Intelligent people and vehicle access control method and system based on optimized scheduling

Abstract

This invention relates to the field of access control technology, specifically to an intelligent pedestrian and vehicle access control method and system based on optimized scheduling. The method includes: acquiring the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density, and equipment response delay between the management terminal and each access gate; performing topological mapping on the status data of the area to be processed based on preset multi-dimensional constraints to obtain the status data of the nodes to be scheduled and a dynamic constraint diagram with multiple constraint dimensions; the preset multi-dimensional constraints include historical passage efficiency profiles of different types of passage subjects and passage priority rules. This invention optimizes pedestrian and vehicle flow by dynamically adjusting the opening duration of access gates or the pointing angle of guiding devices based on real-time traffic flow and equipment response delay, thereby reducing congestion. It can accurately adjust the passage efficiency during peak traffic periods based on specific traffic flow and status data.

Description

Technical Field

[0001] This invention relates to the field of access control technology, specifically to an intelligent human and vehicle access control method and system based on optimized scheduling. Background Technology

[0002] Currently, traditional methods often rely on fixed schedules or static rules to control gates and guidance equipment, which cannot be dynamically adjusted according to real-time traffic flow, equipment response delays, or weather changes, making them prone to congestion during peak periods or in abnormal situations. Moreover, traditional methods usually only focus on the traffic flow at a single entrance or exit or in a local area, lacking global optimization considerations, which can easily lead to excessive congestion in certain areas and low overall traffic efficiency.

[0003] Furthermore, the lack of a flexible response mechanism for emergencies and the resulting delays may lead to prolonged congestion. Moreover, traditional methods lack the ability to learn or gradually optimize based on historical data, making it difficult to improve efficiency and stability in the long run. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent pedestrian and vehicle access control method based on optimized scheduling, comprising: Obtain the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each of the access gates. Based on preset multi-dimensional constraints, the state data of the area to be processed is topologically mapped to obtain the state data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. With minimizing the global congestion cost as the optimization objective, dynamic constraint graphs of different constraint dimensions are correlated and fused to obtain a multi-dimensional fused correlation graph, and the corresponding time-varying directed graph structure is obtained based on the multi-dimensional fused correlation graph. Based on the time-varying directed graph structure and the state data of the nodes to be scheduled corresponding to the preset control response time scale, a scheduling decision is made based on the preset access control model to obtain the access control instructions for people and vehicles in the current scheduling period; the duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the state data of the area to be processed; the access control instructions for people and vehicles are used to adjust the opening duration of at least one of the access gates or the pointing angle of the guiding device.

[0005] Preferably, the status data of the region to be processed in the current scheduling period is obtained, including: Real-time traffic flow and equipment response delay of each access gate are collected according to a preset monitoring time interval; Retrieve historical traffic efficiency profiles for each person or vehicle passing through; these historical traffic efficiency profiles are used to characterize the dynamic correlation between the behavioral norms and traffic speed of the passing subjects during historical traffic processes. Obtain weather change parameters that are individually associated with each region within a preset future time period; Based on the weather change parameters associated with each region, the real-time traffic flow, the device response latency, and the historical traffic efficiency profile, the status data of the region to be processed is generated.

[0006] Preferred methods for constructing multi-dimensional constraints include: The status data of the area to be processed is statistically analyzed according to the main traffic entities to obtain the time series of motor vehicle and pedestrian traffic and the equipment delay time series. The traffic flow time series of all traffic entities are merged according to spatial location to obtain the corresponding regional occupancy time series; Based on the area occupancy time series and the device delay time series, the corresponding traffic demand power spectrum is obtained; Based on the traffic demand power spectrum and the historical traffic efficiency profile, traffic priority rules with different constraint dimensions are obtained.

[0007] Preferably, the state data of the region to be processed is topologically mapped based on preset multi-dimensional constraints to obtain state data of the nodes to be scheduled and a dynamic constraint graph with multiple constraint dimensions, including: The state data of the region to be processed is processed into nodes based on preset multi-dimensional constraints to obtain node state data with multiple constraint dimensions. When the current scheduling period is the first scheduling period, the node state data of each constraint dimension are initialized with graph structure and the constraint matrix is ​​jointly optimized and solved to obtain the node state data to be scheduled and the dynamic constraint graph of the corresponding constraint dimension. When the current scheduling period is not the first scheduling period, the state data of the corresponding node is deduced based on the dynamic constraint graph of each constraint dimension of the previous scheduling period to obtain the state data of the node to be scheduled in the corresponding constraint dimension. Obtain the optimization constraint matrix corresponding to the state data of the node to be scheduled; The optimized constraint matrix and the dynamic constraint graph of the constraint dimension corresponding to the previous scheduling cycle are weighted and fused to obtain the dynamic constraint graph of the constraint dimension corresponding to the current scheduling cycle.

[0008] Preferably, the dynamic constraint graphs of different constraint dimensions are correlated and fused with the goal of minimizing the global congestion cost, resulting in a multi-dimensional fused correlation graph, including: Using the priority weights corresponding to the dynamic constraint graphs of different constraint dimensions as optimization variables, a priority weight optimization model is constructed based on the optimization objective of minimizing the global congestion cost. The priority weight optimization model is then solved according to the state data of the nodes to be scheduled in different constraint dimensions and the corresponding dynamic constraint graphs to obtain the priority weights in different constraint dimensions. The corresponding dynamic constraint graphs are associated and fused based on the priority weights of different constraint dimensions to obtain the multidimensional fused association graph.

[0009] Preferably, based on the multidimensional fusion correlation graph, the corresponding time-varying directed graph structure is obtained, including: Based on the multidimensional fusion association graph and the corresponding node degree distribution, the corresponding undirected weight graph is obtained; Based on the undirected weighted graph, the flow pressure ratio corresponding to each edge of each passage node is obtained, and based on the comparison result of the flow pressure ratio corresponding to each edge with the preset pressure threshold, the one-way passage relationship between the corresponding passage node and other nodes is obtained. Based on the one-way traffic relationship between each passing node and other nodes, the time-varying directed graph structure is generated based on the undirected weighted graph.

[0010] Preferably, the preset access control model includes a graph neural network, an attention mechanism layer, and a decision output network; Based on the time-varying directed graph structure and the state data of the nodes to be scheduled corresponding to the preset control response time scale, scheduling decisions are made based on the preset access control model to obtain the personnel and vehicle access control instructions for the current scheduling period, including: The state data of the node to be scheduled corresponding to the preset control response time scale and the time-varying directed graph structure are input into the graph neural network to extract topological features and obtain the spatiotemporal features of the passage topology. The passage topology spatiotemporal features are input into the attention mechanism layer to extract key nodes and obtain the corresponding key node passage features. The spatiotemporal characteristics of the traffic topology and the traffic characteristics of the key nodes are input into the decision output network to generate a scheduling strategy, thereby obtaining the personnel and vehicle access control instructions for the current scheduling period.

[0011] Preferably, the method further includes: When the status data of the area to be processed triggers a preset congestion judgment rule, abnormal traffic situation data is determined; the abnormal traffic situation data includes gate congestion data and channel blockage data; Based on the abnormal traffic situation data and the historical traffic efficiency profile, the target gate or target channel that caused the abnormality is determined. Based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the personnel and vehicle access control commands.

[0012] Preferably, based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the pedestrian and vehicle access control commands, including: When the abnormal passage situation data only includes the data of people lingering in front of the gate, the passage capacity weight of the node corresponding to the target gate is increased according to the passage efficiency status and the range of the lingering area of ​​the target gate. When the abnormal traffic situation data only includes the channel blockage data, the guidance weight of the downstream node of the target channel is adjusted according to the blockage change rate and the diversion path. When the abnormal traffic situation data includes the gate congestion data and the channel blockage data, the scheduling adjustment priority is calculated based on the congestion occurrence ratio and the congestion change rate, and the gate node weight and the guide node weight are adjusted in sequence according to the priority.

[0013] The intelligent pedestrian and vehicle access control system based on optimized scheduling is applicable to the aforementioned intelligent pedestrian and vehicle access control method based on optimized scheduling, including: The data acquisition unit is used to acquire the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each of the access gates. The topology mapping unit is used to perform topology mapping on the state data of the area to be processed based on preset multi-dimensional constraints to obtain the state data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. The association and fusion unit is used to perform association and fusion on dynamic constraint graphs with different constraint dimensions with the optimization objective of minimizing global congestion cost, to obtain a multi-dimensional fusion association graph, and to obtain the corresponding time-varying directed graph structure based on the multi-dimensional fusion association graph; The access control unit is used to make scheduling decisions based on the time-varying directed graph structure and the status data of the nodes to be scheduled corresponding to the preset control response time scale, and to obtain the access control instructions for people and vehicles in the current scheduling period based on the preset access control model. The duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the status data of the area to be processed. The access control instructions for people and vehicles are used to adjust the opening duration of at least one of the access gates or the pointing angle of the guiding device.

[0014] Compared with the prior art, the beneficial effects of the present invention are: This invention optimizes pedestrian and vehicle flow by dynamically adjusting the opening duration of entry and exit gates or the pointing angle of guiding devices based on real-time traffic flow and device response delay, thereby reducing congestion. It can precisely adjust the passage efficiency during peak traffic periods based on specific traffic flow and status data to ensure efficient passage. Furthermore, by utilizing multi-dimensional information such as historical passage efficiency profiles, real-time traffic flow, device response delay, and weather changes, it constructs dynamic constraint graphs and multi-dimensional fusion association graphs, enabling the system to perform real-time optimization based on external conditions and internal traffic status, thus improving the system's flexibility and adaptability. This invention achieves coordinated scheduling of the entire region or system by minimizing the global congestion cost based on the optimization objective of a multi-dimensional constraint graph. This optimization method can globally consider the state of different entrances and exits, avoiding excessive congestion in certain areas, thereby improving the overall system throughput. Moreover, by combining a time-varying directed graph structure with graph neural networks and attention mechanisms, the invention extracts the traffic characteristics of key nodes, further optimizing scheduling decisions and making control commands more precise. Especially in dynamic situations, it can adjust strategies in a timely manner, avoiding traffic problems caused by human or equipment response delays. This invention enables the system to adjust the scheduling weight of target gates or channels based on historical traffic efficiency profiles and real-time monitoring data when abnormal situations such as congestion or channel blockage are detected. This allows for timely optimization measures to alleviate abnormal situations and ensure that the passage of people and vehicles is not significantly affected. Furthermore, by gradually optimizing the dynamic constraint diagram and control commands, the system can continuously improve its scheduling strategy in multiple scheduling cycles, thereby maintaining high efficiency and stability during long-term operation. Attached Figure Description

[0015] Figure 1 This is a schematic flowchart of the overall method in one embodiment of the present invention; Figure 2 This is a schematic diagram of the overall system architecture in one embodiment of the present invention.

[0016] In the diagram: 1. Data acquisition unit; 2. Topology mapping unit; 3. Association and fusion unit; 4. Input / output control unit. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1, please refer to Figure 1This invention provides a technical solution: an intelligent pedestrian and vehicle access control method based on optimized scheduling, comprising: S1. Obtain the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each access gate. S2. Based on preset multi-dimensional constraints, perform topological mapping on the status data of the area to be processed to obtain the status data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. S3. With minimizing the global congestion cost as the optimization objective, the dynamic constraint graphs of different constraint dimensions are correlated and fused to obtain a multi-dimensional fused correlation graph. Based on the multi-dimensional fused correlation graph, the corresponding time-varying directed graph structure is obtained. S4. Based on the time-varying directed graph structure and the status data of the nodes to be scheduled corresponding to the preset control response time scale, make scheduling decisions based on the preset access control model to obtain the access control instructions for people and vehicles in the current scheduling cycle; the duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the status data of the area to be processed; the access control instructions for people and vehicles are used to adjust the opening duration of at least one access gate or the pointing angle of the guiding device.

[0019] It should be noted that within a specific time period (scheduling cycle), the relevant system will collect various status data within the area to be processed (such as a large shopping mall, airport, or parking lot); these data include: real-time pedestrian density: for example, how many people are entering or leaving at the entrance; real-time vehicle density: for example, how many cars are queuing to enter at the parking lot entrance; equipment response latency: refers to the communication latency between the management system and each gate, for example, the time required to transmit data; The system analyzes the collected data based on preset multi-dimensional constraints (such as historical traffic efficiency and traffic priority). For example, if we find that the peak hours for a certain entrance / exit are from 5 pm to 6 pm, when the flow of people and vehicles is usually large, the system will perform topological mapping on this information to form a dynamic constraint graph, which helps to understand which entrances / exits may experience greater congestion during peak hours. To reduce congestion, the system needs to correlate data from different dimensions. For example, by combining vehicle and pedestrian density data, it can analyze which entrance / exit is most prone to congestion during a certain period of time. The result of this correlation and fusion is a multi-dimensional correlation graph that can more clearly show the relationship between various entrance / exit nodes and potential congestion risks. Based on the multidimensional fusion association graph, the system constructs a time-varying directed graph structure; this structure not only considers real-time data, but also future trends, such as predicting that the flow of people will increase at 5 pm, so as to make adjustments in advance; Access control decisions are made based on time-varying directed graphs and preset response times (e.g., control commands must be generated within 5 seconds). These decisions may include: extending the opening time of an entrance or exit to allow more people and vehicles to pass through smoothly; and adjusting the pointing angle of guidance equipment to help people find the best passage route more quickly. Specific examples: Suppose that during peak hours in a large shopping mall, the pedestrian density at entrance A suddenly increases, while the vehicle density at entrance B is also high. After monitoring this information, the system finds that there are 300 people queuing at entrance A and 50 vehicles waiting to enter at entrance B. Based on historical data, the throughput of entrance A is low during this time period, so it needs special attention. Through topology mapping, the system discovers that opening entrance C (an area with lower pedestrian traffic) at the same time can effectively divert some of the crowd. Ultimately, the system decides to extend the opening time of entrance A by 10 minutes and adjust the guidance equipment to point to entrance C, indicating to customers where to enter, in order to reduce congestion.

[0020] In an optional embodiment, obtaining the status data of the region to be processed in the current scheduling period includes: Real-time traffic flow and equipment response delay of each access gate are collected according to the preset monitoring time interval; Retrieve historical traffic efficiency profiles for each person or vehicle passing through; these historical traffic efficiency profiles are used to characterize the dynamic correlation between the behavioral norms and traffic speed of the passing subjects during historical traffic processes. Obtain weather change parameters that are individually associated with each region within a preset future time period; Based on weather change parameters, real-time traffic flow, equipment response latency, and historical traffic efficiency profiles associated with each region, status data for the regions to be processed is generated.

[0021] It should be noted that the throughput of each access gate (i.e., the number of people or vehicles passing through per unit time) is monitored in real time; this data is collected based on preset time intervals (e.g., every 5 minutes or every 30 seconds); in addition, the system also records the device response delay from the management system to the gate (i.e., the time difference from when the system sends a command to when the gate executes it); this information is crucial for subsequent scheduling decisions. Each person or vehicle has a historical traffic efficiency profile, which is constructed by recording their behavior (such as whether they follow traffic rules and pass through quickly) and passage rate (speed of passing through the gate) during past passages; historical data reflects the passage habits of each subject; for example, some vehicles may often have a slow passage speed due to improper operation by the driver (such as unfamiliarity with the route); while some common people may have a high passage efficiency. Weather conditions can affect the speed of pedestrian and vehicle traffic; for example, rainy or windy weather may cause people to move slowly or vehicles to be congested. The system will obtain weather change information for each area in the future period through weather forecast data, such as whether it will rain, wind speed, temperature, etc., and these factors will be incorporated into subsequent scheduling decisions. By combining all the collected data (such as weather changes, traffic flow, equipment latency, historical traffic efficiency, etc.), the status data of the areas to be processed is generated. This data will reflect the real-time and future traffic conditions of each area, providing a reference for subsequent scheduling decisions. For example, if the future forecast shows that it will rain in a certain area, the system will adjust the entrance and exit scheduling strategy of that area in advance to avoid congestion caused by the increase in people.

[0022] In an optional embodiment, the method for constructing preset multi-dimensional constraints includes: The status data of the area to be processed is classified and statistically analyzed according to the main traffic entities to obtain the time series of motor vehicle and pedestrian traffic and the time series of equipment delay. The traffic flow time series of all traffic entities are merged according to spatial location to obtain the corresponding regional occupancy time series; Based on the area occupancy time series and the equipment delay time series, the corresponding traffic demand power spectrum is obtained; Based on the traffic demand power spectrum and historical traffic efficiency profile, traffic priority rules with different constraint dimensions are obtained.

[0023] It should be noted that the status data of the area to be processed is classified and statistically analyzed according to different traffic subjects, such as motor vehicles and pedestrians. This data includes: Motor vehicle traffic time series: refers to the number of vehicles in a certain area or entrance / exit within a specific time period; for example, the change in vehicle traffic in a parking lot at different time periods; Pedestrian traffic time series: refers to the number of pedestrians passing through a certain area at different time periods; for example, pedestrian traffic data at a shopping mall entrance; In addition, the equipment latency time series records the equipment response time at each traffic point, which is crucial for determining traffic efficiency; equipment latency reflects the processing capacity of the equipment itself and its performance during peak hours. Based on the traffic flow data of the main traffic entities (such as vehicles and pedestrians) in different areas, their time series are merged according to spatial location; the occupancy time series of each area is obtained by summing the traffic flow data of each area; the goal of this step is to understand the traffic demand of a certain area (e.g., a certain entrance of a shopping mall or part of a parking lot) in different time periods; the merged area occupancy time series will show the total occupancy of the area in different time periods, that is, the total amount occupied by vehicles and pedestrians at the same time; By combining the time series of regional occupancy with the time series of equipment delays, the corresponding traffic demand power spectrum can be calculated. The power spectrum is a spectrum describing the fluctuation of system demand, showing the changing trends of traffic and delay over different time periods. The purpose of this step is to understand the peak load and demand of the system at certain points in time. For example, during the lunch rush in a shopping mall, equipment response delay and traffic flow will reach their peak, thus affecting overall traffic efficiency. The analysis of this power spectrum helps to identify high-demand periods that require special attention. Traffic priority rules are formulated based on the traffic demand power spectrum and historical traffic efficiency profiles (such as the historical performance of a particular traffic subject). Priority rules determine which traffic subjects should be given priority during a specific time period by comprehensively considering different constraints (such as traffic subject type, historical performance, traffic density, etc.). For example, it may show that some traffic subjects have performed well and have fast traffic speeds during past peak hours, so these subjects should be given priority in the current time period. If the demand in a certain area fluctuates significantly, priority rules may require priority to be given to areas with high traffic volume and high equipment latency in order to alleviate congestion.

[0024] In an optional embodiment, topological mapping is performed on the state data of the region to be processed based on preset multi-dimensional constraints to obtain state data of the nodes to be scheduled and a dynamic constraint graph with multiple constraint dimensions, including: Based on preset multi-dimensional constraints, the state data of the region to be processed is processed into nodes to obtain node state data with multiple constraint dimensions. When the current scheduling period is the first scheduling period, the graph structure initialization and constraint matrix joint optimization solution are performed on the node state data of each constraint dimension to obtain the node state data to be scheduled and the dynamic constraint graph of the corresponding constraint dimension. When the current scheduling period is not the first scheduling period, the state data of the corresponding node is deduced based on the dynamic constraint graph of each constraint dimension of the previous scheduling period to obtain the state data of the node to be scheduled in the corresponding constraint dimension. Obtain the optimization constraint matrix corresponding to the status data of the node to be scheduled; The dynamic constraint graph of the constraint dimension corresponding to the previous scheduling cycle is obtained by weighted fusion of the optimized constraint matrix and the dynamic constraint graph of the constraint dimension corresponding to the current scheduling cycle.

[0025] It should be noted that, for the state data of the area to be processed, based on the preset multi-dimensional constraints, this data is processed into nodes; the purpose of this process is to transform complex state data into node state data that is easy to manage and analyze, so as to facilitate subsequent scheduling and optimization. If the current scheduling cycle is the first cycle, the system will initialize the graph structure of the node state data for each constraint dimension. This means creating a preliminary graph model for each dimension to represent the relationships and constraints between nodes. At the same time, joint optimization of the constraint matrix will be performed. The goal of this step is to obtain more reasonable node state data and dynamic constraint graphs to be scheduled by adjusting and optimizing these constraints, thereby providing support for the upcoming scheduling decision. In non-first scheduling cycles, the system does not reinitialize the graph structure. Instead, it uses the dynamic constraint graph of the previous scheduling cycle to perform state deduction on the node state data of the current cycle. This means that the system will use the data and graph structure of previous cycles to predict and update the node state of the current cycle, ensuring the continuity and timeliness of information. At this stage, the system extracts optimization constraint matrices related to the state data of the nodes to be scheduled; these matrices contain various restrictions and requirements that the system needs to follow during the scheduling process. The current cycle's optimization constraint matrix is ​​weighted and fused with the dynamic constraint graph corresponding to the previous scheduling cycle. This means combining the information from both to form a new dynamic constraint graph for use in the current scheduling decision. This process ensures that information from different sources can be effectively integrated through weighting, thereby improving the accuracy and effectiveness of scheduling.

[0026] In an optional embodiment, dynamic constraint graphs of different constraint dimensions are correlated and fused with the goal of minimizing the global congestion cost to obtain a multi-dimensional fused correlation graph, including: Using the priority weights corresponding to the dynamic constraint graphs of different constraint dimensions as optimization variables, a priority weight optimization model is constructed based on the optimization objective of minimizing the global congestion cost. The priority weight optimization model is then solved according to the state data of the nodes to be scheduled in different constraint dimensions and the corresponding dynamic constraint graphs to obtain the priority weights in different constraint dimensions. Based on the priority weights of different constraint dimensions, the corresponding dynamic constraint graphs are associated and fused to obtain a multi-dimensional fused association graph.

[0027] It should be noted that the priority weights of the dynamic constraint graph corresponding to each constraint dimension should be clearly defined; different constraint dimensions have different importance, so the dynamic constraint graph of each dimension will also have different priority weights; based on the goal of minimizing the global congestion cost, an optimization model is constructed to determine these priority weights; congestion cost refers to the degree of tension or imbalance in the system's operating state during the scheduling process, and the goal is to reduce global congestion by reasonably adjusting priorities, thereby improving scheduling efficiency; Based on the state data of the nodes to be scheduled for each constraint dimension and the corresponding dynamic constraint graph, the priority weight optimization model is solved. This solution process calculates the optimal priority weight for each constraint dimension based on the characteristics of the constraint graph, the node state, and the system requirements. The core purpose of the solution process is to ensure that the weights of each constraint dimension after optimization can reflect the global optimization goal of the system, that is, to reduce the overall congestion cost by adjusting the priority of different constraints. After obtaining the priority weights of each constraint dimension, the next step is to associate and fuse the dynamic constraint graphs of each dimension based on these weights; this means that the graph structures of different constraint dimensions will be adjusted and merged according to the changes in priority weights. By using weighted fusion, dynamic constraint graphs from different dimensions can more reasonably reflect the global state of the system after fusion, ensuring the coordination between the various constraint dimensions and optimizing the overall operation of the system. After priority weight adjustment and correlation fusion, a multi-dimensional fusion correlation graph is obtained. This graph is the result of the synthesis of dynamic constraint graphs of different constraint dimensions. It can effectively integrate the constraint information of each dimension and reflect the interrelationship between multiple dimensions. This multidimensional fusion graph provides comprehensive information support for subsequent scheduling decisions, enabling the system to make decisions and optimize more efficiently when facing multiple constraints, thereby reducing the global congestion cost and improving the overall efficiency of the system.

[0028] In an optional embodiment, a corresponding time-varying directed graph structure is obtained based on the multidimensional fusion correlation graph, including: Based on the multidimensional fusion association graph and the corresponding node degree distribution, the corresponding undirected weight graph is obtained; Based on the undirected weighted graph, the traffic pressure ratio corresponding to each edge of each passage node is obtained, and the one-way passage relationship between the corresponding passage node and other nodes is obtained based on the comparison result of the traffic pressure ratio corresponding to each edge and the preset pressure threshold. Based on the one-way traffic relationship between each passing node and other nodes, a time-varying directed graph structure is generated from an undirected weighted graph.

[0029] It should be noted that, based on the multidimensional fusion association graph and the degree distribution of nodes, an undirected weighted graph is first generated. In this step, the relationships and weights between nodes are obtained by fusing information from multiple constraint dimensions. The degree distribution of nodes refers to the connection between each node and other nodes. The weighted graph reflects the strength of the connection or the closeness of the relationship between different nodes. An undirected graph means that the edges between nodes have no directionality, that is, the two connected nodes do not indicate a clear flow direction in the graph, and the edges are bidirectional. Based on the undirected weighted graph, the next step is to calculate the flow pressure percentage for each edge of each passable node. The flow pressure percentage refers to the proportion of flow through a certain edge to the edge's carrying capacity, reflecting the load situation of that edge. The flow pressure percentage of each edge is compared with the preset pressure threshold. The purpose of this step is to determine whether the flow pressure of a certain edge exceeds the system's capacity or whether congestion has occurred. If the flow pressure exceeds the threshold, it may indicate that the path or connection is congested. Based on the comparison between the traffic pressure ratio and the preset pressure threshold, the one-way traffic relationship between the passing node and other nodes is further obtained. This means that if the traffic pressure on a certain edge is too high, the system may restrict the traffic flow in certain directions. Finally, it will determine which nodes can pass one-way and which paths are restricted. This step is actually regulating the traffic flow between each node to ensure that there is no excessive congestion and to avoid unnecessary traffic flow. Based on the established one-way traffic relationship, a time-varying directed graph structure is generated based on the undirected weighted graph. The key feature of the time-varying directed graph is that its edges are directional, representing the one-way transmission of traffic or information. Unlike the undirected graph, this graph can reflect the dynamically changing system state. This time-varying directed graph will be continuously updated over time, and the relationships between nodes and the direction of edges may change due to changes in traffic pressure or system requirements; this dynamism allows the graph to better adapt to actual scheduling and decision-making needs.

[0030] In one optional embodiment, the preset access control model includes a graph neural network, an attention mechanism layer, and a decision output network; Based on the time-varying directed graph structure and the state data of the nodes to be scheduled corresponding to the preset control response time scale, scheduling decisions are made based on the preset access control model to obtain the personnel and vehicle access control instructions for the current scheduling period, including: The state data of the nodes to be scheduled and the time-varying directed graph structure corresponding to the preset control response time scale are input into the graph neural network to extract topological features and obtain the spatiotemporal features of the passage topology. The spatiotemporal features of the traffic topology are input into the attention mechanism layer to extract key nodes and obtain the corresponding key node traffic features. The spatiotemporal characteristics of the traffic topology and the traffic characteristics of key nodes are input into the decision output network to generate a scheduling strategy, thereby obtaining the personnel and vehicle access control instructions for the current scheduling period.

[0031] It should be noted that the model consists of three main components: a graph neural network, an attention mechanism layer, and a decision output network; the graph neural network is used to process the relationships and structural information between nodes; the attention mechanism layer helps the model focus on important information or nodes; and the decision output network is responsible for generating specific scheduling strategies and control instructions based on the extracted features. The status data of the nodes to be scheduled (such as current traffic flow, waiting time, etc.) corresponding to the preset control response time scale and the time-varying directed graph structure (reflecting the one-way traffic relationship between nodes and its dynamic changes) are input into the graph neural network; these data are the basis for scheduling decisions and provide real-time traffic condition information; After receiving the input, the graph neural network processes the nodes and their connections in the graph to extract the spatiotemporal features of the traffic topology. These features reflect the structural characteristics and flow of the traffic network in a specific time period, including the state of each node and its impact on surrounding nodes. This step ensures that the model has a comprehensive understanding of traffic flow and interactions between nodes. The extracted traffic topology spatiotemporal features are input into the attention mechanism layer. In this layer, the model can identify the key nodes that have the greatest impact on scheduling decisions through the attention mechanism. These key nodes may be nodes with large traffic flow, severe congestion, or special functions (such as intersections, hubs, etc.). By focusing on these important nodes, the model can improve the accuracy and efficiency of decision-making. The spatiotemporal characteristics of the traffic topology and the traffic characteristics of key nodes are input into the decision output network. This network comprehensively considers all extracted features to generate vehicle and pedestrian access control instructions for the current scheduling cycle. These instructions may include measures such as traffic restrictions on specific road sections, traffic light control, and traffic guidance to optimize traffic flow, reduce congestion, and ensure traffic safety.

[0032] In an optional embodiment, the method further includes: When the status data of the area to be processed triggers the preset congestion judgment rule, abnormal traffic situation data is determined; abnormal traffic situation data includes gate congestion data and channel blockage data. Based on abnormal traffic flow data and historical traffic efficiency profiles, the target gate or target channel causing the anomaly can be identified. Based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the personnel and vehicle access control commands.

[0033] It should be noted that the status data of the area to be processed should be continuously monitored, such as the crowd congestion in front of the turnstile, and the vehicle congestion or blockage in the passage. When these real-time data trigger the preset congestion judgment rules, the system will mark the current situation as an abnormal traffic situation. Abnormal traffic situations mainly include two types of data: gate congestion data: reflecting backlog or queues at the gate or entrance, with people or vehicles waiting for too long; channel blockage data: reflecting flow blockage inside the channel or on the passage, affecting the overall flow efficiency. After identifying abnormal data, the system analyzes it in conjunction with historical traffic efficiency profiles; the historical profiles record the traffic efficiency and flow patterns of each gate or channel at different time periods. By comparing abnormal data with historical profiles, the system can pinpoint the specific target gate or channel that caused the anomaly; this step ensures that the control strategy is targeted, rather than blindly adjusting the entire area. Once abnormal nodes (target gates or channels) are identified, the system will monitor the data of these nodes in their current pending scheduling state, including the number of people waiting, vehicle density, traffic pressure, etc. Based on this information, the weights of the corresponding nodes in the time-varying directed graph are adjusted; the essence of weight adjustment is to change the importance or passage priority of the node in the graph, thereby affecting the overall flow path and passage strategy. After the node weights are adjusted, the control system will regenerate or adjust the access control instructions for people and vehicles based on the updated graph structure. These instructions may include measures such as increasing or decreasing the opening time of a gate, guiding vehicles or people to choose different channels, and temporarily adjusting the direction of passage. Through this dynamic adjustment, the system can alleviate abnormal situations, restore traffic efficiency, and ensure the safety and smooth flow of people and vehicles.

[0034] In an optional embodiment, based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the pedestrian and vehicle access control commands, including: When the abnormal passage situation data only includes the data of people stuck in front of the gate, the passage capacity weight of the corresponding node of the target gate is increased according to the passage efficiency status of the target gate and the range of the stuck area. When the abnormal traffic situation data only includes channel congestion data, the guidance weight of the downstream node of the target channel is adjusted according to the congestion change rate and the diversion path. When abnormal traffic situation data includes data on delays in front of gates and data on channel congestion, the scheduling adjustment priority is calculated based on the delay occurrence ratio and the congestion change rate, and the weight of gate nodes and guide nodes are adjusted in sequence according to the priority.

[0035] It should be noted that when the system detects an anomaly manifested only as congestion in front of the turnstile (e.g., a large number of people or vehicles waiting at an entrance), it will adjust its settings based on the target turnstile's throughput efficiency and the extent of the congestion area. The throughput efficiency refers to the target turnstile's current capacity (e.g., how many people or vehicles can pass through per unit of time). If the target turnstile's capacity is low, its weight needs to be increased to enhance its throughput. The extent of congestion refers to how many people or vehicles are affected by the congestion in front of the target turnstile. Based on the size of the congestion area, the urgency of the area can be determined, and the throughput weight of the target turnstile node can be increased accordingly. This allows the turnstile to handle excessive traffic more efficiently, avoiding more severe congestion. When only channel congestion data is available: If the anomaly is caused by channel congestion (e.g., severe blockage of vehicle or pedestrian flow within a channel), the system will adjust based on the congestion change rate and diversion paths. Congestion change rate: This refers to the change in the degree of congestion within a channel over a period of time; if the congestion worsens rapidly, the system will make corresponding adjustments. Diversion paths: Depending on the degree of congestion in a channel, the system may guide traffic to detour via other paths; in this case, the system will adjust the guidance weights of downstream nodes of the target channel to enhance its guidance capacity for other paths; this can help alleviate pressure in congested areas by diverting traffic flow to other paths, preventing further deterioration of congestion. When both data on delayed arrival at turnstiles and data on congestion exist simultaneously: If both types of abnormal data (delay and congestion) are present, the system will calculate the scheduling priority based on the delay occurrence ratio and the congestion change rate. Delay occurrence ratio: This refers to the ratio of the number of people or vehicles delayed at the turnstiles to the number of people or vehicles passing through; this helps to assess the severity of the delay. Congestion change rate: This reflects the rate at which the degree of congestion in the channel changes over time, helping the system determine which area's congestion problem needs to be addressed first. Based on these priorities, the system will adjust the weights of the gate nodes and the guidance nodes in sequence. First, it will prioritize increasing the weight of the nodes that need the most attention based on the situation of congestion and delays, and then gradually adjust the guidance weights of other nodes to optimize the overall traffic flow.

[0036] Example 2, please refer to Figure 2 This invention provides a technical solution: an intelligent pedestrian and vehicle access control system based on optimized scheduling, applicable to the aforementioned intelligent pedestrian and vehicle access control method based on optimized scheduling, comprising: Data acquisition unit 1 is used to acquire the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each access gate. Topology mapping unit 2 is used to perform topology mapping on the state data of the area to be processed based on preset multi-dimensional constraints, so as to obtain the state data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. The association and fusion unit 3 is used to perform association and fusion on dynamic constraint graphs of different constraint dimensions with the optimization objective of minimizing the global congestion cost, to obtain a multi-dimensional fusion association graph, and to obtain the corresponding time-varying directed graph structure based on the multi-dimensional fusion association graph; The access control unit 4 is used to make scheduling decisions based on the time-varying directed graph structure and the status data of the node to be scheduled corresponding to the preset control response time scale, and to obtain the access control command for the current scheduling cycle. The duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the status data of the area to be processed. The access control command is used to adjust the opening duration of at least one access gate or the pointing angle of the guiding device.

[0037] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. An intelligent pedestrian and vehicle access control method based on optimized scheduling, characterized in that, include: Obtain the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each of the access gates. Based on preset multi-dimensional constraints, the state data of the area to be processed is topologically mapped to obtain the state data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. With minimizing the global congestion cost as the optimization objective, dynamic constraint graphs of different constraint dimensions are correlated and fused to obtain a multi-dimensional fused correlation graph, and the corresponding time-varying directed graph structure is obtained based on the multi-dimensional fused correlation graph. Based on the time-varying directed graph structure and the state data of the nodes to be scheduled corresponding to the preset control response time scale, scheduling decisions are made based on the preset access control model to obtain the personnel and vehicle access control instructions for the current scheduling period. The duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the status data of the area to be processed; the personnel and vehicle access control command is used to adjust the opening duration of at least one of the access gates or the pointing angle of the guiding device.

2. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 1, characterized in that, Obtain the status data of the pending areas in the current scheduling cycle, including: Real-time traffic flow and equipment response delay of each access gate are collected according to a preset monitoring time interval; Retrieve historical traffic efficiency profiles for each person or vehicle passing through; these historical traffic efficiency profiles are used to characterize the dynamic correlation between the behavioral norms and traffic speed of the passing subjects during historical traffic processes. Obtain weather change parameters that are individually associated with each region within a preset future time period; Based on the weather change parameters associated with each region, the real-time traffic flow, the device response latency, and the historical traffic efficiency profile, the status data of the region to be processed is generated.

3. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 2, characterized in that, Methods for constructing pre-defined multi-dimensional constraints include: The status data of the area to be processed is statistically analyzed according to the main traffic entities to obtain the time series of motor vehicle and pedestrian traffic and the equipment delay time series. The traffic flow time series of all traffic entities are merged according to spatial location to obtain the corresponding regional occupancy time series; Based on the area occupancy time series and the device delay time series, the corresponding traffic demand power spectrum is obtained; Based on the traffic demand power spectrum and the historical traffic efficiency profile, traffic priority rules with different constraint dimensions are obtained.

4. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 3, characterized in that, Based on preset multi-dimensional constraints, topological mapping is performed on the state data of the region to be processed to obtain state data of the nodes to be scheduled and a dynamic constraint graph with multiple constraint dimensions, including: The state data of the region to be processed is processed into nodes based on preset multi-dimensional constraints to obtain node state data with multiple constraint dimensions. When the current scheduling period is the first scheduling period, the node state data of each constraint dimension are initialized with graph structure and the constraint matrix is ​​jointly optimized and solved to obtain the node state data to be scheduled and the dynamic constraint graph of the corresponding constraint dimension. When the current scheduling period is not the first scheduling period, the state data of the corresponding node is deduced based on the dynamic constraint graph of each constraint dimension of the previous scheduling period to obtain the state data of the node to be scheduled in the corresponding constraint dimension. Obtain the optimization constraint matrix corresponding to the state data of the node to be scheduled; The optimized constraint matrix and the dynamic constraint graph of the constraint dimension corresponding to the previous scheduling cycle are weighted and fused to obtain the dynamic constraint graph of the constraint dimension corresponding to the current scheduling cycle.

5. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 4, characterized in that, With minimizing the global congestion cost as the optimization objective, dynamic constraint graphs of different constraint dimensions are correlated and fused to obtain a multi-dimensional fused correlation graph, including: Using the priority weights corresponding to the dynamic constraint graphs of different constraint dimensions as optimization variables, a priority weight optimization model is constructed based on the optimization objective of minimizing the global congestion cost. The priority weight optimization model is then solved according to the state data of the nodes to be scheduled in different constraint dimensions and the corresponding dynamic constraint graphs to obtain the priority weights in different constraint dimensions. The corresponding dynamic constraint graphs are associated and fused based on the priority weights of different constraint dimensions to obtain the multidimensional fused association graph.

6. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 5, characterized in that, Based on the multidimensional fusion correlation graph, the corresponding time-varying directed graph structure is obtained, including: Based on the multidimensional fusion association graph and the corresponding node degree distribution, the corresponding undirected weight graph is obtained; Based on the undirected weighted graph, the flow pressure ratio corresponding to each edge of each passage node is obtained, and based on the comparison result of the flow pressure ratio corresponding to each edge with the preset pressure threshold, the one-way passage relationship between the corresponding passage node and other nodes is obtained. Based on the one-way traffic relationship between each passing node and other nodes, the time-varying directed graph structure is generated based on the undirected weighted graph.

7. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 6, characterized in that, The preset access control model includes a graph neural network, an attention mechanism layer, and a decision output network; Based on the time-varying directed graph structure and the state data of the nodes to be scheduled corresponding to the preset control response time scale, scheduling decisions are made based on the preset access control model to obtain the personnel and vehicle access control instructions for the current scheduling period, including: The state data of the node to be scheduled corresponding to the preset control response time scale and the time-varying directed graph structure are input into the graph neural network to extract topological features and obtain the spatiotemporal features of the passage topology. The passage topology spatiotemporal features are input into the attention mechanism layer to extract key nodes and obtain the corresponding key node passage features. The spatiotemporal characteristics of the traffic topology and the traffic characteristics of the key nodes are input into the decision output network to generate a scheduling strategy, thereby obtaining the personnel and vehicle access control instructions for the current scheduling period.

8. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 7, characterized in that, The method further includes: When the status data of the area to be processed triggers a preset congestion judgment rule, abnormal traffic situation data is determined; the abnormal traffic situation data includes gate congestion data and channel blockage data; Based on the abnormal traffic situation data and the historical traffic efficiency profile, the target gate or target channel that caused the abnormality is determined. Based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the personnel and vehicle access control commands.

9. The intelligent pedestrian and vehicle access control method based on optimized scheduling according to claim 8, characterized in that, Based on the status data of the nodes to be scheduled for the target gate or target channel, the weights of the corresponding nodes in the time-varying directed graph structure are adjusted to update the pedestrian and vehicle access control commands, including: When the abnormal passage situation data only includes the data of people lingering in front of the gate, the passage capacity weight of the node corresponding to the target gate is increased according to the passage efficiency status and the range of the lingering area of ​​the target gate. When the abnormal traffic situation data only includes the channel blockage data, the guidance weight of the downstream node of the target channel is adjusted according to the blockage change rate and the diversion path. When the abnormal traffic situation data includes the gate congestion data and the channel blockage data, the scheduling adjustment priority is calculated based on the congestion occurrence ratio and the congestion change rate, and the gate node weight and the guide node weight are adjusted in sequence according to the priority.

10. An intelligent pedestrian and vehicle access control system based on optimized scheduling, applicable to the intelligent pedestrian and vehicle access control method based on optimized scheduling as described in any one of claims 1-9, characterized in that, include: The data acquisition unit is used to acquire the status data of the area to be processed in the current scheduling cycle; the status data of the area to be processed includes the real-time pedestrian density, real-time vehicle density and equipment response delay between the management terminal and each of the access gates. The topology mapping unit is used to perform topology mapping on the state data of the area to be processed based on preset multi-dimensional constraints to obtain the state data of the nodes to be scheduled and the dynamic constraint graph with multiple constraint dimensions; the preset multi-dimensional constraints include the historical passage efficiency profiles of different types of passage subjects and passage priority rules. The association and fusion unit is used to perform association and fusion on dynamic constraint graphs with different constraint dimensions with the optimization objective of minimizing global congestion cost, to obtain a multi-dimensional fusion association graph, and to obtain the corresponding time-varying directed graph structure based on the multi-dimensional fusion association graph; The access control unit is used to make scheduling decisions based on the time-varying directed graph structure and the state data of the node to be scheduled corresponding to the preset control response time scale, and to obtain the human and vehicle access control command for the current scheduling period based on the preset access control model. The duration of the preset control response time scale is greater than or equal to the preset monitoring time interval of the status data of the area to be processed; the personnel and vehicle access control command is used to adjust the opening duration of at least one of the access gates or the pointing angle of the guiding device.

Images