Multi-intersection traffic signal light collaborative control system and method

Abstract

The present application provides a multi-intersection traffic light collaborative control system and method, which relates to the field of signal light control technology. The method includes: constructing an intelligent agent to obtain N signal light control intelligent agents; building a traffic data analysis multi-channel based on a set of traffic influencing factor indicators; collecting N traffic condition data streams and mapping them to the traffic data analysis multi-channel for calculation to obtain N traffic operation indicator parameter sets; performing signal light control analysis on the N traffic operation indicator parameter sets based on the N signal light control intelligent agents; performing balanced collaborative optimization on the N signal light control parameter sets, and performing signal light collaborative control through the N signal light optimization control parameter sets. Through this application, the technical goal of intelligent and adaptive optimization scheduling based on the collaborative control of traffic lights at multiple intersections can be achieved, and the technical effect of optimizing the signal light control strategy can be achieved.

Description

Technical Field

[0001] The present application relates to the technical field of signal light control, and in particular to a collaborative control system and method for multi-intersection traffic light systems. Background Art

[0002] With the acceleration of urbanization, the number of vehicles continues to grow, resulting in frequent fluctuations in traffic flow. The traditional fixed-duration traffic light control method can no longer meet the needs of real-time traffic scheduling.

[0003] Currently, traffic light control technology struggles to achieve efficient traffic flow scheduling optimization in large-scale, multi-intersection coordinated control environments. Existing traffic light control technology has limitations in decision-making optimization. Some traffic light control methods based on traffic flow prediction rely on traditional statistical modeling or simple rule-based decision-making, failing to fully utilize large-scale traffic data for deep learning analysis, resulting in insufficient prediction accuracy.

[0004] In summary, the existing technology has technical problems such as the difficulty of achieving efficient traffic flow scheduling optimization in a large-scale, multi-intersection coordinated control environment due to the difficulty of traffic light control technology, which leads to limitations in decision-making optimization, further affecting the accuracy of deep learning analysis based on large-scale traffic data and the overall traffic flow optimization effect. Summary of the Invention

[0005] The purpose of this application is to provide a multi-intersection traffic light collaborative control system and method to solve the technical problem in the existing technology that traffic light control technology is difficult to achieve efficient traffic flow scheduling optimization in a large-scale, multi-intersection collaborative control environment, resulting in limitations in decision-making optimization, further affecting the accuracy of deep learning analysis based on large-scale traffic data and the overall traffic flow optimization effect.

[0006] In view of the above problems, the present application provides a multi-intersection traffic light collaborative control system and method.

[0007] In the first aspect, the present application provides a multi-intersection traffic signal light collaborative control system for executing a multi-intersection traffic signal light collaborative control method, including: an intelligent agent construction module for constructing an intelligent agent based on the distribution design information of N intersections and a historical traffic signal data set to obtain N signal light control intelligent agents; an influencing factor index acquisition module for obtaining a set of traffic influencing factor indicators, wherein the traffic influencing factor index set includes traffic pressure value, lane queue length, traffic density and weather condition influence; a data analysis multi-channel construction module for constructing a traffic data analysis multi-channel based on the traffic influencing factor index set; a channel mapping module , used to capture and collect N traffic condition data streams through cameras set at the N intersections, map the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtain N traffic operation index parameter sets; a signal light control analysis module, used to perform signal light control analysis on the N traffic operation index parameter sets based on the N signal light control intelligent agents, and obtain N signal light control parameter sets; a signal light optimization control module, used to perform balanced collaborative optimization on the N signal light control parameter sets, determine N signal light optimization control parameter sets, and perform signal light collaborative control through the N signal light optimization control parameter sets.

[0008] In a second aspect, the present application provides a method for collaborative control of traffic lights at multiple intersections, which is implemented through a collaborative control system for traffic lights at multiple intersections, including: constructing an intelligent agent based on the distributed design information and historical traffic signal data sets of N intersections to obtain N signal light control intelligent agents; obtaining a set of traffic influencing factor indicators, wherein the traffic influencing factor indicator set includes traffic pressure value, lane queue length, traffic density and weather condition influence; building a traffic data analysis multi-channel based on the traffic influencing factor indicator set; capturing and collecting N traffic condition data streams through cameras set at the N intersections, mapping the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtaining N traffic operation indicator parameter sets; performing signal light control analysis on the N traffic operation indicator parameter sets based on the N signal light control intelligent agents to obtain N signal light control parameter sets; performing balanced collaborative optimization on the N signal light control parameter sets to determine N signal light optimization control parameter sets, and performing signal light collaborative control through the N signal light optimization control parameter sets.

[0009] The technical solution provided in this application has at least the following technical effects or advantages: by realizing the technical goal of intelligent and adaptive optimization scheduling based on coordinated control of traffic lights at multiple intersections, the technical effects of improving traffic flow efficiency, reducing traffic congestion, optimizing signal light control strategies, and enhancing the system's ability to adapt to complex traffic environments are achieved.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, which can be implemented in accordance with the contents of the description, and to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically listed below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in this application or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely exemplary, and a person of ordinary skill in the art can obtain other drawings based on the provided drawings without creative work.

[0012] Figure 1 This is a schematic diagram of the structure of the multi-intersection traffic signal light coordinated control system of this application;

[0013] Figure 2 This is a flow chart of the collaborative control method for multi-intersection traffic lights in this application.

[0014] Explanation of the accompanying symbols: intelligent body construction module 1, influencing factor index acquisition module 2, data analysis multi-channel construction module 3, channel mapping module 4, traffic light control analysis module 5, traffic light optimization control module 6. DETAILED DESCRIPTION

[0015] This application provides a multi-intersection traffic signal collaborative control system and method to address the existing technical issues of traffic signal control, which are the difficulty of achieving efficient traffic flow scheduling optimization in a large-scale, multi-intersection collaborative control environment. This leads to limitations in traffic signal control decision optimization, further affecting the accuracy of deep learning analysis based on large-scale traffic data and the overall traffic flow optimization effect. This application achieves the technical goal of intelligent, adaptive optimization scheduling based on multi-intersection traffic signal collaborative control, achieving the technical effects of improving traffic flow efficiency, reducing traffic congestion, optimizing signal control strategies, and enhancing the system's ability to adapt to complex traffic environments.

[0016] Below, the technical solutions in this application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this application, rather than all the embodiments of this application. It should be understood that this application is not limited to the example embodiments described herein. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application. It should also be noted that, for the convenience of description, only the parts related to this application, rather than all of them, are shown in the accompanying drawings.

[0017] For example, see the attached Figure 1 , this application provides a multi-intersection traffic signal light coordinated control system, specifically including:

[0018] The intelligent agent construction module 1 is used to construct an intelligent agent based on the distribution design information of N intersections and the historical traffic signal data set to obtain N traffic light control intelligent agents.

[0019] Furthermore, the intelligent agent construction module 1 includes: a spatial modeling unit 11, which is used to perform entity extraction and annotation and three-dimensional space modeling based on the distribution design information of the N intersections to generate N three-dimensional models of intersections; a transfer training unit 12, which is used to construct N multi-analysis task lists according to the traffic light control objectives, and obtain N multi-task traffic light controllers through transfer training according to the N multi-analysis task lists; an initialization fusion unit 13, which is used to embed the N multi-task traffic light controllers into the N intersection three-dimensional models for initialization fusion to generate N initial traffic light intelligent agents; a decision mechanism iterative optimization unit 14, which is used to introduce a reinforcement learning mechanism to iteratively optimize the decision mechanism of the N initial traffic light intelligent agents to obtain N traffic light control intelligent agents.

[0020] Furthermore, the transfer training unit 12 includes: a task arrangement channel 121, which is used to arrange the analysis sequence of each analysis task in the N multi-analysis task list to obtain N multi-analysis task sequence sets; an analyzer matching channel 122, which is used to match and select N source domain traffic task analyzer sets according to the analysis characteristic information of the N multi-analysis task sequence sets; an analyzer merging channel 123, which is used to merge the N source domain traffic task analyzer sets in series according to the multi-analysis task sequence sets to generate N initial task traffic light controllers; a transfer training tuning channel 124, which is used to perform transfer training tuning on the N initial task traffic light controllers respectively to obtain N multi-task traffic light controllers.

[0021] Furthermore, the migration training and tuning channel 124 includes: a migration training port 1241, which is used to collect and obtain the target domain traffic data sets of the N intersections, and perform migration training on the N initial task traffic light controllers based on the target domain traffic data sets of the N intersections to obtain N task migration traffic light controllers; a performance verification port 1242, which is used to perform performance verification and evaluation on the N task migration traffic light controllers to obtain N task controller performance parameters; a particle swarm optimization port 1243, which is used to determine N particle swarm optimization directions based on the N task controller performance parameters; and an iterative update port 1244, which is used to evaluate and iteratively update the N task migration traffic light controllers according to the N particle swarm optimization directions until a preset termination condition is met to obtain the N multi-task traffic light controllers.

[0022] Furthermore, the decision mechanism iterative optimization unit 14 includes: an evaluation index decomposition channel 141, which is used to obtain the traffic light control strategy target, decompose the evaluation index and perform effect evaluation fitting on the traffic light control strategy target, and construct an agent scorer; a strategy execution scoring channel 142, which is used to use the agent scorer to perform strategy execution scoring on the N initial traffic light agents respectively to obtain N agent scoring information; a strategy tuning update channel 143, which is used to introduce a reinforcement learning mechanism to feed back the N agent scoring information to the N initial traffic light agents for strategy tuning and updating until the preset convergence conditions are met to obtain the N traffic light control agents.

[0023] Specifically, the distributed design information for N intersections refers to data such as their geographic locations, surrounding road layouts, and traffic flow. Entity extraction and annotation from this distributed design information involves extracting and annotating entity content, such as intersections, traffic facilities, and roads, by labeling their type and design parameters to obtain the annotated information for each of the N intersections. Using 3D modeling technology, a 3D model of each intersection is created based on the annotated information, generating a total of N 3D intersection models for more accurate simulation and analysis of traffic flow.

[0024] The traffic light control objective is to optimize traffic flow and reduce congestion by adjusting traffic light control strategies. Based on this objective, a list of multiple analysis tasks is constructed. Specifically, tasks to be analyzed are listed based on various traffic influencing factors, such as traffic volume and lane occupancy. Examples include traffic pressure analysis, queue length analysis, and signal duration optimization.

[0025] Arrange the analysis tasks in the N multi-analysis task lists in the order of the signal light control logic to obtain a set of N multi-analysis task sequences. For example, the sequence tasks may be arranged as traffic flow prediction task, traffic congestion prediction task, and signal light control strategy task.

[0026] Next, based on the analytical characteristics of each analysis task in the N multi-analysis task sequence sets, N source-domain traffic task analyzer sets are selected to match and process the task. A source-domain traffic task analyzer set refers to analysis tools or algorithms tailored to different traffic task types that have been trained and demonstrated processing capabilities in similar traffic environments. For example, some analysis tasks may require the use of specific traffic prediction algorithms, while others may be better suited to flow calculation methods.

[0027] Then, the N source domain traffic task analyzer sets are serially merged according to the multi-analysis task sequence set to generate N initial task signal light controllers. That is, according to the arrangement of the task sequence, the selected traffic task analyzers are combined to finally form a preliminary controller.

[0028] Collect and obtain the target domain traffic dataset of N intersections. The target domain traffic dataset refers to the traffic data of N intersections, such as traffic volume, speed, queue length, etc., which is used to provide the actual traffic environment information to the signal light controller.

[0029] Based on the target domain traffic dataset of N intersections, each of the N initial task signal light controllers is transferred and trained. That is, by migrating the training data from the source domain (which may be simulated or historical traffic data) to the target domain (the actual data of the current intersection), N task-migrated signal light controllers are obtained, so that the controllers can better adapt to actual traffic conditions.

[0030] Next, the performance of N task migration signal light controllers is verified and evaluated. That is, actual operation tests are carried out to evaluate their effectiveness in controlling traffic flow, such as whether they can effectively reduce congestion and improve traffic efficiency. The performance parameters of N task controllers are obtained, and the quantitative evaluation of the controller effect is obtained.

[0031] Particle swarm optimization (PSO) is an optimization algorithm based on swarm intelligence that finds the best solution to a problem. Based on the performance parameters of N task controllers, N PSO optimization directions are determined. PSO directions are determined by analyzing the performance parameters of each task controller to identify areas for improving the control strategy and optimize the traffic light's control performance.

[0032] Finally, the N task migration traffic light controllers are evaluated and iteratively updated according to the N particle swarm optimization directions. The task migration traffic light controllers are gradually updated and adjusted until the preset termination condition is met (for example, traffic efficiency is maximized), and the optimal N multi-task traffic light controllers are finally obtained.

[0033] N multi-task signal controllers are embedded into N 3D intersection models for initial fusion. The logic and control strategies of the multi-task signal controllers are integrated into the 3D intersection model, allowing the multi-task signal controllers to be run and tested in a virtual environment, generating N initial signal light agents. Ultimately, these N initial signal light agents are fused to possess signal light control capabilities.

[0034] Next, the signal light control strategy objective refers to the expected effect that signal light control needs to achieve, such as reducing vehicle waiting time, improving road capacity, or reducing traffic congestion. The signal light control strategy objective is obtained and then decomposed into evaluation indicators and fitted with effect evaluation. Evaluation indicator decomposition refers to breaking down the overall signal light control strategy objective into specific metrics, such as average vehicle speed, intersection queue length, or green light utilization rate. Effect evaluation fitting compares actual traffic data with the ideal target to determine the deviation between the current signal light control strategy and the signal light control strategy objective. This then constructs an agent scorer to score the control effects of the N initial signal light agents based on specific evaluation indicators to measure whether they have achieved the expected control objectives.

[0035] Next, the agent scorer is a system used to evaluate the performance of the initial signal light agent. It quantitatively assesses the signal light control strategy implemented by the initial signal light agent and provides an optimization basis for reinforcement learning. The agent scorer establishes evaluation metrics such as travel time, traffic flow, average speed, and vehicle queue length to ensure that the scoring results accurately reflect the strengths and weaknesses of the agent's control strategy. The weights of these evaluation metrics can be adjusted by those skilled in the art based on specific application scenarios. The agent scorer is used to score the strategy execution of each of the N initial signal light agents. Based on their performance in simulated environments or actual traffic, the control strategies of the N initial signal light agents are objectively evaluated, resulting in N agent scores that reflect the control effectiveness of each agent in different environments. For example, if an initial signal light agent successfully reduces vehicle queue length by 30% during peak hours, the agent scorer may assign a higher score. However, if an initial signal light agent causes an unreasonable increase in waiting times, the score may be lowered.

[0036] Reinforcement learning is a machine learning method that optimizes control strategies through continuous experimentation and feedback. A reinforcement learning mechanism is introduced to feed back the scoring information of N agents to N initial signal light agents for policy tuning and updating. These N initial signal light agents continuously conduct signal light control experiments in a three-dimensional simulation environment, learning control strategies that improve traffic efficiency. For example, they try different green light durations or adjust red-green light switching rules. The reinforcement learning mechanism is then introduced to iteratively optimize the decision-making mechanisms of each of the N initial signal light agents, allowing the agents to continuously optimize their control strategies based on the scoring information until a preset convergence condition is met. Ultimately, after multiple rounds of optimization, N signal light control agents are obtained, capable of intelligently adapting to varying traffic conditions and making optimal control decisions. After each scoring cycle, the strategies that resulted in high and low scores are analyzed and adjusted accordingly. The preset convergence condition is a predefined convergence condition set by those skilled in the art based on actual conditions. For example, if the score no longer significantly improves after multiple adjustments, it indicates that the initial signal light agent has found a preferred control strategy, and the optimization process can be terminated.

[0037] The influencing factor index acquisition module 2 is used to obtain a traffic influencing factor index set, where the traffic influencing factor index set includes traffic pressure value, lane queue length, traffic density and weather condition impact.

[0038] Specifically, the traffic influencing factor index set refers to a collection of parameters that can influence traffic flow and efficiency. It is used to describe traffic conditions and provide a basis for decision-making in signal control. By obtaining this set of traffic influencing factor indicators, we can more accurately assess current traffic conditions and adjust signal control strategies to optimize traffic efficiency.

[0039] Traffic pressure is a measure of road load, calculated from factors such as traffic volume, speed, and road capacity. For example, during the morning rush hour, if traffic at an intersection approaches its maximum capacity, the traffic pressure is high and more frequent adjustments to traffic light durations may be necessary to alleviate congestion.

[0040] Lane queue length refers to the physical length of the queue of vehicles waiting at a red light. Long lane queues may indicate that current signal timing is inappropriate and that green light durations need to be optimized. For example, if the average queue length in a particular lane is longer than in other lanes, consider shortening the green light duration in other lanes and increasing the green light duration in that lane to reduce congestion.

[0041] Traffic density refers to the number of vehicles per unit length of road and directly reflects the degree of congestion. High density can cause vehicles to slow down or even stop and wait, while low density indicates a smooth road. For example, high traffic density on a particular road may require adjustments to traffic light timing to reduce vehicle delays.

[0042] Weather conditions affect traffic flow. For example, rain, haze, or snow can reduce driver speeds and increase braking distances, thus impacting traffic flow and capacity. For example, in heavy rain, green light durations can be appropriately extended to compensate for the reduced efficiency caused by slower driving speeds. Table 1 shows the most recent data for the traffic influencing factor indicator set, derived based on actual conditions.

[0043] Table 1: The latest indicator data record

[0044]

[0045] The data analysis multi-channel building module 3 is used to build a traffic data analysis multi-channel based on the traffic influencing factor indicator set.

[0046] Furthermore, the data analysis multi-channel building module 3 includes: a data analysis channel design unit 31, which is used to design multiple data analysis channels based on the traffic influencing factor indicator set, and each channel in the multiple data analysis channels corresponds one-to-one to the traffic influencing factor indicator set; an algorithm configuration unit 32, which is used to configure multiple basic channel indicator algorithms based on the analysis objectives of each channel in the multiple data analysis channels, and test and optimize the multiple basic channel indicator algorithms through a data test set to obtain multiple target channel indicator algorithms; a channel mapping configuration unit 33, which is used to perform channel mapping configuration on the multiple data analysis channels based on the multiple target channel indicator algorithms to build the traffic data analysis multi-channel.

[0047] Specifically, based on a set of traffic influencing factor indicators, multiple data analysis channels are designed to facilitate targeted analysis, with each channel corresponding to a set of traffic influencing factor indicators. For example, the data analysis channel corresponding to traffic pressure values ​​can be used to calculate the saturation level of vehicle throughput capacity; the data analysis channel corresponding to lane queue length can be used to monitor queuing at intersections; the data analysis channel corresponding to traffic density can be used to assess the distribution of vehicles per unit length of road; and the data analysis channel corresponding to weather conditions can be used to analyze the impact of weather changes on road capacity.

[0048] Each of the multiple data analysis channels requires a calculation method tailored to its analysis objective, specifically, configuring a basic channel indicator algorithm. For example, the data analysis channel corresponding to traffic pressure values ​​can use the ratio of flow rate to capacity, the data analysis channel corresponding to lane queue length can use the product of the number of vehicles and the average vehicle length, the data analysis channel corresponding to traffic density can calculate the number of vehicles per unit distance using camera-based object detection technology, and the data analysis channel corresponding to weather conditions can use regression analysis methods combining historical data and real-time monitoring data. Different algorithms are suitable for different types of traffic data, making the calculations of the data analysis channels more accurate.

[0049] Multiple basic channel indicator algorithms are tested and optimized using data test sets. The calculation results of the basic channel indicator algorithms are verified and adjusted using real or simulated traffic datasets, resulting in multiple target channel indicator algorithms. For example, when testing the basic channel indicator algorithm corresponding to traffic pressure values, historical traffic flow data and actual road carrying capacity can be compared to adjust the calculation weights to better align with actual road conditions. When testing the basic channel indicator algorithm corresponding to lane queue length, actual queue conditions captured by video surveillance can be compared with the algorithm-calculated queue length for error analysis and boundary conditions optimization. The calculation of the basic channel indicator corresponding to traffic density can be verified using traffic flow data captured by drones to ensure the robustness of the algorithm. The calculation of the basic channel indicator corresponding to weather conditions can be fitted with traffic flow data from different seasons and weather conditions to make the calculation model more adaptable. Through testing and optimization, the basic algorithm is improved, ultimately forming a target channel indicator algorithm suitable for actual traffic environments.

[0050] Based on multiple target channel indicator algorithms, channel mapping is configured for multiple data analysis channels. The optimized target channel indicator algorithms are matched and integrated with the corresponding data analysis channels to achieve overall collaborative computing, thereby building a multi-channel system for traffic data analysis. For example, the channel corresponding to the traffic pressure value will use the optimized flow calculation algorithm and combine it with the calculation results of the lane queue length channel for joint analysis to accurately assess congestion. The channel corresponding to traffic density can be jointly modeled with the weather condition impact channel to predict the potential impact of extreme weather on traffic flow. If the traffic density at a certain intersection is high and the weather condition impact value is large, the signal light control strategy can be automatically adjusted during the channel mapping configuration process. For example, in high-density areas on rainy days, the green light time can be extended to reduce the safety risks caused by frequent starts.

[0051] The channel mapping module 4 is used to capture and collect N traffic condition data streams through cameras set at the N intersections, map the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtain N traffic operation indicator parameter sets.

[0052] Specifically, cameras installed at N intersections capture N traffic data streams to obtain real-time traffic information. These cameras can be high-definition surveillance cameras or visual sensors with intelligent recognition capabilities. They continuously capture information such as vehicles, pedestrians, and road signs, generating a continuous data stream.

[0053] N traffic data streams are mapped to a multi-channel traffic data analysis system for computation, using different calculation methods to extract valuable information. Each traffic data stream may contain multiple information dimensions, such as vehicle speed, traffic volume, lane occupancy, and signal light response. The multi-channel traffic data analysis system can apply different analysis strategies to different data dimensions, resulting in N sets of traffic operation indicator parameters. For example, a traffic data stream captured by a camera can be simultaneously input into a traffic pressure value analysis channel, a lane queue length analysis channel, a traffic density analysis channel, and a weather impact analysis channel. Each channel performs specific computational tasks to extract different traffic characteristics.

[0054] The signal light control parsing module 5 is configured to perform signal light control parsing on the N traffic operation index parameter sets based on the N signal light control agents to obtain N signal light control parameter sets.

[0055] Specifically, by analyzing N traffic operation parameter sets using N traffic light control agents, the system can automatically adjust the signal light control method based on different traffic conditions, guiding actual signal light regulation and generating corresponding signal light control strategies. This results in N signal light control parameter sets, including, for example, adjusting the duration of green, red, and yellow lights, optimizing signal cycles, and even dynamically adjusting the priorities of different directions.

[0056] The traffic light optimization control module 6 is used to perform balanced collaborative optimization on the N traffic light control parameter sets, determine N traffic light optimization control parameter sets, and perform traffic light collaborative control using the N traffic light optimization control parameter sets.

[0057] Furthermore, the traffic light optimization control module 6 includes: a criticality assessment unit 61, used to perform criticality assessment on the N intersections and determine the criticality of the N intersections; an intersection optimization unit 62, used to perform intersection optimization on the N traffic light control parameter sets according to the criticality of the N intersections to obtain a first traffic light control parameter set; a balanced collaborative correction unit 63, used to perform balanced collaborative correction on the remaining traffic light control parameter sets in the N traffic light control parameter sets based on the first traffic light control parameter set to determine N traffic light optimized control parameter sets.

[0058] Furthermore, the balanced collaborative correction unit 63 includes: a remaining parameter evaluation channel 631, which is used to evaluate the influence of the remaining signal light control parameter sets in the N signal light control parameter sets in turn based on the first signal light control parameter set, and obtain N-1 signal light control influence coefficients; a Nash equilibrium correction channel 632, which is used to perform Nash equilibrium correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control influence coefficients, and obtain N-1 signal light optimized control parameter sets; a control parameter combination channel 633, which is used to combine the first signal light control parameter set and the N-1 signal light optimized control parameter sets to determine the N signal light optimized control parameter sets.

[0059] Furthermore, the Nash equilibrium correction channel 632 includes: a strategy adjustment analysis port 6321, which is used to perform strategy adjustment analysis on the remaining signal light control parameter sets in the N signal light control parameter sets according to the N-1 signal light control influence coefficients, and determine the N-1 signal light control parameter adjustment steps; an iterative optimization correction port 6322, which is used to perform Nash equilibrium calculation and iterative optimization correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control parameter adjustment steps, and obtain the N-1 signal light optimized control parameter sets.

[0060] Specifically, a criticality assessment, or importance analysis, is performed on N intersections to identify those with the greatest impact on overall traffic flow. This criticality assessment is based on multiple traffic indicators, such as traffic volume, traffic pressure, lane queue length, traffic density, and accident rate.

[0061] After determining the criticality of N intersections, N signal light control parameter sets are optimized according to their criticality. Specifically, signal light control strategies are prioritized for intersections with higher criticality. Since the optimization of signal light control at more critical intersections significantly improves the overall traffic network, intersection optimization is performed to ensure smooth overall traffic flow, resulting in a first signal light control parameter set. For example, signal light parameters at intersections with high criticality are prioritized to reduce congestion.

[0062] While ensuring that the signal light control parameter set at the intersection with the highest intersection criticality is optimal, the remaining signal light control parameter sets are controlled in a balanced manner.

[0063] The first signal light control parameter set is used as a reference to evaluate the influence of the remaining signal light control parameter sets. That is, the influence of the changes in the signal light parameters of each intersection on the overall traffic flow is analyzed one by one in a certain order. The influence of other signal light control parameter sets on the basis of this optimization is judged, and N-1 signal light control influence coefficients are obtained, which represent the influence coefficients of all signal light control parameter sets except the first signal light control parameter set.

[0064] According to the N-1 signal light control influence coefficients, the remaining signal light control parameter sets in the N signal light control parameter sets are strategically adjusted and analyzed. That is, the remaining signal light control parameter sets are adjusted and analyzed. The intersection with a larger signal light control influence coefficient is positioned closer to the front, and vice versa. The adjustment pace of the N-1 signal light control parameters is then determined to ensure that they tend to be optimal after adjustment to reduce the impact on downstream traffic.

[0065] Based on the N-1 signal control parameter adjustment steps, a Nash equilibrium calculation is performed on the remaining signal control parameter sets within the N signal control parameter sets. This optimization optimizes the signal parameters to achieve equilibrium across all intersections. This Nash equilibrium calculation ensures that the signal control strategy at each intersection is optimal overall, ensuring that adjusting the signal parameters at any one intersection does not degrade overall system performance. For example, if shortening the signal cycle at one intersection causes congestion at downstream intersections, the adjustment plan needs to be optimized based on overall system equilibrium.

[0066] After performing Nash equilibrium calculation on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control parameter adjustment steps, iterative optimization and correction are performed. That is, through multiple rounds of iteration and correction, all signal light control parameters except the first signal light control parameter set are optimized, and N-1 signal light optimized control parameter sets are obtained.

[0067] The first signal light control parameter set and the N-1 signal light optimization control parameter sets are combined to form a complete set of N signal light optimization control parameter sets. Since the first signal light control parameter set is selected first after the intersection criticality assessment, it has a higher global optimization value. The N-1 signal light optimization control parameter sets are obtained based on Nash equilibrium calculation and iterative optimization correction, so it is necessary to ensure coordination between them when combining. For example, if the green light duration of the first signal light is adjusted to 45 seconds, and the green light duration of a downstream intersection is optimized to 50 seconds, it may cause a mismatch in vehicle arrival time. Therefore, it is necessary to further adjust the combination strategy to ensure smoother connection between signal lights. After the combination, the traffic flow changes are analyzed through simulation to observe whether there are new bottlenecks at certain intersections due to improper parameter combination, to ensure that the signal light control parameters of the entire intersection group do not conflict with each other.

[0068] To sum up, the multi-intersection traffic light collaborative control system provided by this application has the following technical effects: by realizing the technical goal of intelligent and adaptive optimization scheduling based on the collaborative control of multi-intersection traffic lights, it achieves the technical effects of improving traffic flow efficiency, reducing traffic congestion, optimizing signal light control strategies and enhancing the system's ability to adapt to complex traffic environments.

[0069] Example 2: Based on the same inventive concept as the multi-intersection traffic signal light cooperative control system in the above embodiment, this application also provides a multi-intersection traffic signal light cooperative control method, please refer to the attached Figure 2 , including: S1: constructing an intelligent agent based on the distribution design information and historical traffic signal data sets of N intersections to obtain N signal light control intelligent agents; S2: obtaining a set of traffic influencing factor indicators, wherein the traffic influencing factor indicator set includes traffic pressure value, lane queue length, traffic density and weather conditions; S3: building a traffic data analysis multi-channel based on the traffic influencing factor indicator set; S4: capturing and collecting N traffic condition data streams through cameras set on the N intersections, mapping the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtaining N traffic operation indicator parameter sets; S5: performing signal light control analysis on the N traffic operation indicator parameter sets based on the N signal light control intelligent agents to obtain N signal light control parameter sets; S6: performing balanced collaborative optimization on the N signal light control parameter sets, determining N signal light optimization control parameter sets, and performing signal light collaborative control through the N signal light optimization control parameter sets.

[0070] Furthermore, the method for collaborative control of traffic lights at multiple intersections also includes: performing entity extraction and annotation and three-dimensional space modeling based on the distribution design information of the N intersections to generate N three-dimensional models of intersections; constructing N multi-analysis task lists according to the traffic light control objectives, and obtaining N multi-task traffic light controllers through migration training based on the N multi-analysis task lists; embedding the N multi-task traffic light controllers into the N three-dimensional models of the intersections for initialization and fusion to generate N initial traffic light intelligent agents; introducing a reinforcement learning mechanism to iteratively optimize the decision-making mechanism of the N initial traffic light intelligent agents to obtain N traffic light control intelligent agents.

[0071] Furthermore, the method for collaborative control of traffic lights at multiple intersections also includes: arranging the analysis sequences of the analysis tasks in the N multi-analysis task lists to obtain N multi-analysis task sequence sets; matching and selecting N source domain traffic task analyzer sets according to the analysis characteristic information of the N multi-analysis task sequence sets; merging the N source domain traffic task analyzer sets in series according to the multi-analysis task sequence sets to generate N initial task signal light controllers; and performing migration training and optimization on the N initial task signal light controllers to obtain N multi-task signal light controllers.

[0072] Furthermore, the multi-intersection traffic signal light collaborative control method also includes: collecting and obtaining the target domain traffic data sets of the N intersections, and performing migration training on the N initial task signal light controllers based on the target domain traffic data sets of the N intersections to obtain N task migration signal light controllers; performing performance verification and evaluation on the N task migration signal light controllers to obtain N task controller performance parameters; determining N particle swarm optimization directions based on the N task controller performance parameters; and evaluating and iteratively updating the N task migration signal light controllers according to the N particle swarm optimization directions until a preset termination condition is met to obtain the N multi-task signal light controllers.

[0073] Furthermore, the multi-intersection traffic light collaborative control method also includes: obtaining a traffic light control strategy target, performing evaluation index decomposition and effect evaluation fitting on the traffic light control strategy target, and constructing an agent scorer; using the agent scorer to perform strategy execution scoring on the N initial traffic light agents respectively to obtain N agent scoring information; introducing a reinforcement learning mechanism to feed back the N agent scoring information to the N initial traffic light agents for strategy tuning and updating until the preset convergence conditions are met, thereby obtaining the N traffic light control agents.

[0074] Furthermore, the method for collaborative control of traffic lights at multiple intersections also includes: designing multiple data analysis channels based on the traffic influencing factor indicator set, each channel in the multiple data analysis channels corresponds one-to-one to the traffic influencing factor indicator set; configuring multiple basic channel indicator algorithms based on the analysis objectives of each channel in the multiple data analysis channels, and testing and optimizing the multiple basic channel indicator algorithms through a data test set to obtain multiple target channel indicator algorithms; performing channel mapping configuration on the multiple data analysis channels based on the multiple target channel indicator algorithms to build the traffic data analysis multi-channel.

[0075] Furthermore, the method for collaborative control of traffic lights at multiple intersections also includes: performing a criticality assessment on the N intersections to determine the criticality of the N intersections; performing intersection optimization on the N traffic light control parameter sets according to the criticality of the N intersections to obtain a first traffic light control parameter set; and performing balanced collaborative correction on the remaining traffic light control parameter sets in the N traffic light control parameter sets based on the first traffic light control parameter set to determine N optimized traffic light control parameter sets.

[0076] Furthermore, the method for coordinated control of traffic lights at multiple intersections also includes: evaluating the degree of influence of the remaining signal light control parameter sets in the N signal light control parameter sets based on the first signal light control parameter set to obtain N-1 signal light control influence coefficients; performing Nash equilibrium correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control influence coefficients to obtain N-1 signal light optimized control parameter sets; and combining the first signal light control parameter set and the N-1 signal light optimized control parameter sets to determine the N signal light optimized control parameter sets.

[0077] Furthermore, the method for coordinated control of traffic lights at multiple intersections also includes: performing strategy adjustment analysis on the remaining signal light control parameter sets in the N signal light control parameter sets according to the N-1 signal light control influence coefficients to determine the N-1 signal light control parameter adjustment paces; performing Nash equilibrium calculation and iterative optimization correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control parameter adjustment paces to obtain the N-1 signal light optimized control parameter sets.

[0078] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The multi-intersection traffic light collaborative control system and specific examples in the aforementioned embodiment one are also applicable to the multi-intersection traffic light collaborative control method of this embodiment. Through the aforementioned detailed description of the multi-intersection traffic light collaborative control system, those skilled in the art can clearly understand the multi-intersection traffic light collaborative control method of this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0079] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

[0080] Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A multi-intersection traffic light coordinated control system, characterized in that: The system comprises: An agent construction module is used to construct an agent based on the distributed design information of N intersections and the historical traffic signal dataset to obtain N signal light control agents; An influencing factor index acquisition module is used to obtain a set of traffic influencing factor indicators, wherein the set of traffic influencing factor indicators includes traffic pressure value, lane queue length, traffic density and weather condition impact; A data analysis multi-channel building module is used to build a traffic data analysis multi-channel based on the traffic influencing factor indicator set; a channel mapping module, configured to capture and collect N traffic condition data streams through cameras installed at the N intersections, map the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtain N traffic operation indicator parameter sets; a signal light control parsing module, configured to perform signal light control parsing on the N traffic operation index parameter sets based on the N signal light control agents to obtain N signal light control parameter sets; a signal light optimization control module, configured to perform balanced collaborative optimization on the N signal light control parameter sets, determine the N signal light optimization control parameter sets, and perform signal light collaborative control using the N signal light optimization control parameter sets; The data analysis multi-channel building module includes: a data analysis channel design unit, configured to design a plurality of data analysis channels according to the traffic influencing factor indicator set, wherein each of the plurality of data analysis channels corresponds one-to-one to the traffic influencing factor indicator set; an algorithm configuration unit, configured to configure a plurality of basic channel indicator algorithms based on the analysis objectives of each channel in the plurality of data analysis channels, and to test and optimize the plurality of basic channel indicator algorithms using a data test set to obtain a plurality of target channel indicator algorithms; A channel mapping configuration unit is used to perform channel mapping configuration on the multiple data analysis channels based on the multiple target channel indicator algorithms to build the traffic data analysis multi-channel.

2. The multi-intersection traffic signal light coordinated control system according to claim 1, characterized in that: The intelligent agent building module includes: A spatial modeling unit, configured to perform entity extraction and annotation and three-dimensional spatial modeling based on the distribution design information of the N intersections, and generate three-dimensional models of the N intersections; A transfer training unit is used to construct N multi-analysis task lists according to the traffic light control target, and obtain N multi-task traffic light controllers through transfer training according to the N multi-analysis task lists; An initialization fusion unit, configured to embed the N multi-task signal light controllers into the N three-dimensional intersection models for initialization fusion, thereby generating N initial signal light intelligent agents; The decision-making mechanism iterative optimization unit is used to introduce a reinforcement learning mechanism to iteratively optimize the decision-making mechanisms of the N initial traffic light agents to obtain N traffic light control agents.

3. The multi-intersection traffic signal light coordinated control system according to claim 2, characterized in that: The migration training unit includes: A task arrangement channel is used to arrange the analysis tasks in the N multi-analysis task lists into analysis sequences to obtain N multi-analysis task sequence sets; An analyzer matching channel, configured to respectively match and select N source domain traffic task analyzer sets according to the analysis characteristic information of the N multiple analysis task sequence sets; An analyzer merging channel, configured to combine the N source domain traffic task analyzer sets in series according to the multiple analysis task sequence sets to generate N initial task signal light controllers; The migration training and tuning channel is used to perform migration training and tuning on the N initial task traffic light controllers respectively to obtain N multi-task traffic light controllers.

4. The multi-intersection traffic signal light coordinated control system according to claim 3, characterized in that: The migration training and tuning channel includes: A migration training port is used to collect and obtain the target domain traffic data set of the N intersections, and perform migration training on the N initial task signal light controllers based on the target domain traffic data set of the N intersections to obtain N task migration signal light controllers; A performance verification port is used to perform performance verification and evaluation on each of the N task migration signal light controllers to obtain performance parameters of the N task controllers; A particle swarm optimization port, used to determine N particle swarm optimization directions according to the N task controller performance parameters; The iterative update port is used to evaluate and iteratively update the N task migration signal light controllers according to the N particle swarm optimization directions until a preset termination condition is met, thereby obtaining the N multi-task signal light controllers.

5. The multi-intersection traffic signal light coordinated control system according to claim 2, characterized in that: The decision-making mechanism iterative optimization unit includes: An evaluation index decomposition channel is used to obtain the traffic light control strategy target, perform evaluation index decomposition and effect evaluation fitting on the traffic light control strategy target, and construct an intelligent agent scorer; a strategy execution scoring channel, configured to use an agent scorer to perform strategy execution scoring on each of the N initial traffic light agents, thereby obtaining N agent scoring information; A strategy tuning and updating channel is used to introduce a reinforcement learning mechanism to feed back the scoring information of the N agents to the N initial traffic light agents for strategy tuning and updating until a preset convergence condition is met, thereby obtaining the N traffic light control agents.

6. The multi-intersection traffic signal light coordinated control system according to claim 1, characterized in that: The traffic light optimization control module includes: A criticality assessment unit, configured to perform a criticality assessment on the N intersections to determine the criticality of the N intersections; an intersection optimization unit, configured to perform intersection optimization on the N signal light control parameter sets according to the N intersection criticalities to obtain a first signal light control parameter set; A balanced collaborative correction unit is configured to perform balanced collaborative correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the first signal light control parameter set to determine N optimized signal light control parameter sets.

7. The multi-intersection traffic signal light coordinated control system according to claim 6, characterized in that: The balanced collaborative correction unit includes: The remaining parameter evaluation channel is used to sequentially evaluate the influence of the remaining signal light control parameter sets in the N signal light control parameter sets based on the first signal light control parameter set, to obtain N-1 signal light control influence coefficients; a Nash equilibrium correction channel, configured to perform Nash equilibrium correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control influence coefficients, to obtain N-1 signal light optimized control parameter sets; The control parameter combination channel is used to combine the first signal light control parameter set and the N-1 signal light optimization control parameter sets to determine the N signal light optimization control parameter sets.

8. The multi-intersection traffic signal light coordinated control system according to claim 7, characterized in that: The Nash equilibrium correction channel includes: a strategy adjustment analysis port, configured to perform strategy adjustment analysis on the remaining signal light control parameter sets in the N signal light control parameter sets according to the N-1 signal light control influence coefficients, and determine adjustment steps for the N-1 signal light control parameters; The iterative optimization correction port is used to perform Nash equilibrium calculation and iterative optimization correction on the remaining signal light control parameter sets in the N signal light control parameter sets based on the N-1 signal light control parameter adjustment steps to obtain the N-1 signal light optimized control parameter sets.

9. A method for collaboratively controlling traffic lights at multiple intersections, characterized in that: The method is implemented by a multi-intersection traffic signal light cooperative control system according to any one of claims 1 to 8, comprising: Based on the distributed design information of N intersections and the historical traffic signal dataset, an intelligent agent is constructed to obtain N signal light control intelligent agents; Obtaining a set of traffic influencing factor indicators, wherein the set of traffic influencing factor indicators includes traffic pressure value, lane queue length, traffic density, and weather condition impact; Building multiple channels for traffic data analysis based on the set of traffic influencing factor indicators; Capturing and collecting N traffic condition data streams by cameras arranged at the N intersections, mapping the N traffic condition data streams to the traffic data analysis multi-channel for calculation, and obtaining N traffic operation indicator parameter sets; performing signal light control analysis on the N traffic operation index parameter sets based on the N signal light control agents to obtain N signal light control parameter sets; Balanced collaborative optimization is performed on the N signal light control parameter sets to determine N signal light optimized control parameter sets, and signal light collaborative control is performed using the N signal light optimized control parameter sets.

Images