Road congestion judging and adjusting method and system based on flow and vehicle speed

Abstract

The invention provides a road congestion judgment and adjustment method and system based on flow and vehicle speed, and relates to the technical field of traffic engineering, and the method comprises the steps: obtaining the traffic flow of a current road section, the average vehicle speed, the historical traffic flow, and the average vehicle speed; based on the historical traffic flow and the average vehicle speed information, probability distribution modeling is carried out on the flow-vehicle speed point cloud by adopting a mixed density network, and a traffic flow-vehicle speed relation curve is obtained; determining the maximum traffic volume and the congestion time period of the current road section based on the traffic flow-vehicle speed curve; traffic jam charging is introduced, the minimum total travel cost of a user is used as an optimization target, historical traffic flow and average vehicle speed are used as constraint conditions, a bottleneck mathematical model of traffic adjustment is constructed and solved, and an optimal solution is used for a corresponding traffic control strategy to endow the user with equivalent travel cost. And the target of corresponding road traffic adjustment is achieved.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of traffic engineering, in particular to a road congestion determination and regulation method and system based on traffic flow and vehicle speed. BACKGROUND

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute the prior art.

[0003] The current traffic infrastructure construction speed is slower than the growth speed of traffic demand, and the continuously increasing traffic flow leads to frequent highway traffic congestion, which in turn causes a series of serious subsequent problems, and it is urgent to regulate and control road traffic flow to govern it.

[0004] Traffic congestion discriminates accurately define the traffic state and grasp the traffic control opportunity, which is an important prerequisite for formulating effective induction strategies. The current evaluation of road traffic operation state is mainly based on the ratio V / C of road traffic flow V and road capacity C, and this method divides the road service state into congestion, stable flow and free flow states by discretization. Although this method has certain reference value in practical application, it generally relies on static data and fixed threshold, and it is difficult to reflect the dynamic change characteristics of traffic operation state, and it is also difficult to effectively identify and analyze the frequently congested area and its time distribution law, so it is difficult to provide strong support for precise traffic regulation and management. With the continuous improvement of the complexity of the traffic system, it is difficult to meet the actual needs of efficient traffic management by relying only on traditional methods.

[0005] Meanwhile, the development of big data technology provides strong support for dynamic perception and accurate analysis of traffic conditions. For example, by analyzing the toll and traffic data collected by the highway gantry system, key parameters such as traffic flow and vehicle speed on the road section can be obtained in real time. This provides data basis and technical support for the upgrading of traffic state recognition methods. However, the above methods are based on manually set traffic characteristics to carry out traffic state prediction research, and the traffic state parameters are not screened during the research, but the traffic state is set as congestion, crowded, and smooth, and the discrete traffic state coding method is used to describe the large amount of collected traffic state data as the corresponding traffic state vector, combined with the deep learning algorithm to propose a multi-classification algorithm for traffic state discrimination, which will lead to a large amount of data calculation, increase the time cost, and cannot quickly obtain the recognition of the road state. In addition, the existing method usually only relies on a single parameter such as flow and speed to fit, ignoring the differences of traffic state under different time, space and environmental conditions. For example, under the same flow condition, the running speed of the road section may be significantly different in the morning peak, at night, in sunny weather, and in rainy weather. If these context factors are not considered, the model may be too idealized and cannot accurately reflect the diversity and non-linear characteristics of traffic operation. In addition, in the method of model optimization solution, the traditional model optimization is usually modeled as a bi-level programming problem: the upper layer takes the minimization of the total travel cost of the system as the objective, and the lower layer takes the user equilibrium as the constraint. Although this method can obtain an equilibrium solution in theory, it has several shortcomings in practical application: first, the model structure is complex, and the solution often relies on large-scale iterative calculation, which is difficult to meet the real-time requirement; second, when the traffic demand has randomness or there are accidents, weather and other uncertain disturbances, the bi-level programming is difficult to adjust in time, resulting in insufficient stability of the results; third, the optimization process needs to make idealized assumptions about the behavior of travelers, which deviates from the actual situation. SUMMARY

[0006] To solve the above problems, the present disclosure proposes a road congestion determination and adjustment method and system based on traffic flow and speed, which fits the traffic flow-speed curve by analyzing real-time traffic data and historical data, determines the service level change and congestion period of the road section, uses an improved bottleneck model, introduces traffic congestion toll, and recalculates the toll scheme by taking the historical road traffic flow data as a new constraint condition, to achieve the corresponding road traffic adjustment target by giving the equivalent travel cost to the user.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions: The road congestion determination and adjustment method based on traffic flow and speed comprises: obtaining the traffic flow, average vehicle speed of the current road section, and historical traffic flow, average vehicle speed; Based on historical traffic flow and vehicle average speed information, a mixed density network is used to model the conditional probability distribution of the flow-speed point cloud, and the traffic flow-speed relationship curve of the current road section is obtained; Based on the traffic flow-speed curve, the maximum traffic volume of the current road section and the congestion time period of the current road section are determined; The traffic congestion charge is introduced, the minimum total travel cost of the user is taken as the optimization objective, the historical traffic flow and the vehicle average speed are taken as the constraint conditions, a bottleneck mathematical model for traffic regulation is constructed, a reinforcement learning method is introduced to solve the bottleneck mathematical model, and the optimal solution is used for the corresponding traffic control strategy, so that the equivalent travel cost is given to the user, and the corresponding road traffic regulation target is achieved.

[0008] According to some embodiments, the present disclosure adopts the technical solutions as follows: A road congestion determination and regulation system based on traffic flow and speed includes: A data acquisition module is configured to acquire the traffic flow, vehicle average speed, historical traffic flow, and vehicle average speed of a current road section; A curve fitting module is configured to model the conditional probability distribution of the flow-speed point cloud based on historical traffic flow and vehicle average speed information using a mixed density network, and obtain the traffic flow-speed relationship curve of the current road section; A judgment module is configured to determine the maximum traffic volume of the current road section and the congestion time period of the current road section based on the traffic flow-speed curve; A regulation module is configured to introduce a traffic congestion charge, take the minimum total travel cost of the user as the optimization objective, take the historical traffic flow and the vehicle average speed as the constraint conditions, construct a bottleneck mathematical model for traffic regulation, introduce a reinforcement learning method to solve the bottleneck mathematical model, and use the optimal solution for the corresponding traffic control strategy to give the equivalent travel cost to the user, and achieve the corresponding road traffic regulation target.

[0009] According to some embodiments, the present disclosure adopts the technical solutions as follows: A computer program product includes a computer program, which, when executed by a processor, implements the road congestion determination and regulation method based on traffic flow and speed.

[0010] According to some embodiments, the present disclosure adopts the technical solutions as follows: A non-transitory computer readable storage medium is configured to store computer instructions, which, when executed by a processor, implement the road congestion determination and regulation method based on traffic flow and speed.

[0011] According to some embodiments, the present disclosure adopts the technical solutions as follows: An electronic device comprises a processor, a memory and a computer program; wherein the processor is connected with the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the road congestion determination and adjustment method based on traffic flow and vehicle speed.

[0012] Compared with the prior art, the beneficial effects of the present disclosure are: The road congestion determination and adjustment method based on traffic flow and vehicle speed of the present disclosure can comprehensively analyze traffic flow and vehicle speed data, learn conditional probability distribution by using a mixed density network to probabilistically model traffic flow-vehicle speed point clouds based on historical traffic flow and vehicle average speed information, and obtain a traffic flow-vehicle speed relationship curve of a current road section. The curve can represent a multimodal speed distribution under different traffic phases, so as to more accurately identify maximum traffic volume and critical speed. The maximum traffic volume of the current road section and the congestion time period of the current road section are determined based on the traffic flow-vehicle speed curve. Accurate fitting of the traffic flow-vehicle speed curve can better understand the road capacity, thereby improving overall traffic efficiency and reducing congestion time. The method can accurately identify frequently occurring congestion nodes and timely discover traffic bottlenecks.

[0013] The road congestion determination and adjustment method based on traffic flow and vehicle speed of the present disclosure introduces traffic congestion charging, takes the minimum total travel cost of users as the optimization objective, and takes historical traffic flow and vehicle average speed as the constraint condition to construct a bottleneck mathematical model for traffic adjustment. Based on the analysis of historical and real-time data, a more accurate traffic flow adjustment scheme can be developed to optimize the service level of the road. The reinforcement learning method is introduced to solve the bottleneck mathematical model, and the optimal solution is used for the corresponding traffic control strategy to give the equivalent travel cost to the user, so as to achieve the corresponding road traffic adjustment target. Through the reinforcement learning-based solving method, the bottleneck model can be adaptively optimized in a complex and uncertain traffic environment. Compared with the traditional mathematical programming method, the method has stronger real-time performance, self-learning ability and generalization ability, thereby providing more efficient and intelligent technical support for traffic congestion adjustment.

[0014] The road congestion determination and adjustment method based on traffic flow and vehicle speed of the present disclosure can use historical and current traffic data to provide early warning of possible congestion and provide a scientific basis for preventive traffic management and diversion. Through effective traffic flow adjustment and congestion management, the occurrence of traffic accidents can be reduced, and the safety of road traffic can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0015] The accompanying drawings, which form a part of the present disclosure, are used to provide a further understanding of the present disclosure, and the schematic embodiments of the present disclosure and the description thereof are used to explain the present disclosure, and do not constitute an improper limitation on the present disclosure.

[0016] Figure 1 Traffic flow-speed relationship curve for an embodiment of the present disclosure; Figure 2 Flow and vehicle speed based road congestion determination and regulation method flowchart for an embodiment of the present disclosure. DETAILED DESCRIPTION

[0017] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0018] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the present disclosure. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this present disclosure belongs.

[0019] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the present disclosure. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, steps, operations, elements, components, and / or groups thereof, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or groups thereof.

[0020] Embodiment 1 A flow and vehicle speed based road congestion determination and regulation method is provided in an embodiment of the present disclosure, the method steps comprising: Step one: obtaining the current traffic flow, vehicle average speed and historical traffic flow, vehicle average speed of the current road section; Step two: based on the historical traffic flow and vehicle average speed information, using a hybrid density network to model the conditional probability distribution of the flow-vehicle speed point cloud, obtaining the traffic flow-vehicle speed relationship curve of the current road section; Step three: determining the maximum traffic volume of the current road section and the congestion time period of the current road section based on the traffic flow-vehicle speed curve; Step four: introducing traffic congestion charges, taking the minimum total travel cost of users as the optimization objective, taking the historical traffic flow and vehicle average speed as the constraint condition, constructing a bottleneck mathematical model for traffic regulation, introducing a reinforcement learning method to solve the bottleneck mathematical model, and using the optimal solution for the corresponding traffic control strategy to give the equivalent travel cost to the users, achieving the corresponding road traffic regulation goal.

[0021] As an embodiment, the traffic flow and vehicle speed based road congestion determination and adjustment method of the present disclosure can accurately determine the service level change and congestion period of the road section by analyzing the real-time data and historical data collected by the traffic detector, fitting the traffic flow-vehicle speed curve, and formulating the targeted traffic adjustment scheme, so as to improve the road passing efficiency and management level. The specific implementation process is as follows: Step 1: Obtain the traffic flow, vehicle average speed of the current road section, and historical traffic flow, vehicle average speed; Specifically, taking the traffic data of a certain highway gantry as an example, the PeMS data set is selected as a substitute. PeMS (Performance Measurement System) is an open-source road traffic data platform, which contains 4,681 detectors and reports data every 30 seconds. The data content of the PeMS system is rich, covers a wide time period, and has high research value. The system compiles data into 30-second interval records and continuously aggregates into 5-minute incremental data, with small detection errors and high intelligence.

[0022] Select a certain traffic node on a certain road and the corresponding time period (from January 2023 to December 2023), and use an automated program to download the historical traffic flow and speed data of the node. Then, sort the traffic data, convert the data obtained at a certain collection frequency (for example, every five minutes) into hourly traffic flow, and calculate the average speed of all vehicles in the current period. Integrate these historical traffic data, aggregate the historical data of each week (such as every Monday), each month (30-day average), and specific holidays together, and calculate the average traffic flow and speed.

[0023] Further, obtain the road traffic data at a certain collection frequency (for example, every five minutes), and convert it into hourly traffic flow and average speed of all vehicles collected in the current period; Step 2: Based on the historical traffic flow and vehicle average speed information, use a hybrid density network to model the conditional probability distribution of the flow-vehicle speed point cloud, and obtain the traffic flow-vehicle speed relationship curve of the current road section; Specifically, the sorted traffic data is plotted as a coordinate system graph with vehicle speed as the vertical axis and traffic flow as the horizontal axis. The historical traffic data of a certain node in a certain time period is represented in the form of points in the coordinate system graph. A hybrid density network is used to model the conditional probability distribution of the flow-vehicle speed point cloud, and learn the conditional probability distribution , to obtain the traffic flow-vehicle speed relationship curve of the current road section, which specifically includes: Taking traffic flow as input features and vehicle average speed as output variables, a neural network outputs the mixing parameters of multiple Gaussian distributions with parameter set wherein, is a weight, is a mean, is a covariance matrix, the model is trained by maximum likelihood method, so as to obtain the conditional probability distribution: (1) wherein, is a weight of the K th Gaussian component, is a mean of the K th Gaussian component, is a covariance matrix of the th Gaussian component under the condition. K In the inference stage, by inputting the given traffic flow

[0024] and the context features , the corresponding speed distribution is generated, and a multimodal traffic flow-speed relationship curve can be drawn. The peak value of the curve and the corresponding speed are used to determine the maximum traffic volume and the critical speed, and different modes correspond to different traffic operation phases (such as free flow, transition flow, and congested flow). The context features here refer to information related to traffic operation, such as time period, spatial section, environmental conditions, real-time operation characteristics, and emergencies. The model not only learns the “average curve”, but also learns a set of different curves under different time, place, and environment, truly achieving “adapting to local conditions and time”.

[0025] Through the traffic flow-speed curve, the peak value of the curve is taken as the maximum traffic volume, and the speed corresponding to the peak value is taken as the critical speed, so as to obtain the maximum traffic capacity of the current section according to the determined maximum traffic volume and the critical speed .

[0026] Step 3: determining the maximum traffic volume of the current section and the congestion time period of the current section based on the traffic flow-speed curve; Specifically, first, the road service level change is classified, and the classification content is shown in Table 1.

[0027] Table 1 Classification content of road service level change

[0028] ​Furthermore, based on the criteria in Table 1, the determined road capacity and traffic flow data for each time period are used to analyze changes in the road's service level every five minutes on a weekly, monthly, and specific holiday basis. Periods with a service level below Level 4 for more than one consecutive hour are identified as congested periods for the current road segment. For these congested periods and road segments, corresponding traffic flow adjustment plans are developed to optimize road efficiency and reduce congestion.

[0029] Step 4: Introduce traffic congestion charging, set the minimum total travel cost of users as the optimization goal, use historical traffic flow and average vehicle speed as constraints, construct a bottleneck mathematical model for traffic regulation, introduce reinforcement learning methods to solve the bottleneck mathematical model, and use the optimal solution for the corresponding traffic control strategy to assign equivalent travel costs to users and achieve the corresponding road traffic regulation goals.

[0030] Specifically, in traditional road traffic regulation schemes based on bottleneck models, the original bottleneck model assumes that travelers have complete departure time information (i.e., users know in advance the road conditions at each time period on the road section), and solves the road traffic flow in equilibrium based on this constraint. This assumption is obviously inconsistent with the actual situation (it is difficult for users to understand the traffic conditions of each section on their chosen path from the moment of departure). Therefore, this disclosure proposes an improved solution, using the compiled historical road traffic flow data as a new constraint, introducing traffic congestion charges, and recalculating the charging scheme. The specific strategy is as follows: The node is known to be passable per unit time vehicles (maximum capacity), when the arrival rate of commuters at the bottleneck exceeds When the bottleneck is reached, a queue will appear on the road. The rate of change at time can be expressed as: (2) Where: for The length of the queue at the bottleneck at any given moment; for The rate of commuters entering the bottleneck at any given moment; is the unit time capacity at the bottleneck. After the introduction of traffic congestion charges, the total cost of a commuter's trip is composed of three parts: the total travel time cost (including queuing and travel time), the penalty for arriving early or late, and the traffic congestion charge. Therefore The total travel cost for a commuter departing at time can be expressed as: (3) Where: is the time value coefficient; Penalty factor for arriving early to work; Penalty coefficient for being late for work; The travel time for commuters from their accommodation to their workplace; It’s time to start working; The earliest departure time for travelers; The latest departure time for travelers; Departure time to arrive at work on time; for The cumulative cost of the moment; For travelers The variable charging costs that should be borne by departure at any time; represents the queuing time of commuters at the bottleneck.

[0031] Use historical traffic flow data as , we can get the additional cost that should be given in a certain period: (4) Combined with the subsequent regulation mechanism, the corresponding traffic control strategy can be used to assign equivalent travel costs to users to achieve the corresponding road traffic regulation goals.

[0032] In some embodiments of the present disclosure, a reinforcement learning method is introduced to the proposed bottleneck mathematical model solution process to achieve intelligent dynamic optimization. Specifically, it includes: (1) Environmental modeling: The bottleneck model is constructed as a reinforcement learning environment, where the environmental state consists of current traffic flow, average vehicle speed, queue length, historical toll levels, and contextual features (such as time period, weather, and road conditions). Environmental transitions are determined by a bottleneck queuing dynamics formula, which updates queue length and delay within each time slice based on vehicle arrival rate, road service capacity, and toll levels.

[0033] (2) Action space: The agent selects the amount to adjust the charging level at each discrete time slice. The action can be discrete (for example, setting a number of charging levels) or continuous (selecting any charging value within the allowed range).

[0034] (3) Reward function: The reward function is the negative of the system's total travel cost, which includes the vehicle's queuing time and travel time, early or late arrival penalties, and congestion charges. Maximizing the cumulative reward is equivalent to minimizing the system's total travel cost.

[0035] (4) Agent training: The agent is trained using a deep reinforcement learning algorithm. For example, a deep Q network (DQN) can be used in a discrete action space, and a proximal policy optimization (PPO), a deep deterministic policy gradient (DDPG), or a soft actor-critic (SAC) method can be used in a continuous action space. Through repeated interaction with the bottleneck model environment, the agent gradually learns the optimal tolling strategy under different traffic conditions.

[0036] (5) Online deployment and rolling optimization: In practical applications, the system obtains the latest traffic data at preset time intervals (e.g., 5 minutes or 15 minutes) and inputs them into the trained reinforcement learning agent, which outputs the tolling strategy in real time. At the same time, combined with online learning and rolling optimization mechanisms, when traffic conditions are abnormal (e.g., accidents, holiday traffic surges), the agent can quickly adapt and update the strategy, thereby maintaining the robustness of the tolling adjustment effect.

[0037] Through the above solution method based on reinforcement learning, the present disclosure can achieve adaptive optimization of the bottleneck model in a complex and uncertain traffic environment, and has stronger real-time performance, self-learning ability, and generalization ability compared to traditional mathematical programming methods, thereby providing more efficient and intelligent technical support for traffic congestion regulation.

[0038] After completing the bottleneck mathematical model solution and traffic congestion cost function design, the key task of the subsequent regulation mechanism is to convert the optimal tolling result and travel cost distribution strategy into an actual executable traffic control means, which specifically includes: (1) Time-of-day dynamic tolling mechanism Based on the optimal congestion tolling values p(t) obtained by model solving, a refined time-of-day dynamic tolling table can be designed, which divides a day into multiple 5-minute or 15-minute intervals, sets a higher fee during peak hours, and sets a low fee or even free during off-peak or smooth hours, guiding travelers to stagger their travel times.

[0039] (2) Induction information publishing system Through channels such as traffic induction screens, navigation applications, and mobile terminal Apps, real-time predictive congestion information and travel cost information (including queue delay, toll amount, etc.) are published to improve user awareness and enable them to adjust their departure times or routes based on the cost minimization principle even without a "god's eye view".

[0040] (3) Linkage signal control and lane allocation strategy When the predicted queue length at a node exceeds a safety threshold, the upstream and downstream signal light durations, the main and auxiliary lane allocation, and the opening of emergency lanes can be linked to release the traffic capacity and relieve the bottleneck pressure without adding infrastructure.

[0041] (4) Feedback correction mechanism The system can continuously collect traffic flow and speed data after implementation, adaptively update the model, dynamically correct the estimation error and congestion pricing strategy, and form a closed loop of "prediction-regulation-feedback-reoptimization".

[0042] Embodiment 2 In an embodiment of the present disclosure, a road congestion determination and regulation system based on traffic flow and vehicle speed is provided, comprising: A data acquisition module for acquiring traffic flow, vehicle average speed, and historical traffic flow and vehicle average speed of a current road section; A curve fitting module for modeling the conditional probability distribution of the flow-vehicle speed point cloud using a hybrid density network based on historical traffic flow and vehicle average speed information, to obtain the traffic flow-vehicle speed relationship curve of the current road section; A judgment module for determining the maximum traffic volume of the current road section and the congestion time period of the current road section based on the traffic flow-vehicle speed curve; A regulation module for introducing traffic congestion charges, taking the minimum total travel cost of users as the optimization objective, taking historical traffic flow and vehicle average speed as constraint conditions, constructing a bottleneck mathematical model for traffic regulation, solving the bottleneck mathematical model, and using the optimal solution for the corresponding traffic control strategy to give users equivalent travel costs, and achieving the goal of corresponding road traffic regulation.

[0043] Embodiment 3 In an embodiment of the present disclosure, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the road congestion determination and regulation method based on traffic flow and vehicle speed.

[0044] Embodiment 4 In an embodiment of the present disclosure, a non-transitory computer readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the road congestion determination and regulation method based on traffic flow and vehicle speed.

[0045] Embodiment 5 In an embodiment of the present disclosure, an electronic device is provided, comprising a processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory; when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes the road congestion determination and regulation method based on traffic flow and vehicle speed.

[0046] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0047] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flow or blocks. Figure 1 one or more flow or blocks.

[0048] Although the present disclosure has been described with reference to the embodiments thereof, it is apparent that a variety of modifications or changes can be made thereto without departing from the scope of the present disclosure.

Claims

1. A method for determining and adjusting road congestion based on traffic flow and vehicle speed, characterized in that: include: Obtain the traffic flow, average vehicle speed of the current road section as well as the historical traffic flow and average vehicle speed; Based on historical traffic flow and average vehicle speed information, a mixed density network is used to perform conditional probability distribution modeling on the traffic flow-speed point cloud to obtain the traffic flow-speed relationship curve of the current road section. Determine the maximum traffic volume of the current road section and the congestion time period of the current road section based on the traffic flow-vehicle speed curve; Traffic congestion charging is introduced, and the minimization of the user's total travel cost is taken as the optimization goal. Historical traffic flow and average vehicle speed are used as constraints. A bottleneck mathematical model of traffic regulation is constructed, and the bottleneck mathematical model is solved. The optimal solution is used for the corresponding traffic control strategy to give users an equivalent travel cost and achieve the corresponding road traffic regulation goal.

2. The method for determining and adjusting road congestion based on traffic flow and vehicle speed according to claim 1, wherein: Obtain road traffic data at the set collection frequency and convert it into hourly traffic flow and the average speed of all vehicles collected during the current period; integrate the acquired historical traffic data, aggregate the historical traffic data corresponding to each week, month, and specific holidays, and calculate the historical average traffic flow and average speed.

3. The method for determining and adjusting road congestion based on traffic flow and vehicle speed according to claim 1, wherein: Based on the historical average traffic flow and average vehicle speed, a mixture density network is used to model the conditional probability distribution of the flow-speed point cloud. The historical average traffic flow is used as the input feature of the mixture density network, and the average vehicle speed is used as the output variable. The neural network is used to output the mixture parameters of multiple Gaussian distributions. The model is trained using the maximum likelihood method to generate the corresponding speed distribution and draw a multimodal traffic flow-speed relationship curve.

4. The method for determining and adjusting road congestion based on traffic flow and vehicle speed according to claim 1, wherein: Based on the maximum capacity of the current road section determined by the multimodal traffic flow-speed relationship curve and the traffic flow data of each time period, the changes in road service levels within the set time of each week, month and specific holidays are analyzed, and the time periods where the service level is lower than the set level for more than the set time period are screened out and identified as the congested period of the current road section.

5. The method for determining and adjusting road congestion based on traffic flow and vehicle speed according to claim 1, wherein: The maximum traffic capacity per unit time at a node is cars, when the arrival rate of users at the bottleneck exceeds When , a queue will be formed on the road section, and the queue length at the bottleneck is calculated. t The rate of change at each moment is calculated, traffic congestion charges are introduced, and the total travel cost of a commuter's trip is minimized as the optimization goal. Historical traffic flow data is used as constraints, an objective function is constructed, and reinforcement learning methods are introduced to solve the additional cost that should be given in a certain period of time.

6. The method for determining and adjusting road congestion based on traffic volume and vehicle speed according to claim 5, characterized in that: The total travel cost of a commuter's trip includes three parts: total travel time cost, early or late arrival penalties, and traffic congestion charges. The total travel time cost includes queuing and driving time.

7. A road congestion determination and adjustment system based on traffic flow and vehicle speed, characterized in that: include: The data acquisition module is used to obtain the traffic flow, average vehicle speed of the current road section and the historical traffic flow and average vehicle speed; The curve fitting module is used to model the conditional probability distribution of the traffic flow-speed point cloud using a mixed density network based on historical traffic flow and average vehicle speed information, and obtain the traffic flow-speed relationship curve of the current road section; A judgment module, configured to determine the maximum traffic volume of a current road section and a congestion time period of the current road section based on a traffic flow-vehicle speed curve; The regulation module is used to introduce traffic congestion charges, take the minimization of users' total travel costs as the optimization goal, use historical traffic flow and average vehicle speed as constraints, construct a bottleneck mathematical model for traffic regulation, introduce reinforcement learning methods to solve the bottleneck mathematical model, and use the optimal solution for the corresponding traffic control strategy to assign equivalent travel costs to users and achieve the corresponding road traffic regulation goals.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the road congestion determination and adjustment method based on traffic flow and vehicle speed according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium, characterized in that The non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by the processor, the road congestion determination and adjustment method based on traffic flow and vehicle speed as described in any one of claims 1 to 6 is implemented.

10. An electronic device, characterized in that: include: A processor, a memory, and a computer program; wherein the processor is connected to the memory, and the computer program is stored in the memory. When the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the road congestion determination and adjustment method based on traffic flow and vehicle speed as described in any one of claims 1 to 6.

Images