Intelligent traffic control method and system based on real-time traffic flow detection

Abstract

The invention provides an intelligent traffic control method and system based on real-time traffic flow detection, belongs to the technical field of intelligent traffic control, and aims to solve the problems of single traffic flow detection, inaccurate data acquisition, lack of overall network analysis, inaccurate traffic flow prediction and unreasonable signal lamp timing in the prior art. Comprising the steps of obtaining vehicle transmission information data, supplementing and correcting the vehicle transmission information data through a mobile terminal and road network monitoring data, performing edge calculation analysis, preprocessing analyzed vehicle information, and constructing a prediction model through a machine learning algorithm to perform traffic flow prediction. A graph theory model is adopted to describe an intersection relation, and a self-organized traffic network is created; defining three indexes through the predicted traffic flow, and constructing a traffic light timing calculation model; and calculating an optimal timing strategy by adopting a chaotic particle swarm algorithm to realize cooperative control. The traffic flow prediction accuracy and the signal lamp timing rationality are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent traffic control, and particularly relates to an intelligent traffic control method and system based on real-time traffic flow detection. Background Art

[0002] The statements in this part only provide background technical information related to the present invention and do not necessarily constitute prior art.

[0003] Currently, traffic lights are installed at a large number of intersections on urban roads to facilitate motor vehicles and pedestrians to pass according to the traffic light rules. However, if the release times of the traffic lights at multiple intersections are set unreasonably, it will lead to road congestion and make the traffic conditions in the whole city very complex.

[0004] The signal timing method of traditional traffic control systems is usually fixed timing, and each intersection often works independently, lacking coordination and optimization, unable to be flexibly adjusted according to the real-time traffic conditions, and easily leading to a decline in traffic efficiency.

[0005] Therefore, with the development of the automotive industry, modern urban traffic has gradually developed towards intelligent traffic, and began to gradually adopt artificial intelligence technology to design signal control systems, and various intelligent algorithms have begun to be applied to traffic control systems.

[0006] However, in the existing intelligent traffic control system technology, the traffic flow detection method is single, the data collection accuracy is not high, and it is difficult to meet the needs of intelligent traffic systems. Traditional systems also fail to comprehensively manage multiple adjacent intersections as a complex network, resulting in inaccurate overall prediction of the system and low traffic management efficiency during peak traffic flow periods. In the existing traffic control technology, it is impossible to ensure the accuracy of traffic flow prediction and the rationality of signal timing, and it is even more impossible to improve traffic efficiency. Summary of the Invention

[0007] To overcome the deficiencies of the above-mentioned prior art, the present invention provides an intelligent traffic control method and system based on real-time traffic flow detection, which breaks the traditional fixed signal timing scheme, associates the signal timing cycle with the traffic flow data in the network, realizes dynamic signal timing based on real-time traffic flow detection, can adjust the signal timing sequence of intersections in real time, and improves traffic efficiency. Multiple means such as road network image recognition, mobile terminal quantity estimation, and vehicle networking data reporting are used for traffic flow measurement. Multiple adjacent intersections are regarded as a complex network for traffic flow data measurement, improving the accuracy of traffic flow prediction and the rationality of signal timing.

[0008] To achieve the above object, one or more embodiments of the present invention provide the following technical solutions:

[0009] The first aspect of the present invention provides an intelligent traffic control method based on real-time traffic flow detection.

[0010] An intelligent traffic control method based on real-time traffic flow detection, comprising:

[0011] Obtaining vehicle transmission information data through an intelligent connected vehicle OBU and performing calibration;

[0012] Performing edge computing analysis on the calibrated vehicle transmission information data and performing preprocessing;

[0013] Based on the preprocessed vehicle information, constructing a prediction model through a machine learning algorithm to predict traffic flow;

[0014] Describing intersection relationships using a graph theory model, analyzing network characteristics, and creating a self-organizing traffic network;

[0015] Defining three indicators of average vehicle queue length, average waiting time, and regional entry-exit ratio based on the predicted traffic flow;

[0016] Based on the self-organizing traffic network and the three indicators, constructing a traffic light timing calculation model;

[0017] Based on the traffic light timing calculation model, using a chaotic particle swarm algorithm to calculate and obtain an optimal timing strategy to perform coordinated control on multiple intersections in the area;

[0018] Wherein, based on the self-organizing traffic network and the three indicators, constructing a traffic light timing calculation model, the formula is;

[0019]

[0020] Wherein, α, β, γ, and δ all represent weight coefficients, represents the average delay time at the intersection; l q (X) is the average vehicle queue length at the intersection; n is the number of intersections; ρ(X) is the regional vehicle entry-exit ratio; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is a regularization term, λ n represents the green signal ratio, the proportion of the green light time in the total signal cycle time; λ avg represents the average green signal ratio.

[0021] As an implementation, calibrating the vehicle transmission information data to obtain the calibrated vehicle transmission information data, the specific process is:

[0022] Collect the terminal connection data of roadside 5G base stations in real time;

[0023] Estimate the regional vehicle density based on the terminal connection data of roadside 5G base stations, and supplement the vehicle transmission information data;

[0024] Calculate the number of road vehicles based on the road network monitoring data using AI algorithms;

[0025] Based on the regional vehicle density and the number of road vehicles, correct the vehicle transmission information data through the space-time alignment method to obtain the corrected vehicle transmission information data.

[0026] As an implementation, based on the preprocessed vehicle transmission information, construct a prediction model through machine learning algorithms for traffic flow prediction. Among them, the prediction model is an LSTM model, specifically:

[0027] Obtain historical period traffic data;

[0028] Input the historical period traffic data into the LSTM model for model fitting training to obtain a trained LSTM model;

[0029] Input the preprocessed vehicle transmission information into the trained LSTM model to predict the traffic flow at the next moment.

[0030] As an implementation, use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network. Specifically, based on the graph theory model to describe intersection relationships, use vehicles and roadside facilities as network nodes, equip communication and computing devices, and use C-V2X and Wi-Fi Direct technologies to dynamically connect and self-organize network topologies.

[0031] As an implementation, the formula for calculating the average vehicle queue length is:

[0032]

[0033] where l q is the average vehicle queue length; n is the number of intersections; represents the remaining queue length of intersection i; represents the current queue length of intersection i; a and b are weight coefficients, representing the contribution ratios of the remaining queue length and the newly added queue length respectively.

[0034] As an implementation, use the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, specifically:

[0035] Based on the self-organizing traffic network, with the shortest average vehicle queue length, the least average waiting time, and the smallest regional entry-exit ratio as the optimization objectives, the chaotic particle swarm optimization algorithm is used to obtain the optimal solution of the traffic signal timing calculation model;

[0036] Convert the optimal solution of the traffic signal timing calculation model into a signal control instruction, that is, obtain the optimal timing strategy.

[0037] The second aspect of the present invention provides an intelligent traffic control system based on real-time traffic flow detection, including:

[0038] A data acquisition module, configured to: obtain vehicle transmission information data through an intelligent networked vehicle OBU and perform calibration;

[0039] An edge computing module, configured to: perform edge computing analysis on the calibrated vehicle transmission information data and perform preprocessing;

[0040] A core computing module, configured to: based on the preprocessed vehicle information, construct a prediction model through a machine learning algorithm to predict traffic flow;

[0041] Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network;

[0042] Through the predicted traffic flow, define three indicators: average vehicle queue length, average waiting time, and regional entry-exit ratio;

[0043] Based on the self-organizing traffic network and the three indicators, construct a traffic signal timing calculation model;

[0044] Based on the traffic signal timing calculation model, use the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, and perform coordinated control on multiple intersections in the region;

[0045] Among them, based on the self-organizing traffic network and the three indicators, construct a traffic signal timing calculation model, and the formula is;

[0046]

[0047] Among them, α, β, γ, and δ all represent weight coefficients, represents the average delay time at the intersection, that is, the average waiting time; l q (X) is the average vehicle queue length at the intersection; n is the number of intersections; ρ(X) is the regional vehicle entry-exit ratio; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is the regularization term, λn It represents the green signal ratio, which is the proportion of the green light time in the total signal cycle time; λ avg It represents the average green signal ratio.

[0048] The third aspect of the present invention provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in a method as described in the first aspect of the present invention.

[0049] The fourth aspect of the present invention provides a computer-readable storage medium, on which a program is stored. When the program is executed by a processor, it implements the steps in a method as described in the first aspect of the present invention.

[0050] The fifth aspect of the present invention provides a computer program product containing instructions. When it runs on a computer, it enables the computer program to implement the steps in a method as described in the first aspect of the present invention when executed by a processor.

[0051] The above one or more technical solutions have the following beneficial effects:

[0052] The present invention adopts a variety of data source collection means, uses 5G intelligent devices as a supplement to traffic flow data, and introduces relatively mature road network monitoring for vehicle flow data correction, improving the accuracy of vehicle flow measurement.

[0053] The present invention enhances vehicle-road cooperation. By adopting the new generation of 5G communication technology and combining with the industry-wide C-V2X vehicle networking technology standard, it realizes real-time communication between vehicles (V2V), between vehicles and roadside facilities (V2I), and between vehicles and pedestrians (V2P). Through V2P and V2I, it can predict traffic control signals, forward traffic flow, speed limit signs, road gradients, etc. in advance, helping adjacent vehicles to reasonably plan the driving route and improving the accuracy of vehicle-road cooperation control.

[0054] The present invention uses a graph theory model to describe the intersection relationship, manages multiple adjacent intersections as a complex network, improves the accuracy of vehicle flow prediction and the rationality of signal timing, enhances the management and optimization ability of the overall traffic network, and realizes the optimization of the traffic network.

[0055] The present invention collects traffic vehicle information, defines three signal timing indicators in signal timing, and uses the chaotic particle swarm algorithm to realize dynamic adjustment of signal timing, reduce urban traffic congestion, improve road traffic capacity, and improve traffic efficiency.

[0056] The overall effect of the present invention is reflected in: under other ideal environmental conditions, the long-distance congestion situation at a single intersection is reduced by 20%, and the average waiting time of a single vehicle is reduced by 15%. It has established communication between basic roadside traffic facilities, vehicles, and pedestrians, realized information sharing, and at the same time laid a foundation for vehicle-road cooperation control.

[0057] Advantages of additional aspects of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments and descriptions thereof of the present invention are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0059] Figure 1 It is a flowchart of an intelligent traffic control method based on real-time traffic flow detection according to Embodiment 1 of the present invention;

[0060] Figure 2 It is a schematic diagram of the framework of an intelligent traffic control system based on real-time traffic flow detection according to Embodiment 2 of the present invention;

[0061] Figure 3 It is a schematic diagram of the obvious congestion situation on the road under fixed timing according to Embodiment 1 of the present invention;

[0062] Figure 4 It is a schematic diagram of a simulation experiment of an intelligent traffic control method based on real-time traffic flow detection according to Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0063] It should be noted that the following detailed description is exemplary and is intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0064] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0065] In the case of no conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other.

[0066] Embodiment 1

[0067] This embodiment discloses an intelligent traffic control method based on real-time traffic flow detection, which collects traffic vehicle information, defines three timing indicators in terms of signal light timing, and uses the chaotic particle swarm optimization algorithm to realize dynamic signal light timing adjustment, solve the traffic congestion problem caused by traditional fixed signal light timing, improve the accuracy of traffic flow measurement, and enhance the management and optimization ability of the overall traffic network, including:

[0068] S1. Obtain vehicle transmission information data through the on-board unit (OBU) of intelligent connected vehicles and perform calibration;

[0069] S2. Perform edge computing analysis on the corrected vehicle transmission information data and conduct preprocessing;

[0070] S3. Based on the preprocessed vehicle information, construct a prediction model through machine learning algorithms for traffic flow prediction;

[0071] S4. Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network;

[0072] S5. Define three indicators: average vehicle queue length, average waiting time, and regional entry / exit ratio based on the predicted traffic flow;

[0073] Based on the self-organizing traffic network and the three indicators, construct a traffic light timing calculation model;

[0074] Adopt the chaotic particle swarm algorithm to calculate and obtain the optimal timing strategy for collaborative control of multiple intersections within the region.

[0075] To more clearly illustrate this embodiment, as Figure 1 shown, the implementation process of an intelligent traffic control method based on real-time traffic flow detection can be specifically described as follows:

[0076] As Figure 1 shown, in step S1, obtain vehicle transmission information data through the intelligent connected vehicle OBU and perform correction to obtain the corrected vehicle transmission information data.

[0077] In this embodiment, data is collected through the intelligent connected vehicle OBU and real-time road network monitoring (implemented by industrial cameras + visual videos), and the collected data is transmitted to the edge computing device.

[0078] S1-1. Obtain vehicle transmission information data through the intelligent connected vehicle OBU.

[0079] In this embodiment, the in-vehicle OBU obtains vehicle transmission information data, including: real-time vehicle position, speed, and vehicle density.

[0080] S1-2. Supplement and correct the vehicle transmission information data to obtain the corrected vehicle transmission information data.

[0081] Since the driving data uploaded by the OBU is the most accurate and reliable data, it is used as the direct data source in this embodiment. However, considering the insufficient number of connected vehicles supporting OBU data transmission on the market, which will cause a large error in data collection, therefore, two other data sources are designed for supplementation and correction.

[0082] Real-time collect the terminal connection data of roadside 5G base stations and obtain real-time road network monitoring data. Respectively, through 5GC positioning enhancement service technology and image recognition target detection technology, supplement and correct the vehicle transmission information data.

[0083] The two methods for supplementing and correcting the data sources specifically include:

[0084] (1) Supplement the vehicle transmission information data.

[0085] Obtain the number of mobile terminals, estimate the regional vehicle density based on the number of mobile phones connected to the 5G base station (assuming the average passenger capacity per vehicle), and supplement the vehicle transmission information data.

[0086] In this embodiment, the collection of the number of intelligent mobile terminals: Considering the problem of insufficient inventory of connected vehicles at the current stage, in practical applications of this embodiment, intelligent mobile terminals are used as supplementary data sources. By using the precise positioning technology in the 5GC-R17 standard, broadcast information to intelligent mobile terminals within the coverage of the 5G base station, and calculate the traffic flow in each direction of the current intersection by measuring the average number of passengers per vehicle and the location of the intelligent terminal during this period, so as to enhance the data accuracy during the period of insufficient connected vehicle numbers.

[0087] Real-time collect the terminal connection data of roadside 5G base stations, achieve high-precision positioning of mobile terminals through 5GC positioning enhancement service, calculate the number of 5G intelligent mobile terminals currently on the road surface, and then estimate the number of road vehicles at this time according to the average number of mobile terminals carried by each vehicle, supplement the vehicle number, and correct the vehicle density. Subsequently, it can be combined with the signal light dynamic timing scheme to provide more appropriate route planning for road vehicles.

[0088] (2) Visual image data correction.

[0089] Obtain the road network monitoring data and process the road network monitoring data. Specifically:

[0090] First, preprocess the collected video stream, perform edge detection using the Canny algorithm, and perform median filtering denoising processing on the collected images. The preliminarily processed images will be subjected to real-time target detection and recognition through the YOLOv5 target detection algorithm to achieve accurate counting and positioning of vehicles. Finally, use the non-maximum suppression (NMS) algorithm to remove overlapping detection frames and reduce the situation of false detection and missed detection.

[0091] Then, directly count the number of road vehicles using the camera and AI algorithm (such as YOLOv5 target detection) to correct the traffic flow.

[0092] In this embodiment, by accessing the existing traffic management system and using real-time road network monitoring images, through image recognition and target detection technology analysis, the risk of untrusted user-side data is avoided. In the simulation stage, industrial cameras can be used to simulate the acquisition of simulated vehicles on the road to correct the traffic flow.

[0093] Specifically, through spatio-temporal alignment, the OBU data, the data collected by intelligent terminals, and the road network monitoring data are mapped into a unified integrated time-series feature vector to make up for the coverage blind spots of a single data source.

[0094] After the above steps, the corrected vehicle transmission information data is finally obtained, making up for the coverage blind spots of a single data source and avoiding errors in data collection.

[0095] Step 2: Perform edge computing analysis on the corrected vehicle transmission information data to obtain the analyzed vehicle information; and perform preprocessing to obtain the preprocessed vehicle information.

[0096] In this embodiment, edge computing devices are deployed on roadside facilities (signal light host computers), which can process and summarize the information sent by vehicles nearby, so as to achieve fast response and real-time control, provide more appropriate route planning for road vehicles, and optimize the traffic flow.

[0097] Specifically, it includes the following steps:

[0098] Step 2-1: Perform edge computing analysis on the corrected vehicle transmission information data to obtain the analyzed vehicle information.

[0099] Receive the vehicle transmission information and perform analysis:

[0100] The edge computing device receives the information sent by the vehicle through the Uu interface and the PC5 interface in the C-V2X standard; then, the received information is analyzed and filtered to obtain the analyzed vehicle information data, and the analyzed data includes vehicle position, speed, and density.

[0101] To ensure the security and reliability of data transmission, in this embodiment, the RabbitMQ message queue is used to cache and load balance the received data packets to reduce packet loss caused by overloading of a single node.

[0102] After the above steps, the accuracy and reliability of the information are ensured.

[0103] Step 2-2: Perform preprocessing on the analyzed vehicle information to obtain the preprocessed vehicle information

[0104] Vehicle information processing: The edge computing device uses its own processing power to further process and calculate the analyzed vehicle information. It mainly includes the statistics and analysis of vehicle density and speed.

[0105] Step 2-3, Data Reporting:

[0106] To improve the service adaptability with edge computing facilities, this project can use the HTTP protocol and the WebSocket protocol to achieve real-time data transmission. At the same time, considering that there is a large amount of duplicate data and high data compressibility, this project performs compression operations on the data before transmission.

[0107] In this embodiment, the cloud computing as a whole uses SpringBoot as the server framework, mainly accepting long connection data sent by the WebSocket protocol and short-term data sent by HTTP.

[0108] Upload the preprocessed vehicle speed, density, and vehicle position.

[0109] Step 3, Based on the preprocessed vehicle information, construct a prediction model through machine learning algorithms for traffic flow prediction.

[0110] In this embodiment, specifically, the LSTM model is used for prediction. The specific prediction process is as follows:

[0111] (1) Use the memory unit to save the traffic data in historical periods and perform model fitting as training data.

[0112] In the traffic flow prediction scenario, the fitting model in this embodiment uses a long short-term memory network (LSTM) model. The historical traffic data is the fusion of in-vehicle OBU data, the number of mobile terminal accesses, and road network monitoring data to construct a dynamic prediction framework for time dependence. The specific process is as follows:

[0113] 1) Construct a prediction model.

[0114] LSTM (long short-term memory network) is a special recurrent neural network (RNN), which solves the problem of insufficient modeling ability of traditional RNN for long-term dependence relationships through internal "memory units" and "gating mechanisms".

[0115] 2) Input the traffic data in historical periods into the LSTM model for model fitting training to obtain a trained LSTM model.

[0116] Input the preprocessed vehicle transmission information into the trained LSTM model to predict the traffic flow at the next moment.

[0117] Among them, through the memory unit, historical key information (such as the periodic traffic flow during morning and evening rush hours) is stored. Through the forget gate / input gate, dynamically determine which information to retain or discard (for example, sudden traffic accidents will weaken the attention to regular traffic flow).

[0118] Through the output gate, the prediction result at the current moment is generated. Thus, the trained prediction model (LSTM model) can be obtained.

[0119] (2) Dynamic prediction and adaptive optimization.

[0120] To achieve real-time response and long-term adaptability, the system adopts a two-stage mechanism:

[0121] Edge computing inference: Deploy the trained LSTM model to the roadside edge server to directly process local data streams and complete the prediction within 5 minutes (reducing cloud transmission latency);

[0122] Online incremental learning: Regularly fine-tune the model parameters with newly added data to continuously improve the prediction ability as the traffic pattern evolves (such as the opening of new roads and changes in traffic flow during holidays).

[0123] (3) Input the preprocessed vehicle information into the trained prediction model for traffic flow prediction.

[0124] Based on the training with historical traffic data (OBU, 5G base stations, road network monitoring), the trained LSTM model is obtained. According to real-time inputs, key indicators such as traffic flow, queue length, and delay time at each intersection within the next 5 - 15 minutes are predicted. That is, the traffic flow at the intersection at the next moment is obtained, expressed as:

[0125] q 预测 = LSTM(vehicle position, vehicle password, vehicle speed).

[0126] Step 4: Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network.

[0127] In this embodiment, the concept of network science is introduced. Each intersection is regarded as a network node, and the connecting roads are regarded as edges to establish a graph theory model to describe intersection relationships. Multiple intersections are combined into a network for research to create a self-organizing traffic network, improving the adaptability to complex road conditions, the reliability of the system, the accuracy of traffic flow prediction, and the rationality of signal timing.

[0128] Vehicles and roadside facilities are used as network nodes, equipped with communication and computing devices, and dynamic connection and self-organizing network topology are realized using technologies such as C-V2X and Wi-Fi Direct. Information such as traffic conditions and accident alerts can be instantaneously exchanged between nodes.

[0129] By analyzing network characteristics, traffic management decisions are supported. The network science method is used to identify key intersections, optimize traffic planning, dynamically adjust signal timing, and reduce congestion. Compared with the traditional centralized network architecture, the self-organizing network can better adapt to complex road conditions and has higher reliability.

[0130] Step 5: Define three indicators, namely the average vehicle queue length, the average waiting time, and the regional entry-exit ratio, based on the predicted traffic flow.

[0131] Based on the self-organizing traffic network and the three indicators, construct a traffic signal timing calculation model.

[0132] Adopt the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, and conduct coordinated control of multiple intersections in the area.

[0133] (1) Define three indicators, namely the average vehicle queue length, the average waiting time, and the regional entry-exit ratio, based on the predicted traffic flow.

[0134] Establish an algorithm with the goal of minimizing the average vehicle queue length, minimizing the average vehicle delay time, and minimizing the regional vehicle entry-exit ratio; construct a timing model based on the chaotic particle swarm optimization algorithm, and achieve the optimal state by optimizing the above indicators.

[0135] In this embodiment, take the average vehicle queue length, the average vehicle delay time, and the regional vehicle entry-exit ratio of every five intersections (including this intersection and the four adjacent intersections in the east, west, south, and north) as input parameters, and establish a traffic signal timing calculation model. Finally, calculate and obtain the optimal timing strategy. The specific model is as follows:

[0136] Calculation of the average vehicle queue length:

[0137]

[0138] Among them, l q is the average vehicle queue length; n is the number of intersections; represents the remaining queue length at intersection i; represents the current queue length at intersection i; a and b are weight coefficients, representing the contribution ratios of the remaining queue length and the newly added queue length respectively. For example, a = 0.7 and b = 0.3. Through weight adjustment, the queue evaluation in different congestion scenarios can be flexibly optimized (such as paying more attention to the newly added queue during peak hours);

[0139]

[0140] l q is the average vehicle queue length; l s is the remaining vehicle queue length; l p is the queue vehicle length; n is the number of intersections; g i1 represents the effective green time of the upstream intersection of the i-th intersection; g i2 represents the effective green time of the downstream intersection of the i-th intersection; v represents the average vehicle driving speed; v t is the stop wave speed; T s= 1 indicates that the last vehicle arrives at this section within the first red - light signal; T s = 2 indicates that the last vehicle arrives at this section within the second red - light signal; v q represents the starting shock - wave speed; T t represents the stop - signal section where the first vehicle arrives at this section; l max represents the maximum vehicle stop - and - queue length.

[0141] The calculation formula for vehicle delay time is:

[0142]

[0143] Among them,

[0144]

[0145] T delay is the delay time of all intersections, n is one of all intersections; k represents an approach of n in the intersection; r n represents all entrances of intersection n; η k represents the saturation degree, s k is the lane saturation flow; q k represents the road traffic flow at the k - th entrance of the intersection; λ k represents the green - signal ratio at the k - th entrance of the intersection; d k (w, λ n , o) represents the delay function, which is used to describe the influence of the turning - proportion at the intersection on the delay (such as the proportion of left - turning vehicles), w represents the turning - proportion, λ n represents the green - signal ratio, the proportion of green - light time in the total signal - cycle time, o represents the relative importance of balancing the green - signal ratio (the proportion of green - light time in the total cycle) and the turning - proportion (w) when calculating the delay of a single approach; represents the signal cycle; q total represents the sum of the traffic flows of all approaches of intersection n, which is used to balance the delay calculation under different traffic - flow scales.

[0146] The calculation formula for the ratio of vehicle inflow to outflow in the area is:

[0147]

[0148] Among them, ρ is the ratio of inflow to outflow in the area, that is, the ratio of the maximum number of output vehicles to the total number of vehicles entering the area, which is set as the objective function, Q is the total number of vehicles entering the area; t = 1 represents the first signal cycle, t = 1,..., n, f = 1 represents the first phase, F is the maximum number of phases; l out is the number of approaches at the boundary area; is the number of vehicles driving out from approach k within t signal cycles.

[0149]

[0150] indicates whether approach k is green at phase, if yes, it is 1, otherwise it is 0; q k q(t) is the road traffic flow at intersection k in the t-th signal cycle. is the effective green time of the f-th phase at intersection n in the t-th signal cycle. The delayed vehicles at approach k in the f-th phase of the t-th signal cycle; s k is the lane saturation flow rate.

[0151] (2) Based on the self-organizing traffic network and three indicators, construct a traffic signal timing calculation model and optimize it.

[0152] In this embodiment, with the minimum average vehicle delay time, the shortest average queue length, and the minimum regional entry-exit ratio as the optimization objectives, a regional traffic signal optimization model is established. Let the number of intersections in the region be n, each intersection has f phases, and the optimization vector is:

[0153] X = [λ1,t1,λ2,t2,…,λ m ,t m (4)

[0154] where, λ m represents the green ratio of the m-th intersection, t m represents the effective green time of intersection m.

[0155] The final optimized model formula is:

[0156]

[0157] where, α, β, γ, and δ all represent weight coefficients, represents the average delay time of intersections; l q l(X) is the average vehicle queue length at intersections; n is the number of intersections; ρ(X) is the regional vehicle entry-exit ratio; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is the regularization term, λ n represents the green ratio, the proportion of the green time in the total signal cycle time; λ avg represents the average green ratio, Regularize to constrain the fluctuation range of the green signal ratio, prevent the green light time at a single intersection in the timing plan from being too long or too short, and improve coordination stability.

[0158] The constraint condition is

[0159] where T represents the traffic light cycle, T min represents the minimum traffic light cycle, that is, the shortest time to complete one traffic light cycle, T max represents the maximum traffic light cycle, that is, the longest time to complete one traffic light cycle, which limits the cycle range of the traffic light; g represents the time when the green light is on, g min represents the shortest time when the green light is on, g max represents the longest time when the green light is on.

[0160] (3) Adopt the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy.

[0161] Deeply integrate the prediction ability of the LSTM model with the dynamic decision-making ability of the traffic light timing model, and improve the real-time response and global optimization ability of the traffic system through multi-dimensional collaboration.

[0162] The LSTM model predicts key indicators such as traffic flow, queue length, and delay time at each intersection within the next 5 - 15 minutes based on historical traffic flow data (OBU, 5G base stations, road network monitoring) and real-time inputs. The traffic light timing model takes the prediction results of the LSTM as inputs, combines the current actual traffic state (such as vehicle queue length lq, delay time Tdelay), and solves the optimal timing plan through the chaotic particle swarm optimization algorithm.

[0163] The optimal timing strategy represents the length of the next cycle timing for multiple directions at the current intersection.

[0164] Traffic light timing is to convert traffic flow data into specific green light durations and phase switching strategies. Each indicator (such as delay time, queue length) in formulas (1), (2), and (3) is directly related to the actual control parameters of the signal lights, and the chaotic particle swarm optimization algorithm finds the best balance point through iterative optimization. Finally, the system realizes real-time and dynamic traffic control through edge computing and V2I communication, significantly improving the road network efficiency.

[0165] Based on the constraint conditions, with the goal of minimizing the average vehicle delay time, the shortest average queue length, and the smallest regional entry - exit ratio, solve the traffic light timing model formula (5) to obtain the optimal solution, that is, obtain the optimal timing strategy, and convert the optimal timing strategy into signal control instructions.

[0166] (4) After obtaining the optimal timing strategy, issue instructions to perform coordinated control on multiple intersections within the area.

[0167] Instruction issuance means that the data center issues the optimal timing strategy (i.e., the optimal timing for the next cycle) obtained through the traffic signal timing calculation model to the edge device, and the edge device issues specific timing adjustment instructions.

[0168] The instructions include operations such as cycle timing adjustment, forced yellow flashing, forced red light, and return to normal, which are used to control the operation of the entire system.

[0169] Cooperative control of multiple intersections in the area is carried out as follows:

[0170] ① Instruction reception: The source of the instruction can be an instruction sent through the communication network or an instruction from the local control center. The instruction issuance module needs to establish a stable communication connection and have an instruction reception and caching mechanism to ensure the reliable transmission of the instruction.

[0171] ② Instruction parsing: After receiving the instruction, the instruction issuance module needs to parse it to extract the operation requirements and parameters of the instruction. The parsing process should adopt standard parsing algorithms and rules to correctly parse the instruction and extract key information. The parsed instruction information will be used for subsequent instruction execution.

[0172] ③ Instruction execution: According to the parsed instruction information, the instruction issuance module needs to call the corresponding control interface to operate the traffic lights. The specific operation method can be to change the traffic signal timing plan, adjust the phase of the traffic lights, or trigger emergency signals, etc. When executing the instruction, it is necessary to ensure the correctness and safety of the operation to avoid having an adverse impact on traffic operation.

[0173] ④ Feedback of execution result: After executing the instruction, the instruction issuance module needs to feedback the execution result to the upper management system. The feedback content should include the execution status of the instruction, execution result, and relevant parameter information, etc. The upper management system can make corresponding decisions and adjustments based on the feedback result.

[0174] To verify the effectiveness of this embodiment, using the existing high-performance computing cluster in the laboratory to simulate the cloud computing center, the data collected by the edge computing server is processed and summarized, and software is used for simulation, such as Figure 3 and Figure 4 As shown, when the traffic flow is large, if the signal duration cannot be effectively adjusted, it will cause congestion on the entire road, resulting in traffic paralysis. Figure 3 As shown is the obvious congestion situation on the road under fixed timing, Figure 4 As shown, after introducing the control method in this embodiment, the vehicle congestion situation is significantly alleviated.

[0175] According to the signal timing algorithm designed in this embodiment, by adjusting the green phase duration in different directions, it is found that the above-mentioned large traffic flow can be reasonably regulated to optimize the road traffic congestion situation, better adapt to the changes in traffic flow, avoid congestion and delay, and improve the utilization rate of the road.

[0176] Embodiment 2

[0177] The purpose of this embodiment is to provide an intelligent transportation control system based on real-time traffic flow detection, including:

[0178] A data acquisition module, configured to: obtain vehicle transmission information data through an intelligent connected vehicle OBU and perform calibration;

[0179] An edge computing module, configured to: perform edge computing analysis on the calibrated vehicle transmission information data and perform preprocessing;

[0180] A core computing module, configured to: based on the preprocessed vehicle information, construct a prediction model through a machine learning algorithm to predict traffic flow;

[0181] Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network;

[0182] Based on the predicted traffic flow, define three indicators: average vehicle queue length, average waiting time, and regional in-out ratio;

[0183] Based on the self-organizing traffic network and the three indicators, construct a traffic signal timing calculation model;

[0184] Based on the traffic signal timing calculation model, use a chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, and perform coordinated control on multiple intersections in the area;

[0185] Among them, based on the self-organizing traffic network and the three indicators, construct a traffic signal timing calculation model, and the formula is;

[0186]

[0187] Among them, α, β, γ, and δ all represent weight coefficients, represents the average delay time at the intersection; l q (X) is the average vehicle queue length at the intersection; n is the number of intersections; ρ(X) is the regional vehicle in-out ratio; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is the regularization term, λ nIndicates the green-to-signal ratio, the ratio of the green light time to the total signal cycle time; λ avg Represents the average green-to-signal ratio.

[0188] The data acquisition module includes vehicle-mounted OBU, 5G smart mobile terminal and image data interface.

[0189] The edge computing module mainly includes roadside edge computing facilities (ECN).

[0190] The main functions of the cloud computing part include data receiving module, core computing module, data storage module, instruction issuing module, and cloud-edge collaborative unloading module.

[0191] The data receiving module has a universal and scalable data access layer, which is compatible with various communication protocols such as Restful and WebSokect, ensuring that all types of data can be seamlessly connected to the cloud. In order to cope with data peaks, Redis is used as a memory cache technology to store part of the received data, ensuring that data is not lost and reducing a lot of processing pressure on the backend. In addition, the message queue Kafka is deployed for asynchronous data processing, realizing data decoupling and peak-shaving.

[0192] In the core computing module, the chaotic particle swarm algorithm is used to define the average queue length, average waiting time, and regional in-and-out ratio of vehicles through the predicted traffic flow; the three indicators are used as input parameters, combined with the self-organizing traffic network, to establish a traffic light timing calculation model, calculate the optimal timing strategy, realize the coordinated control of multiple intersections in the region, and optimize the traffic flow of the entire road network through V2I communication. This ensures that the module can respond efficiently and accurately to complex road conditions, effectively improving the commuting efficiency of intersections.

[0193] In this embodiment, the data storage module is also included to store and manage various data generated during the operation of the system, including traffic flow data collected by the data collection part, the timing strategy of each cycle of the signal light, and the details of the cloud-edge collaborative task issuance. This module is designed to use the government cloud object storage and existing cloud services for specific execution. At the same time, the data security work is completed by the government cloud maintenance party.

[0194] Instruction issuing module: mainly issues instructions or cloud-edge collaborative computing instructions based on the calculation results of the core computing module.

[0195] Cloud-edge collaborative offloading module: The main function of this module is to detect the load details of each edge device, and according to the cloud load and the load of each edge node, send collaborative computing tasks to the edge node computing module on demand to enhance data response speed and processing efficiency.

[0196] In this embodiment, the overall system uses Spring Boot as the server framework, mainly accepting long-connection data sent via the WebSocket protocol and short-term data sent via HTTP.

[0197] Embodiment Three

[0198] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the above method are implemented.

[0199] Embodiment Four

[0200] The purpose of this embodiment is to provide a computer-readable storage medium.

[0201] A computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the steps of the above method are executed.

[0202] Embodiment Five

[0203] The purpose of this embodiment is to provide a computer program product containing instructions. When it runs on a computer, it causes the computer to execute the methods and functions involved in any one of the above embodiments.

[0204] The steps involved in the devices of the above embodiments correspond to those of Method Embodiment One. For specific implementation manners, reference may be made to the relevant description part of Embodiment One. The term "computer-readable storage medium" should be understood to include a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium that can store, encode, or carry an instruction set for execution by a processor and cause the processor to execute any method in the present invention.

[0205] Those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computer device. Optionally, they can be implemented by program codes executable by a computing device. Thus, they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0206] Although the specific implementation manners of the present invention have been described above in conjunction with the accompanying drawings, it is not a limitation to the protection scope of the present invention. Those skilled in the art should understand that based on the technical solutions of the present invention, various modifications or deformations that can be made without creative efforts by those skilled in the art are still within the protection scope of the present invention.

Claims

1. An intelligent traffic control method based on real-time traffic flow detection, characterized in that, Including: Obtain vehicle transmission information data through the intelligent connected vehicle OBU and perform calibration; Perform edge computing analysis on the calibrated vehicle transmission information data and perform preprocessing; Based on the preprocessed vehicle information, construct a prediction model through machine learning algorithms for traffic flow prediction; Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network; Define three indicators: average vehicle queue length, average waiting time, and regional entry-exit ratio through the predicted traffic flow; Based on the self-organizing traffic network and the three indicators, construct a traffic light timing calculation model; Based on the traffic light timing calculation model, use the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy for coordinated control of multiple intersections in the area; Among them, based on the self-organizing traffic network and the three indicators, construct a traffic light timing calculation model, and the formula is; Among them, α, β, γ, and δ all represent weight coefficients, represents the average delay time at intersections; l q (X) is the average queue length of vehicles at intersections; n is the number of intersections; ρ(X) is the ratio of vehicle inflow and outflow in the area; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is the regularization term, λ n represents the green signal ratio, which is the ratio of the green light time to the total signal cycle time; λ avg represents the average green signal ratio.

2. The intelligent traffic control method based on real-time traffic flow detection according to claim 1, wherein, Calibrate the vehicle transmission information data, and the specific process is: Real-time collect terminal connection data of roadside 5G base stations; Estimate the regional vehicle density based on the terminal connection data of roadside 5G base stations and supplement the vehicle transmission information data; Based on road network monitoring data, use AI algorithms to calculate the number of road vehicles; Based on the regional vehicle density and the number of road vehicles, use the spatio-temporal alignment method to calibrate the vehicle transmission information data to obtain the calibrated vehicle transmission information data.

3. An intelligent traffic control method based on real-time traffic flow detection according to claim 1, characterized in that, Based on the preprocessed vehicle transmission information, construct a prediction model through machine learning algorithms for traffic flow prediction. Among them, the prediction model is an LSTM model, specifically: Obtain historical period traffic data; Input the historical period traffic data into the LSTM model for model fitting training to obtain a trained LSTM model; Input the preprocessed vehicle transmission information into the trained LSTM model to predict the traffic flow at the next moment.

4. The intelligent traffic control method based on real-time traffic flow detection according to claim 1, wherein Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network. Specifically, based on the graph theory model to describe intersection relationships, regard vehicles and roadside facilities as network nodes, equip communication and computing devices, and use C-V2X and Wi-FiDirect technologies to dynamically connect and self-organize the network topology.

5. An intelligent traffic control method based on real-time traffic flow detection according to claim 1, characterized in that, The formula for the average vehicle queue length is: where l q is the average queue length of vehicles; n is the number of intersections; represents the remaining queue length at intersection i; represents the current queue length at intersection i; a and b are weight coefficients, representing the contribution ratios of the remaining queue length and the newly added queue length respectively.

6. The intelligent traffic control method based on real-time traffic flow detection according to claim 1, characterized in that Use the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, specifically: Based on the self-organizing traffic network, with the shortest average vehicle queue length, the least average waiting time, and the smallest regional entry-exit ratio as the optimization objectives, use the chaotic particle swarm optimization algorithm to obtain the optimal solution of the traffic light timing calculation model; Convert the optimal solution of the traffic light timing calculation model into a signal control instruction, that is, obtain the optimal timing strategy.

7. An intelligent transportation control system based on real-time traffic flow detection, characterized in that, Including: A data acquisition module configured to: obtain vehicle transmission information data through the intelligent connected vehicle OBU and perform calibration; An edge computing module configured to: perform edge computing analysis on the calibrated vehicle transmission information data and perform preprocessing; A core computing module configured to: based on the preprocessed vehicle information, construct a prediction model through machine learning algorithms for traffic flow prediction; Use a graph theory model to describe intersection relationships, analyze network characteristics, and create a self-organizing traffic network; Define three indicators, namely the average vehicle queue length, the average waiting time, and the ratio of regional entry to exit, based on the predicted traffic flow; Construct a traffic signal timing calculation model based on the self-organizing traffic network and the three indicators; Based on the traffic signal timing calculation model, use the chaotic particle swarm optimization algorithm to calculate and obtain the optimal timing strategy, and conduct coordinated control of multiple intersections in the area; Among them, a traffic signal timing calculation model is constructed based on the self-organizing traffic network and the three indicators, and the formula is; Among them, α, β, γ, and δ all represent weight coefficients, represents the average delay time at intersections; l q (X) is the average vehicle queue length at intersections; n is the number of intersections; ρ(X) is the ratio of vehicle inflow to outflow in the area; q 预测 represents the traffic flow predicted by the LSTM model; q 实际 represents the actual traffic flow; ∈·∑ n (λ n -λ avg ) 2 is the regularization term, λ n represents the green signal ratio, which is the ratio of the green light time to the total signal cycle time; λ avg represents the average green signal ratio.

8. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method described in any one of claims 1-6 above.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it executes the steps of the method described in any one of claims 1-6 above.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method described in any one of claims 1-6.

Images