AI traffic decision management method and system for multi-region collaboration

Abstract

The invention discloses an AI traffic decision management method and system for multi-region collaboration, and relates to the technical field of intelligent traffic systems, and the method comprises the steps: building a region set capable of achieving linkage collaboration; establishing a regional traffic data set, and establishing an edge traffic state recognition result; establishing a short-term driving target of the vehicle according to the regional traffic data set; acquiring a navigation plan of a vehicle, and establishing a long-term driving target according to the navigation plan; and performing traffic decision management according to the collaborative decision result. The technical problem of urban traffic congestion caused by isolated traffic data among multiple areas in a city and lack of comprehensive consideration of long and short term targets of vehicles in the prior art is solved, and the technical effect of improving the urban traffic efficiency is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and particularly to the technical field of AI traffic decision-making management methods and systems for multi-region collaboration. Background Art

[0002] In the modern urban transportation system, the efficient operation of traffic has a crucial impact on the development of the city, the lives of residents, and the conduct of economic activities. Scientific and reasonable traffic decision-making management is the core link to ensure smooth traffic.

[0003] In the prior art, for the problem of traffic decision-making management, traditional traffic management methods mainly rely on manual experience and simple signal control and vehicle scheduling methods. However, with the acceleration of the urbanization process and the rapid growth of the vehicle ownership, it is difficult to maintain efficient traffic order in the increasingly complex traffic conditions.

[0004] During the process of implementing the inventive technical solution in the embodiments of the present application, the inventors of the present application found that the above technologies have at least the following technical problems:

[0005] The traffic flow, road conditions, and travel demands vary significantly among different regions in the city, and the traffic conditions are constantly changing dynamically. Traditional traffic management means are difficult to conduct collaborative analysis and accurate prediction of the traffic conditions in different regions. As a result, there is a lack of scientific basis in formulating traffic decisions, and it is impossible to achieve reasonable allocation and collaborative scheduling of traffic resources among multiple regions, leading to frequent problems such as traffic congestion and vehicle delays. Summary of the Invention

[0006] The present application solves the technical problem in the prior art that traffic data among multiple regions in the city is isolated and the comprehensive consideration of the short-term and long-term goals of vehicles is lacking, which leads to urban traffic congestion. The present application establishes a region set that can be linked and coordinated, arranges multi-source data collection sensors in the region to collect traffic data, uses edge computing nodes to identify traffic states, combines the short-term and long-term driving goals of vehicles, and synchronizes relevant data to the collaborative cloud platform. On the cloud platform, collaborative decisions are made by establishing busy evaluation thresholds, clustering analysis, etc. After generating the decision results, feedback evaluation and real-time optimization are carried out according to the real-time tracked traffic flow data, and a congestion database is established for congestion prediction and pre-emptive guidance, so as to realize the collaborative and efficient management of multi-region traffic, and make the AI traffic decision-making management for multi-region collaboration more scientific, reasonable, accurate, and efficient.

[0007] In view of the above technical problems, the present application proposes a technical solution for an AI traffic decision-making management method and system for multi-region collaboration.

[0008] In a first aspect, the present application provides an AI traffic decision-making management method for multi-region collaboration. The method includes:

[0009] Establish a set of regions that can be linked and collaborated, and deploy multi-source data acquisition sensors within the set of regions. The multi-source data acquisition sensors include cameras and radars; use the multi-source data acquisition sensors to identify traffic data within the region, establish a regional traffic data set, and perform traffic status identification through an edge computing node integrated in the region to establish an edge traffic status identification result; establish a short-term driving goal for the vehicle based on the regional traffic data set; obtain the navigation plan of the vehicle under the condition of obtaining the owner's permission, and establish a long-term driving goal based on the navigation plan; synchronize the edge traffic status identification result, the short-term driving goal, and the long-term driving goal to a collaborative cloud platform, perform collaborative decision-making through the collaborative cloud platform to generate a collaborative decision result, and perform traffic decision-making management based on the collaborative decision result.

[0010] In a second aspect, the present application provides an AI traffic decision-making management system for multi-region collaboration. The system includes:

[0011] A region set establishment module for establishing a set of regions that can be linked and collaborated, and deploying multi-source data acquisition sensors within the set of regions. The multi-source data acquisition sensors include cameras and radars; a traffic data identification module for using the multi-source data acquisition sensors to identify traffic data within the region, establishing a regional traffic data set, and performing traffic status identification through an edge computing node integrated in the region to establish an edge traffic status identification result; a short-term goal establishment module for establishing a short-term driving goal for the vehicle based on the regional traffic data set; a long-term goal establishment module for obtaining the navigation plan of the vehicle under the condition of obtaining the owner's permission, and establishing a long-term driving goal based on the navigation plan; a decision result generation module for synchronizing the edge traffic status identification result, the short-term driving goal, and the long-term driving goal to a collaborative cloud platform, performing collaborative decision-making through the collaborative cloud platform to generate a collaborative decision result, and performing traffic decision-making management based on the collaborative decision result.

[0012] The present application proposes one or more technical solutions, which at least have the following technical effects:

[0013] This application establishes an area set that can be linked and coordinated, and arranges multi-source data acquisition sensors within the area set. The multi-source data acquisition sensors include cameras and radars. Then, the multi-source data acquisition sensors are used to identify traffic data within the area, establish an area traffic data set, and perform traffic state identification through an edge computing node integrated in the area to establish an edge traffic state identification result. Next, a short-term driving target of the vehicle is established based on the area traffic data set, and with the permission of the vehicle owner, the navigation plan of the vehicle is obtained, and a long-term driving target is established based on the navigation plan. Next, the edge traffic state identification result, the short-term driving target, and the long-term driving target are synchronized to the collaborative cloud platform, and collaborative decision-making is performed through the collaborative cloud platform to generate a collaborative decision result. Traffic decision management is carried out according to the collaborative decision result, achieving the technical effect of improving urban traffic efficiency.

[0014] The above content outlines this application's solution for an AI traffic decision management method and system for multi-region collaboration. This application will detail the steps of the technical solution in the following specific embodiments to facilitate a clear and complete understanding of this application by those skilled in the art. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of the AI traffic decision management method for multi-region collaboration provided by an embodiment of this application.

[0017] Figure 2 It is a structural diagram of the AI traffic decision management system for multi-region collaboration provided by an embodiment of this application.

[0018] Description of the reference numerals: Area set establishment module 1, traffic data identification module 2, short-term target establishment module 3, long-term target establishment module 4, decision result generation module 5. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In this application, a set of linkable regions is established, traffic data is collected by multi-source sensors, and short-term and long-term driving targets of vehicles are obtained after being processed by edge computing nodes. The data is synchronized to the collaborative cloud platform, and collaborative decision-making results are generated through various algorithms and clustering analysis. After the decision is executed, traffic flow data is tracked for feedback evaluation and real-time optimization, and a congestion database is also established to predict congestion and conduct pre-emptive dredging, so as to achieve precise and efficient traffic decision-making management for multi-region collaboration, and achieve the technical effect of improving urban traffic efficiency.

[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0021] It should be noted that any variations of the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0022] Embodiment 1, as Figure 1 shown, an AI traffic decision-making management method for multi-region collaboration, wherein the method includes:

[0023] Step A100: Establish a set of regions that can be linked and coordinated, and deploy multi-source data collection sensors within the set of regions. The multi-source data collection sensors include cameras and radars.

[0024] In the embodiments of the present application, the set of regions that can be linked and coordinated refers to the integration of multiple regions with traffic relevance. The multi-source data collection sensors refer to a combination of devices used to collect various traffic-related data, including: millimeter-wave radar, lidar, 360-degree field-of-view 4K cameras, etc.

[0025] Specifically, we first grouped areas with good road connectivity and convenient vehicle traffic according to the degree of interconnection between nodes in the road network, and then referred to historical traffic flow data. If the traffic between regions greatly affects each other, such as congestion in one area will cause traffic changes in another area, we included them in the same set, and combined the two factors to divide the area set that can be coordinated and coordinated. Then, we rationally arranged cameras and radars in these areas, such as deploying millimeter-wave radars and 4K cameras at the vertices of intersections on the road layer; installing laser radars in the central isolation belt of the road test layer to cover key sections and traffic nodes, ensuring real-time and comprehensive collection of traffic data, and providing accurate basis for subsequent traffic decisions.

[0026] By dividing the area and rationally deploying sensors within the area, we can integrate regional traffic resources, comprehensively collect traffic data, and provide data support for multi-regional collaborative traffic decision-making.

[0027] Step A200: Use the multi-source data acquisition sensor to identify traffic data in the area, establish a regional traffic data set, and identify traffic status through edge computing nodes integrated in the area to establish edge traffic status identification results.

[0028] In the embodiments of the present application, the regional traffic data set refers to a data set formed by comprehensively integrating and summarizing the traffic information within a set of regions that can be linked and coordinated. The edge computing node refers to the edge computing gateway deployed in the traffic management system. Traffic state identification refers to the process of determining the current operating status of the traffic management system. The edge traffic state identification result refers to the judgment conclusion about the traffic operation status drawn by the edge computing node after analyzing and processing the traffic data in the region. The traffic management system refers to a system that integrates functions such as data collection, processing and analysis, interaction with vehicle navigation, target setting and dynamic optimization, can obtain traffic information in real time and carry out multiple traffic management tasks based on this, and also has a user interface for all parties to use.

[0029] Optionally, cameras are used to identify vehicle types, quantities, and other data, and radars are used to obtain vehicle speed, distance, and other data. The data collected by the two constitute a regional traffic data set. The edge computing node then analyzes this data locally to avoid delays and bandwidth pressure in remote data transmission. The edge computing node's data analysis process is mainly divided into four steps: data reading, feature extraction, state judgment, and result output:

[0030] First, the traffic data after format conversion and cleaning and denoising is read from the storage device.

[0031] Next, the traffic data is processed by statistical analysis algorithms, such as calculating the average vehicle flow and the standard deviation of vehicle speed over a period of time. The average vehicle flow can reflect the average flow of a certain section during a specific period. If it far exceeds the normal level, it may indicate congestion. The standard deviation reflects the degree of dispersion of vehicle speed. A large dispersion indicates large fluctuations in vehicle speed and unstable traffic flow.

[0032] Then, the extracted features are compared with the preset vehicle flow and speed. Here, the preset vehicle flow and speed are thresholds set by those skilled in the art based on long-term practices in the traffic field. For example, during the peak hours on urban arterial roads, the normal vehicle flow is 1,500 - 2,000 vehicles per hour. If the edge computing node calculates that the vehicle flow at a certain time reaches 2,500 vehicles, exceeding the upper limit, congestion can be preliminarily judged. In terms of vehicle speed, the speed limit on urban ordinary roads is 60 km / h. If the average vehicle speed is continuously lower than 30 km / h for a long time and the vehicle flow exceeds a certain proportion, according to the rules, it can be determined that the traffic state is poor. To judge the traffic state, it is based on vehicle speed and vehicle density to determine whether there is congestion.

[0033] Finally, the judgment results are sorted out and output as the edge traffic state recognition results. The edge computing node can identify the traffic state, such as judging road congestion or accident situations, thus establishing the edge traffic state recognition results.

[0034] By obtaining and analyzing traffic data, the edge traffic state recognition results are established, providing a basis for rapid decision-making in traffic management and improving management efficiency.

[0035] Step A300: Establish short-term driving goals for vehicles based on the regional traffic data set.

[0036] In the embodiments of the present application, the short-term driving goal refers to the driving expectation and action guidance planned for a vehicle in a relatively short time based on the current traffic conditions.

[0037] In an embodiment of the present application, first, the data directly related to vehicles in the regional traffic data set is screened and extracted. First, it is classified according to the data source and type, focusing on vehicle sensor and related monitoring data, and initially screening out the data set that may contain information such as vehicle position, speed, and driving direction. Then, using feature keywords or identifiers such as "vehicle ID", "position coordinates", "speed value", and "driving direction angle", the specific data of each vehicle at a specific moment is accurately extracted. By identifying the position information of each vehicle at a specific moment, the specific lane where it is located is determined. The lane information is directly related to the driving constraint conditions of the vehicle, such as the situation where it is inconvenient to change lanes in a certain lane, which is a short-term constraint.

[0038] Next, analyze the current driving speed and direction of the vehicle. Based on this information, combined with the speed limit regulations of the road where the vehicle is located, intersection passing rules, and traffic flow conditions of the road section ahead, infer and predict the driving trend of the vehicle in the next few to more than ten minutes based on the feature-trend mapping relationship. First, collect the relationships between typical features and corresponding driving trends in a large number of traffic scenarios to build a rule base. The rule base includes the feature-trend mapping relationship. For example, if the vehicle is in a straight lane, the green light is on at the front intersection and the traffic flow is less than a certain threshold (such as less than 10 vehicles passing through each lane per minute), and the current speed of the vehicle is between 80% - 100% of the speed limit, then the rule sets that the vehicle will pass through the intersection at the current speed. Then, match the currently extracted vehicle information, including position, speed, driving direction, lane where the vehicle is located, road speed limit, intersection passing rules, and traffic flow conditions of the road section ahead, with the conditions in the rule base. When the conditions corresponding to the rules in the rule base are met, execute the driving trend prediction result corresponding to the rule.

[0039] Then, comprehensively consider short-term constraint conditions (factors affecting traffic operation in a short period of time) to adjust and optimize the initially set driving goals. If there are signs of congestion ahead in the lane where the vehicle is located, although its original plan is to go straight, due to this short-term constraint of difficult lane change, the goal may need to be adjusted to drive slowly in this lane and wait for the congestion to ease, rather than forcing a lane change.

[0040] Through the above steps, utilize the multi-dimensional information contained in the regional traffic dataset, fully consider the actual traffic environment where the vehicle is located and short-term constraint conditions, so as to establish vehicle short-term driving goals that conform to the actual situation. These goals can provide accurate and practically guiding basic data for subsequent traffic decisions, such as intelligent navigation path planning, traffic signal optimization control, etc., and contribute to improving the operation efficiency and smoothness of the entire traffic management system.

[0041] Step A400: Under the condition of obtaining the owner's permission, obtain the navigation plan of the vehicle, and establish a long-term driving goal according to the navigation plan.

[0042] In the embodiment of the present application, the navigation plan of the vehicle refers to the driving guidance plan formulated by the owner from the current position to the destination. The long-term driving goal refers to a set of goals based on the vehicle's navigation plan, covering a series of coherent action plans to achieve a smooth arrival at the destination.

[0043] Specifically, first, obtaining the owner's permission is a prerequisite for the entire process. Usually, with the help of the in-vehicle intelligent system or the mobile application program bound to the vehicle, in the form of pop-up prompts, agreement signing, etc., explain to the owner in detail the purpose of obtaining navigation plan data, data storage methods, and privacy protection measures. After the owner clearly authorizes and agrees, the system has the permission to obtain relevant data.

[0044] Next, obtain the navigation plan of the vehicle. The navigation system of modern vehicles relies on positioning technologies such as the Global Positioning System (GPS) and the Beidou Satellite Navigation System. It not only includes the coordinate information of the starting point and the ending point, but also covers detailed content such as each key location (i.e., waypoints) passed by the planned driving route, the recommended driving speed, and the estimated driving time.

[0045] Then, establish a long-term driving goal based on the obtained navigation plan. Based on the route information in the navigation plan, determine the driving path goal of the vehicle within a relatively long time span, including the driving direction to be followed in different sections and the turning options at intersections. At the same time, refer to the estimated driving time and the recommended driving speed in the navigation plan, and combine with the real-time change prediction of traffic flow to set the speed goal of the vehicle in each section. For example, if the navigation plan predicts that the traffic flow on a certain section of the highway will be large in the next period of time, the speed goal of this section may be set slightly lower than the speed limit to ensure safe and efficient passage. In addition, considering possible emergencies such as road construction and traffic accidents, the long-term driving goal also needs to have a certain degree of flexibility and adjustability. The system will dynamically optimize and adjust the established long-term driving goal according to the real-time traffic information obtained to ensure that the vehicle always drives towards the destination in the best state.

[0046] Through the above steps, on the premise of respecting the privacy of the vehicle owner and obtaining permission, and making full use of the navigation plan data of the vehicle, a scientific, reasonable and practically guiding long-term driving goal of the vehicle can be established.

[0047] Step A500: Synchronize the edge traffic state recognition result, the short-term driving goal, and the long-term driving goal to the collaborative cloud platform, perform collaborative decision-making through the collaborative cloud platform to generate a collaborative decision result, and perform traffic decision management according to the collaborative decision result.

[0048] In the embodiment of the present application, the collaborative cloud platform refers to a traffic data processing and sharing center built based on cloud computing technology. The collaborative decision result refers to the output result generated by the collaborative decision-making process. Traffic decision management refers to the specific measures taken by the traffic management department to manage traffic according to the collaborative decision result generated by the collaborative cloud platform.

[0049] In an embodiment of the present application, synchronize these edge traffic state recognition results, short-term driving goals, and long-term driving goals to the collaborative cloud platform. Each edge computing node and the vehicle terminal upload the corresponding data to the collaborative cloud platform through network communication technology to achieve centralized aggregation of data.

[0050] Then, collaborative decision-making is carried out on the collaborative cloud platform. The platform uses the method of general busy evaluation threshold to comprehensively analyze the aggregated data. Through this series of analyses, a collaborative decision-making result is generated, which is a comprehensive decision-making suggestion for aspects such as the optimal allocation of traffic resources and the adjustment of vehicle driving strategies.

[0051] Finally, traffic decision management is carried out according to the collaborative decision-making result generated by the collaborative cloud platform. The specific steps are described in detail in A530 - A550. Thus, the effective management of the traffic management system is realized, and the overall traffic operation efficiency and safety are improved.

[0052] By synchronizing the corresponding data to the collaborative cloud platform for collaborative decision-making, the technical effects of improving the operation efficiency, safety and orderliness of the traffic management system and optimizing the allocation of traffic resources are achieved.

[0053] Furthermore, step A500 in the method provided by the embodiment of the present application includes:

[0054] A510: Establish a general busy evaluation threshold, and perform busy screening on the edge traffic state recognition result through the general busy evaluation threshold to establish a true busy state identifier for traffic.

[0055] A520: Construct a first dredging target based on the true busy state identifier, and use the first dredging target to perform collaborative decision-making based on the short-term driving target and the long-term driving target to establish a collaborative decision-making result.

[0056] In the embodiment of the present application, the general busy evaluation threshold refers to a unified quantitative standard preset by those skilled in the art to measure whether the traffic state is busy. When the general busy evaluation threshold is exceeded, the traffic is in a busy state. Busy screening refers to the process of comparing and screening the relevant data in the edge traffic state recognition result by using the general busy evaluation threshold. The true busy state identifier refers to a clear mark of the traffic area determined to be in a busy state based on busy screening. The first dredging target refers to a preliminary action direction and specific goal set to relieve the traffic pressure in the true busy area based on the true busy state identifier.

[0057] Specifically, first, a general busy evaluation threshold is established, which is constructed by those skilled in the art based on a large amount of historical traffic data and professional knowledge in the traffic field. In order to uniformly measure the traffic busyness under the background of multi-region collaboration in different regions and different time periods, such a general standard needs to be formulated. For example, taking the traffic flow reaching 2,000 vehicles per hour and the average vehicle speed being lower than 30 kilometers per hour as a reference index. By comparing this threshold with the relevant data of the edge traffic state recognition results (including traffic flow, vehicle speed, congestion conditions, etc.) analyzed by the edge computing node for the regional traffic data, the regional traffic state where the traffic flow reaches or exceeds 2,000 vehicles per hour and the average vehicle speed is lower than 30 kilometers per hour will be determined as a busy state.

[0058] Next, based on the selected busy state information, a true busy state identifier for the traffic is established. A clear mark is made for the traffic areas in the busy state for subsequent targeted processing. For example, add a specific label "true busy" to the traffic data of each area determined to be in the busy state, and record detailed information such as relevant time, location, and degree of busyness, forming a true busy state identifier, enabling the traffic management system to clearly identify which areas are in a truly busy traffic condition.

[0059] Then, a first dredging target is constructed based on the true busy state identifier. For example, for a certain road section in the true busy state, according to its geographical location, the surrounding road network structure, and the traffic flow distribution, determine a strategy to relieve congestion by guiding vehicles to detour around relatively idle surrounding roads, and transform this strategy into a specific target, such as reducing the traffic flow of this road section by 30% within the next 30 minutes. This is the constructed first dredging target.

[0060] Finally, the first dredging target is used to make a collaborative decision based on short-term driving targets and long-term driving targets, and a collaborative decision result is established. The specific steps are elaborated in detail in A521 - A523.

[0061] By making such collaborative adjustments to the short-term and long-term driving targets of numerous vehicles, a complete decision-making plan is formed, that is, the collaborative decision result. This result reasonably guides the vehicle driving behavior, effectively relieves the traffic pressure in the true busy areas, and realizes the collaborative optimization of multi-region traffic.

[0062] Furthermore, step A520 in the method provided by the embodiment of the present application includes:

[0063] A521: Call the path fitting channel to perform path additional fitting on the short-term driving target and the long-term driving target of each vehicle under the first dredging target, and establish a path additional fitting result.

[0064] A522: Invoke the travel time fitting channel to fit the travel time of each vehicle under the first dredging target, and establish a time savings fitting result.

[0065] A523: Obtain the balance preference of the user mapped to the vehicle, and make a collaborative decision based on the balance preference, the path attachment fitting result, and the time savings fitting result to establish a collaborative decision result.

[0066] In the embodiments of the present application, invoking the distance fitting channel means enabling a functional module dedicated to analyzing and adjusting the vehicle driving path in the traffic management system. The path attachment fitting result refers to the specific changes of the new path after re-planning and integration compared with the original path. The travel time fitting channel refers to a functional module for accurately calculating and analyzing the travel time of vehicles on different driving paths. The time savings fitting result refers to the time reduction of each vehicle passing under the first dredging target compared with the original travel time. The balance preference refers to the tendency degree of the user associated with the vehicle for the two factors of saving time and saving distance during the travel process. The collaborative decision result refers to the vehicle driving strategy obtained after comprehensive analysis and judgment based on the balance preference, the path attachment fitting result, the time savings fitting result, etc.

[0067] Optionally, first, the distance fitting channel can analyze and adjust the vehicle driving path based on information such as the traffic network structure, road connectivity, and real-time traffic conditions. When constructing the path fitting channel, first obtain traffic network structure, connectivity, and real-time condition data from relevant data sources such as the collaborative cloud platform and edge computing nodes, integrate and clean them, and then store them. Then select a path search algorithm and optimize it to make it quickly find alternative routes. Then construct a path evaluation and integration module to determine the factors and weights affecting the path, calculate the impact of the new route on the original target path, and obtain the fitting result. Finally, design the interfaces with the travel time fitting channel, collaborative decision-making, and the system to achieve data interaction. When the first dredging target is to relieve congestion on a certain section of the road, vehicles need to be guided to detour. The distance fitting channel will perform path attachment fitting for the original short-term driving target path and long-term driving target path of each vehicle, combined with the surrounding alternative route conditions.

[0068] Next, the time fitting channel mainly focuses on the time consumption of the vehicle on different driving routes. When constructing the time fitting channel, first connect to multi-source data collection sensors and the collaborative cloud platform to obtain real-time road vehicle speeds, signal timings, traffic flows, etc., and establish a real-time update mechanism. Subsequently, calculate the vehicle's passing time on the section and waiting time at intersections, add them up to obtain the comprehensive passing time, and thus establish a time calculation model. Then compare the passing time of the new route (i.e., the route adjusted by the route fitting channel) with that of the original route, and output the time-saving fitting result. Finally, monitor the calculation error of the time fitting channel by comparing the actual traffic data with the calculation results, optimize the model parameters, and improve the calculation accuracy.

[0069] Then, obtain the balance preference of the user mapped to the vehicle. This information can usually be obtained through the user's historical operation records, set preferences, or questionnaire surveys in traffic-related applications. For example, by analyzing the user's behavior of choosing routes for multiple past trips, if the user often chooses a route with a longer distance but a shorter expected passing time, it can be judged that the user is more inclined to save time; conversely, if the user always chooses a shorter route, even if it may take more time, then the user is more inclined to save distance.

[0070] Finally, the system comprehensively considers the three factors of balance preference, route additional fitting result, and time-saving fitting result. By obtaining the user's balance preference to determine the weights of time and distance, quantify the route additional fitting result as the increased distance and the time-saving fitting result as the saved time, and use the linear weighting method to calculate the comprehensive scores of each alternative route (Comprehensive score = time weight × time savings - distance weight × distance increase). Select the route with the highest score as the optimal driving strategy, and recalculate and adjust according to the real-time traffic conditions during the vehicle's driving, so as to formulate and dynamically optimize the driving plan for each vehicle. For example, for a user who is more inclined to save time, if the route additional fitting result shows that a certain detour route increases a certain distance, but the time-saving fitting result indicates that it can significantly shorten the passing time, the system may give priority to recommending this route as the adjusted driving plan.

[0071] By making such comprehensive decisions for each vehicle based on its user's balance preference, a comprehensive set of collaborative decision results is formed. This result can, while meeting the first dredging goal, maximize the satisfaction of different users' personalized needs in terms of time and distance, and achieve the efficient collaborative optimization of multi-region traffic.

[0072] Furthermore, the method provided in the embodiment of the present application further includes step A600, where the step A600 further includes:

[0073] A610: Performing regional collaborative association clustering on the region set to establish a collaborative clustering cluster, wherein each subcluster in the collaborative clustering cluster can participate in repeated clustering.

[0074] A620: Compare the traffic status within the collaborative cluster according to the regional traffic data set to establish a second traffic dredging target.

[0075] A630: Make collaborative decisions based on the second dredging goal, the short-term driving goal, and the long-term driving goal.

[0076] In the embodiments of the present application, regional collaborative association clustering refers to a clustering operation performed on traffic areas in multiple different geographical locations using data mining and cluster analysis techniques. Collaborative clustering clusters refer to the results formed after the regional collaborative association clustering operation. The second dredging target refers to a specific target formulated after comparing the traffic status of the collaborative clusters, combined with the goals and needs of traffic management, to improve the traffic conditions within the clusters.

[0077] Specifically, first, regional collaborative correlation clustering is performed on the regional set of traffic areas in multiple different geographical locations to establish collaborative clustering clusters. Regional collaborative correlation clustering is based on the operation of data mining and cluster analysis technology, aiming to discover the similarities and correlations in traffic characteristics between different regions.

[0078] By using a density-based spatial clustering algorithm, regions with similar traffic status characteristics (such as traffic flow change trends, vehicle speed distribution, congestion periods, etc.) are divided into the same cluster, thus forming a collaborative cluster:

[0079] First, collect and organize multi-regional traffic data, set parameters such as neighborhood radius and minimum number of points. Then calculate the density with the traffic characteristic data points of each region as the center, and mark the points with qualified density as core points. Starting from the core point, the areas corresponding to the data points connected by density are assigned to the same cluster and continue to expand. The data points with insufficient density and not connected to the core point are regarded as noise points. The multiple cluster sets finally formed are collaborative clusters, in which the traffic status characteristics of the areas within each cluster are similar.

[0080] Subclusters (i.e., individual regions) in a collaborative cluster can participate in clustering repeatedly. Since traffic data changes in real time, if the traffic characteristics of a subcluster change significantly and do not match the overall characteristics of the cluster in which it is located, it will be included in the clustering analysis process and re-clustered to ensure that the clustering results always reflect the current traffic conditions truthfully.

[0081] Next, after obtaining the co-clustering clusters, for each region within each cluster, detailed traffic information data of each region is used to compare traffic states. For example, compare the traffic flow volume, average vehicle speed, and distribution of congested sections in different regions within the cluster during peak hours. Through comparison and analysis, identify regions within the cluster where there are differences in traffic states and problems such as congestion, and in combination with the goals and requirements of traffic management, formulate a second dredging goal aimed at improving the traffic conditions within the cluster.

[0082] Finally, co-decision making is carried out according to the second dredging goal, short-term driving goal, and long-term driving goal. The detailed steps are specifically described in A810 - A840.

[0083] By making such co-adjustments to the driving goals of numerous vehicles, a comprehensive co-decision making scheme that conforms to the actual traffic conditions is formed, achieving efficient co-management of multi-region traffic and improving the operation efficiency and service quality of the entire traffic management system.

[0084] Furthermore, step A800 in the method provided in the embodiments of the present application includes:

[0085] A810: Perform regional competitive association clustering on the region set to establish a competitive association clustering result.

[0086] A820: Analyze the real-time competitive state of the regions according to the competitive association clustering result and configure weak competitive points.

[0087] A830: Generate an optimization strategy set for the traffic lights according to the weak competitive points.

[0088] A840: Use the optimization strategy set to compensate for the co-decision making of the second dredging goal, the short-term driving goal, and the long-term driving goal.

[0089] In the embodiments of the present application, the competitive association clustering result refers to the result obtained by clustering regions with competitive relationships in the region set. The weak competitive point refers to the region position that is relatively disadvantaged or prone to becoming a traffic bottleneck during the traffic resource competition process. The optimization strategy set refers to a set of strategy combinations formulated for weak competitive points to improve traffic conditions. The co-decision making compensation refers to the process of using the optimization strategy set generated for weak competitive points to adjust and supplement the original co-decision making scheme.

[0090] Specifically, the competitive association clustering results are formed by applying the hierarchical clustering algorithm. First, comprehensively collect data such as traffic flow, road capacity, and traffic demand in each region and organize them into a format that can be processed by the algorithm. Then, calculate the similarity between regions based on these data. For numerical data, methods such as Euclidean distance are used to measure the differences, and for non-numerical data, specific matching algorithms are used for calculation. Initially, each region is regarded as an independent cluster, and then, based on the similarity, the two clusters with the highest similarity are continuously merged. For example, the clusters of regions A and B with the highest comprehensive similarity in various aspects of traffic are merged into a new cluster. Continue this merging process until the predetermined number of clusters is reached or the stop condition is satisfied. The final cluster set obtained is the competitive association clustering result.

[0091] Next, the competitive association clustering results present the classification of competitive relationships between different regions. For these clustering results, combined with the current real-time traffic data, such as the real-time traffic flow and vehicle speed changes in each region, deeply analyze the actual state of each region in traffic competition:

[0092] Data collection and integration: Use the multi-source data acquisition sensors in step A200 to collect and establish a regional traffic data set, and integrate these real-time traffic data from multiple sources.

[0093] Comparison benchmark setting: Refer to historical traffic data and traffic planning indicators of each region to set the normal range or ideal values for key indicators such as traffic flow and vehicle speed in each region.

[0094] State evaluation calculation: For each region, compare the real-time traffic flow and vehicle speed data collected with the set benchmark values. Calculate the traffic flow deviation rate, that is, (real-time traffic flow - average value of the normal range of traffic flow) / average value of the normal range of traffic flow × 100%; calculate the vehicle speed deviation rate, that is, (ideal value of vehicle speed - real-time vehicle speed) / ideal value of vehicle speed × 100%.

[0095] Competitive state determination: According to the calculated deviation rates, combined with the traffic association characteristics between regions, determine the traffic competition state. If a certain region has a high traffic flow deviation rate and a large vehicle speed deviation rate, and at the same time there is a close traffic connection between this region and its surrounding regions, such as being in a traffic hub position where a large number of vehicles need to pass through this region to go to other regions, it can be determined that this region is in a high-pressure state in traffic competition and faces great pressure in traffic resource competition.

[0096] In the competitive association clustering, through the competitive state determination, it is found that although the traffic flow of a certain region is not the largest, due to its narrow roads and limited traffic capacity, the vehicle congestion situation is the most serious in the traffic competition with its surrounding regions, and this region is configured as a weak competition point.

[0097] Then, generate specialized signal light optimization strategies for these weak competition points. For example, for an intersection with a weak competition point where the traffic flow is large and congestion is severe, by analyzing the variation law of traffic flow in different directions and applying the signal timing optimization algorithm, optimize the signal timing of the intersection with the weak competition point through the Webster algorithm. First, use devices such as induction coils and cameras to collect data such as the number of vehicles arriving at different lanes of each approach, the length of the red light queue, and the number of vehicles not leaving during the green light within 1 hour during the morning rush hour, and clarify the lane function and maximum vehicle speed. Then, based on the Webster delay formula, combined with the traffic flow of each approach, signal cycle, green light duration, saturation flow, etc., calculate the average delay time in each direction. Subsequently, with the goal of minimizing the total delay, use methods such as the Lagrange multiplier method to take the partial derivative of the green signal ratio in each direction and set it to 0, and solve the system of equations to obtain the optimal green signal ratio. Then, calculate the optimal cycle length according to the Webster optimal cycle length formula, and determine the green light duration in each direction in combination with the optimal green signal ratio. Finally, according to the calculation results, implement strategies such as extending the green light time in a specific direction and reasonably adjusting the phase difference to improve the road traffic capacity in the area of the weak competition point.

[0098] Finally, when making collaborative decisions, considering the complexity of the actual traffic situation, there may be cases where the original collaborative decision-making plan cannot fully meet the traffic demand in some aspects. At this time, the set of signal light optimization strategies plays the role of compensating for collaborative decisions. For example, the original collaborative decision-making plan is to guide some vehicles to avoid congested areas, but during the implementation process, it is found that due to the unreasonable setting of signal lights at some intersections, vehicles are still congested on the new driving route. At this time, use the signal light optimization strategies generated for weak competition points to adjust and compensate the original collaborative decision-making plan, such as adjusting the signal timing at relevant intersections according to the optimization strategy to make the vehicles drive more smoothly.

[0099] By performing competitive association clustering and analysis on the regional set, configuring weak competition points and generating a set of optimization strategies for compensation, so as to better achieve the short-term and long-term driving goals of vehicles while meeting the second dredging goal, and improve the overall effect of multi-regional collaborative traffic management.

[0100] Furthermore, step A500 in the method provided by the embodiments of the present application includes:

[0101] A530: Trigger a traffic flow tracking instruction, where the traffic flow tracking instruction is used to continuously collect traffic flow data after the collaborative decision result is executed, and establish a traffic flow tracking data set.

[0102] A540: Perform a decision feedback evaluation of the collaborative decision result according to the traffic flow tracking data set.

[0103] A550: Generate a real-time optimization plan based on the decision feedback evaluation, and perform traffic decision management through the real-time optimization plan.

[0104] In the embodiments of the present application, the traffic flow tracking instruction refers to an instruction signal used to initiate traffic flow data collection work, which takes effect immediately after the collaborative decision result is executed. The traffic flow tracking data set refers to a set composed of traffic flow-related data continuously collected by devices such as sensors, cameras, and in-vehicle terminals distributed in various traffic regions. The decision feedback evaluation refers to a comprehensive evaluation process of the effects generated by the collaborative decision result in actual traffic operation. The real-time optimization plan refers to targeted improvement measures generated based on the decision feedback evaluation results. Traffic decision management refers to a series of activities covering the formulation of collaborative decision results to the adjustment and optimization of decisions according to actual situations.

[0105] Specifically, when the plan of the collaborative decision result starts to be implemented, the traffic management system's control center sends traffic flow tracking instructions to devices such as sensors, cameras, and in-vehicle terminals distributed in various traffic regions. After receiving the instructions, these devices start to continuously collect traffic flow data, including detailed information such as traffic volume, vehicle speed, and vehicle driving direction on different roads and at different times. The collected data accumulates continuously and gradually forms a traffic flow tracking data set.

[0106] Then, by means of data mining and clustering analysis techniques, such as the specific steps of A600 - A800, the data in the traffic flow tracking data set is deeply mined and analyzed. For example, compare the traffic volume changes in the same area and at the same time before and after the execution of the collaborative decision result. If, after the decision is executed, the traffic volume on the originally congested road section decreases significantly and the vehicle speed increases significantly, it indicates that the decision has achieved positive results in alleviating traffic congestion; on the contrary, if the traffic volume has not improved significantly or even increased and the vehicle speed remains slow, then the rationality of the decision needs to be reflected. At the same time, the transfer of traffic flow between different regions can also be analyzed to determine whether the decision has achieved a reasonable distribution of traffic flow among multiple regions. Through the comprehensive analysis of these data, the collaborative decision result is scored and evaluated from multiple dimensions to form a comprehensive and objective decision feedback evaluation report. This report clearly points out the advantages and disadvantages of the collaborative decision result, providing a clear direction for the next step of optimization.

[0107] Finally, the real-time optimization plan is a targeted improvement measure formulated based on the decision feedback evaluation results. If the decision feedback evaluation shows that the traffic flow in some areas is still unbalanced, or the congestion problem on some sections has not been effectively solved, then the traffic management department and the traffic management system will generate corresponding real-time optimization plans according to the specific problems in the evaluation report. For example, for an area with excessive traffic flow, the signal timing plan may be adjusted to extend the green light time to improve the vehicle passing efficiency; for the unreasonable traffic flow diversion, the vehicle driving route guidance strategy may be re-planned. After generating the real-time optimization plan, it is immediately applied to the actual traffic decision management. Through the control platform of the traffic management system, instructions are sent to relevant traffic facilities and vehicles to implement these optimization measures. At the same time, continuously monitor the changes in traffic flow, and continuously repeat the above processes of traffic flow tracking, decision feedback evaluation, and real-time optimization to form a closed-loop traffic decision management system, ensuring that the traffic management system is always in an efficient and orderly operation state.

[0108] Furthermore, the method provided by the embodiment of the present application further includes step A700, where the step A700 includes:

[0109] A710: Establish a congestion database for the area set, and the congestion database has a time period identifier.

[0110] A720: Perform congestion prediction according to the congestion database, and perform congestion pre-emptive guidance management based on the congestion prediction result.

[0111] In the embodiment of the present application, the congestion database refers to a database specifically used to store information related to traffic congestion. The time period identifier refers to the time mark added to the traffic congestion information recorded in the congestion database. Congestion prediction refers to the process of combining historical congestion data with real-time traffic data to estimate the possible traffic congestion conditions in a specific future time period and area. The congestion prediction result refers to the possible traffic congestion situation in a specific future time period and area. Congestion pre-emptive guidance management refers to the management method of implementing a series of measures in advance to relieve or avoid congestion before it is predicted that traffic congestion may occur.

[0112] In one embodiment, first, a congestion database for the area set is established. The establishment process relies on the widely distributed multi-source data collection devices in the traffic management system. These devices continuously collect traffic data of different regions at different times, including key information such as traffic volume, vehicle speed, and road occupancy. The raw data is analyzed and screened to identify traffic congestion situations and record them in the congestion database. For example, when the vehicle speed on a certain road section continuously drops below a certain threshold (such as 20 kilometers per hour), and the traffic volume exceeds a certain proportion (such as 80%) of the normal carrying capacity of this road section, it is determined that this road section is in a congested state. These congestion information are sorted and classified according to regions and occurrence times and stored in the congestion database. For example, during the morning rush hour (7:00 - 9:00) on weekdays, the main road X in area A of the city is congested, and the congestion duration is 30 minutes. The relevant detailed information such as traffic volume and vehicle speed will be recorded in this database and marked with the corresponding time period and region.

[0113] Next, congestion prediction is performed based on the congestion database. Using the time series analysis algorithm, which is based on the changing pattern of historical congestion data over time, a mathematical model is established to predict the congestion probability and degree in future time periods:

[0114] First, extract the historical traffic congestion data from the congestion database, including information such as whether congestion occurs, the degree of congestion (such as mild, moderate, severe congestion), and congestion duration in different regions at different times.

[0115] Then, clean the extracted data to remove outliers and noise points. For example, some congestion durations with extremely unreasonable records (such as obvious error data exceeding 24 hours) need to be corrected or deleted. Then, perform stationary processing on the data because the time series analysis algorithm usually requires the data to be stationary. The non-stationary time series can be transformed into a stationary sequence through difference operations (such as first-order difference, second-order difference). If there are seasonal fluctuations in the data, seasonal decomposition methods can also be used to separate the trend, seasonal, and residual components.

[0116] Next, select a suitable time series analysis model, such as ARIMA (Autoregressive Integrated Moving Average Model). Determine the orders p (autoregressive order), d (differencing order), and q (moving average order) of the ARIMA model according to the autocorrelation function (ACF) and partial autocorrelation function (PACF) of the data. Then, estimate the model parameters using methods such as maximum likelihood estimation.

[0117] After that, a part of the historical data is used as the training set to train the selected time series model, and the parameters are continuously adjusted to minimize the loss function (such as the mean square error MSE). Then, another part of the data is used as the validation set to evaluate the prediction performance of the model. The accuracy of the model can be judged by calculating indicators such as the mean absolute error (MAE) and the root mean square error (RMSE). If the model performance does not meet the requirements, return to the previous step to adjust the model order or reselect the model.

[0118] Finally, the latest historical congestion data is input into the trained and validated time series model to predict the traffic congestion conditions in each region during specific future periods (such as the next 1 hour, 2 hours, etc.), and the prediction results are output, including the probability of congestion occurrence, the possible degree of congestion, etc.

[0119] According to the congestion prediction results, the traffic management system formulates diversion strategies in advance. For areas where congestion is predicted to occur, the signal timing is adjusted in advance to increase the green light duration to improve the vehicle passing efficiency; traffic warning information is released to drivers through channels such as traffic broadcasts and electronic displays to guide vehicles to choose other routes to detour in advance, so as to effectively relieve traffic pressure before congestion forms.

[0120] Through the pre-congestion diversion management, effective intervention can be carried out when traffic congestion has not yet formed or is in its infancy, thereby improving the operation efficiency of the entire traffic management system, reducing the time waste and energy consumption caused by congestion, and improving the traffic service quality.

[0121] In summary, the AI traffic decision-making management method for multi-region collaboration provided by the embodiments of the present application has the following technical effects:

[0122] In the traffic scenario of multi-region collaboration, the present application collects traffic data in each region, obtains data related to the traffic operation state through data mining, clustering analysis, algorithm operations, etc., calculates information such as the traffic flow change trend and congestion situation, and adjusts the traffic decision-making management strategy by combining the traffic characteristic markers and calculation results in each region, achieving the technical effect of improving urban traffic efficiency.

[0123] Embodiment 2, an AI traffic decision-making management system for multi-region collaboration, as Figure 2 shown, the system includes:

[0124] A regional set establishment module 1, which is used to establish a regional set that can be linked and coordinated, and deploy multi-source data collection sensors within the regional set. The multi-source data collection sensors include cameras and radars.

[0125] Traffic data recognition module 2, which is used to recognize traffic data in the area by using the multi-source data acquisition sensor, establish a regional traffic data set, and recognize the traffic status through the edge computing node integrated in the area to establish an edge traffic status recognition result.

[0126] Short-term goal establishment module 3, which is used to establish the short-term driving goal of the vehicle according to the regional traffic data set.

[0127] Long-term goal establishment module 4, which is used to obtain the navigation plan of the vehicle under the condition of obtaining the owner's permission, and establish the long-term driving goal according to the navigation plan.

[0128] Decision result generation module 5, which is used to synchronize the edge traffic status recognition result, the short-term driving goal, and the long-term driving goal to the collaborative cloud platform, perform collaborative decision-making through the collaborative cloud platform to generate a collaborative decision result, and perform traffic decision management according to the collaborative decision result.

[0129] Further, the decision result generation module 5 is used to perform the following steps:

[0130] Establish a general busy evaluation threshold, perform busy screening on the edge traffic status recognition result through the general busy evaluation threshold, and establish a true busy status identifier for traffic.

[0131] Construct a first dredging goal based on the true busy status identifier, perform collaborative decision-making based on the short-term driving goal and the long-term driving goal by using the first dredging goal, and establish a collaborative decision result.

[0132] Further, the decision result generation module 5 is used to perform the following steps:

[0133] Call the path fitting channel to perform path additional fitting on the short-term driving goal and the long-term driving goal of each vehicle under the first dredging goal, and establish a path additional fitting result.

[0134] Call the time fitting channel to perform passing time fitting on each vehicle under the first dredging goal, and establish a time saving fitting result.

[0135] Obtain the balance preference of the user mapped to the vehicle, and perform collaborative decision-making according to the balance preference, the path additional fitting result, and the time saving fitting result to establish a collaborative decision result.

[0136] Further, the decision result generation module 5 is used to perform the following steps:

[0137] Perform regional collaborative association clustering on the set of regions to establish collaborative clustering clusters, where each sub-cluster in the collaborative clustering clusters can participate in repeated clustering.

[0138] Perform comparison of traffic states within the clusters on the collaborative clustering clusters according to the regional traffic data set to establish a second dredging target.

[0139] Perform collaborative decision-making according to the second dredging target, the short-term driving target, and the long-term driving target.

[0140] Furthermore, the decision result generation module 5 is used to execute the following steps:

[0141] Perform regional competitive association clustering on the set of regions to establish a competitive association clustering result.

[0142] Perform real-time competitive state analysis of the regions according to the competitive association clustering result and configure weak competitive points.

[0143] Generate an optimization strategy set for traffic lights according to the weak competitive points.

[0144] Use the optimization strategy set to compensate for the collaborative decision-making of the second dredging target, the short-term driving target, and the long-term driving target.

[0145] Furthermore, the decision result generation module 5 is used to execute the following steps:

[0146] Trigger a traffic flow tracking instruction, which is used to continuously collect traffic flow data after the collaborative decision result is executed to establish a traffic flow tracking data set.

[0147] Perform decision feedback evaluation of the collaborative decision result according to the traffic flow tracking data set.

[0148] Generate a real-time optimization plan according to the decision feedback evaluation and perform traffic decision management through the real-time optimization plan.

[0149] Furthermore, the decision result generation module 5 is used to execute the following steps:

[0150] Establish a congestion database for the set of regions, and the congestion database has time period identifiers.

[0151] Perform congestion prediction according to the congestion database and perform congestion pre-dredging management based on the congestion prediction result.

[0152] The AI traffic decision management system for multi-region collaboration provided by the embodiments of the present invention can execute the AI traffic decision management method for multi-region collaboration provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0153] Although various references are made to certain modules in the system according to embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The various units and modules included are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0154] The above specific embodiments do not constitute a limitation to the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present application shall be included within the protection scope of the present application. In some cases, the actions or steps recited in the present application can be executed in a different order than in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An AI traffic decision management method for multi-regional collaboration, characterized in that: The method comprises: Establishing a set of regions that can be linked and coordinated, and deploying multi-source data acquisition sensors in the set of regions, wherein the multi-source data acquisition sensors include cameras and radars; Using the multi-source data acquisition sensor to identify traffic data in the area, establish a regional traffic data set, and identify traffic status through edge computing nodes integrated in the area to establish edge traffic status identification results; Establishing a short-term driving target for the vehicle based on the regional traffic dataset; With the permission of the vehicle owner, obtaining the navigation plan of the vehicle, and establishing a long-term driving goal according to the navigation plan; The edge traffic state recognition result, the short-term driving goal, and the long-term driving goal are synchronized to the collaborative cloud platform, collaborative decision-making is performed through the collaborative cloud platform, collaborative decision results are generated, and traffic decision management is performed according to the collaborative decision results.

2. The AI ​​traffic decision management method for multi-regional collaboration according to claim 1, characterized in that: The collaborative decision-making through the collaborative cloud platform includes: Establishing a universal busy evaluation threshold, performing busy screening of the edge traffic state recognition result by using the universal busy evaluation threshold, and establishing a true busy state identification of the traffic; A first unblocking target is constructed based on the true busy state identifier, and the first unblocking target is used to make a collaborative decision based on the short-term driving target and the long-term driving target to establish a collaborative decision result.

3. The AI ​​traffic decision management method for multi-regional collaboration as claimed in claim 2, characterized in that: The using the first dredging target to make a collaborative decision based on the short-term driving target and the long-term driving target includes: Calling a route fitting channel to perform additional path fitting of the short-term driving target and the long-term driving target of each vehicle under the first dredging target, and establishing an additional path fitting result; Calling the time fitting channel to fit the travel time of each vehicle under the first dredging target, and establishing a time saving fitting result; The balance preference of the user mapped with the vehicle is obtained, a collaborative decision is made according to the balance preference, the path addition fitting result, and the time saving fitting result, and a collaborative decision result is established.

4. The AI ​​traffic decision management method for multi-regional collaboration according to claim 1, characterized in that: The collaborative decision making through the collaborative cloud platform also includes: Performing regional collaborative association clustering on the region set to establish a collaborative clustering cluster, wherein each subcluster in the collaborative clustering cluster can participate in repeated clustering; Comparing the traffic status within the collaborative cluster according to the regional traffic data set to establish a second dredging target; A collaborative decision is made based on the second dredging target, the short-term driving target, and the long-term driving target.

5. The AI ​​traffic decision management method for multi-regional collaboration according to claim 4, characterized in that: The collaborative decision-making according to the second dredging target, the short-term driving target, and the long-term driving target includes: Performing regional competition association clustering on the region set to establish a competition association clustering result; Performing real-time competition status analysis of the region according to the competition association clustering results, and configuring weak competition points; Generate an optimization strategy set of traffic lights according to the weak competition points; The optimization strategy set is used to compensate for the collaborative decision-making of the second dredging target, the short-term driving target, and the long-term driving target.

6. The AI ​​traffic decision management method for multi-regional collaboration according to claim 1, characterized in that: The traffic decision management according to the collaborative decision result includes: triggering a traffic tracking instruction, wherein the traffic tracking instruction is used to continuously collect traffic flow data after the collaborative decision result is executed and establish a traffic tracking data set; Performing decision feedback evaluation of the collaborative decision-making result according to the traffic tracking data set; A real-time optimization plan is generated according to the decision feedback evaluation, and traffic decision management is performed through the real-time optimization plan.

7. The AI ​​traffic decision management method for multi-regional collaboration according to claim 1, characterized in that: After the traffic decision management is performed according to the collaborative decision result, the method includes: Establishing a congestion database of a set of regions, wherein the congestion database has a time period identifier; Congestion prediction is performed according to the congestion database, and pre-congestion diversion management is performed based on the congestion prediction result.

8. AI traffic decision management system for multi-regional collaboration, characterized by: For implementing the AI ​​traffic decision management method for multi-regional collaboration according to any one of claims 1 to 7, the system comprises: A region set establishment module, the region set establishment module is used to establish a region set that can be linked and coordinated, and to deploy multi-source data acquisition sensors in the region set, the multi-source data acquisition sensors including cameras and radars; A traffic data identification module, the traffic data identification module is used to use the multi-source data acquisition sensor to identify traffic data in the area, establish a regional traffic data set, and identify traffic status through edge computing nodes integrated in the area to establish edge traffic status identification results; A short-term goal establishing module, the short-term goal establishing module is used to establish a short-term driving goal of the vehicle according to the regional traffic data set; A long-term goal establishing module, the long-term goal establishing module is used to obtain the navigation plan of the vehicle with the permission of the vehicle owner, and establish a long-term driving goal according to the navigation plan; A decision result generation module, the decision result generation module is used to synchronize the edge traffic state recognition result, the short-term driving goal, and the long-term driving goal to the collaborative cloud platform, perform collaborative decision-making through the collaborative cloud platform, generate collaborative decision results, and perform traffic decision management according to the collaborative decision results.

Images