A trunk two-way green wave solving method based on interval strategy search

Abstract

The application discloses a trunk two-way green wave solving method based on interval strategy search, and relates to the field of intelligent traffic, which comprises the following steps: collecting static data and real-time dynamic data of each intersection of a trunk and preprocessing to obtain real-time trunk traffic data; inputting the data into a standard maximum green wave bandwidth algorithm for solving to obtain a maximum green wave solving result, and starting interval strategy search if there is no feasible solution; constructing a constraint interval of each green wave coordination core parameter and discretizing to form a multi-level search space; executing three-level hierarchical search to output a final trunk two-way green wave solving scheme; analyzing the scheme into signal adjustment instructions of each intersection and executing, while monitoring the execution effect and iteratively optimizing the interval search strategy. The application solves the problem that the existing trunk two-way green wave solving method cannot stably obtain an effective green wave scheme, and achieves the effect that a feasible two-way green wave scheme can be stably output when there is no effective solution in the maximum green wave bandwidth algorithm.

Description

Technical Field

[0001] This application relates to the field of intelligent transportation, and in particular to a method for solving bidirectional green wave problems on trunk lines based on interval strategy search. Background Technology

[0002] In urban arterial road traffic signal coordination control, the construction of bidirectional green wave bands plays a fundamental role in improving arterial road traffic efficiency and reducing vehicle delays and stops. Current engineering practice commonly employs maximum bandwidth algorithms based on linear programming to achieve bidirectional green wave coordination. Specifically, this involves using the distance between intersections, average vehicle speed, phase difference, and green light duration as input parameters, and solving a system of linear inequalities to find the optimal timing scheme that maximizes the bidirectional green wave bandwidth. However, such algorithms have inherent limitations in practical applications: when the phase sequence and timing of arterial road intersections are subject to rigid constraints, intersection distances differ significantly, the expected speeds of vehicles in both directions differ greatly, or traffic flow exhibits dynamic fluctuations, the algorithm often outputs no feasible solution due to linear constraint conflicts, or the calculated green wave bandwidth is too small to have practical traffic value, thus failing to generate an effective bidirectional green wave coordination scheme.

[0003] Currently, in related technologies, there is a technical problem that the bidirectional green wave solution for trunk lines cannot reliably obtain an effective green wave scheme. Summary of the Invention

[0004] This application provides a method for solving bidirectional green wave solutions for trunk lines based on interval strategy search. It collects and preprocesses trunk line traffic data, inputs the maximum green wave bandwidth algorithm to determine if it is an unsolvable scenario, and if so, constructs constraint intervals for core parameters and discretizes them into a multi-level search space. Then, it executes a three-level hierarchical search to output the final solution. Finally, it parses the solution into instructions for execution and iteratively optimizes the search strategy based on feedback. These techniques solve the technical problem of existing bidirectional green wave solutions for trunk lines failing to stably obtain effective green wave solutions, achieving the technical effect of stably outputting feasible bidirectional green wave solutions even when the maximum green wave bandwidth algorithm has no effective solution.

[0005] This application provides a method for solving bidirectional green wave traffic on trunk roads based on interval strategy search, including: collecting static data and real-time dynamic data from each intersection of the trunk road and preprocessing them to obtain real-time trunk road traffic data; inputting the real-time trunk road traffic data into a standard maximum green wave bandwidth algorithm for solving to obtain the maximum green wave solution result; if the maximum green wave solution result is a valid solution, then a green wave coordination scheme is directly adopted; if the maximum green wave solution result is an infeasible solution or the green wave bandwidth value is less than a preset valid threshold, then it is determined to be a solutionless scenario requiring interval strategy search; for the solutionless scenario, based on... Based on the real-time arterial traffic data, constraint intervals for each core parameter of green wave coordination are constructed, and these constraint intervals are discretized into a grid to form a multi-level search space. The core parameters of green wave coordination include at least phase difference, green light duration in the arterial direction, and expected vehicle speed in both directions. Based on the multi-level search space, a three-level hierarchical search is performed to output the final solution for the arterial bidirectional green wave. The final solution for the arterial bidirectional green wave is parsed into signal adjustment instructions for each intersection and executed. Simultaneously, the execution effect of the solution is monitored in real time, and the interval search strategy is iteratively optimized based on feedback data.

[0006] In a possible implementation, for the unsolvable scenario, based on the real-time arterial traffic data, a constraint interval for each green wave coordination core parameter is constructed, and the following processing is performed: Based on the intersection spacing and real-time vehicle speed in the real-time arterial traffic data, the theoretical range of phase difference is calculated, and combined with signal timing rules, phase difference values ​​that conflict with phase sequence are removed from the theoretical range of phase difference to obtain the phase difference constraint interval; Based on the minimum green light duration of each intersection in the arterial direction in the real-time arterial traffic data, the maximum extendable duration is determined and limited to not exceeding the green light compression limit of the non-arterial direction to obtain the green light duration constraint interval for the arterial direction; Based on the average speed of the two-way real-time traffic flow in the real-time arterial traffic data, a preset speed fluctuation value is adjusted up and down to obtain the two-way expected speed constraint interval.

[0007] In a possible implementation, the constraint interval is discretized into a grid to form a multi-level search space, and the following processing is performed: the phase difference constraint interval, the trunk direction green light duration constraint interval, and the bidirectional expected vehicle speed constraint interval are discretized into a grid using a three-dimensional interval grid, and basic search nodes are formed with a preset adjustment step size; the trunk start point, trunk end point, and intersections with high traffic congestion are taken as key intersections, and the parameter adjustment range of the key intersections is preferentially limited to form a first-level key intersection layer search space; based on the optimization results of the first-level key intersection layer search space, the parameter adjustment range of the key intersections is extended to the parameter search of all trunk intersections in the basic search nodes to form a second-level trunk overall layer search space.

[0008] In a possible implementation, based on the multi-level search space, a three-level hierarchical search is performed to output the final solution for the bidirectional green wave on the trunk line. The following processing is performed: A first-level search is performed, conducting a search for the maximum green wave bandwidth interval in both the up and down directions. With the goal of maximizing the green wave bandwidth in one direction, the optimal green wave parameter combinations for both the up and down directions are obtained in the first-level key intersection layer search space and the second-level trunk line overall layer search space. The green wave bandwidth, number of vehicle stops, and average delay evaluation index corresponding to each optimal green wave parameter combination are calculated to form a set of optimal solutions for each direction. A second-level search is performed, constructing a bidirectional equilibrium evaluation function with the goal of optimizing the overall traffic efficiency of the trunk line. The weights of the bidirectional equilibrium evaluation function are based on the bidirectional traffic flow of the trunk line. The flow ratio is dynamically adjusted. Using the unidirectional optimal solution set as the initial iteration point, a neighborhood iterative search is performed in the first-layer key intersection layer search space and the second-layer trunk line overall layer search space. In each iteration, the core parameters of the green wave coordination are fine-tuned and the bidirectional equilibrium evaluation function value is calculated. The parameter combination with the optimal bidirectional equilibrium evaluation function value is retained. The iteration stops when the number of iterations reaches the preset iteration threshold or the bidirectional equilibrium evaluation function value tends to stabilize, and a bidirectional equilibrium candidate solution is obtained. A three-level search is performed to perform hard constraint verification on the bidirectional equilibrium candidate solutions. Candidate solutions that do not meet the hard constraint verification conditions are eliminated to form a feasible solution set. The feasible solution set is sorted by multiple indicators to optimize the scheme. The scheme with the best comprehensive index is selected as the final solution scheme for the trunk line bidirectional green wave.

[0009] In a possible implementation, the following processing is performed: the evaluation indicators of the bidirectional equilibrium evaluation function include the average bidirectional green wave bandwidth, the average bidirectional vehicle delay, and the total number of bidirectional stops; the preset iteration number threshold is determined based on the product of the number of trunk intersections and the preset single-intersection benchmark iteration number.

[0010] In a possible implementation, the following processing is performed: the hard constraint verification conditions include that the green light duration at each intersection is not less than the minimum legal duration, there is no conflict between the phase difference and the phase sequence, and the two-way green wave scheme meets the real-time driving requirements of the trunk line traffic flow.

[0011] In possible implementations, the following processing is performed: the core screening criteria for the multi-index ranking are: the average bidirectional bandwidth is greater than or equal to the preset effective bandwidth value, the reduction in average vehicle delay is greater than or equal to the preset delay reduction ratio, and the reduction in the number of stops is greater than or equal to the preset stop reduction ratio; if there is no solution in the feasible solution set that meets the core screening criteria, the second-best solution with the best comprehensive index is selected, and the solution adaptation description is marked.

[0012] In possible implementations, the execution effect of the scheme is monitored in real time, and the interval search strategy is iteratively optimized based on feedback data. The following processes are performed: traffic operation indicators after the implementation of the green wave scheme are monitored in real time through multi-source sensing devices; manual evaluations are collected and effect labels are marked for the solution schemes, including excellent, medium, and poor effect labels; the solution schemes with excellent effect labels and the corresponding unsolvable scenario features and green wave coordination core parameters are combined and supplemented into the trunk green wave knowledge base; the reasons for the solution schemes with medium or poor effect labels are marked and the constraint verification rules are improved; the interval search strategy is optimized using feedback data with effect labels, and the optimization content includes adjusting the search step size, adjusting the iteration threshold, and adjusting the weight of the two-way equilibrium evaluation function.

[0013] This application proposes a method for solving bidirectional green wave traffic on main roads based on interval strategy search. First, static and real-time dynamic data from each intersection on the main road are collected and preprocessed to obtain real-time main road traffic data. Then, this real-time traffic data is input into a standard maximum green wave bandwidth algorithm for solving, yielding the maximum green wave solution. If the maximum green wave solution is valid, a green wave coordination scheme is directly adopted. If the maximum green wave solution is infeasible or the green wave bandwidth value is less than a preset valid threshold, it is determined to be a no-solution scenario requiring interval strategy search. Then, for the no-solution scenario... Based on the real-time arterial traffic data, constraint intervals for each core parameter of green wave coordination are constructed, and these constraint intervals are then discretized into a grid to form a multi-level search space. The core parameters of green wave coordination include at least phase difference, green light duration in the arterial direction, and expected vehicle speed in both directions. Based on this multi-level search space, a three-level hierarchical search is performed to output the final solution for the arterial bidirectional green wave. Finally, the final solution for the arterial bidirectional green wave is parsed into signal adjustment instructions for each intersection and executed. Simultaneously, the execution effect of the solution is monitored in real time, and the interval search strategy is iteratively optimized based on feedback data. Through the above process, the method proposed in this application achieves the technical effect of stably outputting a feasible bidirectional green wave solution even when the maximum green wave bandwidth algorithm has no effective solution. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0015] Figure 1 This is a flowchart illustrating a trunk bidirectional green wave solution method based on interval strategy search, provided in an embodiment of this application.

[0016] Figure 2 This is a flowchart illustrating the three-level hierarchical search process in a trunk bidirectional green wave solution method based on interval strategy search provided in an embodiment of this application. Detailed Implementation

[0017] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0018] This application provides a method for solving bidirectional green wave propagation on trunk lines based on interval strategy search, such as... Figure 1 As shown, the method includes:

[0019] Step S100: Collect static data and real-time dynamic data of each intersection on the trunk line and preprocess them to obtain real-time trunk line traffic data.

[0020] Specifically, static data is collected for each intersection on the main road, including: the physical distance between adjacent intersections, accurate to 1 meter; the channelization of each intersection, specifically the number of approach lanes and the configuration of each turning lane; the signal phase sequence, clearly defining the fixed phase sequence and the allowable range of variable phases; and the minimum green light duration for each phase in each direction of the main road, determined based on the legal pedestrian crossing requirements and the time required for safe vehicle start-stop. Simultaneously, the design speed of the main road and the expected speed range of two-way traffic under normal conditions are collected. All static data, after being associated with the corresponding intersection numbers and location information, are stored in the main road traffic database.

[0021] Multi-source sensing devices are deployed at the entrance lanes and sections of each intersection on the main road to collect real-time dynamic data. Specifically, microwave radar with a detection range of 0-200m and a speed measurement error ≤1km / h is used to collect real-time driving speeds and headway distances of traffic in both directions; geomagnetic detectors with a response time ≤0.1s are used to collect traffic flow and lane saturation in each intersection's main road direction; and the current timing scheme for each intersection, including green light duration, red light duration, and phase difference, is uploaded in real-time through the signal controller. The collected real-time dynamic data is preprocessed: a Kalman filter algorithm is used to remove outliers in the traffic speed and traffic flow data. For example, when a main road speed exceeding 80km / h is detected, the data point is identified as an anomaly and removed. A 5-minute time window moving average is calculated for the two-way traffic speed to obtain the real-time average speed of the main road in both directions. By combining the minimum green light duration constraints of each intersection with the established phases, the adjustable phase difference range and the variable green light duration range are selected, and any timing parameters that conflict with the established phase sequence are removed to obtain the preprocessed real-time trunk traffic data.

[0022] Step S200: Input the real-time trunk traffic data into the standard maximum green wave bandwidth algorithm for solving to obtain the maximum green wave solution result. If the maximum green wave solution result is a valid solution, the green wave coordination scheme is directly adopted. If the maximum green wave solution result is an infeasible solution or the green wave bandwidth value is less than the preset valid threshold, it is determined to be an unsolvable scenario that requires the initiation of interval strategy search.

[0023] Specifically, the Maximum Green Wave Bandwidth Algorithm (Max Band algorithm) is a prior art algorithm based on linear programming used to solve for the maximum possible bandwidth in the coordinated control of green waves in two-way trunk lines. The real-time trunk line traffic data obtained after preprocessing in step S100 is used as input, and the standard Maximum Green Wave Bandwidth Algorithm is called to solve the linear programming problem. The algorithm's operation follows its established mathematical programming process and outputs a green wave bandwidth solution. The result is judged as follows: if the algorithm outputs a solution and the corresponding green wave bandwidth value is ≥ 8 seconds, it is considered to have actual green wave passage effect, and the solution is recognized as valid, with the green wave coordination scheme directly adopted as the final solution. If the algorithm outputs no feasible solution or the green wave bandwidth value corresponding to the output solution is < 8 seconds, the current trunk line traffic state is determined to be an unsolvable scenario, requiring the initiation of the interval strategy search process in subsequent steps of this application. The interval strategy search, in this application, is a complete solution process executed after determining an unsolvable scenario, including constructing parameter constraint intervals, grid-discretizing to form the search space, and a three-level hierarchical search.

[0024] Step S300: For the unsolvable scenario, based on the real-time trunk traffic data, construct the constraint interval of each green wave coordination core parameter, and discretize the constraint interval into a grid to form a multi-level search space. The green wave coordination core parameters include at least the phase difference, the green light duration in the trunk direction, and the expected vehicle speed in both directions.

[0025] Specifically, the core parameters of green wave coordination refer to the three key variables used in this application for interval search: phase difference, green light duration in the arterial direction, and expected vehicle speed in both directions. For scenarios where no solution has been determined, real-time arterial traffic data is used to determine the feasible value ranges, i.e., constraint intervals, for the three core parameters: phase difference, green light duration in the arterial direction, and expected vehicle speed in both directions. These continuous ranges are then discretized into discrete points using a gridding method. Based on this, a hierarchical search space is constructed according to the traffic characteristics of the arterial road.

[0026] In one possible implementation, for the unsolvable scenario, based on the real-time arterial traffic data, a constraint interval for each green wave coordination core parameter is constructed. Step S300 further includes step S310, which calculates the theoretical range of phase difference based on the intersection spacing and real-time vehicle speed in the real-time arterial traffic data. Combined with signal timing rules, phase difference values ​​that conflict with phase sequence are removed from the theoretical range of phase difference to obtain the phase difference constraint interval. Specifically, based on the adjacent intersection spacing and the current real-time average speed of the two-way traffic flow provided in the real-time arterial traffic data, the time range required for a vehicle to travel between two intersections is calculated using kinematic formulas, which serves as the theoretical range of phase difference values. Then, the current signal timing rules for each intersection are obtained, especially the fixed constraints of phase sequence. Based on this, each phase difference value within the theoretical range of phase difference is checked one by one. If a phase difference value causes the green light phase to fail to match the arrival time of the traffic flow, i.e., a phase sequence conflict occurs, the phase difference value is removed from the range. All unremoved phase difference values ​​together constitute the phase difference constraint interval. As an example, the phase difference adjustment step is set to 1 second.

[0027] Step S320: Based on the minimum green light duration for each intersection in the arterial direction from the real-time arterial traffic data, determine the maximum extendable duration and limit it to not exceed the green light compression limit for non-arterial directions, thus obtaining the arterial direction green light duration constraint interval. Specifically, read the minimum green light duration set for each intersection in the arterial direction from the real-time arterial traffic data; this duration is the legally mandated minimum value that cannot be reduced. Then determine the maximum extendable duration for the arterial direction green light duration. The upper limit of this maximum extendable duration is determined by the compression limit of the non-arterial direction green light at the same intersection, i.e., the non-arterial direction green light duration must not be lower than its own minimum safe duration. Use the original minimum green light duration as the lower limit of the constraint interval, and the sum of the minimum green light duration and the maximum extendable duration as the upper limit of the constraint interval, thereby constructing the arterial direction green light duration constraint interval. As an example, the adjustment step size for the green light duration is set to 1 second.

[0028] Step S330: Based on the average speed of the two-way real-time traffic flow in the real-time trunk line traffic data, a preset speed fluctuation value is adjusted upwards and downwards to obtain the two-way expected speed constraint interval. Specifically, the real-time average speeds in the upbound and downbound directions are obtained from the real-time trunk line traffic data preprocessed in step S100. Using the upbound and downbound real-time average speeds as center values, a preset speed fluctuation value is adjusted upwards and downwards respectively to obtain the expected speed constraint intervals for each direction. As an example, the preset speed fluctuation value is 5 km / h, and the adjustment step is set to 1 km / h.

[0029] In one possible implementation, the constraint interval is discretized into a grid to form a multi-level search space. Step S300 further includes step S340, which uses a three-dimensional interval grid to discretize the phase difference constraint interval, the green light duration constraint interval for the main road direction, and the bidirectional expected vehicle speed constraint interval, forming basic search nodes with a preset adjustment step size as the unit. Specifically, the phase difference constraint interval, the green light duration constraint interval for the main road direction, and the bidirectional expected vehicle speed constraint interval obtained in steps S310, S320, and S330 are regarded as three independent dimensions. Using the three-dimensional interval grid method, with a preset adjustment step size corresponding to each parameter, for example, a step size of 1 second for phase difference and green light duration, and a step size of 1 km / h for vehicle speed, as the unit of grid division, each continuous interval is discretized into a series of discrete numerical points. The discrete numerical points in the three dimensions are combined to form a point matrix in three-dimensional space, and each point represents a complete set of parameter combinations, i.e., a basic search node.

[0030] Step S350: Using the starting point and ending point of the trunk line, as well as intersections with high traffic congestion, as key intersections, the parameter adjustment range of these key intersections is preferentially limited to form the first layer of the key intersection search space. Specifically, key intersections of the trunk line are identified and determined, including: the first intersection (starting point), the last intersection (ending point), and one or more intersections with the highest frequency of traffic congestion based on historical traffic data. In the search space composed of the basic search nodes obtained in step S340, the adjustment range of the phase difference and green light duration parameters corresponding to these key intersections is preferentially limited. The principle of limitation is to narrow the search boundary, for example, to search only within a small interval near the current operating value, so as to reduce the complexity of subsequent searches. This set of search nodes, which has been preferentially limited and only contains the parameter adjustment range of key intersections, constitutes the first layer of the key intersection search space.

[0031] Step S360: Based on the optimization results of the first-layer key intersection search space, the parameter adjustment range of the key intersections is expanded to include the parameter search of all trunk line intersections in the basic search node, forming the second-layer trunk line overall search space. Specifically, after completing the search of the first-layer key intersection search space and obtaining the optimized parameter values ​​of the key intersections, the optimization results are used as fixed conditions. Then, the search range is expanded from including only key intersections to all intersections on the trunk line. At this time, the parameters of all intersections, including key intersections and ordinary intersections, are searched within the complete parameter range covered by the basic search node defined in step S340. This search space covering the complete parameter adjustment range of all trunk line intersections is the second-layer trunk line overall search space.

[0032] Step S400: Based on the multi-level search space, perform a three-level hierarchical search and output the final solution for the trunk bidirectional green wave.

[0033] Specifically, the multi-level search space constructed in step S300 is used as the operating platform, including the first-level key intersection layer search space and the second-level trunk line overall layer search space. The first-level, second-level, and third-level searches are executed in sequence from simple to complex and from local to global, and finally a feasible two-way green wave coordination scheme is selected and output.

[0034] like Figure 2 As shown, in one possible implementation, based on the multi-level search space, a three-level hierarchical search is performed to output the final solution for the bidirectional green wave of the trunk line. Step S400 further includes step S410, which involves performing a first-level search, conducting a search for the maximum green wave bandwidth interval in both the up and down directions, with the goal of maximizing the green wave bandwidth in each direction. The optimal green wave parameter combinations for the up and down directions are obtained by searching the first-level key intersection layer search space and the second-level trunk line overall layer search space. The green wave bandwidth, number of vehicle stops, and average delay evaluation index corresponding to each optimal green wave parameter combination are calculated to form a set of optimal solutions for each direction. Specifically, the first-level search is divided into two independent sub-tasks: the search for the maximum green wave bandwidth in the up direction and the search for the maximum green wave bandwidth in the down direction. Each sub-task uses maximizing the green wave bandwidth in its own direction as the sole optimization objective. The search process begins in the first layer, the key intersection layer search space, where the parameters of key intersections are adjusted. After obtaining the locally optimal parameters of the key intersections, these parameters are fixed, and then a complete search is performed on the parameters of all intersections along the entire trunk line in the second layer, the overall trunk line layer search space. Through this process, the parameter combinations that maximize the green wave bandwidth in the upbound direction and the green wave bandwidth in the downbound direction are found. For each optimal parameter combination, three evaluation indicators are obtained through traffic simulation or analytical calculation based on a classical traffic wave model: green wave bandwidth (seconds), average number of vehicle stops (times / vehicle), and average vehicle delay (seconds / vehicle). The optimal parameter combinations and their evaluation indicators for the upbound and downbound directions are summarized to form a set of unidirectional optimal solutions.

[0035] Step S420: Perform a secondary search to construct a bidirectional equilibrium evaluation function with the goal of optimizing the overall traffic efficiency of the trunk line. The weights of the bidirectional equilibrium evaluation function are dynamically adjusted according to the bidirectional traffic flow ratio of the trunk line. Using the set of unidirectional optimal solutions as the initial iteration point, a neighborhood iterative search is performed in the first-level key intersection layer search space and the second-level trunk line overall layer search space. In each iteration, the core parameters of the green wave coordination are fine-tuned and the bidirectional equilibrium evaluation function value is calculated. The parameter combination with the optimal bidirectional equilibrium evaluation function value is retained. The iteration stops when the number of iterations reaches a preset iteration threshold or the bidirectional equilibrium evaluation function value tends to stabilize, thus obtaining a bidirectional equilibrium candidate solution. The evaluation indicators of the bidirectional equilibrium evaluation function include the average bidirectional green wave bandwidth, the average bidirectional vehicle delay, and the total number of bidirectional stops. The preset iteration threshold is determined by the product of the number of trunk line intersections and the preset single-intersection benchmark iteration number. Specifically, a bidirectional equilibrium evaluation function is constructed to evaluate the degree of equilibrium of the candidate scheme in terms of bidirectional traffic efficiency. Its inputs are the average bidirectional green wave bandwidth, the average bidirectional vehicle delay, and the total number of bidirectional stops, and the output is an evaluation value. The mathematical expression is: the evaluation value equals the sum of (the weight of the upbound direction multiplied by the sum of the normalized values ​​of all indicators in the upbound direction) and (the weight of the downbound direction multiplied by the sum of the normalized values ​​of all indicators in the downbound direction). The weights of the upbound and downbound directions are dynamically set based on the real-time monitored ratio of two-way traffic flow: for example, if the current upbound traffic accounts for 60% of the total traffic flow, then the upbound weight is set to 0.6 and the downbound weight is set to 0.4. The evaluation indicators include three: the average two-way green wave bandwidth (the larger the better), the average two-way vehicle delay (the smaller the better), and the average total number of stops in both directions (the smaller the better). During calculation, each indicator needs to be normalized to make it dimensionless. Then, the two sets of parameter combinations in the unidirectional optimal solution set obtained in step S410 are used as the initial iteration points. The iteration process also follows the search order of first the key intersection layer and then the overall trunk line layer. In each iteration, within the neighborhood of the current parameter combination—that is, adjusting the current phase difference, green light duration, or vehicle speed value by one step—a new set of parameters is generated, and its bidirectional equilibrium evaluation function value is calculated. If the new value is better than the currently retained optimal value, it is replaced. Iteration continues until a stopping condition is met. There are two stopping conditions, and stopping occurs when either one is met: first, the number of iterations reaches a preset threshold, calculated by multiplying the total number of arterial intersections by a preset single-intersection baseline iteration count; as an example, each intersection corresponds to 10 iterations; second, after multiple consecutive iterations, the change in the bidirectional equilibrium evaluation function value is less than a preset minimum threshold, such as 0.01%, indicating that the function value has stabilized. When iteration stops, the parameter combination with the best currently retained evaluation function value is the bidirectional equilibrium candidate solution.

[0036] Step S430: Perform a three-level search to perform hard constraint verification on the bidirectional equilibrium candidate solutions, eliminate candidate solutions that do not meet the hard constraint verification conditions, and form a feasible solution set. Then, use multi-index ranking to optimize the scheme, selecting the scheme with the best comprehensive index as the final solution for the bidirectional green wave traffic flow on the trunk line. The hard constraint verification conditions include: green light duration at each intersection not less than the minimum legal duration; no conflict between phase difference and phase sequence; and the bidirectional green wave scheme meeting the real-time traffic flow requirements of the trunk line. The core screening criteria for the multi-index ranking are: average bidirectional bandwidth greater than or equal to a preset effective bandwidth value; average vehicle delay reduction greater than or equal to a preset delay reduction ratio; and reduction in the number of stops greater than or equal to a preset stop reduction ratio. If no scheme in the feasible solution set meets the core screening criteria, the second-best scheme with the best comprehensive index is selected, and a scheme adaptation description is provided. Specifically, three hard constraint verifications are performed on each bidirectional equilibrium candidate solution output in step S420. The first verification step is to check whether the green light duration for the main road direction at each intersection in the solution is greater than or equal to its preset minimum legal duration. The second verification step is to check whether the phase difference set in the solution conflicts with the inherent phase sequence rules of each intersection; for example, whether the green light activation time overlaps with the clearing phase of the other direction. The third verification step is to check whether the expected vehicle speed set in the solution is within a reasonable range that vehicles can actually reach, i.e., whether it meets the current real-time traffic flow requirements. Any candidate solution that does not meet any of the above three criteria is eliminated. All candidate solutions that pass the verification constitute the feasible solution set. Then, each candidate solution in the feasible solution set is ranked according to multiple indicators. The core screening criteria are set to three: average bidirectional bandwidth ≥ 10 seconds; compared with the no-green-wave solution or the existing solution, the average vehicle delay is reduced by ≥ 15%; and the number of stops is reduced by ≥ 20%. First, candidate solutions that simultaneously meet the above three criteria are screened, and the one with the best comprehensive indicators is selected as the final solution; for example, the improvement rates of the three indicators are weighted and summed. If there is no solution in the feasible solution set that can simultaneously meet the three core criteria, then the one with the highest comprehensive index value is selected from the set as the suboptimal solution and an additional text description is generated to indicate the gap between the solution and the core criteria and the limitations of its applicable scenarios.

[0037] Step S500: The final solution for the bidirectional green wave of the trunk line is parsed into signal adjustment instructions for each intersection and executed. At the same time, the execution effect of the solution is monitored in real time, and the interval search strategy is iteratively optimized based on the feedback data.

[0038] Specifically, the final solution for the bidirectional green wave traffic flow on main roads is parsed into phase difference adjustment commands and green light duration adjustment commands for each intersection. These commands are then pushed to the signal controllers at each intersection via PLC controllers for smooth signal timing switching. The switching transition time is ≤ a preset switching transition time, such as 2 seconds, to avoid traffic congestion caused by sudden green light interruptions or abrupt phase difference changes. Simultaneously, green wave driving speed prompts are displayed on traffic guidance screens to guide traffic flow to the desired green wave speed, improving the effectiveness of the green wave solution. A closed-loop feedback mechanism is established to continuously improve future search processes using actual operational data.

[0039] In one possible implementation, the execution effect of the scheme is monitored in real time, and the interval search strategy is iteratively optimized based on feedback data. Step S500 further includes step S510, which involves real-time monitoring of traffic operation indicators after the implementation of the green wave scheme using multi-source sensing devices, collecting manual evaluations, and labeling the solution with effect tags, including excellent, average, and poor. Specifically, after the final solution obtained in step S400 is implemented, the microwave radar and geomagnetic detector deployed in step S100 are reused to monitor and collect traffic operation indicators after implementation in real time. The collected indicators include: actual pass rate of the two-way green wave (the proportion of vehicles that actually pass through multiple green lights consecutively), average vehicle delay, average number of stops, and overall traffic flow on the main road. At the same time, manual evaluations from front-line traffic police are collected through the traffic management platform, such as smooth green wave connection at an intersection or congestion in a certain direction. Based on the monitoring data and manual evaluations, an effect tag is assigned to the currently implemented solution: if all indicators are better than expected, it is labeled as excellent; if the indicators have improved but not met expectations, it is labeled as average; if the indicators deteriorate or new congestion occurs, it is labeled as poor.

[0040] Step S520 involves organizing and supplementing the solution schemes labeled "Excellent" with their corresponding unsolvable scenario features and core parameter combinations for green wave coordination into the trunk green wave knowledge base. For schemes labeled "Medium" or "Poor," the reasons are noted, and constraint verification rules are improved. Specifically, when a solution scheme is labeled "Excellent," all information corresponding to that scheme is structured and organized. This information includes: the complete parameter combination that generated the scheme (phase difference, green light duration, vehicle speed), and the unsolvable scenario features that triggered the scheme (e.g., specific intersection spacing combinations, bidirectional speed differences, phase constraint conditions, etc.). The organized information is added as a knowledge entry to a specially constructed trunk green wave knowledge base. When a scheme is labeled "Medium" or "Poor," the reasons for the poor performance are analyzed, such as an excessively large parameter search step size leading to missing the optimal solution or unreasonable bidirectional weight settings resulting in excessive sacrifice in one direction. These reason analyses are associated with the corresponding schemes and stored. Based on this, the hard constraint verification conditions used in subsequent steps are improved, and new verification rules are added to avoid generating similar unsatisfactory schemes again.

[0041] Step S530 involves optimizing the interval search strategy using feedback data with effect labels. Optimization includes adjusting the search step size, iteration threshold, and weights of the bidirectional equilibrium evaluation function. Specifically, a fixed optimization cycle is set, for example, with strategy optimization performed weekly. At the end of the cycle, feedback data for all labeled effects are collected. Analysis is then performed based on this data: if multiple good-label schemes use smaller parameter adjustment step sizes, the default preset adjustment step size is considered, for example, from 1 second to 0.5 seconds; if poor-label schemes frequently experience prolonged stagnation in the evaluation function before the iteration count reaches the threshold, the preset iteration count threshold is considered for reduction; if poor-label schemes are mostly due to excessively high weights in one direction causing severe congestion in the other, the dynamic adjustment range of the weights in the bidirectional equilibrium evaluation function or the default baseline weight is adjusted. Through these methods, the parameter settings in the interval search strategy are continuously optimized, enabling the method of this application to better adapt to the constantly changing trunk traffic environment during continuous operation.

[0042] This application's embodiments solve the technical problem of existing trunk line bidirectional green wave solutions failing to stably obtain effective green wave solutions. It achieves the technical effect of stably outputting feasible bidirectional green wave solutions even when the maximum green wave bandwidth algorithm has no effective solution. By collecting and preprocessing trunk line traffic data, inputting the maximum green wave bandwidth algorithm to determine whether it is a scenario with no solution, and constructing the constraint interval of the core parameters and discretizing it into a multi-level search space, a three-level hierarchical search is executed to output the final solution, and finally the solution is parsed into instructions for execution and the search strategy is optimized based on feedback iteration.

[0043] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for solving bidirectional green wave propagation on trunk lines based on interval strategy search, characterized in that, The method includes: Collect static and real-time dynamic data from each intersection on the main road and preprocess them to obtain real-time main road traffic data; The real-time trunk traffic data is input into the standard maximum green wave bandwidth algorithm for solution to obtain the maximum green wave solution result. If the maximum green wave solution result is a valid solution, the green wave coordination scheme is directly adopted. If the maximum green wave solution result is an infeasible solution or the green wave bandwidth value is less than the preset valid threshold, it is determined to be an unsolvable scenario that requires the initiation of interval strategy search. For the unsolvable scenario, based on the real-time trunk traffic data, a constraint interval for each green wave coordination core parameter is constructed, and the constraint interval is discretized into a grid to form a multi-level search space. The green wave coordination core parameters include at least the phase difference, the green light duration in the trunk direction, and the expected vehicle speed in both directions. Based on the multi-level search space, a three-level hierarchical search is performed to output the final solution for the trunk bidirectional green wave. The final solution for the bidirectional green wave of the trunk line is parsed into signal adjustment instructions for each intersection and executed. At the same time, the execution effect of the solution is monitored in real time, and the interval search strategy is iteratively optimized based on the feedback data.

2. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 1, characterized in that, For the unsolvable scenario, based on the real-time trunk traffic data, constraint intervals are constructed for each core parameter of green wave coordination, including: Based on the intersection spacing and real-time vehicle speed in the real-time trunk traffic data, the theoretical range of phase difference is calculated, and combined with the signal timing rules, phase difference values ​​that conflict with the phase sequence are removed from the theoretical range of phase difference to obtain the phase difference constraint interval. Based on the minimum green light duration of each intersection in the trunk direction in the real-time trunk traffic data, the maximum extendable duration is determined and limited to not exceeding the green light compression limit of the non-trunk direction, thus obtaining the green light duration constraint range for the trunk direction. Based on the average speed of the two-way real-time traffic flow in the real-time trunk line traffic data, a preset speed fluctuation value is adjusted up and down to obtain the two-way expected speed constraint range.

3. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 2, characterized in that, The constraint interval is discretized into a grid to form a multi-level search space, including: The phase difference constraint interval, the green light duration constraint interval in the main direction, and the bidirectional expected vehicle speed constraint interval are discretized by using a three-dimensional interval grid, and basic search nodes are formed with a preset adjustment step size as the unit. The starting point, ending point, and intersections with high traffic congestion of the main road are designated as key intersections. The parameter adjustment range of the key intersections is limited first to form the first layer of key intersection search space. Based on the optimization results of the first-level key intersection layer search space, the parameter adjustment range of the key intersections is extended to the parameter search of all trunk line intersections in the basic search nodes, forming the second-level trunk line overall layer search space.

4. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 3, characterized in that, Based on the aforementioned multi-level search space, a three-level hierarchical search is performed to output the final solution for the trunk bidirectional green wave, including: Perform a first-level search, and conduct a search for the maximum green wave bandwidth interval in both the up and down directions. With the goal of maximizing the green wave bandwidth in one direction, search the first-level key intersection layer search space and the second-level trunk line overall layer search space to obtain the optimal green wave parameter combination for each direction. Calculate the green wave bandwidth, number of vehicle stops, and average delay evaluation index corresponding to each optimal green wave parameter combination to form a set of optimal solutions in one direction. A secondary search is performed to construct a two-way equilibrium evaluation function with the goal of optimizing the overall traffic efficiency of the trunk line. The weights of the two-way equilibrium evaluation function are dynamically adjusted according to the two-way traffic flow ratio of the trunk line. The set of unidirectional optimal solutions is used as the initial iteration point. Neighborhood iterative search is performed in the first-level key intersection layer search space and the second-level trunk line overall layer search space. In each iteration, the core parameters of the green wave coordination are fine-tuned and the two-way equilibrium evaluation function value is calculated. The parameter combination with the optimal two-way equilibrium evaluation function value is retained. When the number of iterations reaches the preset iteration number threshold or the two-way equilibrium evaluation function value tends to stabilize, the iteration stops and the two-way equilibrium candidate solution is obtained. A three-level search is performed to check the bidirectional equilibrium candidate solutions with hard constraints. Candidate solutions that do not meet the hard constraint check conditions are eliminated to form a feasible solution set. The feasible solution set is then sorted by multiple indicators to optimize the scheme. The scheme with the best comprehensive index is selected as the final solution scheme for the trunk bidirectional green wave.

5. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 4, characterized in that, The evaluation indicators of the bidirectional equilibrium evaluation function include the average bidirectional green wave bandwidth, the average bidirectional vehicle delay, and the total number of bidirectional stops. The preset iteration number threshold is determined by multiplying the number of trunk intersections by the preset single-intersection baseline iteration number.

6. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 4, characterized in that, The hard constraint verification conditions include that the green light duration at each intersection is not less than the minimum legal duration, there is no conflict between the phase difference and the phase sequence, and the two-way green wave scheme meets the real-time driving requirements of the main road traffic flow.

7. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 4, characterized in that, The core screening criteria for the multi-indicator ranking are: The average bidirectional bandwidth is greater than or equal to the preset effective bandwidth value, the average vehicle delay reduction is greater than or equal to the preset delay reduction ratio, and the number of stops is reduced by a factor greater than or equal to the preset stop reduction ratio. If no solution in the feasible solution set meets the core screening criteria, the second-best solution with the best comprehensive index is selected, and the solution adaptation description is marked.

8. The trunk line bidirectional green wave solution method based on interval strategy search as described in claim 1, characterized in that, Simultaneously, the execution effect of the plan is monitored in real time, and the interval search strategy is iteratively optimized based on feedback data, including: The traffic operation indicators after the implementation of the green wave scheme are monitored in real time by multi-source sensing devices. Manual evaluations are collected and effect labels are marked for the solution scheme. The effect labels include excellent, medium and poor. The solution schemes with excellent effect labels and the corresponding unsolvable scenario characteristics, green wave coordination core parameter combinations are compiled and supplemented into the trunk green wave knowledge base. The reasons for the solutions with medium or poor effect labels are marked and the constraint verification rules are improved. The interval search strategy is optimized using feedback data with effect labels. The optimizations include adjusting the search step size, adjusting the iteration threshold, and adjusting the weights of the two-way equilibrium evaluation function.

Images