Tunnel traffic flow conflict management method and system based on dynamic priority

Abstract

This invention relates to the field of tunnel traffic scheduling technology, and in particular to a method and system for managing tunnel traffic flow conflicts based on dynamic priority. The method includes: acquiring a multi-source traffic perception dataset of the target tunnel operation area; generating an initial conflict candidate set including existing and incremental conflict units by combining a conflict identification logic template; splitting the set into static spatiotemporal and dynamic operation dimensions; extracting features using a graph neural network; and performing nonlinear weight fusion and spatial mapping based on traffic baseline data to output a conflict determination parameter set including priority correction coefficients and speed guidance factors, thereby generating an optimized priority control topology map. This invention can solve problems such as imperfect conflict identification and insufficient scheduling flexibility under complex geographical constraints, achieving refined management of tunnel traffic flow and enhancing the continuity of traffic flow in operation units.

Description

Technical Field

[0001] This invention relates to the field of tunnel traffic scheduling technology, and more specifically, to a method and system for managing tunnel traffic flow conflicts based on dynamic priority. Background Technology

[0002] Tunnel construction transportation scheduling is a core element in ensuring the efficiency of large-scale infrastructure construction, involving complex geographical constraints and diverse vehicle operation logics. As a key support for long tunnel construction, traffic flow management establishes models for the route planning and timing allocation of transport vehicles, aiming to ensure the continuity of construction material and excavated soil transportation. With the expansion of projects to high-altitude and deep mountain areas, the extremely narrow internal spaces, limited visibility, and the characteristics of multiple processes and vehicle types operating simultaneously give tunnel traffic networks typical high-dynamic and highly constrained attributes. This places higher demands on the tunnel traffic flow's conflict resolution capabilities and operational accuracy in complex environments. Achieving orderly vehicle scheduling through information technology is not only related to construction progress but also a crucial prerequisite for ensuring operational safety within the tunnel, reducing energy consumption, and optimizing resource allocation, thus playing a significant role in improving the overall intelligent management level of the project.

[0003] Currently, traditional tunnel traffic scheduling schemes have room for optimization when dealing with highly dynamic construction environments. On the one hand, existing technologies mostly focus on basic vehicle obstacle avoidance, and are not yet perfect in systematically identifying and modeling the intersection risks at side road junctions, head-on collisions in narrow sections, and congestion conflicts caused by localized traffic overload. This can easily lead to frequent failures of scheduling schemes during actual implementation. On the other hand, existing systems often lack scientific methods for quantifying the disorderly state of traffic flow, and the accuracy of assessing the severity of conflicts at different spatiotemporal nodes is limited. When facing large-scale road networks, the ability to quickly locate key conflict points needs to be strengthened. In addition, since scheduling strategies are mostly based on fixed priorities or single yielding rules, unnecessary timetable delays or chain congestion may occur when handling multi-vehicle conflicts. There is room for improvement in achieving a dynamic balance between conflict resolution and efficiency loss. In the context of multi-process cross-operations, due to the lack of comprehensive consideration of the urgency of vehicle tasks and path complexity, the flexibility of the system's response mechanism and the accuracy of controlling incremental scheduling costs when performing real-time intervention still have room for improvement.

[0004] Therefore, existing tunnel traffic management technologies still have limitations in handling the coupling problem of multi-dimensional conflict constraints and dynamic priority repair. There is an urgent need for a collaborative management solution that can integrate full-scenario conflict identification, scientifically quantify conflict intensity, and achieve precise fine-tuning of scheduling plans. This would ensure the continuity of high-priority tasks while improving the overall operational efficiency and intelligence level of the tunnel construction and transportation system. By constructing a closed-loop conflict perception and repair system, reliable technical support can be provided for intelligent traffic flow scheduling in complex and enclosed environments, thereby achieving efficient, safe, and orderly flow of construction materials under complex tunnel geographical constraints. Summary of the Invention

[0005] In view of this, the present invention proposes a method and system for managing tunnel traffic flow conflicts based on dynamic priority, aiming to solve the problems of imperfect conflict identification under complex geographical constraints, limited accuracy of quantification of disordered traffic flow, chain congestion caused by fixed priority strategies, and insufficient scheduling flexibility in multi-process cross-operation environments in existing tunnel traffic scheduling schemes.

[0006] This invention proposes a tunnel traffic flow conflict management method based on dynamic priority. This method is applied to a tunnel traffic flow collaborative management platform, which is used to digitally model and dynamically schedule multi-source traffic sensing infrastructure and transport vehicle resources, including the jurisdiction of at least one tunnel construction section.

[0007] Furthermore, the method includes: Obtain a multi-source traffic perception dataset for the target tunnel operation area. The multi-source traffic perception dataset includes real-time vehicle location coordinate vectors, vehicle speed vectors, road network topology constraint parameters, task urgency weights, and tunnel visibility coefficients. Based on the multi-source traffic perception dataset and the pre-set conflict identification logic template, an initial conflict candidate set for the target tunnel operation area is generated, where: The conflict identification logic template is used to support the logical rule expansion of conflict units based on multi-source traffic perception datasets and the automated dynamic generation of conflict units. A conflict unit is defined as a minimum conflict analysis unit in the traffic flow process. The initial conflict candidate set includes existing conflict units that have been mapped and expanded and incremental conflict units generated based on spatiotemporal trajectory prediction algorithms. Based on the initial conflict candidate set and traffic baseline data, an optimized priority control topology map for the target tunnel operation area is generated, where: The initial conflict candidate set is split into a static spatiotemporal dimension and a dynamic task dimension. A pre-trained graph neural network model is used to extract features at multiple scales for the static spatiotemporal dimension and the dynamic task dimension, respectively, to obtain a static spatiotemporal feature vector set and a dynamic task feature vector set. Based on traffic benchmark data, nonlinear weight fusion and spatial mapping are performed on static spatiotemporal feature vector sets and dynamic operation feature vector sets to output a set of conflict determination parameters, including vehicle traffic priority correction coefficient, road segment congestion threshold, intersection yielding duration parameter, speed guidance adjustment factor, and scheduling incremental cost control variable. Furthermore, after obtaining the multi-source traffic perception dataset of the target tunnel operation area, the process also includes: performing one-way hash masking on the specific vehicle identification codes, accurate mileage counts, and underlying communication protocol addresses of the equipment involved in the multi-source traffic perception dataset to generate a desensitized multi-source traffic perception dataset. Furthermore, before generating the initial set of conflict candidates for the target tunnel operation area, the following steps are also included: The system operation preference indicators input by the user are obtained. These indicators include preset construction schedule constraints, energy consumption control levels, safety distance guarantee weights, and equipment access trust coefficients. Based on the system operation preference index and the multi-source traffic perception dataset, the feature input layer of the conflict identification logic template is used to generate an initial conflict candidate set for the target tunnel operation area.

[0008] Furthermore, when generating the initial conflict candidate set for the target tunnel operation area based on the system operation preference index and the multi-source traffic perception dataset inputted into the feature input layer of the conflict identification logic template, the following steps are included: The conflict logic operators integrated within the conflict identification logic template are used to screen and reorganize conflict units, generate specific conflict resolution paths that meet the system operation preference indicators, and integrate the physical constraints corresponding to the system operation preference indicators into the initial conflict candidate set. Furthermore, the optimization priority control topology diagram of the target tunnel operation area exists in the form of a topology diagram, where a node in the topology diagram represents a conflict unit, and the edges in the topology diagram represent the logical coupling strength and influence transmission weight between conflict units. Furthermore, the conflict units include: Branch intersection conflict unit, narrow road collision conflict unit, local flow overload unit, sight-limited avoidance unit, and multi-process cross operation unit; When performing the identification of narrow road collision units, if the real-time distance between oncoming vehicles is detected to be lower than the preset distance threshold and the occupancy status of the passing tunnel in the narrow road section is a preset free status, the topology graph structure will automatically adjust the passage priority adjacency weight of the corresponding vehicles. Furthermore, it also includes: The optimized priority control topology map of the target tunnel operation area is input into the tunnel traffic management knowledge graph, and an interactive traffic flow operation status map is generated based on the 3D visualization engine. The tunnel traffic management knowledge graph includes a construction procedure logic pattern library, a vehicle-task binding relationship library, a historical conflict handling database, and an abnormal congestion response model library. Furthermore, when performing nonlinear weight fusion and spatial mapping on static spatiotemporal feature vector sets and dynamic operation feature vector sets based on traffic benchmark data, the following steps are included: The static spatiotemporal feature vector set includes the road network node connectivity, slope resistance coefficient, and tunnel cross-sectional width feature extraction values. The dynamic operation feature vector set includes the quantified values ​​of vehicle load change rate, remaining fuel / electricity status, and task deadline difference. Furthermore, when performing nonlinear weight fusion and spatial mapping, the following steps are included: The static spatiotemporal feature vector set and the dynamic operation feature vector set are uniformly encoded and normalized to form a standardized input traffic feature vector; The standardized input traffic feature vectors are weighted and summed according to their corresponding weight coefficients, and a bias term is introduced to obtain the fused intermediate priority representation. By mapping the fused intermediate priority representation into the nonlinear activation function, a nonlinear output result representing the relationship between the priority determination threshold and the regulation of traffic environmental factors is obtained. A conflict determination parameter set is generated based on the nonlinear output results. Specifically, the conflict determination parameter set guides the setting of the passage order and the allocation of spatiotemporal resources for each conflict point in tunnel transportation scheduling. Furthermore, when generating the optimized priority control topology map of the target tunnel operation area based on the initial conflict candidate set and traffic baseline data, the following steps are included: Based on the computational resource window planning value, calculate the number of parallel conflict processing units required for the initial conflict candidate set; Based on the scheduling concurrency, the number of conflict units that will be processed in parallel within the same computation window for the initial conflict candidate set is determined. Among them, traffic baseline data operations include calculating the planned value of the resource window and the concurrency of scheduling calculations; Furthermore, the traffic benchmark data is iteratively calibrated using a historical tunnel transportation sample set. The calibration process includes performing consistency analysis between the optimized priority control topology map and manually reviewed scheduling data, calculating the scheduling failure ratio and average delay deviation, and feeding them back into the loss function of the graph neural network model for parameter updates. Furthermore, the training process of the graph neural network model adopts a multi-task learning framework. The main task is to predict the probability of conflict occurrence, and the auxiliary tasks include road segment capacity regression, vehicle trajectory classification, and environmental perception confidence detection. Each task shares the underlying spatial feature encoder, and the upper-layer task heads output independently. Furthermore, the spatiotemporal trajectory prediction algorithm generates a local trajectory uncertainty index based on the vehicle's onboard sensors. This index comprehensively considers the vehicle's mechanical performance, braking distance constant, and current road surface friction coefficient, and is used to dynamically adjust the vehicle's safe following threshold and the time window coverage of the conflict warning. Furthermore, the conflict logic operators in the conflict identification logic template include a spatiotemporal union operator, an influence range expansion operator, a risk mask operator, and a multidimensional weight superposition operator. Each operator forms a traffic conflict analysis pipeline according to a preset execution order. Furthermore, after generating the optimized priority control topology map of the target tunnel operation area, the following steps are also included: The attributes of each node in the optimization priority control topology are converted into execution instructions for the vehicle control terminal. The execution instructions are then sent to the target vehicle through wireless communication base stations deployed in the tunnel. The on-board execution unit of the target vehicle receives the execution command and controls the throttle opening or braking pressure according to the speed guidance adjustment factor in the execution command to maintain the vehicle's driving state within the preset target range. Furthermore, if the feedback signal strength of the wireless communication base station is lower than the preset communication threshold, the on-board execution unit switches to the local autonomous scheduling mode and performs conflict avoidance actions based on the pre-stored static road network topology and real-time perception data. Compared with existing technologies, the advantages of this invention are as follows: By acquiring a multi-source traffic perception dataset of the target tunnel operation area, vehicle status, road network constraints, task weights, and environmental factors are uniformly spatially modeled. This allows the conflict identification process to be combined with the unique geographical constraints and operational logic of the tunnel construction site, reducing the disconnect between a single obstacle avoidance mode and a complex dynamic environment, and improving the spatial consistency of traffic flow under different working conditions. Secondly, by introducing a conflict identification logic template and dynamically generating conflict units based on conflict logic rules and spatiotemporal trajectory prediction algorithms, complex traffic conflicts are decomposed into reusable and scalable minimum analysis units, realizing the automatic construction and dynamic adjustment of the scheduling process. This approach reduces the dependence on fixed scheduling thresholds, improves the standardization and environmental adaptability of conflict identification, and enables differentiated risk characterization of different cross-process scenarios under a unified technical framework. Furthermore, by splitting the initial conflict candidate set into a static spatiotemporal dimension and a dynamic operational dimension, and using a pre-trained graph neural network model to perform multi-scale feature extraction, deep semantic fusion of multi-source heterogeneous traffic data is achieved. By performing nonlinear weight fusion and spatial mapping on two types of feature vectors based on traffic baseline data, the complex vehicle-road-environment interaction relationship can be effectively modeled, improving the accuracy and robustness of priority determination boundary identification. The final output conflict determination parameter set directly includes key decision variables such as priority correction coefficient, congestion threshold, yielding duration, and speed guidance factor, enabling conflict identification results to be directly transformed into input for traffic flow control and risk mitigation, achieving seamless integration of conflict perception and scheduling repair. This improves the refined management level of the tunnel transportation system, enhances the continuity of traffic in key operational units, and provides highly reliable underlying data support for subsequent smart construction site construction, energy efficiency optimization, and automated transportation.

[0009] On the other hand, the present invention also provides a tunnel traffic flow conflict control system based on dynamic priority, comprising: The multi-source sensing and acquisition module connects the tunnel sensor network and the ground server cluster to acquire multi-source traffic sensing datasets for the target tunnel operation area. The conflict generation module, connected to the multi-source sensing and acquisition module, generates an initial set of conflict candidates for the target tunnel operation area based on the multi-source traffic sensing dataset and the preset conflict identification logic template. The priority construction module, connected to the conflict generation module, generates an optimized priority control topology map of the target tunnel operation area based on the initial conflict candidate set and traffic benchmark data. The scheduling execution module, connected to the priority construction module, executes scheduling instructions for tunnel transport vehicles based on the conflict determination parameter set output by the optimized priority control topology graph. Furthermore, when the scheduling execution module performs operations, it transforms the influence propagation weights in the optimization priority control topology graph into control list entries of the scheduling system. When the real-time conflict determination value of a vehicle request exceeds the preset safety risk threshold, the corresponding right-of-way allocation, speed limit guidance, or forced stopping interface is activated.

[0010] It is understood that the tunnel traffic flow conflict control method and system based on dynamic priority in the above embodiments of the present invention have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0011] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 This is a flowchart illustrating the overall process of a tunnel traffic flow conflict control method based on dynamic priority provided in this application embodiment. Figure 2 This is a schematic diagram of the system composition of the tunnel traffic flow collaborative management and control platform provided in the embodiments of this application; Figure 3 This is a schematic diagram illustrating the interactive scenario in this embodiment of the application, showing how a user inputs system operation preference indicators and generates an initial set of conflict candidates. Figure 4 This is a schematic diagram of the processing flow of the tunnel traffic flow collaborative management and control platform in this application, which executes conflict risk assessment and dispatch instructions. Figure 5 This is a flowchart of the method for generating an optimized priority control topology graph based on the fusion of graph neural networks and multi-dimensional features in the embodiments of this application. Detailed Implementation

[0012] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0013] Before introducing the technical solution protected by this invention, several technical terms involved in the technical solution protected by this invention will be explained in advance, namely: The tunnel traffic flow collaborative management and control platform is a complex integrated architecture encompassing multiple traffic sensing infrastructures, vehicle-mounted resources, and a scheduling computing engine. Its configuration is used for digital modeling and traffic timing allocation of multi-source elements within the tunnel construction section. The platform typically includes elements such as a LiDAR sensor array, vehicle-mounted positioning terminals, roadside communication units, a construction progress database, a conflict identification logic operator library, a graph neural network model, a computing resource scheduler, and a 3D visualization engine. The operation of the platform is divided into several stages, including multi-source data acquisition, data anonymization and preprocessing, initial conflict identification, priority dynamic optimization, risk level assessment, continuous model iteration and accuracy monitoring, manual review and feedback and parameter correction, multi-terminal control command switching, containerized secure computing, and scheduling task completion. Among these, the continuous model iteration and accuracy monitoring stage is crucial for achieving precise traffic flow management. Through automated algorithms and process optimization, it accelerates the entire production and delivery process from multi-source sensing input to the execution of highly reliable scheduling commands. The practical content of this stage mainly includes automated feature construction, automated scheduling failure ratio testing, and automated management and control strategy deployment.

[0014] The target tunnel operation area refers to the specific operation site where the tunnel traffic flow collaborative management and control platform is deployed in construction and transportation management. This includes specific tunnel entrance dispatch centers, intersections of branch roads inside the tunnel, narrow road sections, vehicle-passing tunnels or mobile operation branches, as well as the vehicle database and traffic flow processing environment associated with these locations.

[0015] The conflict resolution parameter set details the priority configuration operations required to transform the current traffic flow state into the target orderly traffic state. Priority configuration operations can include CREATE, DELETE, and UPDATE operations. CREATE operations add new traffic constraint paths. DELETE operations remove redundant avoidance factors. UPDATE operations modify existing vehicle priority correction factors or road segment congestion thresholds. These operations ensure that changes to traffic flow conflict resolution are executed precisely as planned.

[0016] Next, the technical solution provided by this invention will be introduced.

[0017] Generally, extracting transportation conflicts in complex tunnel construction environments and designing scheduling procedures are extremely complex. During continuous model iteration and accuracy monitoring, analysts must have a deep understanding of several key factors, including the coupling between various factors in the traffic environment, the limitations of wireless signal fluctuations within the tunnel, the temporal nature of driving trajectories, the performance standards of onboard terminals, and the technical limitations of various trajectory prediction algorithms. Furthermore, they need to combine this with information on the actual construction procedures in the target area to accurately design the scheduling process. This process is highly manpower-intensive and typically requires the participation of senior traffic engineering experts or construction scheduling specialists. Faced with numerous tunnel operation areas with different procedures and geographical isolation, expert resources become a bottleneck, which restricts the efficiency of transportation operations.

[0018] In view of this, embodiments of the present invention provide a tunnel traffic flow conflict management method based on dynamic priority. This method utilizes a conflict identification logic template designed by technical experts based on traffic benchmark data and scheduling modeling experience. Then, it renders the conflict identification logic template using received traffic perception data of the target area to generate an initial conflict candidate set for the specific area. This method configures a design path for a complex conflict analysis process, reduces the consumption of expert resources, and minimizes analysis costs.

[0019] This method also utilizes traffic baseline data to automatically schedule multiple analytical steps in the dispatching process for each region, creating an optimized priority control topology that comprehensively considers factors such as factor coupling, environmental constraints, time gradients, equipment standards, and algorithm limitations. This optimized priority control topology executes an efficient and accurate ownership allocation procedure.

[0020] like Figure 1 As shown, specifically, a tunnel traffic flow conflict management method based on dynamic priority includes: Step S100: Obtain the multi-source traffic perception dataset of the target tunnel operation area. The multi-source traffic perception dataset includes the real-time vehicle location coordinate vector, vehicle speed vector, road network topology constraint parameters, task urgency weight, and tunnel visibility coefficient stored in the target tunnel operation area.

[0021] Specifically, after obtaining the multi-source traffic perception dataset of the target tunnel operation area, the process also includes: performing one-way hash masking on the specific vehicle identification codes, accurate mileage counts, and underlying communication protocol addresses of the equipment involved in the multi-source traffic perception dataset to generate a desensitized multi-source traffic perception dataset.

[0022] Understandably, during conflict management or scheduling authorization, the first step is to systematically acquire and integrate vehicle and facility-related information in the target tunnel operation area, enabling dynamic location, velocity vectors, physical constraints, and task weights to form a complete data foundation. Based on this, sensitive information such as vehicle identification and mileage offset is masked or anonymized, generating an anonymized multi-source traffic perception dataset for subsequent graph neural network modeling and access decision-making, while maintaining the data's logical structure and analytical validity. This achieves a balance between data privacy protection and real-time management requirements.

[0023] Step S200: Based on the multi-source traffic perception dataset and the preset conflict identification logic template, generate an initial conflict candidate set for the target tunnel operation area. The conflict identification logic template is used to support the logical rule expansion of conflict units based on the multi-source traffic perception dataset and the automated dynamic generation of conflict units. A conflict unit is defined as a minimum conflict analysis unit in the traffic flow process. The initial conflict candidate set includes existing conflict units that have been mapped and expanded, and incremental conflict units generated based on the spatiotemporal trajectory prediction algorithm.

[0024] Specifically, before generating the initial conflict candidate set for the target tunnel operation area, the process includes: obtaining system operation preference indicators input by the user, which include preset construction schedule constraints, energy consumption control levels, safety distance guarantee weights, and equipment access trust coefficients; and generating the initial conflict candidate set for the target tunnel operation area by inputting the system operation preference indicators and multi-source traffic perception datasets into the feature input layer of the conflict identification logic template.

[0025] Specifically, when generating the initial conflict candidate set for the target tunnel operation area based on the system operation preference index and the multi-source traffic perception dataset inputted into the feature input layer of the conflict identification logic template, the process includes: filtering and reorganizing conflict units based on the conflict logic operator integrated within the conflict identification logic template, generating specific conflict resolution paths that conform to the system operation preference index, and incorporating the physical constraints corresponding to the system operation preference index into the initial conflict candidate set.

[0026] Specifically, the optimization priority control topology diagram of the target tunnel operation area exists in the form of a topology diagram, where a node in the topology diagram represents a conflict unit, and the edges in the topology diagram represent the logical coupling strength and influence transmission weight between conflict units.

[0027] Specifically, the conflict units include: branch road intersection conflict units, narrow road head-on conflict units, local traffic overload units, sight-limited avoidance units, and multi-process cross-operation units; when performing the identification of narrow road head-on conflict units, if the real-time distance between oncoming vehicles is detected to be lower than a preset distance threshold and the occupancy status of the passing tunnel in the narrow road section is a preset free status, the topology graph structure automatically adjusts the traffic priority adjacency weight of the corresponding vehicles.

[0028] Understandably, by using vehicle and road network information of the target tunnel operation area and user-inputted operation preference indicators as joint inputs, feature analysis and logical deduction are performed through a preset conflict identification logic template. The conflict logic operators and prediction algorithms integrated within the template are used to screen, reorganize, and dynamically generate the smallest conflict analysis units in the identification process, thereby forming a scheduling process that meets specific safety objectives. This process is represented in a structured form as a topological graph, describing conflict units and their interactions through nodes and edges. During execution, the weights are adjusted in real time based on vehicle spacing and the availability of bypass tunnels, enabling the analysis path to adapt to environmental fluctuations and physical constraints, and performing dynamic optimization and controllable operation of the scheduling analysis process.

[0029] Step S300: Based on the initial conflict candidate set and traffic benchmark data, generate an optimized priority control topology map of the target tunnel operation area, wherein: the initial conflict candidate set is split into a static spatiotemporal dimension and a dynamic operation dimension, and a pre-trained graph neural network model is used to extract multi-scale features from the static spatiotemporal dimension and the dynamic operation dimension respectively to obtain a static spatiotemporal feature vector set and a dynamic operation feature vector set.

[0030] Specifically, when generating the optimized priority control topology map of the target tunnel operation area based on the initial conflict candidate set and traffic reference data, the process includes: calculating the number of parallel conflict processing units required for the initial conflict candidate set based on the planned value of the computing resource window; and determining the number of conflict unit processing units that can be executed in parallel within the same computing window based on the scheduling computing concurrency. The traffic reference data includes the planned value of the computing resource window and the scheduling computing concurrency.

[0031] Understandably, by using the initial conflict candidate set as the analysis framework and introducing traffic benchmark data to constrain scheduling rhythm and computational parallelism, the control process is transformed from a single static description into a schedulable computational plan. By splitting the candidate set into static spatiotemporal dimensions and dynamic operational dimensions, multi-scale feature extraction is performed using a pre-trained graph neural network model to form a feature vector set that can characterize road network attributes and operational habits. Based on this, the required number of processing units is calculated and the number of conflict units that can be executed in parallel within each computational window is determined by combining the computational resource window planning value and concurrency parameter, thus performing collaborative processing of control results in the time and computational resource dimensions.

[0032] Step S400: Input the optimized priority control topology map of the target tunnel operation area into the tunnel traffic management knowledge graph, and generate an interactive traffic flow operation status map based on the three-dimensional visualization engine.

[0033] Specifically, based on traffic benchmark data, nonlinear weight fusion and spatial mapping are performed on static spatiotemporal feature vector sets and dynamic operation feature vector sets to output a conflict determination parameter set including vehicle traffic priority correction coefficient, road segment congestion threshold, intersection yielding duration parameter, speed guidance adjustment factor, and scheduling incremental cost control variable.

[0034] Specifically, the optimized priority control topology map of the target tunnel operation area is input into the tunnel traffic management knowledge graph, and an interactive traffic flow operation status map is generated based on the 3D visualization engine. The tunnel traffic management knowledge graph includes a construction procedure logic pattern library, a vehicle-task binding relationship library, a historical conflict handling database, and an abnormal congestion response model library.

[0035] Specifically, when performing nonlinear weight fusion and spatial mapping on static spatiotemporal feature vector sets and dynamic operation feature vector sets based on traffic benchmark data, the static spatiotemporal feature vector set includes the road network node connectivity, slope resistance coefficient, and tunnel cross-section width feature extraction values; the dynamic operation feature vector set includes the vehicle load change rate, remaining fuel / electricity status, and the quantitative values ​​of the task deadline difference.

[0036] Specifically, the nonlinear weight fusion and spatial mapping process includes: uniformly encoding and normalizing the static spatiotemporal feature vector set and the dynamic operation feature vector set to form a standardized input traffic feature vector; weighting and summing the standardized input traffic feature vectors according to the corresponding weight coefficients and introducing a bias term to obtain the fusion intermediate priority representation; mapping the fusion intermediate priority representation into a nonlinear activation function to obtain a nonlinear output result representing the relationship between the priority judgment threshold and the traffic environment factor regulation; and generating a conflict judgment parameter set based on the nonlinear output result, wherein the conflict judgment parameter set specifically guides the setting of the passage order and the allocation of spatiotemporal resources at each conflict point in tunnel transportation scheduling.

[0037] Understandably, by introducing traffic benchmark data as constraints and references, vehicle physical information and operational environment information are uniformly mapped to the same computational space, and multi-source traffic features are standardized and nonlinearly fused and modeled. Specifically, a feature vector set is constructed based on key physical parameters such as slope and width, and quantitative indicators such as load and power. After unified encoding, normalization, and weighted summation, a bias term is introduced and spatial mapping is completed through a nonlinear activation function, thereby characterizing the complex nonlinear relationship between traffic factors and traffic boundaries, and outputting a set of judgment parameters including priority and adjustment factors. At the same time, the result is input into a knowledge graph that integrates a process pattern library, a binding relationship library, and a historical conflict database, and an interactive situation map is generated with the help of a 3D visualization engine, thus performing the visualization expression and practical application of the control results.

[0038] To enable those skilled in the art to better understand and implement this invention, the following is combined with... Figures 2-5 The tunnel traffic flow collaborative management platform specifically applies the tunnel traffic flow conflict management method based on dynamic priority in this application, further supplementing the specific implementation principle of the present invention.

[0039] like Figure 2 As shown, the tunnel traffic flow collaborative management and control platform may include a multi-source sensing and acquisition module, a conflict generation module, and a priority construction module.

[0040] The multi-source sensing acquisition module is used to collect local traffic information required for scheduling design. Local traffic information can be multi-source traffic sensing datasets already stored in the tunnel operation area, such as vehicle sensor sample lists, vehicle terminal batch numbers, operation coverage areas, communication data formats, and non-standard sensing hardware parameters. Local traffic information is stored in a vehicle database, or in certain metadata files or other traffic management systems. The vehicle sensor sample list refers to a list of all registered vehicle LiDAR and camera data in the tunnel operation area, ensuring the analysis team understands all legitimate entities in the system. Batch numbers record the network access time information of each vehicle terminal in the scheduling system. Operation coverage areas describe the vehicle's jurisdiction section, tunnel index, etc. Data formats specify the characteristics of the data, such as location coordinates, velocity vectors, hexadecimal sensing streams, or structured scheduling logs. Non-standard sensing hardware parameters record the non-standard instruction sets of the vehicle systems in the tunnel operation area; these may be customized to meet specific tunnel operation requirements.

[0041] For example, such as Figure 2 As shown, the multi-source sensing and acquisition module may include a data access framework, a data processing script, and a data desensitization module.

[0042] The data access framework provides a systematic approach to accessing, processing, and storing traffic data. It integrates multiple data sources, such as LiDAR databases, vehicle terminal interfaces, and process management systems, providing a unified access point for data collection. The framework is suitable for tools handling large-scale scheduling and verification flows and is configured for stable data transmission. Developers can add or modify data access and preprocessing components as needed to adapt to different traffic information collection scenarios. Furthermore, the framework incorporates fault tolerance mechanisms to ensure recovery in the event of failures during data access, thereby guaranteeing data integrity. Before data enters the storage system, the framework performs preprocessing operations such as one-way hashing, trajectory extraction, and coordinate alignment. Additionally, the framework includes functions to monitor the data access status and log data for troubleshooting and analysis.

[0043] In this embodiment of the invention, the tunnel traffic flow collaborative management and control platform uses a custom data access framework to instruct each sub-module in the scheduling system to provide the data sources it needs to collect, such as local traffic information, and then uniformly schedule and execute them.

[0044] Data processing scripts are custom-written to meet specific scheduling and analysis needs, enabling personalized processing of specific traffic data sources. They can be configured to run automatically, thus minimizing manual intervention. Data processing scripts can include complex logic to handle various conflict scenarios, including spacing checks and topology correlations. They can also parse and convert data in different formats, such as HEX and Protobuf. Furthermore, they can include error handling mechanisms, such as retry logic, to address temporary network failures that may occur during data access. Data processing scripts can also be used to test and validate traffic analysis processes, ensuring that the collected data is both accurate and effective.

[0045] In this embodiment of the invention, the data access framework provides a stable and scalable platform to support complex traffic data access tasks, while the data processing scripts offer flexibility and customization capabilities to adapt to specific traffic data needs. The combined use of these two components completes the collection of traffic information at various locations.

[0046] Upon receiving traffic information from a local station, the data anonymization module performs anonymization processing. For example, the module filters out sensitive information from the traffic information to prevent the leakage of sensitive vehicle operation data. This sensitive information may include specific vehicle identification codes, precise mileage counts, and the underlying communication protocol addresses of core equipment.

[0047] The conflict generation module is used to generate a set of conflict determination parameters for a region based on local traffic information, conflict identification logic templates, and traffic reference data.

[0048] For example, such as Figure 2 As shown, the conflict generation module may include a template management module, a baseline data management module, a logical orchestration module, and a plan generation module.

[0049] The template management module manages conflict identification logic templates and provides them to the logic orchestration module. A conflict identification logic template is a standardized document designed to guide the analysis or organization of each step in traffic risk management. It helps ensure the transparency, consistency, and standardization of the identification process, thereby determining the impact of changes on organization and construction operations. Conflict identification logic templates may include dispatch request forms, traffic management processes, risk management system templates, analysis project management table templates, priority change process templates, dispatch management plan templates, route map templates, right-of-way request templates, risk control process templates, etc. By using these templates, the conflict generation module can more effectively manage and control changes in the project, thereby achieving project objectives.

[0050] In this embodiment of the invention, the conflict identification logic template is designed by technical experts based on traffic benchmark data and safety analysis experience. The conflict identification logic template supports determining whether to expand or generate a specific conflict unit based on local traffic information. A conflict unit represents an analysis step in the dispatch system during delivery. For example, the conflict identification logic template can support determining whether to display the corresponding conflict unit based on the vehicle load field and traffic benchmark data in the local traffic information. As another example, the conflict identification logic template can support generating multiple conflict units for a corresponding sub-module covering multiple construction sections based on local traffic information.

[0051] The baseline data management module manages pre-set traffic baseline data, such as dynamically adjusting it adaptively based on changes in tunnel environmental quality, and providing traffic baseline data to the planning generation module. In traffic risk management and systems engineering, traffic baseline data refers to identifying the stable state of a project at a specific point in time. This state encompasses multiple aspects, including traffic documentation, design, parameters, configurations, and experimental data. Traffic baselines are a crucial tool for ensuring project consistency and stability; they provide a reference point and baseline, helping to control project changes and risks.

[0052] In this embodiment of the invention, traffic baseline data is used to instruct conflict units in the control process to schedule tasks. Traffic baseline data may include scheduling iteration paths (describing which source priority can be escalated to which target priority), compatibility baselines between factors, estimated computation time for each submodule, performance baselines of onboard equipment, dependencies between factors, change order between multiple logical modules or conflict identification steps, and process data. Traffic baseline data is typically carried out through unstructured construction documents and cannot be consumed by code.

[0053] Traffic baseline data may also include information such as planned computational resource windows and scheduling computation concurrency. The planned computational resource window typically refers to a computational window reserved for change operations in traffic scheduling management and system maintenance. This time window is used to plan and execute system changes to determine their impact on construction. For example, in the maintenance of traffic information processing or analysis, the computational resource window might be during off-peak hours, allowing for priority upgrades or configuration changes without affecting normal output. Planning the computational resource window needs to consider factors such as computational load, system stability, and the workload of the analysis team.

[0054] Concurrency in scheduling computation refers to the number of conflicting verification requests a system can handle simultaneously, reflecting the system's load capacity and performance. In multi-threaded scheduling verification and system design, high concurrency means that more processing threads or processes can be managed and scheduled for execution simultaneously by system resources (such as the CPU and GPU). The level of concurrency directly affects the system's performance and responsiveness. For example, in performance testing, simulating high-concurrency scenarios can evaluate the system's performance under massive scheduling requests, including response time and throughput. Concurrency optimization can be achieved by increasing hardware resources, optimizing code logic, and using concurrency control mechanisms.

[0055] After receiving system operation preference indicators input by the user and local traffic information collected by the multi-source sensing and acquisition module, the logic orchestration module can input these indicators into the conflict identification logic template, render the template, and determine which conflict units are expanded and which are generated. Based on the expanded spatial units and the generated conflict units, the logic orchestration module can construct the regional control process.

[0056] The control process can exist in the form of a "scheduling analysis package," allowing it to be imported into the scheduling execution module of the tunnel operation area for analysis and implementation. In some tunnel scenarios, the network environment inside the tunnel operation area and the service that generates the decision parameter set are not in the same network environment, so it is necessary to use an offline package for transmission.

[0057] If the conflict generation module receives system operation preference indicators input by the user, it can pass these indicators to the logic orchestration module. The logic orchestration module can then input the system operation preference indicators and local traffic information into the conflict identification logic template to obtain the regional control process. Optionally, after orchestrating the control process, the logic orchestration module can integrate the physical constraints corresponding to the system operation preference indicators into the control process, thus obtaining a merged control process.

[0058] The planning generation module utilizes traffic baseline data to automatically schedule multiple conflict units (i.e., analysis steps) in the regional control process, generating a set of decision parameters. The specific implementation process is as follows: The planning generation module can intersect the element list in the traffic baseline data with the element list of the region in the control process to obtain the element list involved in this analysis. The planning generation module can compare the source priority of the iteration path in the traffic baseline data with the actual priority of the region in the control process to determine whether the dispatch system deployed in the region needs to be upgraded or changed. Based on the compatibility baseline and dependencies between factors in the traffic baseline data, the planning generation module can generate the change order of each factor. Based on the performance baseline of on-board equipment in the traffic baseline data, the planning generation module can generate parallel or serial rules between steps. Based on the estimated computation time and computational resource window planning value of each submodule in the traffic baseline data, the planning generation module can calculate the number of computation windows required and the analysis steps to be executed in each computation window.

[0059] The plan generation module can summarize the information involved in this analysis, such as the list of elements, whether they have been upgraded or changed, the order of changes to each factor, the parallel or serial rules between steps, the number of calculation windows required, and the analysis steps to be executed in each calculation window, into a set of decision parameters in the form of a topology graph based on the scheduling calculation concurrency. By utilizing the logical association and time sorting characteristics of the topology graph, it can effectively represent and process complex dependencies.

[0060] In this invention, each node in the topology graph represents a conflict unit. Each edge of the topology graph represents the sequence of steps. The analysis steps to be performed in each calculation window are marked with wireframes.

[0061] The scheduling execution module is used to guide the scheduling system deployed in the region to authorize or intercept based on the set of judgment parameters.

[0062] The scheduling execution module can pre-build a structured traffic data model, which defines the various metadata required for the delivery of complex scheduling systems. This metadata is used by the tunnel traffic flow collaborative management and control platform to generate wizard-style management processes, and can also be used to automatically create documentation for technical personnel. In this way, a consistent source of data is achieved across multiple systems.

[0063] In this invention, the scheduling execution module can convert the set of decision parameters into a wizard-style conflict identification flowchart, enabling implementers to complete the automated upgrade or modification of a complex scheduling system at low cost based on the guidance on the identification flowchart.

[0064] It should be understood that the functional modules and components involved in the aforementioned tunnel traffic flow collaborative management and control platform can all be implemented through software or hardware, depending on the actual situation, and are not limited here. Furthermore, the functional modules and components involved in the aforementioned tunnel traffic flow collaborative management and control platform can be arranged individually or integrated, and are not limited here.

[0065] The above is an introduction to the tunnel traffic flow collaborative management and control platform provided by the embodiments of the present invention. It is understood that the aforementioned tunnel traffic flow collaborative management and control platform can be configured on a regional scheduling and management platform, for example, deployed on at least one instance such as a computing node or container, so that the regional scheduling and management platform can provide conflict identification and design services.

[0066] To enable those skilled in the art to better understand and implement this invention, the following is combined with... Figures 2 to 5 The specific implementation principle of the tunnel traffic flow conflict control method based on dynamic priority provided by the present invention is further supplemented.

[0067] During the data access and preprocessing stage, the tunnel traffic flow collaborative management and control platform accesses vehicle position coordinate vectors, speed vectors, road network topology constraint parameters, task urgency weights, and tunnel visibility coefficients from the target tunnel operation area through the data access framework in the multi-source sensing and acquisition module. Specifically, the data processing script performs one-way hashing, trajectory extraction, and format alignment operations on the acquired heterogeneous traffic data, unifying traffic information from different sources to a preset analysis scale. Subsequently, the data anonymization module performs masking processing on sensitive fields involving vehicle identification codes, precise mileage counts, and device underlying communication protocol addresses, generating an anonymized multi-source traffic sensing dataset.

[0068] In the initial conflict generation phase, the logic orchestration module within the conflict generation module receives user-inputted system operation preference indicators. These indicators specifically encompass preset construction schedule constraints, energy consumption control levels, safety distance guarantee weights, and equipment access trust coefficients. The logic orchestration module inputs these system operation preference indicators along with the anonymized multi-source traffic perception dataset into the feature input layer of the conflict identification logic template. The conflict logic operators integrated within the conflict identification logic template perform filtering and reorganization actions on conflict units, generating resolution paths that conform to specific operational preferences. The spatiotemporal trajectory prediction algorithm generates a local trajectory uncertainty index based on onboard sensors and, combined with mechanical performance parameters, braking distance constants, and road friction coefficients, dynamically adjusts the vehicle safe following threshold and the time window coverage of conflict warnings, thereby generating incremental conflict units from the initial conflict candidate set.

[0069] In the feature extraction and nonlinear modeling stages, the plan generation module splits the initial conflict candidate set into a static spatiotemporal dimension and a dynamic operation dimension. A pre-trained graph neural network model performs multi-scale feature extraction on the static spatiotemporal dimension, generating a static spatiotemporal feature vector set containing feature extraction values ​​for road network node connectivity, slope resistance coefficient, and tunnel cross-section width. Simultaneously, the graph neural network model performs feature extraction on the dynamic operation dimension, generating a dynamic operation feature vector set containing quantized values ​​for load change rate, remaining fuel consumption status, and task deadline difference. During the training of the graph neural network model, a multi-task learning framework synchronously executes the main task of conflict probability prediction and auxiliary tasks such as road segment capacity regression, driving trajectory classification, and environmental perception confidence detection, with each task sharing the underlying spatial feature encoder.

[0070] In the priority optimization and parameter generation phase, the tunnel traffic flow collaborative management platform performs nonlinear weighted fusion of static spatiotemporal feature vector sets and dynamic operational feature vector sets based on traffic benchmark data. Specifically, the system performs unified encoding and normalization on the two types of feature vectors to form standardized input traffic feature vectors, and performs weighted summation according to the corresponding weight coefficients. After introducing a bias term, a fused intermediate priority representation is obtained. A nonlinear activation function performs mapping processing on the fused intermediate priority representation to obtain a nonlinear output result representing the relationship between the priority determination threshold and traffic environment factor regulation. Based on this output result, the system generates a conflict determination parameter set, including vehicle passage priority correction coefficient, road segment congestion threshold, intersection yielding duration parameter, speed guidance adjustment factor, and scheduling incremental cost control variable.

[0071] During the resource scheduling and control execution phase, the plan generation module calculates the number of parallel conflict processing units required for the initial conflict candidate set based on the computational resource window planning value, and determines the number of conflict unit processing units to be executed in parallel within the same computational window based on the scheduling computation concurrency. The tunnel traffic flow collaborative control platform performs priority topology optimization, inputting the optimized topology graph structure into the tunnel traffic control knowledge graph. When the real-time distance between oncoming vehicles is detected to be lower than a preset distance threshold and the occupancy status of the passing tunnel in the narrow section is a preset free marker, the topology graph structure automatically triggers the corresponding vehicle's passage priority adjacency weight adjustment action.

[0072] In the visualization and feedback phase, the 3D visualization engine receives the optimized priority control topology map and, combined with the construction procedure logic pattern library, vehicle-task binding relationship library, historical conflict handling database, and abnormal congestion response model library from the tunnel traffic management knowledge graph, generates an interactive traffic flow operation status map. The system uses a historical tunnel transportation sample set to perform iterative calibration on the traffic benchmark data, performs consistency analysis between the optimized priority control topology map and manually reviewed scheduling data, and calculates the scheduling failure ratio and average delay deviation. These error data are fed back into the loss function of the graph neural network model, and the system automatically updates the model's parameters.

[0073] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further supplemented below with a specific application scenario.

[0074] When carrying out material transportation or spoil removal tasks within the jurisdiction of the tunnel construction section, the multi-source sensing and acquisition module first establishes a communication link with the mobile transport vehicles inside the tunnel and the roadside sensing units. The module then uses vehicle-mounted LiDAR and a positioning chip to extract a multi-source traffic sensing dataset in real time. This dataset includes vehicle position coordinate vectors, speed vectors, road network topology constraints, task weights, and tunnel visibility parameters. For the specific vehicle identification codes and underlying communication addresses collected, the module performs one-way hash masking processing, transforming the sensitive raw information into a fixed-length irreversible feature summary, generating a de-sensitized multi-source traffic sensing dataset.

[0075] During the conflict credential construction phase, the conflict generation module acquires system operation preference indicators input by the user through the interactive scenario interface. These indicators include current construction schedule constraints, energy consumption control levels, and safety distance guarantee weights. The conflict generation module inputs these operation preference indicators along with the anonymized feature dataset into the feature input layer of the conflict identification logic template. Utilizing the spatiotemporal union operator and risk mask operator integrated within the template, the conflict generation module performs logical rule expansion on the smallest conflict analysis unit in the traffic conflict identification process. Simultaneously, the spatiotemporal trajectory prediction algorithm calls upon vehicle-mounted sensor data, integrating mechanical braking performance and road surface friction coefficient to generate a local trajectory uncertainty index. Based on this index, the conflict generation module dynamically generates incremental conflict units and integrates them with existing conflict units obtained through mapping expansion, forming the initial conflict candidate set for the target tunnel operation area.

[0076] In the topology optimization and feature extraction phase, the priority construction module splits the initial conflict candidate set into a static spatiotemporal dimension and a dynamic task dimension. A pre-trained graph neural network model performs multi-scale feature extraction on these two dimensions, generating a static spatiotemporal feature vector set containing node connectivity and cross-sectional width, and a dynamic task feature vector set containing load change rate and task deadline difference. Based on the computational resource window planning value in the traffic baseline data, the priority construction module calculates the number of parallel conflict processing units required for the candidate set. Subsequently, the system determines the scale of conflict units to be executed in parallel within the same computational window based on the scheduling computation concurrency, and establishes the logical coupling strength and influence propagation weights between conflict units in the form of a topology graph, generating an optimized priority control topology graph.

[0077] During the scheduling execution and risk mitigation phases, the scheduling execution module inputs the optimized priority control topology map into the tunnel traffic management knowledge graph. This knowledge graph integrates a process logic pattern library, a vehicle-task binding relationship library, and a historical conflict handling database. The scheduling execution module performs unified encoding and normalization on the static spatiotemporal feature vector set and the dynamic operation feature vector set to form a standardized input traffic feature vector. The system performs weighted summation of the vectors according to weight coefficients and introduces a bias term. Through nonlinear activation function mapping, it outputs a conflict determination parameter set, which includes priority correction coefficients, congestion thresholds, yielding duration parameters, and speed guidance adjustment factors.

[0078] During dynamic execution in actual operating environments, if the narrow road collision avoidance unit detects that the real-time distance between oncoming vehicles is lower than a preset distance threshold, and the adjacent passing tunnel is marked as empty, the optimization priority control topology graph structure automatically adjusts the adjacent weights of the corresponding vehicles' passage priorities. The scheduling execution module converts the influence propagation weights in the topology graph into control list entries for the scheduling system based on the conflict determination parameter set. When the real-time determination value requested by a vehicle exceeds a preset safety risk threshold, the scheduling execution module activates the corresponding right-of-way allocation, speed limit guidance, or forced yielding interface. The 3D visualization engine simultaneously receives topology graph data and knowledge graph feedback, generating an interactive traffic flow status map to accurately define transportation permissions.

[0079] Furthermore, the system uses a historical tunnel transportation sample set to iteratively calibrate traffic benchmark data. The calibration procedure performs consistency analysis between the results of optimizing the priority control topology map and manually reviewed scheduling data, calculating the scheduling failure ratio and average delay deviation. These error data are fed back into the loss function of the graph neural network model, enabling the system to automatically update the model parameters and complete the closed-loop optimization of traffic flow conflict identification accuracy.

[0080] It is understood that the tunnel traffic flow conflict control method and system based on dynamic priority in the above embodiments of the present invention have the same beneficial effects, and will not be described in detail here.

[0081] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0082] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0083] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0084] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for managing tunnel traffic flow conflicts based on dynamic priority, characterized in that, The methods include: Obtain a multi-source traffic perception dataset for the target tunnel operation area; Based on the dataset and the conflict identification logic template, an initial conflict candidate set is generated. The conflict unit is defined as the smallest conflict analysis unit in traffic flow operation. The initial conflict candidate set includes existing conflict units and incremental conflict units. Based on the initial conflict candidate set and traffic baseline data, an optimized priority control topology map is generated, including: using a graph neural network model to extract features from the initial conflict candidate set to obtain a static spatiotemporal feature vector set and a dynamic operation feature vector set; Based on traffic baseline data, the feature vector set is weighted and spatially mapped to output a conflict determination parameter set including vehicle traffic priority correction coefficient, road segment congestion threshold, yielding duration parameter, and speed guidance adjustment factor.

2. The tunnel traffic flow conflict control method based on dynamic priority as described in claim 1, characterized in that, When acquiring the multi-source traffic perception dataset for the target tunnel operation area, the following is also included: A one-way hash masking process is performed on the specific vehicle identification codes, accurate mileage counts, and underlying communication protocol addresses of the devices involved in the multi-source traffic perception dataset to generate a de-identified multi-source traffic perception dataset.

3. The tunnel traffic flow conflict control method based on dynamic priority as described in claim 1, characterized in that, When generating the initial conflict candidate set for the target tunnel operation area, the following are included: The system operation preference indicators input by the user are obtained. These indicators include preset construction schedule constraints, energy consumption control levels, safety distance guarantee weights, and equipment access trust coefficients. The system operation preference index and the multi-source traffic perception dataset are input into the feature input layer of the conflict identification logic template to generate an initial conflict candidate set that conforms to the system operation preference index.

4. The tunnel traffic flow conflict management method based on dynamic priority according to claim 3, characterized in that, The conflict identification logic template integrates conflict logic operators, including spatiotemporal union operator, influence range expansion operator, risk mask operator, and multidimensional weight superposition operator; Generating the initial conflict candidate set includes: screening and reorganizing conflict units based on conflict logic operators, generating specific conflict resolution paths that meet the system operation preference indicators, and incorporating the physical constraints corresponding to the system operation preference indicators into the initial conflict candidate set.

5. The tunnel traffic flow conflict management method based on dynamic priority according to claim 1, characterized in that, The conflict unit includes: Branch intersection conflict unit, narrow road collision conflict unit, local flow overload unit, sight-limited avoidance unit, and multi-process cross operation unit; When performing the identification of narrow road collision units, if the real-time distance between oncoming vehicles is detected to be lower than a preset distance threshold and the occupancy status of the passing tunnel in the narrow road section is a preset free status, the priority adjustment topology map will automatically adjust the adjacent weight of the corresponding vehicle's passage priority.

6. The tunnel traffic flow conflict control method based on dynamic priority as described in claim 1, characterized in that: The static spatiotemporal feature vector set includes the road network node connectivity, slope resistance coefficient, and tunnel cross-sectional width feature extraction values. The dynamic operation feature vector set includes quantified values ​​of vehicle load change rate, remaining fuel consumption or battery status, and the difference between task deadlines.

7. The dynamic priority based tunnel traffic flow conflict management method of claim 6, wherein, When performing nonlinear weight fusion and spatial mapping, the following are included: The static spatiotemporal feature vector set and the dynamic operation feature vector set are uniformly encoded and normalized to form a standardized input traffic feature vector; The standardized input traffic feature vectors are weighted and summed according to their corresponding weight coefficients, and a bias term is introduced to obtain the fused intermediate priority representation. By mapping the fused intermediate priority representation into the nonlinear activation function, a nonlinear output result representing the relationship between the priority determination threshold and the regulation of traffic environmental factors is obtained. A set of conflict determination parameters is generated based on the nonlinear output results, which is used to guide the setting of the passage order and the allocation of spatiotemporal resources at each conflict point in tunnel transportation scheduling. 8.The dynamic priority based tunnel traffic flow conflict management method of claim 1, wherein, When generating the optimization priority control topology graph, the following is also included: Based on the computational resource window planning value in the traffic baseline data, calculate the number of parallel conflict processing units required for the initial conflict candidate set; Based on the scheduling concurrency in traffic baseline data, the number of conflict units to be processed in parallel within the same computation window is determined.

9. The dynamic priority based tunnel traffic flow conflict management method of claim 1, wherein, Also includes: The optimization priority control topology map is input into the tunnel traffic management knowledge graph, and an interactive traffic flow operation status map is generated based on the 3D visualization engine. The tunnel traffic management knowledge graph includes a construction procedure logic pattern library, a vehicle-task binding relationship library, a historical conflict handling database, and an abnormal congestion response model library. Historical tunnel transportation sample sets are used to iteratively calibrate traffic benchmark data. The iterative calibration includes: performing consistency analysis between the optimized priority control topology map and manually reviewed scheduling data, calculating the scheduling failure ratio and average delay deviation, and feeding them back into the loss function of the graph neural network model for parameter updates.

10. A dynamic priority based tunnel traffic flow conflict management system, applied to the dynamic priority based tunnel traffic flow conflict management method of any one of claims 1-9, characterized in that, include: The multi-source sensing acquisition module is used to acquire multi-source traffic sensing datasets for the target tunnel operation area. The datasets include real-time vehicle position coordinate vectors, vehicle speed vectors, road network topology constraint parameters, task urgency weights, and tunnel environment visibility coefficients. The conflict generation module is used to generate an initial set of conflict candidates for the target tunnel operation area based on the multi-source traffic perception dataset and the preset conflict identification logic template. The priority construction module is used to generate an optimized priority control topology map of the target tunnel operation area based on the initial conflict candidate set and traffic benchmark data; The scheduling execution module is used to issue scheduling instructions to tunnel transport vehicles based on the conflict determination parameter set output by the optimization priority control topology graph. The scheduling execution module converts the influence transmission weights in the optimization priority control topology graph into control list entries of the scheduling system. When the real-time conflict determination value requested by a vehicle exceeds the preset safety risk threshold, the corresponding right-of-way allocation, speed limit guidance, or forced stopping interface is activated.

Images