A traffic distribution and maximum pressure signal control collaborative optimization method fusing cognitive levels under an intelligent network connection environment

Abstract

The present application relates to a kind of traffic distribution and maximum pressure signal control collaborative optimization method of fusion cognitive level under intelligent network connection environment, belong to traffic management control technical field.The method includes: the traveler in road network is divided into three levels according to cognitive ability, the traveler of each level updates path selection according to the next day traffic flow prediction value, obtains the next day path planning result and optimizes the signal timing of each intersection, completes the signal timing strategy of each intersection, while the traveler of each level introduces the deviation cost to the prediction calculation prediction travel cost of next day flow and updates path flow, sets the shortest and longest green time for each signal phase, and introduces fixed phase sequence constraint signal phase switching logic and execution order, complete the collaborative optimization of traffic distribution and maximum pressure signal control.The present application is aimed at solving the technical problems that the existing technology causes traffic information utilization degree different to affect travel route selection and the application of CAV prediction pressure is insufficient.

Description

Technical Field

[0001] This invention relates to a collaborative optimization method for traffic allocation and maximum pressure signal control that integrates cognitive levels in an intelligent connected environment, belonging to the field of traffic management and control technology. Background Technology

[0002] With the continuous advancement and widespread application of communication and sensor technologies, intelligent connected vehicles (CAVs) integrating modern communication and environmental perception capabilities will gradually enter a stage of large-scale development. In the long term, intelligent connected vehicles with information interaction and collaborative control capabilities are expected to replace traditional manually driven vehicles. However, considering factors such as technological maturity and market acceptance, this replacement process will undergo a relatively long and gradual transition. During this period, intelligent connected vehicles and manually driven vehicles will coexist in the same traffic environment for a considerable time, forming a typical mixed traffic flow scenario.

[0003] Currently, the collaborative optimization of traffic assignment and signal control methods is a hybrid optimization approach to alleviate urban traffic congestion. To address the shortcomings of current research in reflecting how varying levels of traffic information utilization among travelers influence their route choices, and the limitations of current maximum pressure signal control's ability to alleviate congestion due to insufficient application of CAV (Constant Adaptive Voltage) to predictive pressure, this invention combines maximum pressure traffic control with a traditional diurnal variation model and introduces the concept of cognitive hierarchy. This approach is of great significance for the collaborative optimization of traffic assignment and signal control methods. Summary of the Invention

[0004] The purpose of this invention is to provide a collaborative optimization method for traffic allocation and maximum pressure signal control that integrates cognitive levels in an intelligent connected environment. This method aims to solve the technical problems of existing technologies, such as the impact of varying levels of traffic information utilization on travel route selection and insufficient application of CAV (Conditional Ability Volume) to predict pressure.

[0005] To achieve the above objectives, the technical solution of the present invention is: a collaborative optimization method for traffic assignment and maximum pressure signal control in an intelligent connected environment that integrates cognitive levels, comprising the following steps:

[0006] S1: Construct a road network that includes a set of nodes, a set of road segments, and a set of origin-destination pairs. Divide travelers in the road network into three levels according to their cognitive abilities. For each level of travelers, obtain the traffic flow forecast for the next day based on the exponential movement coefficient. Update the route selection based on the traffic flow forecast for the next day to obtain the route planning results for each level of travelers.

[0007] S2: Based on the route planning results for the next day, optimize the signal timing of each intersection in the road network to complete the signal timing strategy for each intersection. Based on the signal timing strategy, at the end of the day, feed back the traffic situation of each intersection to travelers at each level. The optimization involves dividing the intersection pressure detection into real-time detected vehicles and predicted intelligent connected vehicles arriving at the intersection in the next cycle. The signal pressure of each intersection is calculated by the number of real-time detected vehicles and the number of predicted vehicles.

[0008] S3: Based on the traffic conditions at each intersection, travelers at each level introduce deviation costs that deviate from the existing routes while minimizing travel costs. The predicted travel costs are calculated based on the deviation costs and the predicted traffic flow for the next day. The path flow is updated based on the predicted travel costs to obtain the updated path flow for travelers at each level.

[0009] S4: Based on the updated traveler path flow at each level, set the shortest and longest green light times for each signal phase, and introduce a fixed phase sequence to constrain the switching logic and execution order of the signal phases, thereby completing the collaborative optimization of traffic allocation and maximum pressure signal control.

[0010] Optionally, S1 specifically includes:

[0011] Construct a road network Where N is the set of nodes, L is the set of road segments, and W is the set of OD pairs. Any OD pair is defined as... Travel demand is The set of feasible paths is ;

[0012] Travelers in the road network are classified into K levels based on their cognitive ability, where K=3, and the proportion of travelers in each of the K levels is defined as follows: and satisfy Define the set of feasible path flows for k types of travelers. for:

[0013]

[0014]

[0015]

[0016] in, Let d be the path flow vector of type k, and d be the OD demand vector. This is the OD path association matrix. For travelers of class k, select the traffic flow from route r to OD to w, where T represents the transpose operation;

[0017] On day t, the path flow of the k-th type of traveler is defined as follows: In the road network, the paths of all travelers are ;

[0018] Travelers get Then, based on its own cognitive level, it forms a prediction of traffic flow in the road network on day t+1. And obtain the corresponding predicted travel cost as This allows us to obtain an update of the travel path of the k-th type of traveler on day t+1. for:

[0019]

[0020] in, Let y be the projection operator. The target vehicle path flow in the middle, Adjust the sensitivity parameter for the path flow of a single positive real scalar, and , It is the exponential moving average, and ;

[0021] in:

[0022]

[0023]

[0024] In the formula, Let k be the set of feasible route flows for travelers. It is an empty set;

[0025] This allows us to obtain route planning information for the next day's traffic patterns from travelers with different cognitive abilities, specifically:

[0026] First, we define the traffic pattern of travelers with k=0 on day t as unchanged on day t+1, and define travelers with k=0 as non-connected HDV travelers. When all travelers in the road network are k=0 travelers, the road network becomes the traditional case without intelligent connected vehicles (CAVs), expressed as:

[0027]

[0028] in, For the traveler at k=0, predict the traffic flow in the road network on day t+1;

[0029] Then, when k≥1, the k-traveler plans its own travel route by predicting the reactions of travelers at lower levels to the current path pattern. Travelers with k≥1 are defined as CAV (Consumer-Aided Vehicle) travelers, thus obtaining the normalized ratio of k-travelers to ℎ-travelers. for:

[0030]

[0031] In the formula, and This represents the percentage of travelers at the current traveler level.

[0032] A traveler with k=1 considers all other travelers in the road network to be travelers with k=0, thus obtaining the traveler's prediction of traffic flow in the road network on day t+1. for:

[0033]

[0034] In the formula, and For projection operators, To normalize the proportion, For the predicted travel costs, and For prediction coefficients;

[0035] This allows us to obtain the travel route update for the traveler with k=1 on day t+1. for:

[0036]

[0037] In the formula, For projection operators, The path flow for the k=1 type of traveler;

[0038] Finally, the traveler with k=2 simultaneously predicts the responses of both the traveler with k=0 and the traveler with k=1, thus obtaining the normalized proportions as follows: and Define the traveler with k=2 to predict traffic flow in the road network on day t+1. for:

[0039]

[0040] In the formula, and For projection operators;

[0041] This allows us to obtain the travel route update for the traveler with k=2 on day t+1. for:

[0042]

[0043] In the formula, For projection operators, This is the path flow for the k=2 type of traveler.

[0044] Optionally, S2 specifically includes:

[0045] The signal timing at each intersection in the road network is optimized using the following expression:

[0046]

[0047] In the formula, This represents the weight difference between road segment l and road segment m during time period t. This represents the queue length between road segment l and road segment m during time period t; Let m be the set of all road segments ending at point m. This represents the proportion of traffic flow from road segment m to road segment v during time period t. This represents the queue length between road segment m and road segment v during time period t;

[0048] In a connected vehicle environment, intersections simultaneously contain both connected vehicle (CAV) and non-connected vehicle (HDV) traffic. Based on the predictability of CAV paths, intersection pressure detection is divided into real-time detection of vehicles at the intersection and prediction of CAV arrival times in the next cycle. The expression for the arrival time of a CAV at the intersection is as follows:

[0049]

[0050]

[0051] In the formula, To predict the number of steps, Let n be the predicted speed of the nth vehicle at a future time. Let be the remaining distance of the nth vehicle from the intersection at the current time t. The period length, Let n be the estimated arrival time of the nth vehicle at the intersection;

[0052] Thus, the signal phase is obtained. Number of intelligent connected vehicles (CAVs) arriving at the intersection for:

[0053]

[0054] After predicting the arrival times of intelligent connected vehicles (CAVs) at intersections in different signal phases, and combining the path planning of different cognitive levels on day t+1, the signal timing strategies for each intersection on day t+1 are completed by combining the number of vehicles detected in real time at the intersections with the number of vehicles predicted by the intelligent connected vehicles (CAVs). After day t+1 ends, the traffic conditions of each intersection on that day are returned to travelers at different cognitive levels, and travelers at different cognitive levels optimize their travel routes for the next day based on the traffic conditions of the previous day.

[0055] Optionally, S3 specifically includes:

[0056] While minimizing travel costs, travelers need to incur deviation costs for deviating from their established routes, expressed as:

[0057]

[0058]

[0059] In the formula, This represents the projection operator for solving optimization problems. This represents the deviation cost from the existing path, and , The set of feasible path flows. Set of feasible path flows Candidate path flow vectors in; The path flow vector to be projected;

[0060] The updated traffic flow predictions for the next day are thus obtained for the three levels of travelers:

[0061]

[0062]

[0063]

[0064] In the formula, , , The predictions of travelers with k=0, k=1, and k=2 regarding traffic flow in the road network on day t+1 are given. , , To solve for the optimized projection operator, The deviation cost from the existing path after the update;

[0065] The k-th level traveler calculates the predicted travel cost based on the traffic flow forecast for the next day, and updates the path flow based on the predicted travel cost. The updated path flow is an adjustment rule in the form of projection. Specifically, the traveler updates the path flow along the direction of improvement of the predicted travel cost, and the projection operator ensures that the update result satisfies the non-negativity and demand conservation constraints.

[0066] Optionally, setting the shortest and longest green light times for each signal phase specifically involves:

[0067]

[0068] In the formula, The shortest green light time is determined by the length of the intersection. The actual green light time for the i-th signal phase. The longest green light time, The total available green light time within a signal cycle. Number the signal phase.

[0069] The beneficial effects of this invention are as follows: This invention simultaneously considers the optimization methods of traffic assignment and signal control for urban roads, which can effectively reduce the average queue length of the road network and increase the overall number of vehicles served by the road network, thus having practical significance for alleviating traffic congestion. Furthermore, this invention achieves the goal of alleviating urban congestion from the perspective of traffic guidance, combining cognitive hierarchy and maximum pressure signal control to overcome the limitations of existing technologies where traffic congestion occurs on other uncongested road sections after initial assignment, and where the optimization effect decreases when queue overflow occurs in the maximum pressure signal control model. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating the present invention;

[0071] Figure 2 This is a schematic diagram of the maximum pressure signal control for predicting CAV according to the present invention;

[0072] Figure 3 This is the simulated road network diagram of the present invention;

[0073] Figure 4 This is a diagram showing the average queue length when the system is not saturated.

[0074] Figure 5 This is a diagram showing the average number of stops in the road network when the road is not saturated, according to the present invention.

[0075] Figure 6 This is a diagram showing the average queue length when the present invention is near saturation;

[0076] Figure 7 This is a diagram showing the average number of stops on the road network when the road is near saturation, according to the present invention.

[0077] Figure 8 This is a diagram showing the average queue length under oversaturation conditions according to the present invention;

[0078] Figure 9 This is a diagram showing the average number of stops in the road network under oversaturation conditions according to the present invention;

[0079] Figure 10 This is a diagram showing the number of service vehicles under different saturation levels according to the present invention. Detailed Implementation

[0080] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0081] Example 1: As Figure 1 As shown, a collaborative optimization method for traffic assignment and maximum pressure signal control in an intelligent connected environment that integrates cognitive levels includes the following steps:

[0082] S1: Construct a road network that includes a set of nodes, a set of road segments, and a set of origin-destination pairs. Divide travelers in the road network into three levels according to their cognitive abilities. For each level of travelers, obtain the traffic flow forecast for the next day based on the exponential movement coefficient. Update the route selection based on the traffic flow forecast for the next day to obtain the route planning results for each level of travelers.

[0083] Optionally, a road network can be constructed. Where N is the node set, L is the road segment set, and W is the origin-destination (OD) pair set, and any OD pair is defined as... Travel demand is The set of feasible paths is ;

[0084] Travelers in the road network are classified into K levels based on their cognitive ability, where K=3, and the proportion of travelers in each of the K levels is defined as follows: and satisfy Assuming the proportion of each type of traveler is consistent across all OD pairs, define the set of feasible path flows for k types of travelers. for:

[0085]

[0086]

[0087]

[0088] in, Let d be the path flow vector of type k, and d be the OD demand vector. This is the OD path association matrix. For travelers of class k, select the traffic flow from route r to OD to w, where T represents the transpose operation;

[0089] On day t, the path flow of the k-th type of traveler is defined as follows: In the road network, the paths of all travelers are ;

[0090] Travelers get Then, based on its own cognitive level, it forms a prediction of traffic flow in the road network on day t+1. And obtain the corresponding predicted travel cost as This allows us to obtain an update of the travel path of the k-th type of traveler on day t+1. for:

[0091]

[0092] in, Let y be the projection operator. The target vehicle path flow in the middle, A sensitivity parameter is adjusted for the path flow of a single positive real scalar to regulate the strength of the impact of predicted travel costs on the target vehicle's path flow. , The exponential moving average reflects the traveler's inertia or unwillingness to change. A smaller value indicates a weaker willingness to adjust, meaning travelers are more likely to stick to their original routes; a larger value indicates a greater willingness to switch routes. ;

[0093] in:

[0094]

[0095]

[0096] In the formula, Let k be the set of feasible route flows for travelers. It is an empty set;

[0097] It is important to understand that behavioral game theory experiments show that most individuals' reasoning depth usually does not exceed two steps. Therefore, in practical analysis, we often focus on cases where the number of levels does not exceed three. Thus, in this embodiment, we take K=3. K=3 can be regarded as a simplified heuristic for the strategic thinking process, used to characterize the behavioral mechanism at the group level. It can also be understood as the collective cognition of a certain type of traveler, rather than the precise cognition of an individual. In addition, different cognitive levels can correspond to traffic decision-making subjects with different predictive abilities. For example, basic drivers or systems that make decisions based solely on real-time information can be regarded as k=0, while navigation systems or traffic management platforms that can predict congestion trends can be regarded as higher levels.

[0098] Optionally, this embodiment further simulates how travelers with different cognitive abilities form their path planning for the next day's traffic patterns based on K=3. Specifically, the predictions of future traffic by travelers at each level are formed recursively. Travelers at k=1 assume that other travelers are of type k=0 and predict their adjusted traffic; travelers at k=2 simultaneously predict the travel behaviors of travelers at k=0 and k=1, and these predictions are weighted and summed according to their belief weights. In this embodiment, travelers of type k recursively predict the travel behaviors of all lower-level travelers and sum them in a weighted manner to obtain the overall traffic prediction for the next day. That is, higher-level travelers form their judgment on the future state of the system by predicting the behavioral responses of lower-level travelers, specifically as follows:

[0099] First, travelers with k=0 do not engage in strategic thinking. Therefore, the traffic pattern of travelers with k=0 on day t remains unchanged on day t+1. Furthermore, travelers with k=0 are non-connected vehicle (HDV) travelers. When all travelers in the road network are k=0 travelers, the road network becomes a traditional case without connected vehicles (CAVs), expressed as:

[0100]

[0101] in, For the traveler at k=0, predict the traffic flow in the road network on day t+1;

[0102] Then, when k≥1, the k-traveler plans its own travel route by predicting the reactions of travelers at lower levels to the current path pattern. Travelers with k≥1 are defined as CAV travelers, thus obtaining the normalized ratio of k-travelers to ℎ-travelers. for:

[0103]

[0104] In the formula, and This represents the percentage of travelers at the current traveler level.

[0105] A traveler with k=1 considers all other travelers in the road network to be travelers with k=0, thus obtaining the traveler's prediction of traffic flow in the road network on day t+1. for:

[0106]

[0107] In the formula, and For projection operators, To normalize the proportion, For the predicted travel costs, and For prediction coefficients;

[0108] This allows us to obtain the travel route update for the traveler with k=1 on day t+1. for:

[0109]

[0110] In the formula, For projection operators, The path flow for the k=1 type of traveler;

[0111] Finally, the traveler with k=2 simultaneously predicts the responses of both the traveler with k=0 and the traveler with k=1, thus obtaining the normalized proportions as follows: and Define the traveler with k=2 to predict traffic flow in the road network on day t+1. for:

[0112]

[0113] In the formula, and For projection operators;

[0114] This allows us to obtain the travel route update for the traveler with k=2 on day t+1. for:

[0115]

[0116] In the formula, For projection operators, This is the path flow for the k=2 type of traveler.

[0117] S2: Based on the route planning results for the next day, optimize the signal timing of each intersection in the road network to complete the signal timing strategy for each intersection. Based on the signal timing strategy, at the end of the day, feed back the traffic situation of each intersection to travelers at each level. The optimization involves dividing the intersection pressure detection into real-time detected vehicles and predicted intelligent connected vehicles arriving at the intersection in the next cycle. The signal pressure of each intersection is calculated by the number of real-time detected vehicles and the number of predicted vehicles.

[0118] Optionally, the signal timing at each intersection in the road network can be optimized, as shown in the expression:

[0119]

[0120] In the formula, This represents the weight difference between road segment l and road segment m during time period t. This represents the queue length between road segment l and road segment m during time period t; Let m be the set of all road segments ending at point m. This represents the proportion of traffic flow from road segment m to road segment v during time period t. This represents the queue length between road segment m and road segment v during time period t;

[0121] Furthermore, such as Figure 2 As shown, in a connected environment, intersections simultaneously contain both CAVs (Conducting Vehicles) and HDVs (Unconnected Vehicles). The detection of intersection pressure based on the predictability of CAV paths is divided into real-time detection of vehicles at the intersection and prediction of CAVs arriving at the intersection in the next cycle. The expression for the CAV arrival time at the intersection is:

[0122]

[0123]

[0124] In the formula, To predict the number of steps, Let n be the predicted speed of the nth vehicle at a future time. Let be the remaining distance of the nth vehicle from the intersection at the current time t. The period length, Let n be the estimated arrival time of the nth vehicle at the intersection;

[0125] Thus, the signal phase is obtained. Number of CAV vehicles arriving at the intersection for:

[0126]

[0127] After predicting the arrival of CAVs at intersections in different signal phases, the system combines the path planning of different cognitive levels on day t+1 with the real-time vehicle count at the intersections and the CAV-predicted vehicle count to complete the signal timing strategy for each intersection on day t+1. After day t+1 ends, the traffic conditions of each intersection on that day are returned to travelers at different cognitive levels, and travelers at different cognitive levels optimize their travel routes for the next day based on the traffic conditions of the previous day.

[0128] S3: Based on the traffic conditions at each intersection, travelers at each level introduce deviation costs that deviate from the existing routes while minimizing travel costs. The predicted travel costs are calculated based on the deviation costs and the predicted traffic flow for the next day. The path flow is updated based on the predicted travel costs to obtain the updated path flow for travelers at each level.

[0129] Alternatively, while minimizing travel costs, travelers need to incur deviation costs for deviating from the existing route, expressed as:

[0130]

[0131]

[0132] In the formula, This represents the projection operator for solving optimization problems. This represents the deviation cost from the existing path, and , The set of feasible path flows. Set of feasible path flows Candidate path flow vectors in; The path flow vector to be projected;

[0133] It is understandable that this embodiment introduces deviation costs. Used to indicate path switching tendency or adjustment step size. When the distance is large, travelers are more likely to change their route choices; when When travel is less frequent, travelers are more likely to continue along their original routes. The basis for travelers at different levels to adjust their routes is not current cost, but rather cost induced by predicted traffic flow.

[0134] The updated traffic flow predictions for the next day are thus obtained for the three levels of travelers:

[0135]

[0136]

[0137]

[0138] In the formula, , , The predictions of travelers with k=0, k=1, and k=2 regarding traffic flow in the road network on day t+1 are given. , , To solve for the optimized projection operator, The deviation cost from the existing path after the update;

[0139] In the path selection incorporating real-time traffic dynamics, the k-th level traveler calculates the predicted travel cost based on the traffic flow forecast for the next day, and updates the path flow based on the predicted travel cost. The updated path flow is an adjustment rule in the form of projection, specifically, the traveler updates the path flow along the direction of improvement of the predicted travel cost, and the projection operator ensures that the update result satisfies the non-negativity and demand conservation constraints.

[0140] S4: Based on the updated traveler path flow at each level, set the shortest and longest green light times for each signal phase, and introduce a fixed phase sequence to constrain the switching logic and execution order of the signal phases, thereby completing the collaborative optimization of traffic allocation and maximum pressure signal control.

[0141] Optionally, setting the shortest and longest green light times for each signal phase specifically involves:

[0142]

[0143] In the formula, For the shortest green light time, The actual green light time for the i-th signal phase. The longest green light time, The total available green light time within a signal cycle. Number the signal phase.

[0144] Optionally, in this embodiment, considering the actual pedestrian crossing needs at intersections, The value of is related to the length of the intersection, that is, The green light time is allocated according to the speed; at the same time, in order to prevent Occupying too much green time results in other phases not having enough green time allocated, therefore, The value of needs to reserve at least the minimum green light time for other phases.

[0145] Furthermore, under low traffic conditions, without constraining the execution order of signal phases, the maximum pressure control algorithm may exhibit disordered phase switching and logical confusion. For example, if the current executing phase is east-west straight-through, the next phase might jump directly to north-south left-turn. Such unconventional switching methods significantly conflict with drivers' general driving expectations and traffic habits, reducing the orderly flow of traffic at the intersection. To effectively avoid such disordered switching problems, this embodiment introduces a fixed phase sequence. By constraining the switching logic and execution order of signal phases, it makes them more consistent with actual traffic flow patterns and driver behavior characteristics, thereby improving the safety, stability, and overall traffic efficiency of the intersection while ensuring the flexibility of the control strategy.

[0146] Based on the specific implementation details, the effectiveness of the technical solution of the present invention will be demonstrated through experiments.

[0147] Specifically, such as Figure 3 As shown, this experiment develops a simulation environment based on the actual road network, and creates traffic flow files based on synthetic traffic flow data and real traffic flow data. Parameters are then set for the simulation environment, and the simulation environment is finally created.

[0148] Furthermore, the simulation evaluates vehicle operation by distinguishing different saturation states. The unsaturated state of this invention is defined as a traffic condition with a saturation level less than 0.9; the near-saturated state is defined as a traffic condition with a saturation level equal to 0.9; and the oversaturated state is defined as a traffic condition with a saturation level greater than 0.9. Under different conditions, the average queue length, average vehicle delay, and service traffic volume of vehicles in the road network are statistically analyzed and used as evaluation indicators. To verify whether the maximum pressure signal control fused with cognitive levels can improve the overall traffic efficiency of the road network after travelers at different levels change their travel routes based on received traffic information, this experiment also uses the average number of stops as an evaluation indicator. The feasibility of the proposed maximum pressure signal control fused with cognitive levels (CH-MP) is verified by comparing it with fixed signal control, original maximum pressure signal control (MP), Dijkstra's algorithm, cognitive level algorithm (CH), and maximum pressure signal control combined with Dijkstra's algorithm (Dijkstra-MP). It should be noted that the vehicle composition selected in this invention is a mixed traffic flow of CAV and HDV. Travelers with a cognitive level k≥1 are CAVs who can adjust their travel routes by utilizing real-time changes in traffic conditions.

[0149] Specifically, by Figure 4 , Figure 5 , Figure 10 As shown in Table 1, when the road network is in an unsaturated state, compared with fixed signal timing control, the present invention reduces the average queue length by 58.30%, the average vehicle delay by 80.52%, increases the number of vehicles served by 66.61%, and reduces the average number of stops by 53.54%; compared with the original maximum pressure signal control, the average queue length is reduced by 54.02%, the average vehicle delay by 70.22%, the number of vehicles served by 53.40%, and the average number of stops by 49.57%; compared with Dijkstra's algorithm, the average queue length is reduced by 24.94%. Compared to the cognitive hierarchy algorithm, the average queue length decreased by 58.30%, the average vehicle delay decreased by 72.59%, the number of vehicles served increased by 35.37%, and the average number of stops decreased by 3.07%. Compared to the maximum pressure signal control combined with Dijkstra's algorithm, the average queue length decreased by 3.53%, the average vehicle delay decreased by 40.74%, the number of vehicles served increased by 9.82%, and the average number of stops decreased by 15.57%.

[0150] Table 1 Simulation results of each scheme when unsaturated

[0151]

[0152] Furthermore, by Figure 6 , Figure 7 , Figure 10 As shown in Table 2, when the road network is near saturation, compared to fixed signal timing control, this invention reduces the average queue length by 77.61%, the average vehicle delay by 83.49%, increases the number of vehicles served by 76.37%, and reduces the average number of stops by 74.29%; compared to the original maximum pressure signal control, it reduces the average queue length by 71.87%, the average vehicle delay by 49.81%, increases the number of vehicles served by 60.24%, and reduces the average number of stops by 68.37%; compared to Dijkstra's algorithm, it reduces the average queue length by 35.83%. The average vehicle delay decreased by 75.69%, the number of vehicles served increased by 65.55%, and the average number of stops decreased by 33.34%. Compared to the cognitive hierarchy algorithm, the average queue length decreased by 32.46%, the average vehicle delay decreased by 69.26%, the number of vehicles served increased by 54.11%, and the average number of stops decreased by 18.37%. Compared to the maximum pressure signal control combined with Dijkstra's algorithm, the average queue length decreased by 19.95%, the average vehicle delay decreased by 2.54%, the number of vehicles served increased by 21.86%, and the average number of stops decreased by 30.13%.

[0153] Table 2 Simulation results of various schemes when near saturation

[0154]

[0155] Furthermore, by Figure 8 , Figure 9 , Figure 10 As shown in Table 3, when the road network is in an oversaturated state, compared with fixed signal timing control, the present invention reduces the average queue length by 32.65%, the average vehicle delay by 86.16%, increases the number of vehicles served by 78.99%, and reduces the average number of stops by 27.07%; compared with the original maximum pressure signal control, the average queue length is reduced by 41.90%, the average vehicle delay by 72.41%, the number of vehicles served by 61.72%, and the average number of stops by 36.74%; compared with Dijkstra's algorithm, the average queue length is reduced by 28.40%. The average vehicle delay decreased by 81.61%, the number of vehicles served increased by 66.86%, and the average number of stops decreased by 26.07%. Compared to the cognitive hierarchy algorithm, the average queue length decreased by 27.90%, the average vehicle delay decreased by 74.37%, the number of vehicles served increased by 52.16%, and the average number of stops decreased by 18.26%. Compared to the maximum pressure signal control combined with Dijkstra's algorithm, the average queue length decreased by 5.18%, the average vehicle delay decreased by 37.49%, the number of vehicles served increased by 21.44%, and the average number of stops decreased by 25.69%.

[0156] Table 3 Simulation results of various schemes under oversaturation

[0157]

[0158] In summary, this invention first divides travelers into three cognitive levels, enabling travelers at different levels to predict the traffic flow for the next day based on the previous day's traffic conditions, thereby forming corresponding route selection strategies. Secondly, a deviation cost parameter is introduced during the route update process, allowing travelers at cognitive levels k≥1 to dynamically adjust their travel routes using real-time traffic information from traffic activity assessment (CAV). Furthermore, leveraging the predictable arrival time at intersections provided by CAV, the CAV arrival time prediction is integrated into the pressure detection process of maximum pressure signal control, constructing a collaborative optimization process that combines cognitive level route allocation with maximum pressure signal control, thus coupling the traffic allocation model with the signal control strategy. Finally, the effectiveness of the model is verified through simulation and real-world analysis. This invention can be used for collaborative optimization of road network traffic conditions in mixed traffic environments with CAV, alleviating traffic pressure on urban arterial roads from a traffic guidance perspective. It overcomes the limitations of existing technologies where traffic congestion occurs on uncongested road sections after initial allocation, and the reduced optimization effect of the maximum pressure signal control model when queue overflow occurs. It exhibits strong adaptability to complex and changing traffic conditions.

[0159] The specific embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A collaborative optimization method for traffic assignment and maximum pressure signal control in an intelligent connected environment that integrates cognitive levels, characterized in that, The method includes the following steps: S1: Construct a road network that includes a set of nodes, a set of road segments, and a set of origin-destination pairs. Divide travelers in the road network into three levels according to their cognitive abilities. For each level of travelers, obtain the traffic flow forecast for the next day based on the exponential movement coefficient. Update the route selection based on the traffic flow forecast for the next day to obtain the route planning results for each level of travelers. S2: Based on the route planning results for the next day, optimize the signal timing of each intersection in the road network to complete the signal timing strategy for each intersection. Based on the signal timing strategy, at the end of the day, feed back the traffic situation of each intersection to travelers at each level. The optimization involves dividing the intersection pressure detection into real-time detected vehicles and predicted intelligent connected vehicles arriving at the intersection in the next cycle. The signal pressure of each intersection is calculated by the number of real-time detected vehicles and the number of predicted vehicles. S3: Based on the traffic conditions at each intersection, travelers at each level introduce deviation costs that deviate from the existing routes while minimizing travel costs. The predicted travel costs are calculated based on the deviation costs and the predicted traffic flow for the next day. The path flow is updated based on the predicted travel costs to obtain the updated path flow for travelers at each level. S4: Based on the updated traveler path flow at each level, set the shortest and longest green light times for each signal phase, and introduce a fixed phase sequence to constrain the switching logic and execution order of the signal phases, thereby completing the collaborative optimization of traffic allocation and maximum pressure signal control.

2. The method for coordinated optimization of traffic allocation and maximum pressure signal control in an intelligent connected environment based on integrated cognitive levels, as described in claim 1, is characterized in that... Specifically, S1 is: Construct a road network Where N is the set of nodes, L is the set of road segments, and W is the set of OD pairs. Any OD pair is defined as... Travel demand is The set of feasible paths is ; Travelers in the road network are classified into K levels based on their cognitive ability, where K=3, and the proportion of travelers in each of the K levels is defined as follows: and satisfy Define the set of feasible path flows for k types of travelers. for: ； ； ； in, Let d be the path flow vector of type k, and d be the OD demand vector. This is the OD path association matrix. For travelers of class k, select the traffic flow from route r to OD to w, where T represents the transpose operation; On day t, the path flow of the k-th type of traveler is defined as follows: In the road network, the paths of all travelers are ; Travelers get Then, based on its own cognitive level, it forms a prediction of traffic flow in the road network on day t+1. And obtain the corresponding predicted travel cost as This allows us to obtain an update of the travel path of the k-th type of traveler on day t+1. for: ； in, Let y be the projection operator. The target vehicle path flow in the middle, Adjust the sensitivity parameter for the path flow of a single positive real scalar, and , It is the exponential moving average, and ; in: ； ； In the formula, Let k be the set of feasible route flows for travelers. It is an empty set; This allows us to obtain route planning information for the next day's traffic patterns from travelers with different cognitive abilities, specifically: First, we define the traffic pattern of travelers with k=0 on day t as unchanged on day t+1, and define travelers with k=0 as non-connected HDV travelers. When all travelers in the road network are k=0 travelers, the road network becomes the traditional case without intelligent connected vehicles (CAVs), expressed as: ； in, For the traveler at k=0, predict the traffic flow in the road network on day t+1; Then, when k≥1, the k-traveler plans its own travel route by predicting the reactions of travelers at lower levels to the current path pattern. Travelers with k≥1 are defined as CAV (Consumer-Aided Vehicle) travelers, thus obtaining the normalized ratio of k-travelers to ℎ-travelers. for: ； In the formula, and This represents the percentage of travelers at the current traveler level. A traveler with k=1 considers all other travelers in the road network to be travelers with k=0, thus obtaining the traveler's prediction of traffic flow in the road network on day t+1. for: ； In the formula, and For projection operators, To normalize the proportion, For the predicted travel costs, and For prediction coefficients; This allows us to obtain the travel route update for the traveler with k=1 on day t+1. for: ； In the formula, For projection operators, The path flow for the k=1 type of traveler; Finally, the traveler with k=2 simultaneously predicts the responses of both the traveler with k=0 and the traveler with k=1, thus obtaining the normalized proportions as follows: and Define the traveler with k=2 to predict traffic flow in the road network on day t+1. for: ； In the formula, and For projection operators; This allows us to obtain the travel route update for the traveler with k=2 on day t+1. for: ； In the formula, For projection operators, This is the path flow for the k=2 type of traveler.

3. The method for coordinated optimization of traffic allocation and maximum pressure signal control in an intelligent connected environment based on integrated cognitive levels, as described in claim 1, is characterized in that... Specifically, S2 is: The signal timing at each intersection in the road network is optimized using the following expression: ； In the formula, This represents the weight difference between road segment l and road segment m during time period t. This represents the queue length between road segment l and road segment m during time period t; Let m be the set of all road segments ending at point m. This represents the proportion of traffic flow from road segment m to road segment v during time period t. This represents the queue length between road segment m and road segment v during time period t; In a connected vehicle environment, intersections simultaneously contain both connected vehicle (CAV) and non-connected vehicle (HDV) traffic. Based on the predictability of CAV paths, intersection pressure detection is divided into real-time detection of vehicles at the intersection and prediction of CAV arrival times in the next cycle. The expression for the arrival time of a CAV at the intersection is as follows: ； ； In the formula, To predict the number of steps, Let n be the predicted speed of the nth vehicle at a future time. Let be the remaining distance of the nth vehicle from the intersection at the current time t. The period length, Let n be the estimated arrival time of the nth vehicle at the intersection; Thus, the signal phase is obtained. Number of intelligent connected vehicles (CAVs) arriving at the intersection for: ； After predicting the arrival times of intelligent connected vehicles (CAVs) at intersections in different signal phases, and combining the path planning of different cognitive levels on day t+1, the signal timing strategies for each intersection on day t+1 are completed by combining the number of vehicles detected in real time at the intersections with the number of vehicles predicted by the intelligent connected vehicles (CAVs). After day t+1 ends, the traffic conditions of each intersection on that day are returned to travelers at different cognitive levels, and travelers at different cognitive levels optimize their travel routes for the next day based on the traffic conditions of the previous day.

4. The method for coordinated optimization of traffic allocation and maximum pressure signal control in an intelligent connected environment based on integrated cognitive levels, as described in claim 2, is characterized in that... Specifically, S3 is: While minimizing travel costs, travelers need to incur deviation costs for deviating from their established routes, expressed as: ； ； In the formula, This represents the projection operator for solving the optimization problem. This represents the deviation cost from the existing path, and , The set of feasible path flows. Set of feasible path flows Candidate path flow vectors in; The path flow vector to be projected; The updated traffic flow predictions for the next day are thus obtained for the three levels of travelers: ； ； ； In the formula, , , The predictions of travelers with k=0, k=1, and k=2 regarding traffic flow in the road network on day t+1 are given. , , To solve for the optimized projection operator, The deviation cost from the existing path after the update; The k-th level traveler calculates the predicted travel cost based on the traffic flow forecast for the next day, and updates the path flow based on the predicted travel cost. The updated path flow is an adjustment rule in the form of projection. Specifically, the traveler updates the path flow along the direction of improvement of the predicted travel cost, and the projection operator ensures that the update result satisfies the non-negativity and demand conservation constraints.

5. The method for coordinated optimization of traffic allocation and maximum pressure signal control in an intelligent connected environment based on integrated cognitive levels, as described in claim 1, is characterized in that... The specific steps for setting the shortest and longest green light times for each signal phase are as follows: ； In the formula, The shortest green light time is determined by the length of the intersection. The actual green light time for the i-th signal phase. The longest green light time, The total available green light time within a signal cycle. Number the signal phase.

Images