Intelligent traffic light control method and control device

Abstract

This invention relates to the field of traffic control technology, and particularly to an intelligent traffic light control method and device. The intelligent traffic light control method provided by this invention, when traffic conditions result in congestion in multiple directions, recursively performs congestion detection on the intersections before and after the target intersection. If the detection result is no congestion, the green light time for the uncongested forward or backward direction is increased through an optimization algorithm. By recursively performing congestion detection on the intersections before and after the target intersection, the set of congested intersections related to the target intersection is traversed. The green light time is increased through the optimization algorithm, quickly identifying and alleviating congestion points in the traffic network, preventing congestion from spreading, and gradually solving the entire congested traffic network through a divide-and-conquer approach. This yields the optimal solution for vehicle passage in all directions at all congested intersections, avoiding local optima traps, improving the utilization rate of green light time resources, and solving the global congestion problem.

Description

[Technical Field] This invention relates to the field of traffic control technology, and in particular to an intelligent traffic light control method and control device. [Background Technology] Existing traffic light control methods primarily focus on optimizing timing at individual intersections, neglecting the impact of such optimization on other intersections. However, in real-world traffic networks, congestion at intersections is often interconnected. Based on dynamic game theory and congestion propagation mechanisms, optimizing a local intersection may reduce the efficiency of the overall traffic network. For example, congestion at one intersection can lead to vehicle backlogs at upstream or downstream intersections, further affecting more distant intersections. Therefore, optimizing a single intersection is merely a stopgap measure, causing resource imbalances and failing to address the root cause of global congestion. [Summary of the Invention] To address the aforementioned problems, this invention provides an intelligent traffic light control method and control device.

[0001] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an intelligent traffic light control method, comprising the following steps: collecting image information of a target traffic intersection, obtaining traffic condition parameters based on the image information; judging the traffic situation result based on the traffic condition parameters, and executing a corresponding traffic light control method according to the traffic situation result; if the traffic situation result is multi-directional congestion, recursively performing congestion detection on the traffic intersections before and after the target traffic intersection; if the detection result is no congestion, increasing the green light time for the non-congested forward or backward direction through an optimization algorithm.

[0002] Preferably, if the traffic situation result indicates multi-directional congestion, recursively performing congestion detection on the traffic intersections before and after the target traffic intersection, and if the detection result indicates no congestion, increasing the green light time for the non-congested forward or backward direction through an optimization algorithm specifically includes the following steps: obtaining the traffic situation results of the adjacent traffic intersections of the target traffic intersection along the forward and / or backward directions, and determining whether they are congested; if so, taking the adjacent traffic intersection as the new target traffic intersection, and repeating the above steps; if not, repeating the optimization algorithm for the target traffic intersection and backtracking to the previous target traffic intersection, and when backtracking to the initial target traffic intersection, performing the optimization algorithm on the initial target traffic intersection and ending the backtracking.

[0003] Preferably, the optimization algorithm specifically includes the following steps: Defining congestion Construct the objective function for total congestion. Where n represents the total number of target traffic intersections, i represents the i-th target traffic intersection, and d represents the d-th direction of the target traffic intersection. This represents the green light time for the d-th direction at the i-th target traffic intersection. Represent the vehicle arrival rate in the d-th direction at the i-th target traffic intersection; introduce constraints. Where T represents the total green light time at the target traffic intersection, and a Lagrange function is constructed. ,in Constraints Lagrange daily numbers; for Taking the partial derivative yields ;Will Substitute constraints After normalization, we get ;in, Let represent the optimal solution for the green light time in the d-th direction at the i-th target traffic intersection.

[0004] Preferably, when the target traffic intersection includes both straight-ahead and left-turn directions, the green light time for the left-turn direction at the i-th target traffic intersection is defined as follows: The green light time for the straight-ahead direction at the i-th target traffic intersection is... ,when hour, ,when hour, ;in This represents the vehicle arrival rate in the left-turn direction at the i-th target traffic intersection. This represents the vehicle arrival rate in the straight-ahead direction at the i-th target traffic intersection.

[0005] Preferably, when based on the coordinated control of multiple target traffic intersections, a traffic flow propagation model is constructed. ,in, The attenuation coefficient is given, and the average travel time from the i-th target intersection to the (i+1)-th target intersection is τ; constraints are introduced. ;in, This represents the local reach rate of the target traffic intersection i. Represent the global reachability of the target traffic intersection i; construct an augmented Lagrangian function, wherein the augmented Lagrangian function is... Alternately perform local optimization and global coordination to iteratively update the green light time of each target traffic intersection. and The process continues until a preset convergence condition is met, where local optimization refers to fixing the target traffic intersections. Update based on the augmented Lagrange function and Global coordination refers to updating based on the traffic flow propagation model. .

[0006] Preferably, the formula for local optimization is: , ;in The penalty parameter is dynamically adjusted.

[0007] Preferably, the method of judging the traffic situation result based on the traffic condition parameters and executing the corresponding traffic light control according to the traffic situation result further includes: judging whether there is a special situation in the traffic situation result; if not, executing the driving timing optimization method; if so, judging whether the preceding vehicle of the special situation vehicle has a chance to give way; if so, keeping the traffic light state unchanged; if not, adjusting the traffic light to green and issuing a prompt message.

[0008] Preferably, the method for optimizing vehicle timing specifically includes the following steps: determining whether the target traffic intersection is congested; if not, executing a pedestrian waiting time optimization method; if yes, determining whether the traffic situation result indicates multi-directional congestion; if not, increasing the green light time for the congested direction; if yes, determining whether the traffic intersections adjacent to the target traffic intersection are congested; if yes, recursively performing congestion detection on the traffic intersections in the direction forward and / or backward of the target traffic intersection; if the detection result is no congestion, recursively backtracking and executing the optimization algorithm.

[0009] Preferably, the step of obtaining traffic condition parameters based on the image information specifically includes the following steps: processing the image information based on a preset image processing method; identifying traffic condition parameter information in the image information based on a pre-trained deep learning model, wherein the traffic condition parameter information includes at least one of vehicle type, license plate, and color; the congestion degree of each direction of the target traffic intersection; and at least one of the following: the number of pedestrians, whether pedestrians are waiting to cross the road, and the waiting time of pedestrians.

[0010] To address the aforementioned technical problems, this invention provides another technical solution as follows: a control device for implementing any of the above-mentioned intelligent traffic light control methods. The control device includes the following modules: a screen information acquisition module for acquiring screen information of the target traffic intersection; an image processing and analysis module for obtaining traffic condition parameters based on the screen information; a judgment and execution module for judging the traffic situation result based on the traffic condition parameters and executing the corresponding traffic light control method according to the traffic situation result; and a recursive module for recursively performing congestion detection on the traffic intersections before and after the target traffic intersection when the traffic situation result indicates multi-directional congestion, and increasing the green light time for the non-congested forward or backward direction through an optimization algorithm if the detection result indicates no congestion.

[0011] Compared with the prior art, the intelligent traffic light control method and control device provided by the present invention have the following beneficial effects: 1. This invention provides an intelligent traffic light control method. The method includes first collecting image information of a target traffic intersection, obtaining traffic condition parameters based on the image information, judging the traffic situation result based on the traffic condition parameters, and executing a corresponding traffic light control method according to the traffic situation result. When the traffic situation result indicates multi-directional congestion, congestion detection is recursively performed on the traffic intersections before and after the target intersection. If the detection result is no congestion, the green light time for the non-congested forward or backward direction is increased through an optimization algorithm. Since traffic congestion is complex when the traffic situation result indicates multi-directional congestion, optimization of a single traffic intersection cannot solve the global congestion problem; it can only achieve a local optimum and cannot solve the congestion problem of the entire congested traffic network. By recursively performing congestion detection on the traffic intersections before and after the target intersection, traversing the set of congested traffic intersections associated with the target intersection, and increasing the green light time for the non-congested forward or backward direction through an optimization algorithm, congestion points in the traffic network can be quickly identified and alleviated, preventing congestion from spreading. This divide-and-conquer approach gradually solves the entire congested traffic network. When the detection result indicates no congestion, the green light time for the uncongested forward or backward directions is increased through an optimization algorithm. This proactively cuts off congestion propagation paths. Recursively executing congestion detection gives the control logic a layered expansion capability. If the direct forward or backward intersections of the target intersection are not congested, the green light time for the uncongested forward or backward directions can be directly increased through the optimization algorithm. If the directly related intersections are also congested, the system recursively checks upstream or downstream intersections until an adjustable non-congested node is found, and then the green light time is increased through the optimization algorithm. This layer-by-layer in-depth investigation method allows the system to dynamically adjust the control granularity according to the actual congestion range. This avoids both a one-size-fits-all, coarse-grained control approach and the short-sighted problem of focusing only on adjacent intersections while ignoring the impact on more distant areas, thus enhancing the strategy adaptability under complex traffic networks. By recursively breaking down the traffic light timing optimization problem at multiple congested intersections into several similar sub-problems, traffic light control is upgraded from local optimization at a single node to global collaborative control of the traffic network. Through recursive detection and dynamic resource allocation at associated intersections, the optimal solution for vehicle passage in all directions at all congested intersections is finally obtained from the local to the global perspective. This avoids falling into local optima traps, improves the utilization rate of green light time resources, solves the global congestion problem, and improves the overall traffic efficiency and stability of the traffic network.

[0012] 2. In this embodiment of the invention, if the traffic situation result indicates multi-directional congestion, the traffic situation results of adjacent traffic intersections of the target traffic intersection are obtained along the forward and / or backward directions, and it is determined whether they are congested. If congested, the adjacent traffic intersections are used as new target traffic intersections for repeated congestion determination. If not congested, it means that all congested traffic intersections have been found. The optimization algorithm is repeatedly executed for the target traffic intersections and backtracked to the previous target traffic intersection until all target traffic intersections are optimized and the backtracking ends. After obtaining all congested traffic intersections recursively, the green light time is first adjusted for individual traffic intersections based on the current congestion situation. Then, the traffic light timing of the entire congested traffic network is gradually adjusted through recursive backtracking to gradually alleviate the congestion of the entire traffic network, thereby ultimately forming an overall signal light timing optimization scheme for multiple congested traffic intersections. By recursively assessing the congestion status of adjacent intersections at the target intersection and triggering backtracking optimization, this approach avoids the chain reaction of congestion that might occur when optimizing a single intersection. It enhances regional traffic coordination, optimizes the allocation of green light times at multiple congested intersections, and improves overall traffic flow efficiency. By using forward and / or backward directions to traverse all congested intersections and setting the point where the intersection is no longer congested as the termination condition for recursion, it avoids endless recursion.

[0013] 3. Formulas provided in the embodiments of the present invention middle, Reflecting traffic flow pressure, It reflects the supply of travel time. This indicates that congestion requires increasing green light time. Objective function To ensure the total congestion is minimized, the constraints are as follows: To ensure a constant total green light time, the Lagrange multiplier method is used to calculate the partial derivatives of the objective function, which strictly satisfies the constraints and ensures the mathematical optimality of the solution. By calculating the partial derivatives and normalizing, the optimal solution for the green light time is obtained. This achieves non-linear fair allocation. It is a square root function with a gradually slowing growth rate. Compared with linear allocation, the square root allocation increases the green light time of high arrival rate directions more gradually. By reducing the marginal gain of high arrival rate directions, it balances the efficiency of each direction and avoids excessive concentration of green light time. Low arrival rate directions can also obtain a certain amount of green light time, avoiding being completely squeezed out by high arrival rate directions. This allows high arrival rate directions to still obtain more green light time, but without excessive tilting, thus balancing the green light time allocation and traffic efficiency overall.

[0014] 4. In this embodiment of the invention, when the target traffic intersection includes both straight-ahead and left-turn directions, when... , This allows for the allocation of green light time according to the square root ratio, minimizing congestion. When When left-turn traffic is too high, the left-turn time is forcibly limited to not exceeding the straight-ahead time. Since straight-ahead vehicles typically have a larger volume and a more significant impact on traffic efficiency, ensuring the priority of straight-ahead traffic reduces overall congestion caused by left-turning vehicles excessively occupying green light time. This allows for adjusting green light times based on real-time traffic flow, granting more time to high-volume directions, while also implementing priority control. The straight-ahead priority strategy avoids excessively long left-turn times and reduces congestion on main roads.

[0015] 5. In this embodiment of the invention, when based on the coordinated control of multiple target traffic intersections, a traffic flow propagation model is constructed. The green light time at the upstream intersection Vehicle reach rate at downstream intersections Dynamic correlation: the arrival rate at downstream intersections is directly affected by the green light time and vehicle arrival rate at upstream intersections, and this effect increases with time delay. The propagation mechanism accurately reflects the spatiotemporal propagation process of vehicles from upstream release to downstream arrival, avoiding downstream congestion caused by local optimization of a single target traffic intersection. For example, prolonged upstream release of through traffic can overload downstream intersections. This mechanism forcibly considers the mutual influence between intersections, achieving collaborative optimization of multiple congested traffic intersections. By introducing... It characterizes traffic flow losses in real traffic, such as vehicle lane changes, road congestion, or diversion, making the traffic flow propagation model more closely resemble real-world scenarios. This represents the local reach rate of the target traffic intersection i, i.e., the vehicle throughput demand corresponding to that intersection itself. This represents the global reachability of the target traffic intersection i, including vehicle throughput demand propagated from upstream intersections. This is achieved by setting... This ensures that the total arrival rate of locally generated and upstream propagated traffic at a single target intersection matches its actual processing capacity, forming a consistent constraint to coordinate upstream and downstream traffic flow and avoid downstream congestion caused by local over-optimization. By constructing an augmented Lagrangian function, the original objective function seeking to minimize total congestion is combined with the coupling constraint function of the traffic flow propagation model. The second term in the formula... This is a penalty term used to force downstream arrival rates to match upstream green light times, thereby decomposing the complex multi-intersection optimization problem into local subproblems where each intersection is independently optimized and global coordination is used to update the global arrival rate, reducing computational complexity. This is achieved by alternately optimizing local variables. and and global variables The algorithm gradually converges to the global optimal solution that satisfies all constraints, ensuring stable termination and outputting a feasible and efficient traffic light timing optimization scheme.

[0016] 6. In this embodiment of the invention, the local optimization formula balances the congestion level of a single target traffic intersection with the downstream traffic flow pressure by introducing a penalty parameter β, thereby reducing the downstream congestion state through... and Feedback to the upstream optimization process affects the green light time allocation of upstream target traffic intersections, enabling coordinated control of multiple congested target traffic intersections and preventing a single target traffic intersection from only considering its own optimization and causing negative impacts on other target traffic intersections, thus preventing chain congestion.

[0017] 7. In this embodiment of the invention, the processing mechanism for judging traffic situation results based on traffic condition parameters and executing corresponding traffic light control methods according to the traffic situation results includes a special situation vehicle handling method and a traffic timing optimization method. The special situation vehicle handling method has a higher priority than the traffic timing optimization method, which ensures that special situation vehicles can quickly pass through the intersection when a special situation occurs. The traffic timing optimization method will only be executed normally when there is no special situation. By judging the yield opportunity of preceding vehicles or directly adjusting the green light, priority is given to the passage of emergency vehicles, improving the emergency response capability of the traffic system, reducing the risk of traffic paralysis caused by the congestion of special situation vehicles, and avoiding delays in handling special situations.

[0018] 8. In this embodiment of the invention, differentiated timing optimization strategies are implemented based on whether the target traffic intersection is congested and whether the congestion is in one or multiple directions, taking into account both vehicle and pedestrian needs. When congestion is detected at the target traffic intersection, priority is given to alleviating vehicle traffic pressure. If the congestion is in multiple directions, recursive optimization is triggered, linking all adjacent congested intersections to achieve coordinated allocation of green light times in the area. This serves as the optimal solution considering the passage of vehicles in all directions at all intersections within the set of congested intersections, preventing the spread of local congestion and improving the overall traffic network efficiency. When the target traffic intersection is clear, the system performs pedestrian waiting time optimization, dynamically shortening pedestrian red light time to improve pedestrian passage efficiency and safety.

[0019] 9. The method for obtaining traffic condition parameters based on image information provided in this embodiment of the invention specifically includes image processing of image information based on a preset image processing method; and identification of traffic condition parameter information in the image information based on a pre-trained deep learning model. The traffic condition parameter information includes at least one of vehicle type, license plate, and color; congestion levels in each direction of the target intersection; and at least one of the following: number of pedestrians, whether pedestrians are waiting to cross the road, and pedestrian waiting time. Image processing of image information based on the preset image processing method can reduce image noise and improve image quality. When cropping the analysis region of the image information, unnecessary computation can be reduced. When using a pre-trained deep learning model for identification, it can accurately identify multi-dimensional traffic condition parameters such as vehicle type, license plate, color, congestion levels in each direction of the target intersection, and pedestrian status, providing high-precision, real-time data support for subsequent decision-making and enhancing the system's perception capabilities and decision reliability.

[0020] 10. The present invention also provides a control device that has the same beneficial effects as the above-described intelligent traffic light control method, and will not be described in detail here. [Attached Image Description] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of the steps of the intelligent traffic light control method provided in the first embodiment of the present invention.

[0022] Figure 2 This is a flowchart of step S3 of the intelligent traffic light control method provided in the first embodiment of the present invention.

[0023] Figure 3 This is a schematic diagram of the green light time allocation for all directions in the intelligent traffic light control method provided in the first embodiment of the present invention.

[0024] Figure 4 This is a flowchart of step S2 of the intelligent traffic light control method provided in the first embodiment of the present invention.

[0025] Figure 5 This is a flowchart of step S22 of the intelligent traffic light control method provided in the first embodiment of the present invention.

[0026] Figure 6 This is a flowchart of step S1 of the intelligent traffic light control method provided in the first embodiment of the present invention.

[0027] Figure 7 This is a structural block diagram of the control device provided in the second embodiment of the present invention.

[0028] Explanation of reference numerals in the attached diagram: 100. Control device; 101. Image information acquisition module; 102. Image processing and analysis module; 103. Judgment and execution module; 104. Recursion module.

Detailed Implementation Methods

[0029] Please see Figure 1 The first embodiment of the present invention provides an intelligent traffic light control method, comprising the following steps: Step S1: Collect video information of the target traffic intersection and obtain traffic condition parameters based on the video information; Step S2: Determine the traffic situation result based on traffic condition parameters, and execute the corresponding traffic light control method according to the traffic situation result; Step S3: If the traffic situation result is multi-directional congestion, recursively perform congestion detection on the traffic intersections before and after the target traffic intersection. If the detection result is no congestion, increase the green light time for the non-congested forward or backward direction through the optimization algorithm.

[0030] Understandably, given the complex nature of traffic congestion in multiple directions, optimizing a single intersection cannot solve the global congestion problem; it can only achieve a local optimum and cannot resolve the congestion across the entire traffic network. This solution addresses this by recursively performing congestion detection on intersections before and after the target intersection, traversing the set of congested intersections associated with the target intersection, and then using an optimization algorithm to increase the green light time in the uncongested directions. This allows for the rapid identification and mitigation of congestion points within the traffic network, preventing congestion from spreading, and gradually resolving the overall congestion through a divide-and-conquer approach. When the detection result indicates no congestion, the green light time for the uncongested forward or backward directions is increased through an optimization algorithm. This proactively cuts off congestion propagation paths. Recursively executing congestion detection gives the control logic a layered expansion capability. If the direct forward or backward intersections of the target intersection are not congested, the green light time for the uncongested forward or backward directions can be directly increased through the optimization algorithm. If the directly related intersections are also congested, the system recursively checks upstream or downstream intersections until an adjustable non-congested node is found, and then the green light time is increased through the optimization algorithm. This layer-by-layer in-depth investigation method allows the system to dynamically adjust the control granularity according to the actual congestion range. This avoids both a one-size-fits-all, coarse-grained control approach and the short-sighted problem of focusing only on adjacent intersections while ignoring the impact on more distant areas, thus enhancing the strategy adaptability under complex traffic networks. By recursively breaking down the traffic light timing optimization problem at multiple congested intersections into several similar sub-problems, traffic light control is upgraded from local optimization at a single node to global collaborative control of the traffic network. Through recursive detection and dynamic resource allocation at associated intersections, the optimal solution for vehicle passage in all directions at all congested intersections is finally obtained from the local to the global perspective. This avoids falling into local optima traps, improves the utilization rate of green light time resources, solves the global congestion problem, and improves the overall traffic efficiency and stability of the traffic network.

[0031] It should be noted that the target traffic intersection can be any type of intersection equipped with traffic lights, such as a T-junction, crossroads, or star-shaped intersection. "Forward" refers to the direction of traffic flow, and "backward" refers to the direction opposite to the direction of traffic flow. As an optional implementation, step S3 specifically involves calling a recursive program along the direction of traffic flow for the target traffic intersection A to detect the congestion situation at traffic intersection B. If traffic intersection B is found to be congested, further detection of traffic intersections C, D, and E reveals that they are all congested. When traffic intersection F is detected and found to be uncongested, the recursive backtracking condition is triggered, and the process returns to the congested traffic intersections E, D, C, B, and A in sequence, executing optimization algorithms to increase green light time and optimize the traffic light timing at each intersection. Understandably, in this scenario, the forward uncongested direction is from A to E, and increasing the green light time for the forward uncongested direction means sequentially increasing the green light time at each congested intersection along the congestion detection direction; the backward uncongested direction is from E to A, and increasing the green light time for the backward uncongested direction means sequentially increasing the green light time at each congested intersection along the opposite direction of congestion detection. Other implementation methods are not limited here.

[0032] Optionally, traffic condition parameters may include traffic flow, vehicle speed, congestion index, and time-space occupancy; traffic situation results may include special situations, pedestrian waiting, congestion at the target intersection, one-way congestion at the target intersection, and multi-way congestion at the target intersection.

[0033] Please see Figure 2 Furthermore, step S3 specifically includes the following steps: Step S31: Obtain the traffic situation results of the adjacent traffic intersections of the target traffic intersection along the forward and / or backward directions, and determine whether there is congestion; Step S32: If yes, take the adjacent traffic intersection as the new target traffic intersection and repeat the above steps; Step S33: If not, repeat the optimization algorithm for the target traffic intersection and backtrack to the previous target traffic intersection. When backtracking to the initial target traffic intersection, execute the optimization algorithm for the initial target traffic intersection and end the backtracking.

[0034] Understandably, after recursively identifying all congested intersections, the system first optimizes individual intersections by adjusting green light times based on current congestion levels. Then, through recursive backtracking, it gradually adjusts the traffic light timings across the entire congested traffic network, progressively alleviating congestion and ultimately creating an overall optimized signal timing scheme for multiple congested intersections. Repeatedly executing the optimization algorithm on the target intersection and backtracking to the previous target intersection means that after executing the optimization algorithm on the current target intersection and backtracking to the previous target intersection, the optimization algorithm and backtracking are repeated on the previous target intersection. By recursively judging the congestion status of adjacent intersections and triggering backtracking optimization, the system avoids the chain reaction of congestion that might occur with optimizing a single intersection, enhances regional traffic coordination, optimizes the allocation of green light times across multiple congested intersections, and improves the overall traffic flow efficiency. Use forward and / or backward directions to traverse all congested traffic intersections, and use the cessation of congestion at the intersection as the termination condition for the recursion to avoid endless recursion.

[0035] It should be noted that the above process is an iterative and continuous adjustment process. After the longest traffic light cycle, the congestion situation of all traffic intersections will be reassessed, and the next round of optimization will begin.

[0036] Furthermore, the optimization algorithm includes: Defining congestion Construct the objective function for total congestion. Where n represents the total number of target traffic intersections, i represents the i-th target traffic intersection, and d represents the d-th direction of the target traffic intersection. This represents the green light duration for the d-th direction at the i-th target traffic intersection. This represents the vehicle arrival rate in the d-th direction at the i-th target traffic intersection; Introducing constraints Where T represents the total green light time at the target traffic intersection, construct the Lagrange function. ,in Constraints Lagrange daily numbers; right Taking the partial derivative yields ; Will Substitute constraints After normalization, we get ;in, This represents the optimal solution for the green light time in the d-th direction at the i-th target traffic intersection.

[0037] Understandably, the formula middle, Reflecting traffic flow pressure, it indicates the number of vehicles passing through a target traffic intersection per unit of time. It reflects the supply of travel time. This indicates that congestion requires increasing green light time. Objective function To ensure the total congestion is minimized, the constraints are as follows: To ensure a constant total green light time, the Lagrange multiplier method is used to calculate the partial derivatives of the objective function, which strictly satisfies the constraints and ensures the mathematical optimality of the solution. By calculating the partial derivatives and normalizing, the optimal solution for the green light time is obtained. This achieves non-linear fair allocation. It is a square root function with a gradually slowing growth rate. Compared with linear allocation, the square root allocation increases the green light time of high arrival rate directions more gradually. By reducing the marginal gain of high arrival rate directions, it balances the efficiency of each direction and avoids excessive concentration of green light time. Low arrival rate directions can also obtain a certain amount of green light time, avoiding being completely squeezed out by high arrival rate directions. This allows high arrival rate directions to still obtain more green light time, but without excessive tilting, thus balancing the green light time allocation and traffic efficiency overall.

[0038] Optionally, as a specific implementation, the optimization algorithm is performed within a local road network covering a finite number of target traffic intersections. Optionally, this finite number of target traffic intersections can be a 5×5 local road network. Here, "5×5" is an example representing a local road network covering a finite number of target traffic intersections, containing 25 intersections in 5 rows and 5 columns. 3×4, 4×5, or other grids can also be used, the purpose being to control the computational load of each local road network. This is because if all intersections in the entire city are optimized together, the computational load would be extremely large, requiring the computer to process massive amounts of data, resulting in slow computation speeds and even the inability to complete the task in real time. Furthermore, global optimization requires considering the mutual influence of all intersections, making it difficult to find the optimal solution. By limiting the size of the local road network where the optimization algorithm is applied, a divide-and-conquer approach can be achieved, reducing the computational load and allowing each local optimization to be completed quickly. Therefore, the overall congestion objective function constructed by the optimization algorithm... In reality, it's about the total congestion of the local road network. Based on the total congestion of each local road network, the total green light time for each local road network is allocated at the city-wide level, with more congested local road networks receiving more total green light time. By macroscopically allocating this global parameter of total green light time for each local road network, the priorities of each local road network are coordinated, preventing local optimization from disrupting the global balance. Then, based on the total green light time of the local road network, an optimization algorithm is recursively executed to allocate the total green light time T for each target traffic intersection, obtaining the optimal combination of green light times for each target traffic intersection, thereby improving overall traffic efficiency.

[0039] Understandably, by adopting a divide-and-conquer strategy, the complex global problem is broken down into multiple simple local problems, reducing computational difficulty. Based on the total congestion of each local road network, the total green light time for each local road network is allocated at a macro level. The internal optimization of each local road network determines the combination of green light times for each target traffic intersection within that local road network. The combination of these two approaches achieves hierarchical optimization, balancing optimization efficiency and effectiveness, improving optimization timeliness and stability, and lowering the computational resource requirements and server load.

[0040] Please see Figure 3 Furthermore, when the target traffic intersection includes both straight-ahead and left-turn directions, the green light time for the left-turn direction at the i-th target traffic intersection is defined as follows: The green light time for the straight-ahead direction at the i-th target intersection is ,when hour, ,when hour, ;in This represents the vehicle arrival rate in the left-turn direction at the i-th target traffic intersection. This represents the vehicle arrival rate in the straight-ahead direction at the i-th target traffic intersection.

[0041] It should be noted that the total signal cycle duration for each target traffic intersection is T, which needs to be allocated to all directions. When the target traffic intersection is divided into two groups, the straight-ahead group... : Responsible for east-west or north-south straight traffic; left-turn group : Responsible for left turns from east / west or north / south. Therefore, for east / west directions, , Total green light time for east-west directions =T / 2, the same applies to north-south directions. Figure 3 One intersection displays all east-west traffic directions, and the other intersection displays all north-south traffic directions. By default, right turns are made directly and are not included in the traffic light green light time allocation.

[0042] Understandably, when the target traffic intersection includes both straight-ahead and left-turn directions, when , This allows for the allocation of green light time according to the square root ratio, minimizing congestion. When When left-turn traffic is too high, the left-turn time is forcibly limited to not exceeding the straight-ahead time. Since straight-ahead vehicles typically have a larger volume and a more significant impact on traffic efficiency, ensuring the priority of straight-ahead traffic reduces overall congestion caused by left-turning vehicles excessively occupying green light time. This allows for adjusting green light times based on real-time traffic flow, granting more time to high-volume directions, while also implementing priority control. The straight-ahead priority strategy avoids excessively long left-turn times and reduces congestion on main roads.

[0043] Furthermore, when based on the coordinated control of multiple target traffic intersections, a traffic flow propagation model is constructed. ,in, The attenuation coefficient is τ, and the average travel time for a vehicle from the i-th target traffic intersection to reach the (i+1)-th target traffic intersection is τ. Introducing constraints ;in, This represents the local reach rate of the target traffic intersection i. This represents the global reachability of the target traffic intersection i; Construct the augmented Lagrangian function, which is as follows: ; Alternately perform local optimization and global coordination, iteratively updating the green light time at each target traffic intersection. and The process continues until the preset convergence condition is met, where local optimization refers to fixing the target traffic intersections. Update based on augmented Lagrange function and Global coordination refers to updating traffic flow propagation models. .

[0044] Understandably, when constructing a traffic flow propagation model based on the coordinated control of multiple target traffic intersections, this is necessary. The green light time at the upstream intersection Vehicle reach rate at downstream intersections Dynamic correlation: the arrival rate at downstream intersections is directly affected by the green light time and vehicle arrival rate at upstream intersections, and this effect increases with time delay. The propagation mechanism accurately reflects the spatiotemporal propagation process of vehicles from upstream release to downstream arrival, avoiding downstream congestion caused by local optimization of a single target traffic intersection. For example, prolonged upstream release of through traffic can overload downstream intersections. This mechanism forcibly considers the mutual influence between intersections, achieving collaborative optimization of multiple congested traffic intersections. By introducing... It characterizes traffic flow losses in real traffic, such as vehicle lane changes, road congestion, or diversion, making the traffic flow propagation model more closely resemble real-world scenarios. This represents the local reach rate of the target traffic intersection i, i.e., the vehicle throughput demand corresponding to that intersection itself. This represents the global reachability of the target traffic intersection i, including vehicle throughput demand propagated from upstream intersections. This is achieved by setting... This ensures that the total arrival rate of a single target traffic intersection, generated locally and propagated upstream, matches its actual processing capacity, forming a consistent constraint to coordinate upstream and downstream traffic flow and avoid downstream congestion caused by local over-optimization. An augmented Lagrangian function is constructed to combine the original objective function, which seeks to minimize total congestion, with the coupling constraint function of the traffic flow propagation model. In the formula, N represents the number of all congested target traffic intersections, and the second term... This is a penalty term used to force downstream arrival rates to match upstream green light times, thereby decomposing the complex multi-intersection optimization problem into local subproblems where each intersection is independently optimized and global coordination for updating the global arrival rate, reducing computational complexity. This is achieved by alternately optimizing local variables. and and global variables By balancing local and global requirements, the algorithm gradually converges to approach the global optimal solution that satisfies all constraints, ensuring stable termination and outputting a feasible and efficient traffic light timing optimization scheme.

[0045] It should be noted that the augmented Lagrangian function consists of the following two parts: the original congestion minimization term: This represents the summation of congestion in the straight-ahead (S) and left-turn (L) directions over all target intersections, along with the penalty term for traffic flow propagation constraints. This is used to force the downstream intersection arrival rate and the upstream green light time to satisfy the propagation model, where The penalty coefficient controls the strictness of traffic flow propagation constraints. When performing local optimization, the arrival rate of the current downstream traffic intersection is input. , Based on the previous iteration, for each intersection i, a local optimum is found using the augmented Lagrangian function, and the updated green light time is output. and When performing global coordination to update the arrival rates at downstream traffic intersections, input the current green light time at the upstream traffic intersection. and arrival rate Recalculate downstream arrival rate based on traffic flow propagation model The updated This is used for the next round of local optimization.

[0046] Furthermore, the formula for local optimization is: , ;in The penalty parameter is dynamically adjusted.

[0047] Understandably, the local optimization formula balances the congestion level of a single target traffic intersection with the downstream traffic flow pressure by introducing a penalty parameter β, thereby mitigating the downstream congestion state. and Feedback to the upstream optimization process affects the green light time allocation of upstream target traffic intersections, enabling coordinated control of multiple congested target traffic intersections and preventing a single target traffic intersection from only considering its own optimization and causing negative impacts on other target traffic intersections, thus preventing chain congestion.

[0048] Please see Figure 4 Furthermore, step S2, which determines the traffic situation result based on traffic condition parameters and executes the corresponding traffic light control method according to the traffic situation result, also includes: Step S21: Determine if there are any special circumstances in the traffic situation results; Step S22: If not, execute the driving timing optimization method; Step S23: If yes, determine whether the preceding vehicle of the emergency vehicle has a chance to yield. If yes, keep the traffic light status unchanged. If no, adjust the traffic light to green and issue a prompt message.

[0049] Understandably, the traffic situation assessment mechanism, which determines the traffic situation based on traffic condition parameters and then executes corresponding traffic light control methods accordingly, includes emergency vehicle handling methods and traffic timing optimization methods. The emergency vehicle handling methods have a higher priority than the traffic timing optimization methods. This ensures that emergency vehicles can quickly pass through intersections when emergencies occur, while the traffic timing optimization methods are only executed normally when there are no emergencies. By determining the yield opportunity for preceding vehicles or directly adjusting the green light, priority is given to ensuring the passage of emergency vehicles, improving the emergency response capability of the traffic system, reducing the risk of traffic paralysis caused by emergency vehicle congestion, and avoiding delays in handling emergency situations.

[0050] As an alternative implementation, the presence of a special situation in the traffic situation result can be identified by a pre-trained deep learning model through license plate, vehicle type, color, etc., or it can be based on V2I communication, where the vehicle in distress actively sends vehicle information such as license plate number, color, and vehicle type to the system through in-vehicle communication equipment, navigation software, and 110 emergency call platform. The system analyzes and processes the collected information and sends the data to the server through the network, and receives confirmation information returned by the server, thereby determining that there is a special situation in the traffic situation result.

[0051] Please see Figure 5 Furthermore, step S22, which involves optimizing the driving timing method, specifically includes the following steps: Step S221: Determine whether the target traffic intersection is congested. If not, execute the pedestrian waiting time optimization method. If yes, determine whether the traffic situation result is multi-directional congestion. Step S222: If not, increase the green light time for the congested direction; if yes, determine whether the traffic intersections adjacent to the target traffic intersection are congested. Step S223: If yes, recursively perform congestion detection on the traffic intersections in front of and / or behind the target traffic intersection. If the detection result is no congestion, recursively backtrack and execute the optimization algorithm.

[0052] Understandably, differentiated timing optimization strategies are implemented based on whether the target intersection is congested and whether the congestion is unidirectional or multidirectional, taking into account both vehicle and pedestrian needs. When congestion is detected at the target intersection, priority is given to alleviating vehicle traffic pressure. If the congestion is multidirectional, recursive optimization is triggered, starting step S3, which coordinates with all adjacent congested intersections to achieve coordinated allocation of green light times in the area. This serves as the optimal solution for all directions of vehicle passage at all intersections within the congested intersection set, preventing the spread of local congestion and improving the overall traffic network efficiency. When the target intersection is clear, the system optimizes pedestrian waiting time, dynamically shortening pedestrian red light times to improve pedestrian crossing efficiency and safety.

[0053] As an optional implementation method, the pedestrian waiting time optimization method is as follows: determine whether there are pedestrians waiting at the target traffic intersection; if so, reduce the green light time for vehicles obstructing pedestrian crossing; otherwise, keep the traffic light status unchanged. Under non-congested conditions, by detecting pedestrian waiting status, the green light time for vehicles obstructing pedestrian crossing is dynamically shortened, balancing the green light passage time for corresponding vehicles and pedestrians, improving the overall traffic efficiency of the traffic intersection, and providing convenience for pedestrians to cross the traffic intersection without affecting congestion.

[0054] Please see Figure 6 Furthermore, step S1, which involves obtaining traffic condition parameters based on the image information, specifically includes the following steps: Step S11: Perform image processing on the image information based on a preset image processing method; Step S12: Based on the pre-trained deep learning model, identify traffic condition parameters in the image information. The traffic condition parameters include at least one of the following: vehicle type, license plate, and color; congestion level in each direction of the target intersection; and at least one of the following: number of pedestrians, whether pedestrians are waiting to cross the road, and pedestrian waiting time.

[0055] Understandably, processing image information using pre-defined image processing methods can reduce image noise and improve image quality. Cropping the analysis region of the image information reduces unnecessary computation. When using a pre-trained deep learning model for recognition, it can accurately identify multi-dimensional traffic condition parameters such as vehicle type, license plate, color, congestion levels in all directions of the target intersection, and pedestrian status, providing high-precision, real-time data support for subsequent decision-making and enhancing the system's perception capabilities and decision reliability.

[0056] As an optional implementation, the image processing method involves the server preprocessing the received video feed from the surveillance system. This preprocesses the image to remove noise, improve image quality, and crop the area to be analyzed, reducing unnecessary computation. The image pixel values ​​are then normalized to a uniform range, and key local features such as edges and corners, as well as overall features such as color histograms and texture features, are extracted. Image recognition algorithms such as YOLO, SSD, or Faster R-CNN are used. A classifier is employed to categorize vehicle types and detect and label pedestrian locations in the video. Pedestrian waiting times and behavioral patterns are analyzed to output traffic condition parameters.

[0057] Please see Figure 7 The second embodiment of the present invention provides a control device 100 for implementing the intelligent traffic light control method of the first embodiment of the present invention. The control device 100 includes the following modules: The image information acquisition module 101 is used to acquire image information of the target traffic intersection. Image processing and analysis module 102 is used to obtain traffic condition parameters based on image information; The judgment and execution module 103 is used to judge the traffic situation result based on traffic condition parameters and execute the corresponding traffic light control method according to the traffic situation result; The recursive module 104 is used to recursively perform congestion detection on the traffic intersections before and after the target traffic intersection when the traffic situation result is multi-directional congestion. If the detection result is no congestion, the green light time of the non-congested forward or backward direction is increased through optimization algorithm.

[0058] Understandably, the control device 100 has the same beneficial effects as the aforementioned intelligent traffic light control method, which will not be elaborated here.

[0059] In the embodiments provided by this invention, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0060] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. Those skilled in the art should also recognize that the embodiments described in the specification are optional embodiments, and the actions and modules involved are not necessarily essential to the invention.

[0061] In various embodiments of the present invention, it should be understood that the sequence number of each process does not necessarily imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0062] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It is particularly important to note that each block in a block diagram and / or flowchart, or a combination of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0063] Compared with the prior art, the intelligent traffic light control method and control device provided by the present invention have the following beneficial effects: 1. This invention provides an intelligent traffic light control method. The method includes first collecting image information of a target traffic intersection, obtaining traffic condition parameters based on the image information, judging the traffic situation result based on the traffic condition parameters, and executing a corresponding traffic light control method according to the traffic situation result. When the traffic situation result indicates multi-directional congestion, congestion detection is recursively performed on the traffic intersections before and after the target intersection. If the detection result is no congestion, the green light time for the non-congested forward or backward direction is increased through an optimization algorithm. Since traffic congestion is complex when the traffic situation result indicates multi-directional congestion, optimization of a single traffic intersection cannot solve the global congestion problem; it can only achieve a local optimum and cannot solve the congestion problem of the entire congested traffic network. By recursively performing congestion detection on the traffic intersections before and after the target intersection, traversing the set of congested traffic intersections associated with the target intersection, and increasing the green light time for the non-congested forward or backward direction through an optimization algorithm, congestion points in the traffic network can be quickly identified and alleviated, preventing congestion from spreading. This divide-and-conquer approach gradually solves the entire congested traffic network. When the detection result indicates no congestion, the green light time for the uncongested forward or backward directions is increased through an optimization algorithm. This proactively cuts off congestion propagation paths. Recursively executing congestion detection gives the control logic a layered expansion capability. If the direct forward or backward intersections of the target intersection are not congested, the green light time for the uncongested forward or backward directions can be directly increased through the optimization algorithm. If the directly related intersections are also congested, the system recursively checks upstream or downstream intersections until an adjustable non-congested node is found, and then the green light time is increased through the optimization algorithm. This layer-by-layer in-depth investigation method allows the system to dynamically adjust the control granularity according to the actual congestion range. This avoids both a one-size-fits-all, coarse-grained control approach and the short-sighted problem of focusing only on adjacent intersections while ignoring the impact on more distant areas, thus enhancing the strategy adaptability under complex traffic networks. By recursively breaking down the traffic light timing optimization problem at multiple congested intersections into several similar sub-problems, traffic light control is upgraded from local optimization at a single node to global collaborative control of the traffic network. Through recursive detection and dynamic resource allocation at associated intersections, the optimal solution for vehicle passage in all directions at all congested intersections is finally obtained from the local to the global perspective. This avoids falling into local optima traps, improves the utilization rate of green light time resources, solves the global congestion problem, and improves the overall traffic efficiency and stability of the traffic network.

[0064] 2. In this embodiment of the invention, if the traffic situation result indicates multi-directional congestion, the traffic situation results of adjacent traffic intersections of the target traffic intersection are obtained along the forward and / or backward directions, and it is determined whether they are congested. If congested, the adjacent traffic intersections are used as new target traffic intersections for repeated congestion determination. If not congested, it means that all congested traffic intersections have been found. The optimization algorithm is repeatedly executed for the target traffic intersections and backtracked to the previous target traffic intersection until all target traffic intersections are optimized and the backtracking ends. After obtaining all congested traffic intersections recursively, the green light time is first adjusted for individual traffic intersections based on the current congestion situation. Then, the traffic light timing of the entire congested traffic network is gradually adjusted through recursive backtracking to gradually alleviate the congestion of the entire traffic network, thereby ultimately forming an overall signal light timing optimization scheme for multiple congested traffic intersections. By recursively assessing the congestion status of adjacent intersections at the target intersection and triggering backtracking optimization, this approach avoids the chain reaction of congestion that might occur when optimizing a single intersection. It enhances regional traffic coordination, optimizes the allocation of green light times at multiple congested intersections, and improves overall traffic flow efficiency. By using forward and / or backward directions to traverse all congested intersections and setting the point where the intersection is no longer congested as the termination condition for recursion, it avoids endless recursion.

[0065] 3. Formulas provided in the embodiments of the present invention middle, Reflecting traffic flow pressure, It reflects the supply of travel time. This indicates that congestion requires increasing green light time. Objective function To ensure the total congestion is minimized, the constraints are as follows: To ensure a constant total green light time, the Lagrange multiplier method is used to calculate the partial derivatives of the objective function, which strictly satisfies the constraints and ensures the mathematical optimality of the solution. By calculating the partial derivatives and normalizing, the optimal solution for the green light time is obtained. This achieves non-linear fair allocation. It is a square root function with a gradually slowing growth rate. Compared with linear allocation, the square root allocation increases the green light time of high arrival rate directions more gradually. By reducing the marginal gain of high arrival rate directions, it balances the efficiency of each direction and avoids excessive concentration of green light time. Low arrival rate directions can also obtain a certain amount of green light time, avoiding being completely squeezed out by high arrival rate directions. This allows high arrival rate directions to still obtain more green light time, but without excessive tilting, thus balancing the green light time allocation and traffic efficiency overall.

[0066] 4. In this embodiment of the invention, when the target traffic intersection includes both straight-ahead and left-turn directions, when... , This allows for the allocation of green light time according to the square root ratio, minimizing congestion. When When left-turn traffic is too high, the left-turn time is forcibly limited to not exceeding the straight-ahead time. Since straight-ahead vehicles typically have a larger volume and a more significant impact on traffic efficiency, ensuring the priority of straight-ahead traffic reduces overall congestion caused by left-turning vehicles excessively occupying green light time. This allows for adjusting green light times based on real-time traffic flow, granting more time to high-volume directions, while also implementing priority control. The straight-ahead priority strategy avoids excessively long left-turn times and reduces congestion on main roads.

[0067] 5. In this embodiment of the invention, when based on the coordinated control of multiple target traffic intersections, a traffic flow propagation model is constructed. The green light time at the upstream intersection Vehicle reach rate at downstream intersections Dynamic correlation: the arrival rate at downstream intersections is directly affected by the green light time and vehicle arrival rate at upstream intersections, and this effect increases with time delay. The propagation mechanism accurately reflects the spatiotemporal propagation process of vehicles from upstream release to downstream arrival, avoiding downstream congestion caused by local optimization of a single target traffic intersection. For example, prolonged upstream release of through traffic can overload downstream intersections. This mechanism forcibly considers the mutual influence between intersections, achieving collaborative optimization of multiple congested traffic intersections. By introducing... It characterizes traffic flow losses in real traffic, such as vehicle lane changes, road congestion, or diversion, making the traffic flow propagation model more closely resemble real-world scenarios. This represents the local reach rate of the target traffic intersection i, i.e., the vehicle throughput demand corresponding to that intersection itself. This represents the global reachability of the target traffic intersection i, including vehicle throughput demand propagated from upstream intersections. This is achieved by setting... This ensures that the total arrival rate of locally generated and upstream propagated traffic at a single target intersection matches its actual processing capacity, forming a consistent constraint to coordinate upstream and downstream traffic flow and avoid downstream congestion caused by local over-optimization. By constructing an augmented Lagrangian function, the original objective function seeking to minimize total congestion is combined with the coupling constraint function of the traffic flow propagation model. The second term in the formula... This is a penalty term used to force downstream arrival rates to match upstream green light times, thereby decomposing the complex multi-intersection optimization problem into local subproblems where each intersection is independently optimized and global coordination for updating the global arrival rate, reducing computational complexity. This is achieved by alternately optimizing local variables. and and global variables The algorithm gradually converges to the global optimal solution that satisfies all constraints, ensuring stable termination and outputting a feasible and efficient traffic light timing optimization scheme.

[0068] 6. In this embodiment of the invention, the local optimization formula balances the congestion level of a single target traffic intersection with the downstream traffic flow pressure by introducing a penalty parameter β, thereby reducing the downstream congestion state through... and Feedback to the upstream optimization process affects the green light time allocation of upstream target traffic intersections, enabling coordinated control of multiple congested target traffic intersections and preventing a single target traffic intersection from only considering its own optimization and causing negative impacts on other target traffic intersections, thus preventing chain congestion.

[0069] 7. In this embodiment of the invention, the processing mechanism for judging traffic situation results based on traffic condition parameters and executing corresponding traffic light control methods according to the traffic situation results includes a special situation vehicle handling method and a traffic timing optimization method. The special situation vehicle handling method has a higher priority than the traffic timing optimization method, which ensures that special situation vehicles can quickly pass through the intersection when a special situation occurs. The traffic timing optimization method will only be executed normally when there is no special situation. By judging the yield opportunity of preceding vehicles or directly adjusting the green light, priority is given to the passage of emergency vehicles, improving the emergency response capability of the traffic system, reducing the risk of traffic paralysis caused by the congestion of special situation vehicles, and avoiding delays in handling special situations.

[0070] 8. In this embodiment of the invention, differentiated timing optimization strategies are implemented based on whether the target traffic intersection is congested and whether the congestion is in one or multiple directions, taking into account both vehicle and pedestrian needs. When congestion is detected at the target traffic intersection, priority is given to alleviating vehicle traffic pressure. If the congestion is in multiple directions, recursive optimization is triggered, linking all adjacent congested intersections to achieve coordinated allocation of green light times in the area. This serves as the optimal solution considering the passage of vehicles in all directions at all intersections within the set of congested intersections, preventing the spread of local congestion and improving the overall traffic network efficiency. When the target traffic intersection is clear, the system performs pedestrian waiting time optimization, dynamically shortening pedestrian red light time to improve pedestrian passage efficiency and safety.

[0071] 9. The method for obtaining traffic condition parameters based on image information provided in this embodiment of the invention specifically includes image processing of image information based on a preset image processing method; and identification of traffic condition parameter information in the image information based on a pre-trained deep learning model. The traffic condition parameter information includes at least one of vehicle type, license plate, and color; congestion levels in each direction of the target intersection; and at least one of the following: number of pedestrians, whether pedestrians are waiting to cross the road, and pedestrian waiting time. Image processing of image information based on the preset image processing method can reduce image noise and improve image quality. When cropping the analysis region of the image information, unnecessary computation can be reduced. When using a pre-trained deep learning model for identification, it can accurately identify multi-dimensional traffic condition parameters such as vehicle type, license plate, color, congestion levels in each direction of the target intersection, and pedestrian status, providing high-precision, real-time data support for subsequent decision-making and enhancing the system's perception capabilities and decision reliability.

[0072] 10. The present invention also provides a control device that has the same beneficial effects as the above-described intelligent traffic light control method, and will not be described in detail here.

[0073] The above provides a detailed description of an intelligent traffic light control method and control device disclosed in the embodiments of the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention. Any modifications, equivalent substitutions, and improvements made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling intelligent traffic lights, characterized in that, Includes the following steps: Collect video information from the target traffic intersection and obtain traffic condition parameters based on the video information; Based on the traffic condition parameters, the traffic situation is determined, and the corresponding traffic light control method is executed according to the traffic situation result. If the traffic situation result indicates congestion in multiple directions, congestion detection is recursively performed on the traffic intersections before and after the target traffic intersection. If the detection result indicates no congestion, the green light time for the uncongested forward or backward direction is increased through an optimized algorithm.

2. The intelligent traffic light control method as described in claim 1, characterized in that: If the traffic situation result indicates multi-directional congestion, congestion detection is recursively performed on the traffic intersections before and after the target intersection. If the detection result indicates no congestion, the green light time for the uncongested forward or backward direction is increased through an optimized algorithm, specifically including the following steps: Obtain the traffic situation results of adjacent traffic intersections of the target traffic intersection along the forward and / or backward directions, and determine whether there is congestion; If so, the adjacent traffic intersection is taken as the new target traffic intersection, and the above steps are repeated; If not, repeat the optimization algorithm for the target traffic intersection and backtrack to the previous target traffic intersection. When backtracking to the initial target traffic intersection, execute the optimization algorithm for the initial target traffic intersection and then end the backtracking.

3. The intelligent traffic light control method as described in claim 2, characterized in that: The optimization algorithm includes: Defining congestion Construct the objective function for total congestion. Where n represents the total number of target traffic intersections, i represents the i-th target traffic intersection, and d represents the d-th direction of the target traffic intersection. This represents the green light time for the d-th direction at the i-th target traffic intersection. This represents the vehicle arrival rate in the d-th direction at the i-th target traffic intersection; Introducing constraints Where T represents the total green light time at the target traffic intersection, and a Lagrange function is constructed. ,in Constraints Lagrange daily numbers; right Taking the partial derivative yields ; Will Substitute constraints After normalization, we get ;in, Let represent the optimal solution for the green light time in the d-th direction at the i-th target traffic intersection.

4. The intelligent traffic light control method as described in claim 3, characterized in that: When the target traffic intersection includes both straight-ahead and left-turn directions, the green light time for the left-turn direction at the i-th target traffic intersection is defined as follows: The green light time for the straight-ahead direction at the i-th target traffic intersection is ,when hour, ,when hour, ;in This represents the vehicle arrival rate in the left-turn direction at the i-th target traffic intersection. This represents the vehicle arrival rate in the straight-ahead direction at the i-th target traffic intersection.

5. The intelligent traffic light control method as described in claim 4, characterized in that: When coordinating control based on multiple target traffic intersections, a traffic flow propagation model is constructed. ,in, The attenuation coefficient is τ, and the average travel time for a vehicle from the i-th target traffic intersection to reach the (i+1)-th target traffic intersection is τ. Introducing constraints ;in, This represents the local reach rate of the target traffic intersection i. This represents the global reachability of the target traffic intersection i; Construct an augmented Lagrangian function, wherein the augmented Lagrangian function is: ; Alternately perform local optimization and global coordination to iteratively update the green light time of each target traffic intersection. and The process continues until a preset convergence condition is met, where local optimization refers to fixing the target traffic intersections. Update based on the augmented Lagrange function and Global coordination refers to updating based on the traffic flow propagation model. .

6. The intelligent traffic light control method as described in claim 5, characterized in that: The formula for local optimization is: ， ;in The penalty parameter is dynamically adjusted.

7. The intelligent traffic light control method as described in claim 1, characterized in that: The method of determining the traffic situation result based on the traffic condition parameters and executing the corresponding traffic light control according to the traffic situation result further includes: Determine whether the traffic situation results indicate any special circumstances; If not, implement the driving timing optimization method; If so, determine whether the vehicle preceding the emergency vehicle has a chance to yield. If so, keep the traffic light status unchanged; if not, adjust the traffic light to green and issue a warning message.

8. The intelligent traffic light control method as described in claim 7, characterized in that: The method for optimizing vehicle timing specifically includes the following steps: Determine whether the target traffic intersection is congested. If not, implement the pedestrian waiting time optimization method. If yes, determine whether the traffic situation result indicates multi-directional congestion. If not, increase the green light time for the congested direction; if yes, determine whether the traffic intersections adjacent to the target traffic intersection are congested. If so, recursively perform congestion detection on the traffic intersections in the direction of the target traffic intersection and / or the direction of the traffic intersection. If the detection result is no congestion, recursively backtrack and execute the optimization algorithm.

9. The intelligent traffic light control method as described in claim 1, characterized in that: The process of obtaining traffic condition parameters based on the image information specifically includes the following steps: The image information is processed based on a preset image processing method; Based on a pre-trained deep learning model, traffic condition parameters in the image information are identified. These parameters include at least one of the following: vehicle type, license plate, and color; congestion level in each direction of the target intersection; and at least one of the following: number of pedestrians, whether pedestrians are waiting to cross the road, and pedestrian waiting time.

10. A control device for implementing the intelligent traffic light control method according to any one of claims 1 to 9, characterized in that, The control device includes the following modules: The image information acquisition module is used to acquire image information of the target traffic intersection; The image processing and analysis module is used to obtain traffic condition parameters based on the image information; The judgment and execution module is used to judge the traffic situation result based on the traffic condition parameters, and execute the corresponding traffic light control method according to the traffic situation result; The recursive module is used to recursively perform congestion detection on the traffic intersections before and after the target traffic intersection when the traffic situation result is multi-directional congestion. If the detection result is no congestion, the green light time of the non-congested forward or backward direction is increased through optimization algorithm.

Images