Method, device and system for controlling highway traffic flow

Abstract

The invention discloses an expressway traffic flow control method, device and system, and the method comprises the steps: extracting a preset key feature from obtained historical road traffic data, and constructing a feature vector; sending the feature vector to a pre-trained traffic state prediction model, and predicting a current traffic state; the hyper-parameters of the traffic state prediction model are obtained through PSO algorithm optimization; sending the predicted current traffic state to a pre-trained control strategy model to obtain an optimal control strategy; the control strategy model is established and obtained by taking maximization of traffic flow efficiency, minimization of congestion degree and minimization of accident risk as targets. According to the method, the generalization ability and the prediction precision of the model can be improved, so that a solid technical support is provided for active control of the highway traffic flow.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent transportation systems, and in particular relates to a method, device and system for controlling highway traffic flow. Background Art

[0002] With the continuous acceleration of urbanization, the number of cars has increased dramatically, and the pressure on road traffic has increased day by day. Especially on important traffic arteries such as highways, traffic congestion occurs frequently, which not only seriously affects people's travel efficiency, but also increases the risk of traffic accidents.

[0003] 1. Limitations of traditional traffic control methods

[0004] Traditional traffic control methods mainly rely on manual experience and simple mathematical models to formulate control strategies. Although these methods can alleviate traffic congestion to a certain extent, they have obvious limitations in dealing with emergencies and complex traffic patterns. For example, signal light control strategies based on fixed rules are difficult to adapt to the dynamic changes of traffic flow, resulting in poor control effects. In addition, traditional traffic monitoring methods such as video monitoring and manual patrols cannot obtain real-time status information of large-scale traffic networks, making it difficult to achieve refined management.

[0005] 2. Data-driven intelligent transportation systems

[0006] In recent years, with the development of technologies such as the Internet of Things, big data, and cloud computing, the Intelligent Traffic System (ITS) has gradually become an effective means of solving traffic problems. The Intelligent Traffic System uses advanced information technology to achieve real-time collection, transmission, processing, and control of traffic information, which can significantly improve the level of intelligent traffic management. Specifically, by deploying various sensors (such as cameras, radars, GPS positioning devices, etc.), it is possible to monitor traffic flow, vehicle speed, road conditions, and other information in real time, and upload these data to the cloud for analysis and processing.

[0007] 3. Application of machine learning in traffic prediction

[0008] In intelligent transportation systems, machine learning technology is widely used in traffic prediction and control strategy formulation. Support vector regression (SVR) is an effective regression analysis method that can handle nonlinear problems well and has good generalization ability. By training the SVR model, the future traffic status can be accurately predicted, providing a basis for formulating reasonable control strategies. However, the effect of the SVR model depends largely on the selection of its hyperparameters. Traditional methods such as grid search are time-consuming and may not necessarily find the optimal solution.

[0009] 4. Advantages of Particle Swarm Optimization Algorithm

[0010] As a group intelligence optimization method, the particle swarm optimization (PSO) algorithm has been widely used in many fields due to its advantages such as ease of use and fast convergence speed. It is particularly good at optimizing complex functions and multi-objective problems. By optimizing the hyperparameters of the SVR model with the PSO algorithm, a set of ideal parameter combinations can be found in a relatively short time, thereby improving the prediction accuracy and control effect.

[0011] 5. The need for multi-objective optimization

[0012] In highway traffic flow control, it is often necessary to consider multiple objectives at the same time, such as maximizing traffic flow efficiency, minimizing congestion, reducing accident risks, etc. There may be conflicts between these objectives, so a multi-objective optimization method is needed to comprehensively consider the relationship between the objectives and find one or more compromise solutions. Support Vector Machine (SVM) is an effective classification tool that can be used to identify the optimal control strategy under different traffic conditions, thereby achieving dynamic control of traffic flow.

[0013] Therefore, how to effectively manage and optimize highway traffic flow has become one of the key issues that need to be urgently addressed in the current field of transportation engineering. Summary of the invention

[0014] In view of the above problems, the present invention proposes a method, device and system for controlling highway traffic flow, which realizes high-precision prediction of traffic status and can dynamically adjust the optimal control strategy according to real-time data. At the same time, through multi-objective optimization, it ensures the maximization of traffic flow efficiency, minimization of congestion and reduction of accident risks, thereby effectively improving the traffic capacity and driving safety of highways, and meeting the high requirements of modern intelligent transportation systems for real-time, robustness and adaptability.

[0015] In order to achieve the above technical objectives and the above technical effects, the present invention is implemented through the following technical solutions:

[0016] In a first aspect, the present invention provides a method for controlling traffic flow on a highway, comprising:

[0017] Extracting preset key features from the acquired historical road traffic data to construct a feature vector; sending the feature vector to a pre-trained traffic state prediction model to predict the current traffic state; and optimizing the hyperparameters of the traffic state prediction model through a PSO algorithm;

[0018] The predicted current traffic state is sent to a pre-trained control strategy model to obtain the optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion and minimizing accident risks.

[0019] In combination with the first aspect, optionally, the method for acquiring historical road traffic data includes:

[0020] Obtaining original road traffic data;

[0021] For the original road traffic data, a comprehensive Z-score method is used to identify and mark outliers, and finally the outliers are removed to obtain first road traffic data;

[0022] Performing data padding on the first road traffic data to obtain second road traffic data;

[0023] The second road traffic data is fused and corrected using a Kalman filter sensor to obtain historical road traffic data.

[0024] In combination with the first aspect, optionally, the calculation formula used by the comprehensive Z-score method is:

[0025] ;

[0026] in, Indicates Data points The Z-score value of ∣> preset value, then the data point is considered is an outlier, It is a sliding window The average of the inner data points; It is a sliding window The standard deviation of the internal data points; It is Number of data points The weight factor of is the smoothing factor, It is Data points The weight factor of It is Data points The time decay factor of is the parameter that controls the decay speed. is the timestamp of the data point.

[0027] In combination with the first aspect, optionally, when performing data filling, the filling formula used is:

[0028] ;

[0029] in, Indicates the data points that need to be filled The value of Represents data points The value of is a known value; Represents data points The value of is a known value; Represents data points timestamp, Represents data points timestamp, Represents data points The timestamp of the

[0030] In combination with the first aspect, optionally, when the second road traffic data is fused and corrected using a Kalman filter sensor, the formula used for fusion and correction is:

[0031] ;

[0032] in, It is at the moment The prior state estimate of It is at the moment The posterior state estimate of is the state transition matrix, is the control input model; is the control input, i.e., the second road traffic data; It is at the moment The prediction error covariance matrix is It is at the moment The posterior error covariance matrix of is the state transition matrix The transpose of is the process noise covariance matrix.

[0033] In combination with the first aspect, optionally, the preset key features include historical traffic flow, weather conditions, holiday information, and vehicle type distribution; and the traffic status prediction model is a support vector regression model with an RBF kernel function.

[0034] In combination with the first aspect, optionally, the hyperparameters of the support vector regression model are obtained by the following method:

[0035] Consider each hyperparameter as a particle in a particle swarm;

[0036] Initialize the size of the particle swarm, and initialize the position and velocity of each particle: the position of each particle can be regarded as a vector, which contains the hyperparameters of the support vector regression model, that is, the position of each particle =[ ],in , , is the initial value of the hyperparameter of the support vector regression model, the particle speed is initialized to a random number close to zero; initialize the global optimal position and the local optimal position ;

[0037] According to the position of the particle, the prediction error of the support vector regression model based on the RBF kernel function is calculated;

[0038] Update the global optimal position and the local optimal position ;

[0039] Update the particle's velocity and position:

[0040] ;

[0041] in, It is The particle in The speed in the dimension is The value at the iteration, It is The particle in The position in the dimension is The value at the iteration, is the inertia weight, and is the acceleration constant, and is a random number uniformly distributed in the interval [0,1]; The adaptive inertia weight is introduced to make it change with the number of iterations. Dynamic adjustment, where and are the maximum and minimum values ​​of the inertia weight; is the maximum number of iterations;

[0042] Until the preset loop conditions are met, the optimal set of particles is output as the optimal hyperparameters.

[0043] In combination with the first aspect, optionally, the control strategy model is a support vector machine model, and the objective function of the control strategy model is:

[0044] ;

[0045] in:

[0046] ;

[0047] ;

[0048] in, represents the objective function, , , are weighting coefficients, represents maximizing traffic flow efficiency, It is predicted by the traffic status prediction model Traffic flow at any time, It is predicted by the traffic status prediction model Average speed at the time; is the minimum effective speed; Represents the control strategy vector, including Control strategies , each control strategy Also includes Control variables ; is a set of time windows; It is an indicator function that returns 1 if the condition in the brackets is true, otherwise it returns 0; Represents minimizing congestion, is the congestion threshold speed, It is predicted by the traffic status prediction model The queue length at the time; Represents minimizing the risk of accidents, is the accident probability for a given speed and vehicle length, It is The length of the vehicle, is a collection of vehicles;

[0049] The constraints of the control strategy model include:

[0050] Speed ​​limit constraints: ;

[0051] Safety distance constraints: ;

[0052] in, Indicates the maximum permissible speed of the vehicle. Indicates the minimum safe distance; Indicates safe distance. is the average reaction time of the driver under normal circumstances; is a constant, which indicates the influence of vehicle length on safety distance.

[0053] In a second aspect, the present invention provides a control device for highway traffic flow, comprising:

[0054] The current traffic status prediction module is used to extract preset key features from the acquired historical road traffic data and construct a feature vector; send the feature vector to a pre-trained traffic status prediction model to predict the current traffic status; the hyperparameters of the traffic status prediction model are obtained by optimizing the PSO algorithm;

[0055] The control strategy acquisition module is used to send the predicted current traffic status to a pre-trained control strategy model to obtain the optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion and minimizing accident risks.

[0056] In a third aspect, the present invention provides a control system for highway traffic flow, including a storage medium and a processor;

[0057] The storage medium is used to store instructions;

[0058] The processor is configured to operate according to the instructions to execute the method according to any one of the first aspects.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] The present invention optimizes the hyperparameters of the traffic state prediction model through the PSO algorithm, and can more accurately predict future traffic conditions, such as vehicle flow, vehicle speed, queue length, etc. High-precision prediction helps to prepare countermeasures in advance and reduce improper traffic management caused by prediction errors. During peak hours in the morning and evening, this method can significantly improve the accuracy of predictions, accurately predict changes in vehicle flow, help traffic management departments deploy response strategies in advance, open or close certain lanes in a timely manner, and reasonably allocate emergency resources.

[0061] Based on the real-time estimated traffic status, the present invention uses the support vector machine (SVM) model to determine the optimal control strategy, which can quickly respond to changes in traffic conditions and achieve dynamic adjustments. In the face of emergencies (such as traffic accidents, bad weather, etc.), the method can quickly adjust the control strategy, adjust the timing of traffic lights, open emergency lanes, issue real-time navigation suggestions, etc., effectively alleviating traffic congestion and reducing traffic delays.

[0062] The present invention ensures that the control strategy achieves the optimal balance between multiple objectives by defining multiple objective functions (such as maximizing traffic flow efficiency, minimizing congestion, reducing accident risks, etc.) and constraints (such as speed limit, safety distance, etc.). In practical applications, this method can comprehensively consider various factors, not only improve the road capacity, but also reduce the probability of accidents, and achieve overall optimization of the traffic system. By coordinating the traffic flow of different sections, local overload can be avoided and the traffic efficiency of the overall road network can be improved.

[0063] The present invention enhances the robustness and adaptability of the system by introducing multi-source data fusion technology, so that it can still maintain good performance in complex and changeable traffic environments. In extreme weather or emergencies, this method can make full use of various data sources (such as real-time sensor data, historical data, weather forecasts, etc.) to ensure the effectiveness and stability of the control strategy. In foggy weather, the system can combine meteorological data and real-time traffic data to adjust speed limits and guidance measures in a timely manner to ensure driving safety.

[0064] When a traffic accident or other emergency occurs, the method of the present invention can respond quickly, guide vehicles to be reasonably diverted, and reduce the impact of the accident. In practical applications, once an accident occurs, the system can immediately activate the emergency plan, send real-time road condition information to the driver through an information release platform (such as a mobile phone APP, a car navigation system), guide him to avoid the accident section, and notify the rescue department to arrive at the scene quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative labor, among which:

[0066] Figure 1 A framework diagram of a method for controlling highway traffic flow according to an embodiment of the present invention;

[0067] Figure 2 The figure is a schematic diagram of a data collection process according to an embodiment of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0069] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0070] The method for controlling the highway traffic flow in the embodiment of the present invention is described in detail below in conjunction with a specific implementation.

[0071] Example 1

[0072] An embodiment of the present invention provides a method for controlling traffic flow on a highway, comprising the following steps:

[0073] (1) Extracting preset key features from the acquired historical road traffic data and constructing a feature vector; sending the feature vector to a pre-trained traffic state prediction model to predict the current traffic state; and optimizing the hyperparameters of the traffic state prediction model through the PSO algorithm;

[0074] (2) Sending the predicted current traffic state to a pre-trained control strategy model to obtain an optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion, and minimizing accident risks.

[0075] The present invention can improve the generalization ability and prediction accuracy of the model, thereby providing solid technical support for the active control of highway traffic flow.

[0076] In a specific implementation of the embodiment of the present invention, the method for acquiring historical road traffic data includes:

[0077] Obtaining original road traffic data; in the specific implementation process, the original road traffic data can be collected by: deploying a variety of advanced sensors, high-definition cameras and vehicle communication equipment (V2X technology) along the highway to collect road traffic data in real time, and the road traffic data covers vehicle volume, average vehicle speed, vehicle type distribution, etc.; meteorological stations installed in various regions can provide real-time weather data, including rainfall, wind speed, temperature, etc.; statutory holiday information can be directly obtained through public calendar services or API interfaces; different types of vehicles can be distinguished by image recognition technology in video surveillance systems to collect vehicle type distribution;

[0078] For the original road traffic data, a comprehensive Z-score method is used to identify and mark outliers, and finally the outliers are removed to obtain first road traffic data;

[0079] Performing data padding on the first road traffic data to obtain second road traffic data;

[0080] The second road traffic data is fused and corrected using a Kalman filter sensor to obtain historical road traffic data.

[0081] In a specific implementation of the embodiment of the present invention, considering the dynamic changes of data, the importance of data points, the time decay effect and the adaptive threshold and other factors, the outlier detection can use the sliding window technology to dynamically update the mean and standard deviation, and introduce the weight factor To emphasize the importance of recent data, a time decay factor is introduced To reflect the fact that the importance of data points decreases over time, the calculation formula used in the comprehensive Z-score method is:

[0082] ;

[0083] in, Indicates The Z-score value of a data point, if | ∣> preset value, then the data point is considered is an outlier, It is a sliding window The average value of the data points in the sliding window The value obtained by averaging all data points in the It is a sliding window The standard deviation of the data points in the window is calculated by comparing all the data points in the window with their mean value. The square root of the value obtained by taking the mean of the squared deviations; It is The weight factor for the number of data points, is the smoothing factor, It is The weight of each data point; It is The time decay factor for each data point; is the parameter that controls the decay speed. is the timestamp of the data point.

[0084] In a specific implementation of the embodiment of the present invention, in order to eliminate the dimensional influence between different features, the mean or median of the feature may be used to fill in the missing data in data formatting; when performing data filling, the filling formula used is:

[0085]

[0086] in, Indicates the data points that need to be filled The value of Represents data points The value of is a known value; Represents data points The value of is a known value; Represents data points timestamp, Represents data points timestamp, Represents data points The timestamp of the

[0087] In a specific implementation of the embodiment of the present invention, in order to realize the calibration of the sensor reading, it is necessary to introduce the measured value to correct the predicted value. For this reason, when the Kalman filter sensor is used to fuse and correct the second road traffic data in the embodiment of the present invention, the formula used for fusion and correction is:

[0088] ;

[0089] in, It is at the moment The prior state estimate of It is at the moment The posterior state estimate of is the state transition matrix, is the control input model; is the control input, i.e., the second road traffic data; It is at the moment The prediction error covariance matrix is It is at the moment The posterior error covariance matrix of is the state transition matrix The transpose of is the process noise covariance matrix.

[0090] In a specific implementation of the embodiment of the present invention, in the specific implementation process, the preset key features include historical traffic flow, weather conditions, holiday information, and vehicle type distribution; the traffic state prediction model is a support vector regression model with an RBF kernel function;

[0091] The feature vector is sent to a support vector regression model with an RBF kernel function. When the support vector regression model is trained, the objective function used is:

[0092]

[0093] The constraints adopted are:

[0094]

[0095] in, is the number of training samples; Indicates the training data set Samples, subscript Used to identify the data points and corresponding labels of each training sample, the value range is from 1 to ; is the weight vector, is the weight vector The transpose of is the bias term, is the maximum allowed prediction error; and It is The slack variables of samples are: To handle positive errors, Used to handle negative errors; is the regularization parameter, which is used to balance the importance of the regularization term and the loss term; It is the feature vector mapped to the high-dimensional space by the RBF kernel function. For the The feature vector of samples; It is The true label of each sample is the actual traffic status.

[0096] In the specific implementation process, the training process of the support vector regression model with RBF kernel function is:

[0097] The collected raw data will be preprocessed, and key features that are helpful for prediction will be extracted from the preprocessed data. The key features include: historical traffic flow, weather conditions, holiday information, and vehicle type distribution.

[0098] Construct a feature vector using the key features ,in, represents the historical traffic flow, Indicates weather conditions, Indicates holiday information. Represent the vehicle type distribution as training data to train the support vector regression (SVR) model with RBF kernel function;

[0099] Once the support vector regression (SVR) model is trained, the trained SVR model can be used to estimate traffic status.

[0100] In a specific implementation of the embodiment of the present invention, the hyperparameters of the support vector regression model are obtained by the following method:

[0101] Consider each hyperparameter as a particle in a particle swarm;

[0102] Initialize the size of the particle swarm, and initialize the position and velocity of each particle: the position of each particle can be regarded as a vector, which contains the hyperparameters of the support vector regression model, that is, the position of each particle =[ ],in , , is the initial value of the hyperparameter of the support vector regression model, the particle speed is initialized to a random number close to zero; initialize the global optimal position and the local optimal position ;

[0103] According to the position of the particle, the prediction error of the support vector regression model based on the RBF kernel function is calculated;

[0104] Update the global optimal position and the local optimal position ;

[0105] Update the particle's velocity and position:

[0106]

[0107] in, It is The particle in The speed in the dimension is The value at the iteration, It is The particle in The position in the dimension is The value at the iteration, is the inertia weight, and is the acceleration constant, and is a random number uniformly distributed in the interval [0,1]; The adaptive inertia weight is introduced to make it change with the number of iterations. Dynamic adjustment, where and are the maximum and minimum values ​​of the inertia weight; is the maximum number of iterations;

[0108] Until the preset loop conditions are met, the optimal set of particles is output as the optimal hyperparameters.

[0109] In a specific implementation of the embodiment of the present invention, the control strategy model is a support vector machine model, and the objective function of the control strategy model is:

[0110] ;

[0111] in:

[0112] ;

[0113] ;

[0114] in, represents the objective function, , , are weighting coefficients, represents maximizing traffic flow efficiency, It is predicted by the traffic status prediction model Traffic flow at any time, It is predicted by the traffic status prediction model Average speed at the time; is the minimum effective speed; Represents the control strategy vector, including Control strategies , each control strategy Also includes Control variables ; is a set of time windows; It is an indicator function that returns 1 if the condition in the brackets is true, otherwise it returns 0; Represents minimizing congestion, is the congestion threshold speed, It is predicted by the traffic status prediction model The queue length at the time; Represents minimizing the risk of accidents, is the accident probability for a given speed and vehicle length, It is The length of the vehicle, is a collection of vehicles;

[0115] The constraints of the control strategy model include:

[0116] Speed ​​limit constraints: ;

[0117] Safety distance constraints: ;

[0118] in, Indicates the maximum permissible speed of the vehicle. Indicates the minimum safe distance; It's a safe distance. is the average reaction time of the driver under normal circumstances, generally set at 1.5 seconds; is a constant of 0.75, which represents the effect of vehicle length on safety distance.

[0119] In the specific implementation process, for example, the control strategy vector can be defined including: Changing lane assignments: increasing or decreasing the number of lanes in a certain direction; Adjust the timing of traffic lights: optimize the time distribution of traffic lights; Issue dynamic traffic instructions: advise drivers to avoid congested roads; Initiate emergency measures: such as arranging emergency rescue vehicles on standby to respond to emergencies. Encode these strategies as integer labels: .

[0120] For the above four different control strategies, they can be expressed as a discrete set We encode these policies as integer labels, i.e. ,in The index of the policy.

[0121] When using the SVM model to predict the optimal control strategy, the process can be expressed as: , which means that we need to find a comprehensive objective function Maximum control strategy .

[0122] After obtaining the optimal control strategy vector, a series of control instructions for the current traffic conditions are generated according to the various control variables in the optimal control strategy vector. Multi-source real-time data can be integrated through the central control system. After strict instruction verification to ensure its safety and compliance, the automated execution mechanism is used to adjust lane allocation, control entrance ramp flow and other measures. The execution effect is monitored in real time through a closed-loop feedback system to ensure that the control instructions work as expected. At the same time, its effect is regularly evaluated and the strategy is dynamically adjusted according to the actual road conditions to deal with emergencies, ultimately achieving efficient and intelligent traffic flow management.

[0123] The control method in the embodiment of the present invention is described in detail below in conjunction with a specific implementation manner.

[0124] In this specific implementation, the data collection step collects road traffic data in real time by deploying a variety of advanced sensors, high-definition cameras, and vehicle communication equipment (V2X technology) along the highway, and ensures the accuracy and real-time nature of the data. Figure 2 As shown in the figure, the data collection process specifically includes:

[0125] a1. Deploy sensors: Vehicle detectors - installed on important sections of highways to detect the speed, type and other information of passing vehicles in real time; Weather stations - installed along highways to collect weather information such as temperature, humidity, wind speed, etc.; High-definition cameras - installed at important intersections and sections to monitor traffic flow, vehicle speed and other conditions in real time.

[0126] a2. Data Collection Process

[0127] a2.1. Real-time data collection. Sensors collect data in real time, including information such as vehicle speed, vehicle type, and vehicle passing time detected by vehicle detectors; high-definition cameras capture real-time video streams and extract information such as traffic flow and vehicle type distribution through image processing technology; V2X technology receives real-time information such as location, speed, direction, etc. reported by vehicles; meteorological stations collect weather data such as temperature, humidity, wind speed, etc. in real time.

[0128] a2.2. Data preprocessing: use the comprehensive Z-score method to perform preliminary cleaning of the collected data, remove outliers and invalid data, format the data, and perform data filling.

[0129] a2.3. Kalman filter fusion: Use Kalman filter to fuse and correct sensor data. The basic principle of Kalman filter is to estimate the state of the system and reduce the impact of measurement noise through two stages: prediction and update.

[0130] a3. Data storage and transmission: the fused data is stored in the database and transmitted to the central processing server via the network.

[0131] The data obtained on a highway section in a certain city according to the above steps are shown in Table 1.

[0132] Table 1

[0133] Timestamp Vehicle ID Vehicle speed (km / h) Model Location coordinates Road Section Traffic volume (vehicles / hour) Weather conditions Temperature (℃) Wind speed (m / s) 2023-09-20 10:00:00 001 80 Sedan (31.2304, 121.4737) L001 1200 clear 25 5 2023-09-20 10:01:00 002 75 truck (31.2310, 121.4740) L001 1250 clear 25 5 2023-09-20 10:02:00 003 85 Sedan (31.2315, 121.4743) L001 1300 clear 25 5 2023-09-20 10:03:00 004 70 truck (31.2320, 121.4746) L001 1350 clear 25 5

[0134] Traffic state estimation is performed using the RBF kernel function-based support vector regression (SVR) model, including:

[0135] a1. Data preparation. The collected data is shown in Table 1.

[0136] a2. Data preprocessing: converting data into a format suitable for model training.

[0137] a2.1. Feature selection: extract key features that are helpful for prediction from the preprocessed data, such as historical traffic flow, weather conditions, holiday information, etc. (If predicting traffic volume, select vehicle speed, vehicle type, weather conditions, etc. as features).

[0138] a2.2. Data segmentation: Divide the dataset into training dataset and test set.

[0139] a2.3.Data standardization: Standardize the features so that they are of the same magnitude.

[0140] a3. Use PSO to optimize the hyperparameters of the SVR model: The first step is to initialize the size of the particle swarm, the position and speed of each particle, and initialize the global optimal position and the local optimal position ; The second step is to calculate the prediction error based on the RBF kernel function SVR model according to the particle position; The third step is to update the local optimal position and the global optimal position ; Step 4: Update the velocity and position of the particle; Step 5: Repeat steps 2 to 4 until the maximum number of iterations is reached or the convergence condition is met.

[0141] a4. Use the optimized SVR model to estimate traffic status.

[0142] Table 2 shows some results of traffic state estimation using the control method proposed in the embodiment of the present invention on a highway section in a certain city.

[0143] Table 2

[0144] Timestamp Estimated traffic volume (vehicles / hour) 2023-09-20 10:00:00 1200 2023-09-20 10:01:00 1250 2023-09-20 10:02:00 1310 2023-09-20 10:03:00 1350

[0145] The current traffic status is estimated based on the optimized SVR model, and the optimal control strategy is determined using the support vector machine (SVM) model, including:

[0146] a1. Data preparation: The collected data are shown in Table 1.

[0147] a2. Feature selection and data preprocessing. The real-time traffic state estimation result (Table 2) output by the SVR model in step 4 above is used as one of the input features.

[0148] a3. Use the SVM model to predict the current optimal control strategy as shown in Table 3.

[0149] Table 3

[0150] Timestamp Vehicle ID Vehicle speed (km / h) Model Location coordinates Road Section Traffic volume (vehicles / hour) Weather conditions Temperature (℃) Wind speed (m / s) Control strategy 2023-09-2010:00:00 001 80 Sedan (31.2304,121.4737) L001 1200 clear 25 5 Maintaining the status quo 2023-09-2010:01:00 002 75 truck (31.2310,121.4740) L001 1250 clear 25 5 Open emergency lane 2023-09-2010:02:00 003 85 Sedan (31.2315,121.4743) L001 1300 clear 25 5 Reduce speed limits 2023-09-2010:03:00 004 70 truck (31.2320,121.4746) L001 1350 clear 25 5 Increase speed limit

[0151] Example 2

[0152] A control device for highway traffic flow, characterized by comprising:

[0153] The current traffic status prediction module is used to extract preset key features from the acquired historical road traffic data and construct a feature vector; send the feature vector to a pre-trained traffic status prediction model to predict the current traffic status; the hyperparameters of the traffic status prediction model are obtained by optimizing the PSO algorithm;

[0154] The control strategy acquisition module is used to send the predicted current traffic status to a pre-trained control strategy model to obtain the optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion and minimizing accident risks.

[0155] The rest are the same as in Example 1.

[0156] Example 3

[0157] An embodiment of the present invention provides a control system for highway traffic flow, including a storage medium and a processor;

[0158] The storage medium is used to store instructions;

[0159] The processor is configured to operate according to the instructions to execute the method according to any one of the embodiments 1.

[0160] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0161] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0162] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0164] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

[0165] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for controlling traffic flow on a highway, characterized in that: include: Extract preset key features from the acquired historical road traffic data and construct feature vectors; The feature vector is sent to a pre-trained traffic state prediction model to predict the current traffic state; the hyperparameters of the traffic state prediction model are obtained by optimizing the PSO algorithm; The predicted current traffic state is sent to a pre-trained control strategy model to obtain the optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion and minimizing accident risks.

2. A method for controlling highway traffic flow according to claim 1, characterized in that: The method for acquiring historical road traffic data comprises: Obtaining original road traffic data; For the original road traffic data, a comprehensive Z-score method is used to identify and mark outliers, and finally the outliers are removed to obtain first road traffic data; Performing data padding on the first road traffic data to obtain second road traffic data; The second road traffic data is fused and corrected using a Kalman filter sensor to obtain historical road traffic data.

3. A method for controlling highway traffic flow according to claim 2, characterized in that: The calculation formula used in the comprehensive Z-score method is: ； in, Indicates Data points The Z-score value of ∣> preset value, then the data point is considered is an outlier, It is a sliding window The average of the inner data points; It is a sliding window The standard deviation of the internal data points; It is Number of data points The weight factor of is the smoothing factor, It is Data points The weight factor of It is Data points The time decay factor of is the parameter that controls the decay speed. is the timestamp of the data point.

4. A method for controlling highway traffic flow according to claim 2, characterized in that ,When filling in data, the filling formula used is: ； in, Indicates the data points that need to be filled The value of Represents data points The value of is a known value; Represents data points The value of is a known value; Represents data points timestamp, Represents data points timestamp, Represents data points The timestamp of the 5. A method for controlling highway traffic flow according to claim 2, characterized in that: When the Kalman filter sensor is used to fuse and correct the second road traffic data, the formula used for fusion and correction is: ； in, It is at the moment The prior state estimate of It is at the moment The posterior state estimate of is the state transition matrix, is the control input model; is the control input, i.e., the second road traffic data; It is at the moment The prediction error covariance matrix is It is at the moment The posterior error covariance matrix of is the state transition matrix The transpose of is the process noise covariance matrix.

6. A method for controlling highway traffic flow according to claim 1, characterized in that: The preset key features include historical traffic flow, weather conditions, holiday information, and vehicle type distribution; the traffic state prediction model is a support vector regression model with an RBF kernel function.

7. A method for controlling highway traffic flow according to claim 6, characterized in that: The hyperparameters of the support vector regression model are obtained by the following method: Consider each hyperparameter as a particle in a particle swarm; Initialize the size of the particle swarm, and initialize the position and velocity of each particle: the position of each particle can be regarded as a vector, which contains the hyperparameters of the support vector regression model, that is, the position of each particle =[ ],in , , is the initial value of the hyperparameter of the support vector regression model, the particle speed is initialized to a random number close to zero; initialize the global optimal position and the local optimal position ; According to the position of the particle, the prediction error of the support vector regression model based on the RBF kernel function is calculated; Update the global optimal position and the local optimal position ; Update the particle's velocity and position: ； in, It is The particle in The speed in the dimension is The value at the iteration, It is The particle in The position in the dimension is The value at the iteration, is the inertia weight, and is the acceleration constant, and is a random number uniformly distributed in the interval [0,1]; The adaptive inertia weight is introduced to make it change with the number of iterations. Dynamic adjustment, where and are the maximum and minimum values ​​of the inertia weight; is the maximum number of iterations; Until the preset loop conditions are met, the optimal set of particles is output as the optimal hyperparameters.

8. A method for controlling highway traffic flow according to claim 1, characterized in that: The control strategy model is a support vector machine model, and the objective function of the control strategy model is: ； in: ； ； in, represents the objective function, , , are weighting coefficients, represents maximizing traffic flow efficiency, It is predicted by the traffic status prediction model Traffic flow at any time, It is predicted by the traffic status prediction model Average speed at the time; is the minimum effective speed; Represents the control strategy vector, including Control strategies , each control strategy Also includes Control variables ; is a set of time windows; It is an indicator function that returns 1 if the condition in the brackets is true, otherwise it returns 0; Represents minimizing congestion, is the congestion threshold speed, It is predicted by the traffic status prediction model The queue length at the time; Represents minimizing the risk of accidents, is the accident probability for a given speed and vehicle length, It is The length of the vehicle, is a collection of vehicles; The constraints of the control strategy model include: Speed ​​limit constraints: ; Safety distance constraints: ; in, Indicates the maximum permissible speed of the vehicle. Indicates the minimum safe distance; Indicates safe distance. is the average reaction time of the driver under normal circumstances; is a constant, which indicates the influence of vehicle length on safety distance.

9. A control device for highway traffic flow, characterized in that: include: The current traffic status prediction module is used to extract preset key features from the acquired historical road traffic data and construct a feature vector; The feature vector is sent to a pre-trained traffic state prediction model to predict the current traffic state; the hyperparameters of the traffic state prediction model are obtained by optimizing the PSO algorithm; The control strategy acquisition module is used to send the predicted current traffic status to a pre-trained control strategy model to obtain the optimal control strategy; the control strategy model is established with the goal of maximizing traffic flow efficiency, minimizing congestion and minimizing accident risks.

10. A control system for highway traffic flow, characterized in that: including storage media and processors; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the method according to any one of claims 1-8.

Images