example, the first-type inspection device currently executes a high-altitude traffic macro-monitoring task based on the first inspection strategy issued from the cloud. At this time, if the second-type inspection device identifies a high-risk behavior event during the ground-based close-range perception process and reports it to the cloud or directly forms a risk treatment strategy and sends it to the drone, the first-type inspection device can adjust the flight path, lower the cruising altitude, change the observation angle, or increase the observation frequency accordingly to more accurately perform tasks such as local image acquisition, structural reconstruction, and high-altitude call intervention in the event area. Similarly, if the first-type inspection device identifies a potential risk in a specific road section or area during high-altitude monitoring, such as a sharp increase in traffic density or abnormal stagnation, its risk treatment strategy can also be sent to the second-type inspection device via the cloud or direct broadcast, guiding it to the designated area for ground micro-behavior sampling and on-site intervention, thereby automatically introducing new path nodes, increasing recognition density, or adjusting task priorities in the second inspection strategy. That is, the inspection strategy is no longer static and unidirectional, but can be dynamically adjusted and multi-directionally coordinated between devices based on multi-source risk judgment results, enabling the system to allocate perception resources more efficiently and realize an integrated air-ground and adaptively coordinated traffic risk linkage response system.

[0059] Compared with the traditional model that relies on centralized cloud processing, the first type of inspection equipment and the second type of inspection equipment in this application both have local preliminary analysis, model building and risk identification capabilities, and no longer rely entirely on the cloud to complete the entire process judgment, thereby reducing the security risks caused by communication interruptions, cloud delays or computing bottlenecks. Equipment at all levels can carry out two-way or multi-directional collaborative processing based on risk level, spatial location and data characteristics. For example, when the first type of inspection equipment identifies a high-risk spatial situation, it can directly link the second type of inspection equipment to issue a behavior identification task, and vice versa, to build a collaborative mechanism with transferable tasks, complementary information and shared responses; at the same time, when the cloud cannot respond to the uploaded task in time, the system also supports matching strategies such as spatial distance priority, and dynamically sends the model and data tasks to nearby or capability-matched devices for assistance in processing, thereby ensuring the continuity and real-time nature of the overall monitoring and early warning process. This edge-cloud collaborative and end-to-end complementary system architecture significantly improves the system's autonomous adaptability and fault tolerance in complex and changeable traffic scenarios. It is a traffic risk management solution with highly distributed intelligence and collaborative flexibility.

[0060] Type-one inspection equipment primarily monitors large-scale traffic conditions at a macro level, while type-two inspection equipment focuses on identifying and analyzing micro-level behaviors in key areas and critical nodes. These two types complement each other in terms of functionality, enabling real-time information sharing and intelligent scheduling during task execution. When type-one equipment identifies risk trends such as abnormal traffic density or sudden changes in spatial structure in a specific area, it can proactively issue behavioral supplementary instructions to type-two equipment to further confirm the localized risk characteristics. Furthermore, when type-two equipment detects unusual traffic patterns but lacks environmental context, it can request macro-level contextual data from type-one equipment to enhance its judgment. Furthermore, when cloud processing capacity is limited or response times are delayed, devices can forward modeling tasks or receive data packets to each other based on matching strategies such as spatial distance, task pressure, or resource capacity, enabling collaborative, highly resilient operations even when the cloud is offline. Risk management devices (such as quadruped robots) are also integrated into the collaborative system, receiving risk strategies from any inspection equipment and rapidly intervening at the scene of an anomaly based on the urgency of the task. This collaborative system, in which each level of equipment has independent processing capabilities, serves as redundant support units for each other, and can be uniformly dispatched by the cloud, not only improves the system's response speed and reliability, but also significantly reduces its dependence on cloud stability and continuous bandwidth, realizing an integrated closed loop of distributed traffic intelligent perception, linkage warning and immediate disposal.

[0061] S203: The cloud constructs a digital twin target model of the area to be inspected based on the digital twin initial model and multi-source heterogeneous data.

[0062] In this embodiment, after receiving the digital twin initial model and the corresponding multi-source heterogeneous perception data uploaded from the first type of inspection equipment and the second type of inspection equipment, the cloud first performs unified formatting, noise suppression and semantic fusion processing on the uploaded data to improve the temporal and spatial alignment accuracy and information consistency of the data. On this basis, the cloud further performs fine reconstruction and semantic enhancement on the initial model to construct a more complete, real-time and predictive digital twin target model. This target model not only retains the spatial structure characteristics and traffic behavior elements reflected in the initial model, but also realizes the modeling and prediction of dynamic traffic evolution trends through multimodal data fusion, machine learning algorithms and time series analysis models. The digital twin target model can support three-dimensional scene restoration, traffic participant behavior simulation, event tracing analysis and multi-dimensional risk level assessment, and provide panoramic and real-time virtual-to-real mapping support for subsequent intelligent scheduling, emergency response, policy simulation and traffic optimization. The specific steps may include: S2031: performing spatial coordinate alignment and timestamp synchronization on the three-dimensional scene structure model uploaded by the first type of inspection device and the traffic behavior semantic model uploaded by the second type of inspection device; S2032: Based on the aligned 3D scene structure model and traffic behavior semantic model, determine the behavioral trajectory of traffic participants in specific spatial locations, the relationship between state changes and environmental factors, and generate a structure-behavior semantic mapping dataset; S2033: Jointly model the structure-behavior semantic mapping data with multi-source macro traffic situation data and multi-source micro traffic situation data to construct a digital twin target model of the area to be inspected.

[0063] In the implementation of S2031 to S2032, due to differences in spatial resolution, coordinate system type, and temporal sampling frequency between the aerial and ground-based data sources, the cloud first needs to unify them. Spatial coordinate alignment is typically achieved through SLAM algorithms, GPS-IMU fusion, or spatial anchoring based on known landmarks (such as road markings, traffic lights, and building outlines), enabling precise mapping of the two models in three-dimensional space. Timestamp synchronization involves temporal interpolation or keyframe registration of the acquisition times associated with the sensory data, ensuring that behavioral characteristics and structural changes are integrated and analyzed at the same temporal scale. The spatial trajectories of the behavioral entities (e.g., motor vehicles, non-motor vehicles, and pedestrians) in the three-dimensional scene are then mapped to semantic behavioral labels (e.g., wrong-way driving, congestion source, illegal lane change), and their interactions with road structure (e.g., lane type, traffic light location, blind spots) and the external environment (e.g., lighting, weather, and visibility). This semantic mapping dataset not only presents the spatial context in which the behavior occurs, but also expresses the logical path of behavioral inducements, risk characteristics and environmental coupling. It is the core intermediate layer for building high-quality digital twin models.

[0064] Specifically, the cloud uses multimodal fusion modeling methods (such as neural networks, Bayesian networks, or spatiotemporal causal modeling frameworks) to extract features and jointly learn from these three types of data. This enables the model to not only accurately restore scene structure, but also describe traffic behavior evolution in real time, predict risk trends, and form a global perception and dynamic feedback capability for system operating status. The resulting digital twin target model is highly realistic, time-consistent, and intelligently predictive, capable of supporting a variety of intelligent decision-making tasks such as subsequent scheduling optimization, strategy generation, and real-world simulation.

[0065] S204: The cloud determines the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected.

[0066] In this implementation, the cloud comprehensively assesses the traffic risk level of the area being inspected by analyzing the state of traffic elements, their behavioral evolution trends, and their interactions with environmental factors as reflected in the digital twin target model. Traffic risk levels can be divided into multiple levels (e.g., low, medium, relatively high, high, and very high), each corresponding to a specific scoring range and pre-set response strategy. Specific steps may include: S2041: Based on the digital twin target model of the area to be inspected, predict the set of traffic element change factors for the next inspection cycle of the area to be inspected. The set of traffic element change factors includes at least spatial structure characteristics, traffic density characteristics, and traffic participant behavior characteristics. S2042: Inputting a set of traffic element change factors for the next inspection cycle of the inspection area into a preset traffic risk time series prediction model to calculate a comprehensive traffic risk score; S2043: Determine the traffic risk level of the area to be inspected based on the matching result of the comprehensive traffic risk score and the preset risk level classification standard.

[0067] In the implementation of steps S2034 to S2043, the cloud first performs a short-term traffic evolution trend forecast based on the constructed digital twin target model of the inspection area. Specifically, in S2041, the system analyzes the three-dimensional spatial structure, traffic flow status, semantic labels of traffic participant behavior, and external environmental factors integrated into the current model to extract variables that will critically influence the future evolution of traffic conditions. Based on traffic dynamics models, behavior propagation models, and historical correlation patterns, the system performs short-term time series deduction on these variables to predict a set of traffic elements likely to change in the inspection area during the next inspection cycle, i.e., a set of traffic element change factors. This set is not a simple description of the current state, but rather a prediction of traffic trends over the next cycle. It includes at least spatial structural characteristics (e.g., a prediction that construction fencing will expand its impact in a few minutes), traffic density characteristics (e.g., a prediction that traffic flow in high-risk areas will continue to increase during peak hours, leading to critical traffic capacity), and traffic participant behavior characteristics (e.g., a prediction that conflicting behaviors will cluster in specific areas or that the frequency of abnormal behaviors will increase).

[0068] The cloud then feeds the predicted set of traffic element change factors into a pre-trained traffic risk time series prediction model. This model can utilize a long-short-term memory network, a Transformer architecture, or a hybrid model combining causal inference with graph neural networks. Based on the time series trends of future changes in these factors, their spatial topological associations, and their statistical correlations with historical high-risk events, the model outputs a comprehensive traffic risk score, quantitatively reflecting the likelihood and estimated severity of potential risk events occurring in the inspected area within the next inspection cycle.

[0069] Finally, the cloud matches the score with the preset risk level classification standards, and uses threshold classification, segment mapping or fuzzy membership function to map the comprehensive score into a specific risk level. The level can adopt a five-level system (such as safety, attention, warning, warning, high risk) or set a more detailed grading system based on specific business needs.

[0070] For example, at a major road intersection, the currently constructed digital twin target model indicates that the area is shrinking due to construction fencing. Combining historical patterns with on-site situation predictions, the system infers that traffic volume will continue to increase by approximately 65% over the next detection cycle, with the average speed dropping to 15 km / h. Furthermore, the drone predicts that multiple instances of non-motorized vehicles crossing the lane will occur within the next 10 minutes. Historical data also shows that such incidents are highly correlated with traffic accidents. The system then predicts a set of traffic element change factors: construction obstruction trend = increasing, predicted traffic flow increase = +65%, predicted average speed = 15 km / h, non-motorized vehicle violation trend = increasing, and historical accident correlation intensity = high. This set of prediction factors is fed into a time-series risk assessment model built using LSTM and graph neural networks, resulting in a comprehensive traffic risk score of 77. The cloud determines that the score falls at the "warning" level (for example, the threshold range is 61-80), and ultimately determines that the intersection area has a high risk in the next detection cycle, thereby triggering the corresponding medium-level graded response strategy and issuing corresponding risk response instructions to various inspection equipment and risk management equipment.

[0071] After determining the traffic risk level of the area to be inspected, the cloud needs to issue risk management strategies that match the traffic risk level to the inspection equipment and risk management equipment. Specifically, the cloud first selects a response strategy that matches the current risk level, traffic element change characteristics, regional spatial attributes, and traffic participant behavior patterns based on the preset strategy library corresponding to the divided traffic risk level. This includes but is not limited to patrol frequency adjustment, perception range optimization, weighted inspection of key areas, risk target tracking, voice persuasion triggering, on-site control deployment and other sub-strategy combinations. Then, based on the functional positioning and spatial distribution of various types of equipment, combined with the real-time location and impact range of the risk event, the cloud parses the selected strategy into specific instruction packages for different devices through the task scheduling module, and issues it to the corresponding first-type patrol equipment, second-type patrol equipment, and risk management equipment in accordance with the principles of spatial distance priority and minimization of task redundancy. For example, for identified "warning" level risks, the cloud may send a command package to a nearby multi-rotor drone, requiring it to lower its flight altitude and increase the frequency of image acquisition, and at the same time send instructions to the ground transportation service robot to adjust its path to approach the risk area and activate the voice prompt function; if it is a "high-risk" risk, instructions such as entering the scene, executing environmental mapping and intervention preparation will be sent to the quadruped robot at the same time, so as to achieve a rapid coordinated response of the air and ground, thereby effectively mitigating or controlling the further spread of the risk situation.

[0072] The embodiment of the present application provides a traffic risk management method based on a spatial intelligent scenario. The cloud first intelligently generates and issues inspection strategies based on the traffic monitoring needs, historical risk distribution, and dynamic traffic environment of the area to be inspected, thereby realizing on-demand allocation of inspection resources and route optimization, avoiding blind spots in monitoring coverage caused by blind or repeated inspections in traditional methods, and thus improving the spatial and temporal integrity of data collection. Based on the inspection strategy, the inspection equipment collects multi-source heterogeneous traffic situation data, including macro (such as traffic flow, road status) and micro (such as behavioral characteristics, conflict events), and quickly builds a digital twin initial model locally based on the collected data to achieve low-latency and high-frequency perception of the traffic environment. This local modeling mechanism makes up for the problem that traditional methods are insufficient in capturing micro-behaviors in complex scenarios. Model upload and fusion modeling (cloud): After receiving the initial model and raw data, the cloud fuses the structural models and behavioral semantic models from different devices, uniformly aligns the spatial coordinates and time labels, and then constructs a digital twin target model with higher accuracy and more detailed performance. This step achieves semantic hierarchical integration of heterogeneous information, significantly improving the modeling accuracy and predictive capabilities of dynamic traffic situations and addressing the limited model representation of existing approaches. Based on the constructed target model, the system employs a time-series prediction model to predict the evolutionary trends of traffic elements during the next inspection cycle, deriving a comprehensive traffic risk score and assigning risk levels based on the score and pre-set criteria. This approach enhances the forward-looking and quantitative capabilities of risk identification, effectively overcoming the limitations of traditional approaches that rely on static rules and fail to respond promptly to sudden risks. Based on the determined risk level, the cloud matches and distributes differentiated risk management strategies to inspection equipment and risk management devices (such as quadruped robots), achieving a closed-loop linkage of perception, identification, and response. Strategies can encompass equipment path reconstruction, regional inspection densification, and on-site intervention instructions, improving the accuracy and timeliness of risk response and overcoming the lag and crude handling inherent in traditional systems.

[0073] The present application also provides a traffic risk management system based on a spatial intelligent scenario, referring to Figure 3 , shows a functional module diagram of a traffic risk management system 400 based on a spatial intelligent scenario of the present application, which may include the following modules: The first policy issuing module 401 is used to send the inspection policy from the cloud to the inspection device. The inspection device obtains multi-source heterogeneous data of the inspection area according to the inspection policy. The inspection policy is used to indicate the initial inspection route and data collection rules of the inspection device. The data acquisition module 402 is used for the inspection equipment to construct a digital twin initial model of the inspection area based on multi-source heterogeneous data, and upload the digital twin initial model and multi-source heterogeneous data to the cloud; The model construction module 403 is used to construct a digital twin target model of the area to be inspected based on the digital twin initial model and multi-source heterogeneous data in the cloud; The risk processing module 404 is used in the cloud to determine the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, and to issue a risk processing strategy that matches the traffic risk level to the inspection equipment and risk processing equipment.

[0074] In an optional implementation, the first policy issuing module includes: A first strategy delivery submodule is configured to transmit a first inspection strategy from the cloud to a first type of inspection device. The first type of inspection device obtains multi-source macro traffic situation data of the inspection area according to the first inspection strategy. The first inspection strategy is configured to indicate an initial inspection route and data collection rules for the first type of inspection device. The second strategy sending sub-module is used to send the second inspection strategy to the second type of inspection equipment in the cloud. The second type of inspection equipment obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy. The second inspection strategy is used to indicate the initial inspection route and data collection rules of the second type of inspection equipment.

[0075] In an optional embodiment, the data acquisition module includes: The first data processing submodule is used for the first type of inspection equipment to pre-process the multi-source macro traffic situation data to obtain the key spatial structure characteristics and dynamic traffic distribution characteristics of the area to be inspected; The first model building submodule is used to build a three-dimensional scene structure model of the area to be detected based on key spatial structure characteristics and dynamic traffic distribution characteristics; A second data processing submodule is used for the second type of inspection equipment to pre-process the multi-source microscopic traffic situation data to obtain the spatiotemporal behavior characteristics of traffic participants; The second model building submodule is used to build a traffic behavior semantic model of the area to be detected based on the spatiotemporal behavior characteristics of traffic participants.

[0076] In an optional embodiment, the data acquisition module further includes: A first data uploading submodule is used for the first type of inspection equipment to determine the spatial traffic risk level of the inspection area based on the three-dimensional scene structure model, and upload the three-dimensional scene structure model and multi-source macro traffic situation data to the cloud when the spatial traffic risk level is less than a threshold; The second data uploading submodule is used for the second type of inspection equipment to determine the behavioral traffic risk level of the area to be inspected based on the traffic behavior semantic model. When the behavioral traffic risk level is less than the threshold, the semantic model and multi-source micro-traffic situation data are uploaded to the cloud.

[0077] In an optional implementation, the system further includes a second policy issuing module, which includes: The third strategy issuing submodule is configured to, when the spatial traffic risk level is greater than or equal to the threshold, cause the first type inspection device to issue a risk handling strategy that matches the spatial traffic risk level to the risk handling device and the second type inspection device; The fourth strategy issuing submodule is used for, when the behavior traffic risk level is greater than or equal to the threshold, the second type inspection device to issue a risk handling strategy matching the behavior traffic risk level to the risk handling device and the first type inspection device.

[0078] In an optional embodiment, the system further includes an inspection strategy adjustment module, and the inspection strategy adjustment includes: A first policy issuing submodule is configured for the first type of inspection device to adjust the first inspection policy according to the risk handling policy issued by the second type of inspection device or the cloud; The second strategy issuing submodule is used for the second type of inspection device to adjust the second inspection strategy according to the risk handling strategy issued by the first type of inspection device or the cloud.

[0079] In an optional embodiment, the model building module includes: A data processing submodule, configured to align spatial coordinates and synchronize timestamps of the three-dimensional scene structure model uploaded by the first type of inspection equipment and the traffic behavior semantic model uploaded by the second type of inspection equipment; The implicit relationship processing submodule is used to determine the behavioral trajectories, state changes, and correlations between environmental factors of traffic participants in specific spatial locations based on the aligned 3D scene structure model and traffic behavior semantic model, and generate a structure-behavior semantic mapping dataset. The fusion submodule is used to jointly model the structure-behavior semantic mapping data with multi-source macro traffic situation data and multi-source micro traffic situation data to construct a digital twin target model of the area to be inspected.

[0080] In an optional embodiment, the risk processing module includes: The prediction submodule is used to predict the set of traffic element change factors of the next detection cycle in the detection area based on the digital twin target model of the detection area. The set of traffic element change factors includes at least spatial structure characteristics, traffic density characteristics and traffic participant behavior characteristics; The evaluation submodule is used to input the set of traffic element change factors of the next detection cycle in the detection area into the preset traffic risk time series prediction model to calculate the comprehensive traffic risk score; The risk determination submodule is used to determine the traffic risk level of the area to be inspected based on the matching results of the comprehensive traffic risk score and the preset risk level classification standards.

[0081] In this embodiment, an embodiment of the present application further provides an electronic device, which may include a memory and one or more processors. The memory and processors are coupled. The memory is configured to store computer program code, which includes computer instructions. When the processor executes the computer instructions, the electronic device may perform the functions or steps described in the above method embodiments.

[0082] This embodiment further provides a computer-readable storage medium, in which computer instructions are stored. When the computer instructions are executed on an electronic device, the electronic device executes each function or step in the above method embodiment.

[0083] This embodiment further provides a computer program product. When the computer program product is run on a computer, it enables the computer to execute each function or step in the above method embodiment.

[0084] Among them, the electronic device, computer-readable storage medium, and computer program product provided in this embodiment are all used to execute the corresponding methods provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0085] The above embodiments can be implemented in whole or in part via software, hardware (e.g., circuits), firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions according to the embodiments of the present application are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable system. The computer program or instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired means (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0086] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0087] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0088] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0089] In the several embodiments provided in this application, it should be understood that the disclosed systems, systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the system or unit can be electrical, mechanical or other forms.

[0090] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0091] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0092] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the existing technology, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0093] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A traffic risk management method based on spatial intelligent scenarios, characterized in that: The method comprises: The cloud sends an inspection strategy to the inspection device, and the inspection device obtains multi-source heterogeneous data of the area to be inspected according to the inspection strategy. The inspection strategy is used to indicate the initial inspection route and data collection rules of the inspection device; The inspection device constructs a digital twin initial model of the area to be inspected based on the multi-source heterogeneous data, and uploads the digital twin initial model and the multi-source heterogeneous data to the cloud; The cloud constructs a digital twin target model of the area to be inspected based on the digital twin initial model and the multi-source heterogeneous data; The cloud determines the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, and sends a risk handling strategy matching the traffic risk level to the inspection equipment and risk handling equipment.

2. The traffic risk management method based on spatial intelligent scene according to claim 1 is characterized in that: The inspection device includes a first type of inspection device and a second type of inspection device, the cloud sends an inspection strategy to the inspection device, and the inspection device obtains multi-source heterogeneous data of the area to be inspected according to the inspection strategy, including: The cloud sends a first inspection strategy to the first type of inspection device, and the first type of inspection device obtains multi-source macro traffic situation data of the to-be-inspected area according to the first inspection strategy, wherein the first inspection strategy is used to indicate an initial inspection route and data collection rules for the first type of inspection device; The cloud sends a second inspection strategy to the second type of inspection device, and the second type of inspection device obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy. The second inspection strategy is used to indicate the initial inspection route and data collection rules of the second type of inspection device.

3. The traffic risk management method based on spatial intelligent scene according to claim 2 is characterized in that: The inspection device constructs a digital twin initial model of the area to be inspected based on the multi-source heterogeneous data, including: The first type of inspection equipment pre-processes the multi-source macro traffic situation data to obtain key spatial structure characteristics and dynamic traffic distribution characteristics of the area to be inspected; Constructing a three-dimensional scene structure model of the area to be detected based on the key spatial structure characteristics and dynamic traffic distribution characteristics; The second type of inspection equipment pre-processes the multi-source microscopic traffic situation data to obtain spatiotemporal behavior characteristics of traffic participants; A traffic behavior semantic model of the area to be detected is constructed according to the spatiotemporal behavior characteristics of the traffic participants.

4. The traffic risk management method based on spatial intelligent scene according to claim 3 is characterized in that: The uploading of the digital twin initial model and the multi-source heterogeneous data to the cloud includes: The first type of inspection equipment determines the spatial traffic risk level of the to-be-inspected area based on the three-dimensional scene structure model, and uploads the three-dimensional scene structure model and the multi-source macroscopic traffic situation data to the cloud when the spatial traffic risk level is less than a threshold; The second type of inspection equipment determines the behavioral traffic risk level of the area to be inspected based on the traffic behavior semantic model, and uploads the semantic model and the multi-source micro-traffic situation data to the cloud when the behavioral traffic risk level is less than a threshold.

5. The traffic risk management method based on spatial intelligent scene according to claim 4 is characterized in that: The method further comprises: When the spatial traffic risk level is greater than or equal to a threshold, the first type inspection device sends a risk handling strategy matching the spatial traffic risk level to the risk handling device and the second type inspection device; When the behavior traffic risk level is greater than or equal to a threshold, the second-type patrol device issues a risk handling strategy that matches the behavior traffic risk level to the risk handling device and the first-type patrol device.

6. The traffic risk management method based on spatial intelligent scene according to claim 5 is characterized in that: The method further comprises: The first type of inspection device adjusts the first inspection strategy according to the risk management strategy issued by the second type of inspection device or the cloud; The second type of inspection device adjusts the second inspection strategy according to the risk handling strategy issued by the first type of inspection device or the cloud.

7. The traffic risk management method based on spatial intelligent scene according to claim 1 is characterized in that: The cloud constructs a digital twin target model of the area to be inspected based on the digital twin initial model and the multi-source heterogeneous data, including: Perform spatial coordinate alignment and timestamp synchronization on the three-dimensional scene structure model uploaded by the first type of inspection equipment and the traffic behavior semantic model uploaded by the second type of inspection equipment; Based on the aligned three-dimensional scene structure model and the traffic behavior semantic model, determining the association between the behavior trajectory, state changes and environmental elements of the traffic participants in a specific spatial location, and generating a structure-behavior semantic mapping dataset; The structure-behavior semantic mapping data is jointly modeled with multi-source macro traffic situation data and multi-source micro traffic situation data to construct a digital twin target model of the area to be inspected.

8. The traffic risk management method based on spatial intelligent scene according to claim 1 is characterized in that: The cloud determines the traffic risk level of the area to be inspected based on the digital twin target model of the area to be inspected, including: Based on the digital twin target model of the area to be inspected, predict a set of traffic element change factors for the next inspection cycle of the area to be inspected, where the set of traffic element change factors includes at least spatial structure characteristics, traffic density characteristics, and traffic participant behavior characteristics; Inputting the traffic element change factor set of the next detection cycle of the to-be-detected area into a preset traffic risk time series prediction model to calculate a comprehensive traffic risk score; The traffic risk level of the area to be inspected is determined based on a matching result between the comprehensive traffic risk score and a preset risk level classification standard.

9. A traffic risk management system based on spatial intelligent scenarios, characterized in that: For implementing the method according to any one of claims 1 to 8, the system comprises: A first strategy issuing module is used to send an inspection strategy from the cloud to the inspection device, and the inspection device obtains multi-source heterogeneous data of the area to be inspected according to the inspection strategy. The inspection strategy is used to indicate the initial inspection route and data collection rules of the inspection device; A data acquisition module is used for the inspection equipment to construct a digital twin initial model of the area to be inspected based on the multi-source heterogeneous data, and upload the digital twin initial model and the multi-source heterogeneous data to the cloud; A model construction module is used in the cloud to construct a digital twin target model of the area to be inspected based on the digital twin initial model and the multi-source heterogeneous data; The risk processing module is used in the cloud to determine the traffic risk level of the area to be detected based on the digital twin target model of the area to be detected, and to issue a risk processing strategy that matches the traffic risk level to the inspection equipment and risk processing equipment.

10. The traffic risk management system based on spatial intelligent scene according to claim 9 is characterized in that: The first policy issuing module includes: A first strategy sending submodule is configured for the cloud to send a first inspection strategy to a first type of inspection device, whereby the first type of inspection device obtains multi-source macroscopic traffic situation data of the area to be inspected according to the first inspection strategy, wherein the first inspection strategy is used to indicate an initial inspection route and data collection rules for the first type of inspection device; The second strategy sending sub-module is used for the cloud to send the second inspection strategy to the second type of inspection equipment. The second type of inspection equipment obtains multi-source micro-traffic situation data of the area to be inspected according to the second inspection strategy. The second inspection strategy is used to indicate the initial inspection route and data collection rules of the second type of inspection equipment.

Images