Traffic flow simulation parameter calibration method and device, equipment and storage medium

Abstract

The embodiment of the invention provides a traffic flow simulation parameter calibration method and device, equipment and a storage medium, and relates to the technical field of data processing. According to the method, track data and current speed data of corresponding vehicles are obtained, lane data obtained by lane area division at an intersection are obtained, multiple groups of car-following vehicle pairs are selected from the vehicles according to passing timestamps of the vehicles passing through lane stop lines, and the car-following vehicle pairs are selected according to the passing timestamps of the vehicles passing through the lane stop lines. And inputting the current radar data into the car-following model for data processing to obtain predicted speed data of a rear car, performing error statistics according to the current speed data and the predicted speed data corresponding to the rear car by using an orthogonal analysis method to obtain an error result, and calibrating simulation parameters in the car-following model based on the error result to obtain target simulation parameters. A high-precision map does not need to be introduced, multiple groups of following vehicle pairs are selected by using timestamps, adverse effects of heterogeneous data on result accuracy are effectively avoided, and the accuracy of a calibration result is improved by using an orthogonal analysis method.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular to a traffic flow simulation parameter calibration method, device, equipment, and storage medium. Background Art

[0002] Traffic microscopic simulation reproduces the characteristics of various actual traffic conditions through a simulation system model, and plays an important role in the research and application of intelligent transportation systems. In traffic microscopic simulation, the credibility of the simulation results depends to a large extent on the matching degree between the model-related parameters and the actual microscopic traffic flow scenario. The intersection area in the traffic scenario is a key node of road traffic, and it is particularly important to conduct traffic microscopic simulation on it. During the simulation process, in order to restore the actual operation status of the intersection to the greatest extent, it is necessary to accurately calibrate the parameters of the model.

[0003] In related technologies, a large amount of visual acquisition information is usually combined with high-precision maps to perform vehicle data simulation, and optimization algorithms such as reinforcement learning are used to optimize and solve some parameters in the related models. However, this method has a high degree of dependence on high-precision maps, and the cost of map acquisition itself is high. In addition, due to the existence of a large amount of low-quality data in the collected data, even after simple denoising operations, the accuracy of the obtained simulation results is low, and the calibration accuracy of the simulation parameters is poor. Summary of the Invention

[0004] The main purpose of the embodiments of this application is to propose a traffic flow simulation parameter calibration method, device, equipment, and storage medium, which do not require a high-precision map during the simulation parameter calibration process, and at the same time improve the calibration accuracy of the simulation parameters.

[0005] To achieve the above object, the first aspect of the embodiments of this application proposes a traffic flow simulation parameter calibration method, including:

[0006] Obtain the current radar data of each vehicle within a preset time period in the intersection, and obtain the trajectory data and current speed data of the corresponding vehicle according to the current radar data; and obtain the lane data obtained by dividing the lane area of the intersection, the lane data is obtained according to the current radar data or historical radar data of the intersection, and the lane data includes at least lane stop lines;

[0007] Determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple groups of following vehicle pairs from the vehicles based on the passing timestamp, the following vehicle pairs include a leading vehicle and a following vehicle;

[0008] For the following vehicle pairs, input the current radar data into a following model for data processing to obtain the predicted speed data of the following vehicle, and the following model includes preset simulation parameters;

[0009] Using the method of orthogonal analysis, error statistics is performed based on the current speed data and the predicted speed data corresponding to the following vehicle to obtain an error result, and the simulation parameters in the car-following model are calibrated based on the error result to obtain the target simulation parameters corresponding to the simulation parameters.

[0010] In some embodiments, obtaining lane data by dividing the lane area of the intersection according to the current radar data or historical radar data includes:

[0011] Taking the current radar data or historical radar data as lane division data, and obtaining the reference vehicle coordinates and reference vehicle speed from the lane division data;

[0012] Performing clustering according to the reference vehicle coordinates to obtain the lane centerlines of at least one lane;

[0013] Obtaining the lane width, and expanding the lane centerlines according to the lane width to obtain the lane boundary lines corresponding to each lane;

[0014] Selecting the vehicles with the reference vehicle speed less than the first preset speed as low-speed vehicles, and clustering the reference vehicle coordinates of the low-speed vehicles to obtain the lane stop line.

[0015] In some embodiments, the lane data further includes lane boundary lines. Determining the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and selecting multiple groups of following vehicle pairs from the vehicles based on the passing timestamp includes:

[0016] Determining the vehicles located in the same lane as lane vehicles based on the lane boundary lines;

[0017] Obtaining the timestamp when each lane vehicle passes the lane stop line as the passing timestamp, and arranging the passing timestamps in ascending order;

[0018] According to two consecutive passing timestamps, obtaining the time difference corresponding to the passing timestamps. If the time difference is less than the preset time, determining the passing vehicle identifiers of the corresponding two lane vehicles as consecutive passing vehicle identifiers, and determining multiple groups of following vehicle pairs based on the consecutive passing vehicle identifiers.

[0019] In some embodiments, the current speed data includes vehicle speed. Determining multiple groups of following vehicle pairs based on the consecutive passing vehicle identifiers includes:

[0020] Selecting the two lane vehicles corresponding to the consecutive passing vehicle identifiers as the initial vehicle pair, and the initial vehicle pair includes an initial leading vehicle and an initial following vehicle;

[0021] Traverse all the initial vehicle pairs, obtain the recording times of the current radar data corresponding to the initial leading vehicle and the initial following vehicle respectively, and / or, at each moment, determine the speed change amounts of the initial leading vehicle and the initial following vehicle respectively according to the corresponding vehicle speeds, and / or, obtain the distance differences between the initial leading vehicle and the initial following vehicle corresponding to each moment according to the trajectory data;

[0022] If the recording times are all greater than or equal to a preset number of times, and / or, the speed change amounts are all greater than or equal to a preset speed change amount, and / or, the distance differences are all less than a preset distance difference, regard the initial vehicle pair as the following vehicle pair.

[0023] In some embodiments, the simulation parameters at least include: a safe following time headway and a stopping distance. The inputting the current radar data into the following model for data processing to obtain the predicted speed data of the following vehicle includes:

[0024] Obtain multiple parameter sets, and the first parameter values of the safe following time headway and the second parameter values of the stopping distance in different parameter sets are not exactly the same;

[0025] Substitute the parameter sets into the following model to obtain multiple corresponding following simulation models;

[0026] For each following simulation model, input the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle.

[0027] In some embodiments, the inputting the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle includes:

[0028] Obtain the current speed and the maximum acceleration according to the current radar data of the following vehicle;

[0029] Obtain the speed difference between the front and rear vehicles and the distance between the front and rear vehicles based on the current radar data corresponding to the leading vehicle and the following vehicle respectively;

[0030] Obtain the desired speed of the following vehicle, and calculate the acceleration data of the following vehicle at least according to the desired speed, the current speed, the maximum acceleration, the speed difference between the front and rear vehicles and the distance between the front and rear vehicles, and use the acceleration data as the predicted speed data.

[0031] In some embodiments, the current speed data includes the vehicle speed, and the current speed data further includes the vehicle acceleration. The obtaining the desired speed of the following vehicle includes:

[0032] Obtain the vehicle speeds of all the vehicles within the preset time period, and use the vehicles whose vehicle speeds are less than the second preset speed at least twice as start-stop vehicles;

[0033] For the start-stop vehicles, expand the two speed timestamps corresponding to the second preset speed in the vehicle speeds forward and backward to obtain a start-stop time period;

[0034] Select at least the maximum value of the vehicle acceleration in the current radar data corresponding to the start-stop time period, use the moment corresponding to the maximum value as the start moment, and obtain the start speed of the start-stop vehicle according to the start moment, where the maximum value is a positive value;

[0035] Obtain the expected speed according to the average value of the start speeds of all the start-stop vehicles within the preset time period.

[0036] In some embodiments, calibrating the simulation parameters in the car-following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters includes:

[0037] Select the parameter set with the smallest error result as the target parameter set;

[0038] Use the first parameter value in the target parameter set as the target safe following headway, use the second parameter value in the target parameter set as the target stopping distance, and obtain the target simulation parameters according to the target safe following headway and the target stopping distance.

[0039] In some embodiments, after determining that the passing vehicle identifiers of the two corresponding lanes are consecutive passing vehicle identifiers, the method further includes:

[0040] For each lane, traverse the passing vehicle identifiers of the lane vehicles. If the passing vehicle identifiers of a preset number of consecutive vehicles are all the consecutive passing vehicle identifiers, use the corresponding lane vehicles as a consecutive passing vehicle group;

[0041] Calculate the headway corresponding to the consecutive passing vehicle group, and calculate the average value of all the headways to obtain the saturation flow rate corresponding to the lane.

[0042] In some embodiments, after selecting multiple pairs of car-following vehicles from the vehicles based on the passing timestamp, the method further includes:

[0043] For the car-following vehicle pairs, obtain the speed difference sequence and the rear vehicle acceleration sequence corresponding to the leading vehicle and the rear vehicle within the preset time period;

[0044] Calculate the cross-correlation coefficient at each moment according to the speed difference sequence and the following vehicle acceleration sequence, and select the moment corresponding to the maximum value of the cross-correlation coefficient as the reaction time corresponding to the following vehicle;

[0045] Calculate the average value of the reaction times of all the following vehicles within the preset time period to obtain the average reaction time, and generate a plurality of the first parameter values according to the average reaction time.

[0046] To achieve the above object, a second aspect of the embodiments of the present application proposes a traffic flow simulation parameter calibration device, including:

[0047] Data acquisition module: configured to acquire the current radar data of each vehicle within a preset time period at an intersection, and obtain the trajectory data and the current speed data of the corresponding vehicle according to the current radar data; and acquire the lane data obtained by dividing the lane area of the intersection, where the lane data is obtained according to the current radar data or historical radar data of the intersection, and the lane data includes at least lane stop lines;

[0048] Following confirmation module: configured to determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple sets of following vehicle pairs from the vehicles based on the passing timestamp, where the following vehicle pairs include a leading vehicle and a following vehicle;

[0049] Model prediction module: configured to input the current radar data into a following model for data processing for the following vehicle pair to obtain the predicted speed data of the following vehicle, where the following model includes preset simulation parameters;

[0050] Parameter calibration module: configured to use the method of orthogonal analysis to perform error statistics according to the current speed data and the predicted speed data corresponding to the following vehicle to obtain an error result, and calibrate the simulation parameters in the following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters.

[0051] To achieve the above object, a third aspect of the embodiments of the present application proposes an electronic device, where the electronic device includes a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the method described in the first aspect above is implemented.

[0052] To achieve the above object, a fourth aspect of the embodiments of the present application proposes a storage medium, where the storage medium is a storage medium, the storage medium stores a computer program, and when the computer program is executed by a processor, the method described in the first aspect above is implemented.

[0053] The traffic flow simulation parameter calibration method, device, equipment, and storage medium proposed in the embodiments of this application obtain the current radar data of each vehicle within a preset time period at an intersection, obtain the trajectory data and current speed data of the corresponding vehicle based on the current radar data, and obtain the lane data obtained by dividing the lane area at the intersection, where the lane data is obtained based on the current radar data or historical radar data of the intersection, and the lane data includes at least lane stop lines. Then, determine the passing timestamp when the vehicle passes the lane stop line based on the trajectory data, and select multiple groups of following vehicle pairs from the vehicles based on the passing timestamp, where the following vehicle pair includes a leading vehicle and a following vehicle. Next, for the following vehicle pair, input the current radar data into the following model for data processing to obtain the predicted speed data of the following vehicle. The following model includes preset simulation parameters. Finally, use the method of orthogonal analysis to perform error statistics based on the current speed data and predicted speed data corresponding to the following vehicle to obtain an error result, and calibrate the simulation parameters in the following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters. In the embodiments of this application, based on the aggregation characteristics of radar trajectories at intersections, radar data is used to simulate the lanes at intersections. This simulation method can be directly applied to intersections in different scenarios without introducing high-precision maps, thereby reducing the simulation cost. Since at intersections, radar perception is easily interfered by the surrounding environment, resulting in a certain proportion of abnormal data in the collected data. Therefore, in this embodiment, timestamps are used to select multiple groups of following vehicle pairs, and the object of the simulation data is converted from the complete vehicle data set to the following vehicle pairs, so as to focus on the following behavior that has a greater impact on intersection traffic, effectively avoiding the adverse impact of low-quality heterogeneous data on the result accuracy, and reducing the requirement for data quality. Compared with the filtering method used in reinforcement learning in related technologies, the orthogonal analysis method used in this embodiment can effectively avoid the problem that the calibrated parameters deviate from the normal range due to poor data quality, and improve the accuracy of the calibration result. Description of the Drawings

[0054] Figure 1 is a flowchart of the traffic flow simulation parameter calibration method provided by the embodiments of this application.

[0055] Figure 2 is a flowchart of obtaining lane data obtained by dividing the lane area at the intersection based on the current radar data or historical radar data provided by the embodiments of this application.

[0056] Figure 3 is a schematic diagram of the generation of lane boundary lines provided by the embodiments of this application.

[0057] Figure 4 is a schematic diagram of the lane area division at the intersection provided by the embodiments of this application.

[0058] Figure 5It is a flowchart for determining the passing timestamp when a vehicle passes the lane stop line according to trajectory data and selecting multiple groups of following vehicle pairs from the vehicle based on the passing timestamp in the embodiment of the present application.

[0059] Figure 6 It is a flowchart for determining multiple groups of following vehicle pairs based on continuous vehicle passing identification in the embodiment of the present application.

[0060] Figure 7 It is a flowchart for calculating the saturation flow rate corresponding to each lane in the embodiment of the present application.

[0061] Figure 8 It is a schematic diagram for traversing continuous vehicle passing identification in the embodiment of the present application.

[0062] Figure 9 It is a flowchart for inputting current radar data into a following vehicle model for data processing to obtain the predicted speed data of the following vehicle in the embodiment of the present application.

[0063] Figure 10 It is a flowchart for determining the first parameter value in the embodiment of the present application.

[0064] Figure 11 It is a flowchart for inputting current radar data into a following vehicle simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle in the embodiment of the present application.

[0065] Figure 12 It is a flowchart for obtaining the desired speed of the following vehicle in the embodiment of the present application.

[0066] Figure 13 It is a schematic diagram for lane area division in the embodiment of the present application.

[0067] Figure 14 It is a schematic diagram for cross-correlation coefficient in the embodiment of the present application.

[0068] Figure 15 It is a structural block diagram of a traffic flow simulation parameter calibration device provided in another embodiment of the present application.

[0069] Figure 16 It is a schematic diagram of the hardware structure of an electronic device provided in the embodiment of the present application. Specific Embodiments

[0070] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0071] It should be noted that although the functional modules are divided in the device schematic diagram and the logical sequence is shown in the flowchart, in some cases, the steps shown or described can be executed in a different module division from that in the device or a different order from that in the flowchart.

[0072] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0073] Traffic micro-simulation reproduces the characteristics of various actual traffic conditions through a simulation system model and plays an important role in the research and application of intelligent transportation systems. In traffic micro-simulation, the credibility of the simulation results depends to a large extent on the matching degree between the model-related parameters and the actual micro-traffic flow scenario. The intersection area in the traffic scenario is a key node of road traffic, and it is particularly important to conduct traffic micro-simulation on it. During the simulation process, in order to restore the actual operating conditions of the intersection to the greatest extent, it is necessary to accurately calibrate the parameters of the model.

[0074] In the intersection scenario of related technologies, radar devices are the main data collection tools. Compared with other roadside sensing devices such as cameras, current radar data has many advantages. It has higher real-time performance, larger information capacity, and more general data formats. Therefore, current radar data can be used as a data source for extracting traffic flow parameters.

[0075] However, due to cost limitations, most of the radar sensing devices at intersections choose millimeter-wave radars. This type of device has the problem that the target is easily lost when identifying static or low-speed objects, which makes the quality of current radar data uneven and unable to accurately and effectively restore the state information of the intersection. Therefore, when conducting traffic micro-simulation, it is necessary to rely on manual input of high-precision map information for assistance. However, the acquisition cost of high-precision maps is relatively high, and the proportion of intersections with ready-made intersection structure information in practice is not high.

[0076] When high-precision maps are available, a large amount of visual acquisition information is usually combined with high-precision maps for vehicle data simulation, and optimization algorithms such as reinforcement learning are used to optimize and solve some parameters in the relevant models. However, due to the existence of a large amount of low-quality data in the originally collected data, even after denoising operations, the accuracy of the simulation results obtained in this way is still relatively low, and the calibration accuracy of the simulation parameters is not high.

[0077] Based on this, the embodiments of the present application provide a traffic flow simulation parameter calibration method, device, equipment, and storage medium. Based on the aggregation characteristics of radar trajectories at intersections, radar data is used to simulate intersection lanes. This simulation method can be directly applied to intersections in different scenarios without introducing high-precision maps, thereby reducing the simulation cost. Since at intersections, radar perception is easily interfered by the surrounding environment, a certain proportion of abnormal data exists in the collected data. Therefore, in this embodiment, timestamps are used to select multiple groups of following vehicle pairs, and the object of the simulation data is converted from the complete vehicle data set to the following vehicle pairs, so as to focus on the following behavior that has a greater impact on intersection traffic, effectively avoiding the adverse effects of low-quality heterogeneous data on the result accuracy, and reducing the requirement for data quality. Compared with the filtering method used in reinforcement learning in the related art, the orthogonal analysis method used in this embodiment can effectively avoid the problem that the calibration parameters deviate from the normal range due to poor data quality, and improve the accuracy of the calibration result.

[0078] The embodiments of the present application provide a traffic flow simulation parameter calibration method, device, equipment, and storage medium, which will be specifically described through the following embodiments. First, the traffic flow simulation parameter calibration method in the embodiments of the present application will be described.

[0079] The traffic flow simulation parameter calibration method provided by the embodiments of the present application relates to the technical field of data processing. The traffic flow simulation parameter calibration method provided by the embodiments of the present application can be applied to a terminal, or to a server, or can be a computer program running on a terminal or a server. For example, the computer program can be a native program or a software module in an operating system; it can be a local (Native) application (Application, APP), that is, a program that needs to be installed in the operating system to run, such as a client that supports traffic flow simulation parameter calibration, that is, a program that only needs to be downloaded to the browser environment to run; it can also be a small program that can be embedded in any APP. In short, the above computer program can be any form of application program, module, or plug-in. Among them, the terminal communicates with the server through a network. This traffic flow simulation parameter calibration method can be executed by the terminal or the server, or jointly executed by the terminal and the server.

[0080] In some embodiments, the terminal may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart watch, etc. In addition, the terminal may also be an intelligent vehicle-mounted device. The intelligent vehicle-mounted device applies the traffic flow simulation parameter calibration method of this embodiment to provide relevant services and improve the driving experience. The server may be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms; it may also be a service node in a blockchain system, and the service nodes in the blockchain system form a Peer To Peer (P2P) network, and the P2P protocol is an application layer protocol running on top of the Transmission Control Protocol (TCP) protocol. The terminal and the server can be connected through communication connection methods such as Bluetooth, Universal Serial Bus (USB), or network, and this embodiment does not limit this here.

[0081] This application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet-type devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This application can also be practiced in a distributed computing environment where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0082] It should be noted that in each specific embodiment of the present application, when it comes to performing relevant processing based on data related to the user's identity or characteristics, such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first. Moreover, the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiments of the present application need to obtain the user's sensitive personal information, the user's separate permission or separate consent will be obtained through methods such as pop-up windows or redirecting to a confirmation page. After clearly obtaining the user's separate permission or separate consent, the necessary user-related data for enabling the normal operation of the embodiments of the present application will be obtained.

[0083] The following describes the traffic flow simulation parameter calibration method in the embodiments of the present application.

[0084] Figure 1 It is an optional flowchart of the traffic flow simulation parameter calibration method provided by the embodiments of the present application. Figure 1 The method in may include but is not limited to steps 110 to 140. At the same time, it can be understood that the present embodiment does not specifically limit the order of steps 110 to 140 in, and the order of steps can be adjusted according to actual needs, or some steps can be reduced or added. Figure 1 In, the order of steps 110 to 140 is not specifically limited, and the order of steps can be adjusted according to actual needs, or some steps can be reduced or added.

[0085] Step 110: Obtain the current radar data of each vehicle within a preset time period at the intersection, obtain the trajectory data and current speed data of the corresponding vehicle according to the current radar data, and obtain the lane data obtained by dividing the lane area at the intersection.

[0086] In one embodiment, first, select an intersection from the traffic roads as the object to be analyzed. Here, the intersection is the junction of roads in different directions. Then, select a time period for analysis, determine it as the preset time period, and obtain the current radar data corresponding to the vehicles that appear at this intersection within the preset time period.

[0087] It can be understood that the length of the preset time period will affect the accuracy of subsequent data analysis, so it should be set according to actual needs. The preset time period can be dozens of minutes, several hours, or even several days, etc. Set multiple moments within the preset time period. For each vehicle, obtain the current radar data corresponding to each moment to form the continuous trajectory data of the vehicle during this period.

[0088] In one embodiment, there are at least two ways to obtain the current radar data. It can be the radar data reported by the vehicle itself, or it can also be the radar data collected in real time by the radar devices deployed at intersections. By deploying the radar devices at various positions such as the entrance lanes, exit lanes, and crosswalks, information such as the position, speed, and movement direction of the vehicle can be collected in real time. In this embodiment, there is no limitation on the source of the current radar data.

[0089] In one embodiment, after obtaining the current radar data, the trajectory data and the current speed data of the corresponding vehicle can also be obtained from the current radar data. Among them, the trajectory data refers to the specific position information of the corresponding vehicle at each moment within a preset time period, and the movement path of the vehicle is represented by a series of consecutive vehicle coordinates, where the longitude and latitude are used to represent the vehicle coordinates. And the current speed data is the vehicle speed of the corresponding vehicle at each moment.

[0090] In one embodiment, the current radar data includes information such as vehicle ID, longitude, latitude, vehicle type, and timestamp information. Among them, the longitude and latitude can be the position coordinates of the center point of the vehicle head or the position coordinates of the vehicle center. This embodiment does not limit this. At the same time, based on the longitude, latitude, and timestamp information in the current radar data, the vehicle speed and vehicle acceleration at each moment of the vehicle can be further calculated.

[0091] For example, for a certain vehicle i, the longitudes and latitudes corresponding to the vehicle at two consecutive frames, time T0 and time T1, are (X1, Y1) and (X2, Y2) respectively. Assuming that R is the radius of the earth = 6371 km, the displacement distance dis of the vehicle within the duration of T1 - T0 can be calculated as follows:

[0092]

[0093] Next, based on the displacement distance dis, the vehicle speed of vehicle i corresponding to time T0 can be calculated Expressed as:

[0094]

[0095] Furthermore, the vehicle acceleration of vehicle i at time T0 can be obtained Expressed as:

[0096]

[0097] According to the above calculation process, the vehicle speed and vehicle acceleration corresponding to each vehicle at different times are also written into the corresponding current radar data.

[0098] In one embodiment, it is next necessary to obtain lane data obtained by performing lane area division on an intersection. The lane data is information related to the lanes in the corresponding area of the intersection obtained after performing lane area division on the intersection. The lane data includes the number of lanes, the left and right lane boundary lines of each lane, the lane width, the lane stop line, and the driving direction of the lane, etc. Here, the lane data can be obtained by pre-collecting historical radar data and performing advance analysis, or lane area division can be performed in real time and analyzed using current radar data.

[0099] In one embodiment, with reference to Figure 2 , Figure 2 is a flowchart for obtaining lane data obtained by performing lane area division on an intersection based on current radar data or historical radar data provided by an embodiment of the present application, specifically including the following steps:

[0100] Step 210: Use the current radar data or historical radar data as lane division data, and obtain the reference vehicle coordinates and reference vehicle speed from the lane division data.

[0101] In one embodiment, due to different actual requirements, the data used for lane area division is different. It can be current radar data or historical radar data. Therefore, whether it is current radar data or historical radar data, the selected data will be used as lane division data. That is to say, if it is necessary to perform lane area division on the intersection in advance, the historical radar data will be used as lane division data. If it is necessary to perform lane area division on the intersection in real time, the current radar data will be used as lane division data. It can be understood that the lane division data also includes the radar data corresponding to multiple vehicles respectively within a certain period of time.

[0102] Next, according to the above calculation method of vehicle speed, obtain the reference vehicle speed corresponding to each vehicle at different times from the lane division data, and at the same time obtain the vehicle coordinates at each moment as the reference vehicle coordinates.

[0103] Step 220: Cluster according to the reference vehicle coordinates to obtain the lane centerlines of at least one lane.

[0104] In one embodiment, the lane centerline is a virtual line representing the middle position of the lane. Since vehicles are driving in the lane, usually the reference vehicle coordinates are near the lane centerline. Therefore, cluster the reference vehicle coordinates to obtain the lane centerline of the corresponding vehicle in a certain lane. It can be understood that when clustering, for the same lane, the lane centerline of the lane can be obtained by clustering the reference vehicle coordinates of all vehicles.

[0105] Taking the K-Means algorithm as an example, the specific clustering process can be described as follows: First, the number of lanes can be determined according to prior data, empirical data, or a visual analysis model, and this number is the same as the number of clusters. Then, randomly select initial cluster centers equal to the number of lanes, and assign each reference vehicle coordinate to the category where the nearest cluster center is located. Next, recalculate the center position of each cluster, and continuously repeat this process of assignment and center update until the cluster centers no longer change significantly or reach the preset number of iterations. In this way, the reference vehicle coordinates will be divided into different categories, and each category corresponds to a lane.

[0106] Next, for each lane category obtained by clustering, calculate the statistical features of all reference vehicle coordinates in this category, such as the mean, median, etc., and connect these feature points in a certain order to obtain the lane centerline. It can be understood that during the generation process, it is also necessary to denoise and smooth the lane centerline obtained by clustering to eliminate discontinuities or non-smoothness caused by data fluctuations, etc., so that it better conforms to the shape of the actual lane.

[0107] Step 230: Obtain the lane width, and expand the lane centerline according to the lane width to obtain the lane boundary lines corresponding to each lane.

[0108] In one embodiment, referring to Figure 3 , Figure 3 is a schematic diagram of the generation of the lane boundary lines provided by the embodiment of the present application. The preset lane centerline is located at the center position of the corresponding lane. Therefore, after having the lane width, expand left and right along the lane centerline by half of the lane width respectively to obtain the lane boundary lines on the left and right sides corresponding to each lane. Among them, the lane width can be obtained from prior knowledge.

[0109] According to the above process, two lane boundary lines corresponding to each lane are obtained. In addition, there may be a situation where at least one lane boundary line coincides between two adjacent lanes in the same direction.

[0110] Step 240: Select the vehicles with the reference vehicle speed less than the first preset speed as low-speed vehicles, and cluster the reference vehicle coordinates of the low-speed vehicles to obtain the lane stop line.

[0111] In one embodiment, after the lane boundary lines are obtained, for the intersection area, there is usually also a lane stop line. The lane stop line is used to indicate the red-light restricted area of vehicles in the corresponding lane. That is to say, when vehicles in different lanes pass through the intersection, they probably need to wait at the position of the lane stop line at the same time. Therefore, deceleration occurs before the lane stop line. At this time, vehicles with a reference vehicle speed less than the first preset speed are selected as low-speed vehicles. Here, the first preset speed can be set according to actual needs. For example, it can be set to 2 km / h. After obtaining the low-speed vehicles, along the direction perpendicular to the lane boundary lines, all low-speed vehicles at the same moment are clustered, and a line segment perpendicular to the lane boundary lines can be obtained. Extend it along all lane boundary lines to obtain the corresponding lane stop line.

[0112] In one embodiment, after obtaining the lane boundary lines and the lane stop line, they need to be expressed in the form of mathematical formulas for subsequent calculation and analysis. The mathematical formulas here mainly use the method of determining a straight line by two points. Refer to Figure 4 , Figure 4 which is a schematic diagram of the lane area division of the intersection provided by the embodiment of the present application.

[0113] First, perform trajectory clustering on the reference vehicle coordinates of the vehicles to obtain the lane center line of each lane. Then, according to the lane width, move to the left and right sides along the lane center line respectively to obtain the lane boundary lines on the left and right sides. For example, Figure 4 shows the lane boundary line L1 and the lane boundary line L2 corresponding to a certain lane. Next, perform reference vehicle coordinate clustering according to the low-speed vehicles to obtain the lane stop line, and the intersection points of the lane stop line and the lane boundary line can be obtained. For example, Figure 4 shows the intersection points p1 and p2. Then, take points at a certain distance in the opposite direction of the lane boundary line to obtain the intersection points p3 and p4. Then, align the road forward direction with the north direction of geography, and arrange the four intersection points in order according to the naming method of upper left, upper right, lower right, and lower left, which are the intersection points p1, p2, p3, and p4 respectively. Next, perform line segment fitting on the two groups of points of the intersection points p1 and p4, and the intersection points p2 and p3, and the formulas of the lane boundary line L1 and the lane boundary line L2 corresponding to the lane can be obtained respectively, and the formula L3 of the lane stop line can be obtained from the intersection points p1 and p2.

[0114] It can be understood that according to the above process, by performing lane area mapping on the intersection, each lane in this area can be expressed in a mathematical form.

[0115] Compared with the method of carrying out data analysis by introducing high-precision maps in related technologies, due to the high cost of making and maintaining high-precision maps and the need for continuous updating to fit the changes in roads, the simulation cost remains high. However, in the embodiments of the present application, there is no need to introduce high-precision maps. Based on the clustering characteristics presented by radar trajectories at intersections, a method of simulating intersection lanes using radar data is adopted. This method directly uses the real-time obtained radar data for simulation operations, and can quickly and accurately complete the simulation work for intersections in different scenarios, not only effectively reducing the simulation cost, but also significantly improving the flexibility and real-time performance of the simulation.

[0116] Step 120: Determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple groups of following vehicle pairs from the vehicles based on the passing timestamp.

[0117] In one embodiment, in the intersection environment, radar perception is extremely vulnerable to interference from surrounding environmental factors. This results in a certain proportion of abnormal data in the collected data, which are specifically manifested as errors in vehicle position information, abnormal fluctuations in speed, unreasonable driving trajectories, etc. Therefore, in this embodiment, multiple groups of following vehicle pairs are selected by using timestamps, so that the object of the simulation data is changed from the original complete vehicle data set to following vehicle pairs. The following vehicle pair here refers to a pair of vehicle combinations composed of two vehicles that closely follow each other in the traffic flow. The following vehicle pair includes the leading vehicle and the following vehicle. In this combination, the driving behaviors of the two vehicles are highly correlated, and the driving decision of the following vehicle is often affected by the leading vehicle. For example, the following vehicle will adjust its speed and control the following distance according to the situation of the leading vehicle. By changing the object of the simulation data from the complete vehicle data set to following vehicle pairs, this embodiment can effectively focus on the following behaviors that have a greater impact on intersection traffic. In this way, the adverse effects caused by low-quality heterogeneous data on the result accuracy can be avoided, the overall requirement for data quality is reduced, and the accuracy of subsequent calculations is improved.

[0118] In one embodiment, refer to Figure 5 , Figure 5 is the flowchart of determining the passing timestamp when the vehicle passes the lane stop line according to the trajectory data and selecting multiple groups of following vehicle pairs from the vehicles provided by the embodiments of the present application, which specifically includes the following steps:

[0119] Step 510: Determine the vehicles located in the same lane as lane vehicles based on the lane boundary lines.

[0120] In one embodiment, lane attributes are added to all vehicles within a preset time period according to lane data. The vehicle IDs corresponding to the longitude and latitude coordinates of the vehicle coordinates in the current radar data that are within the same set of lane boundary lines are added with the corresponding lane numbers as lane attributes. Combining Figure 4 , taking lane boundary line L1 and lane boundary line L2 as examples, the vehicle IDs corresponding to the vehicles whose vehicle coordinates are between the two are added with the corresponding lane attributes. For example, if this lane is the first lane at the south entrance and the corresponding lane number is S1, the lane attribute corresponding to the vehicles on this lane can be: vehicle ID (S1). If the second lane at the south entrance corresponds to the lane number S2, the lane attribute corresponding to the vehicles on this lane can be: vehicle ID (S2), and so on. Lane attributes are added to each vehicle.

[0121] Next, since the following vehicle pairs are selected from two consecutive vehicles in the same lane, vehicles in the same lane are selected as lane vehicles based on the left and right lane boundary lines, that is, vehicles with the same lane number in the lane attribute are taken as a group of lane vehicles.

[0122] Step 520: Obtain the timestamp when each lane vehicle passes the lane stop line as the passing timestamp, and sort the passing timestamps in ascending order.

[0123] In one embodiment, according to the current radar data corresponding to the lane vehicles, the running trajectories within a preset time period can be obtained. At this time, the running trajectories usually pass through this intersection, that is, they will cross the lane stop line. At this time, the timestamp when each lane vehicle passes the lane stop line is obtained as the passing timestamp, and this passing timestamp is used to record the specific moment when the corresponding lane vehicle passes the lane stop line. Since vehicles in the same lane cannot pass the lane stop line at the same time, the passing timestamps corresponding to each lane vehicle must have a sequence. Sorting these passing timestamps in ascending order can clearly determine the sequence of vehicles passing the lane stop line on each lane. At the same time, the use of lane IDs can clearly distinguish vehicles in different lanes, making the sorting result more accurate and organized.

[0124] For example, lane vehicle A passes the lane stop line at 7:15:20, and its passing timestamp is recorded as 7:15:20; lane vehicle B passes the lane stop line at 7:15:22, and the passing timestamp is 7:15:35; lane vehicle C passes the lane stop line at 7:15:19, and the passing timestamp is 7:15:19. Sorting the passing timestamps of these lane vehicles in ascending order, the obtained order is lane vehicle C (7:15:19), lane vehicle A (7:15:20), lane vehicle B (7:15:22), indicating that on this lane, lane vehicle C passes the lane stop line first, followed by lane vehicle A, and finally lane vehicle B. In this way, combined with the lane ID, the passing order and time interval of the lane vehicles can be accurately understood.

[0125] Step 530: According to two consecutive passing timestamps, obtain the time difference corresponding to the passing timestamps. If the time difference is less than the preset time, determine that the passing identifiers of the corresponding two lane vehicles are consecutive passing identifiers, and determine multiple groups of following vehicle pairs based on the consecutive passing identifiers.

[0126] In one embodiment, in the order from front to back, select two passing timestamps in sequence. For each group of selected timestamps, calculate the time difference between them. For example, the passing timestamp of the first vehicle is 7:15:20, and the passing timestamp of the second vehicle is 7:15:22, then the time difference between them is 2 seconds. In this way, calculate the time difference for all adjacent vehicles in the entire sorting result.

[0127] Next, compare the calculated time difference with the preset time. If the time difference is less than the preset time (such as 3 seconds), then determine that the passing identifiers of the corresponding two lane vehicles are consecutive passing identifiers. This indicates that these two vehicles pass the lane stop line successively within a short time, and there may be a certain correlation in their driving behaviors, and the following vehicle may be driving closely behind the preceding vehicle. On the contrary, if the time difference is greater than or equal to the preset time, then the passing identifiers of these two lane vehicles are determined to be independent passing identifiers, indicating that their passing behaviors are relatively independent and the following vehicle does not closely follow the preceding vehicle.

[0128] Taking the sorting result: lane vehicle C (7:15:19), lane vehicle A (7:15:20), lane vehicle B (7:15:22) as an example. At this time, the time difference between lane vehicle C and lane vehicle A is 1s, which is less than the preset time. Therefore, the passing identifiers corresponding to lane vehicle C and lane vehicle A are consecutive passing identifiers. The time difference between lane vehicle A and lane vehicle B is 2s, which is less than the preset time. Therefore, the passing identifiers corresponding to lane vehicle A and lane vehicle B are consecutive passing identifiers.

[0129] In one embodiment, after obtaining the passing vehicle identifiers in pairs, multiple groups of following vehicle pairs are determined based on the consecutive passing vehicle identifiers. Refer to Figure 6 , Figure 6 FIG. Figure 6 is a flowchart for determining multiple groups of following vehicle pairs based on consecutive passing vehicle identifiers provided by an embodiment of the present application, which specifically includes the following steps:

[0130] Step 610: Select two-lane vehicles corresponding to the consecutive passing vehicle identifiers as the initial vehicle pairs.

[0131] In one embodiment, not all consecutive passing vehicles are following vehicles and further differentiation is required. At this time, two-lane vehicles corresponding to the consecutive passing vehicle identifiers are used as the initial vehicle pairs, the leading vehicle in the initial vehicle pair is used as the initial leading vehicle, and the trailing vehicle is used as the initial trailing vehicle.

[0132] Step 620: Traverse all the initial vehicle pairs to obtain the recorded times of the current radar data corresponding to the initial leading vehicle and the initial trailing vehicle respectively, and / or, at each moment, determine the speed change amounts of the initial leading vehicle and the initial trailing vehicle respectively according to the corresponding vehicle speeds, and / or, obtain the distance differences corresponding to the initial leading vehicle and the initial trailing vehicle at each moment according to the trajectory data.

[0133] In one embodiment, all the initial vehicle pairs are judged one by one, and one or more of the following three parameters can be used in this embodiment for judgment.

[0134] Among them, the recorded times of the current radar data corresponding to the initial leading vehicle and the initial trailing vehicle can be obtained. For the initial leading vehicle and the initial trailing vehicle, the current radar data records corresponding to the initial leading vehicle and the initial trailing vehicle are respectively queried to obtain the recorded times of the corresponding current radar data, that is, the cumulative times that the initial leading vehicle and the initial trailing vehicle are respectively collected by the radar within a preset time period. For example, within one hour of monitoring time, the current radar data of a certain initial leading vehicle is recorded 300 times, and the current radar data of the initial trailing vehicle is recorded 280 times.

[0135] In addition, the speed change amounts of the initial leading vehicle and the initial trailing vehicle can also be determined at each moment according to the corresponding vehicle speeds. Specifically: at each moment, the speed difference between the current moment and the previous moment of the initial leading vehicle is calculated according to the vehicle speed included in the current radar data, and this difference is the speed change amount. Similarly, the speed change amount of the initial trailing vehicle is obtained. For example, at a certain moment, the vehicle speed of the initial leading vehicle is 35 km / h, and the vehicle speed at the previous moment is 30 km / h, then its speed change amount is 5 km / h; at the same time, the vehicle speed of the initial trailing vehicle changes from 28 km / h to 32 km / h at the corresponding moment, and the speed change amount is 4 km / h.

[0136] In one embodiment, the distance difference corresponding to the initial leading vehicle and the initial following vehicle can also be obtained according to the trajectory data, where the distance difference refers to the distance between the initial leading vehicle and the initial following vehicle. For example, at a certain collection time, the vehicle coordinates of the initial leading vehicle are (100, 200), and the vehicle coordinates of the initial following vehicle are (95, 195). Therefore, the distance difference between the two vehicles is calculated to be approximately 7.07 meters through the distance formula.

[0137] Step 630: If the recording times are all greater than or equal to a preset number, and / or the speed change amounts are all greater than or equal to a preset speed change amount, and / or the distance differences are all less than a preset distance difference, the initial vehicle pair is used as the following vehicle pair.

[0138] In one embodiment, if the recording times of both are greater than or equal to a preset number, such as 40 times. The speed change amounts of both are greater than or equal to a preset speed change amount, such as 10 km / h. The distance difference at each moment is less than a preset distance difference, such as 50 m. If one or more of these three conditions are met, it can be determined that the initial vehicle pair is the following vehicle pair, and the initial leading vehicle and the initial following vehicle are the leading vehicle and the following vehicle in the following state.

[0139] According to the above method, multiple groups of following vehicle pairs are obtained as samples required for the following parameter calibration.

[0140] In one embodiment, the saturated flow rate of a single lane at the intersection can also be determined through continuous vehicle passing. The saturated flow rate of a single lane refers to the maximum number of vehicles that can pass through a single lane per unit time (usually in hours) when vehicles pass through the stop line of the lane in a closely continuous manner at a certain specific intersection. The unit is generally "vehicles per hour (veh / h)". For example, for a certain lane on an urban arterial road, under the condition of smooth traffic and reasonable signal timing, its saturated flow rate of a single lane may be 1800 veh / h, which means that theoretically, at most 1800 vehicles can pass through this lane per hour. The embodiments of the present application calculate the saturated flow rate corresponding to each lane at the intersection based on the continuous vehicle passing identifier. Refer to Figure 7 , Figure 7 is a flowchart for calculating the saturated flow rate corresponding to each lane in the embodiments of the present application, which specifically includes the following steps:

[0141] Step 710: For each lane, traverse the passing identifiers of the lane vehicles. If the passing identifiers of a continuous preset number are all continuous passing identifiers, the corresponding lane vehicles are used as a continuous passing group.

[0142] In one embodiment, the preset quantity can be set to 5 vehicles, which is used to determine whether there is a situation of continuous vehicle passing in the lane. If the number of continuous passing vehicle identifiers reaches or exceeds the preset quantity, it is considered that there is a phenomenon of continuous vehicle passing. For the lane vehicles in a certain lane, traverse the passing vehicle identifiers of the lane vehicles. If the passing vehicle identifiers of the continuous preset quantity are all continuous passing vehicle identifiers, the corresponding lane vehicles are used as a continuous passing vehicle group. For example, lane vehicle I n and lane vehicle I n+1 , if the passing vehicle identifiers of the two vehicles are continuous passing vehicle identifiers, then further traverse to the next lane vehicle I n+2 , and judge lane vehicle I n+1 and lane vehicle I n+2 are still continuous passing vehicle identifiers, then these 3 consecutive vehicles are continuous passing vehicles, and so on.

[0143] In one embodiment, referring to Figure 8 , Figure 8 is a schematic diagram for traversing continuous passing vehicle identifiers provided by an embodiment of the present application. In the figure, i is the number of records corresponding to the timestamps when each lane vehicle in the selected lane passes the lane stop line, d is an increment identifier, and D indicates the continuous passing vehicle quantity of the lane. At this time, starting from the passing vehicle identifier corresponding to the (i + d)-th record, if the time difference between two consecutive lane vehicles is less than 2 s, it is considered that the two are continuous passing vehicle identifiers, and at this time the continuous passing vehicle quantity D is incremented by one. If they are not continuous passing vehicle identifiers, select the next lane vehicle for judgment. If the continuous passing vehicle quantity D is greater than 5, obtain all the timestamps of continuous passing vehicles in this record.

[0144] Step 720: Calculate the headway corresponding to the continuous passing vehicle group, and calculate the average value of all headways to obtain the saturation flow rate corresponding to the lane.

[0145] In one embodiment, if it is continuous passing from lane vehicle I n to lane vehicle I n+i , after lane vehicle I n+i , the continuous passing stops, and when the number i of continuous passing vehicles exceeds 5, the continuous passing vehicle group can be recorded. Next, calculate the headway between every two vehicles in the continuous passing vehicle group, which is expressed as:

[0146]

[0147] Among them, represents the timestamp when lane vehicle I n+i passes the lane stop line, represents the timestamp when lane vehicle passes the lane stop line, and i - 3 represents lane vehicle I n+i and lane vehicle The number of vehicles between, TH ′ Indicates the lane vehicle I n+i And the lane vehicle The headway between.

[0148] It can be understood that the reason for selecting 3 vehicle intervals here is to consider that there may be a relatively large reaction time and starting process when the first vehicle starts. The second vehicle may also be affected by the start of the first vehicle, while the vehicles after the third vehicle are relatively less affected by the starting factors and can better reflect the stable traffic flow characteristics and the mutual relationship between vehicles.

[0149] Next, obtain the average value of the headway at this intersection, which is expressed as:

[0150]

[0151] Among them, TH represents the average value.

[0152] Therefore, according to the average value, obtain the single-lane saturation flow rate of this intersection, which is expressed as:

[0153] V = 3600 / TH (veh / h)

[0154] If the calculated average headway is 2 seconds, then the saturation flow rate, that is, at most 1800 vehicles can pass through per hour. According to the calculated single-lane saturation flow rate, it can be judged whether the traffic capacity of the current intersection meets the traffic demand. If the saturation flow rate is low, measures such as adding lanes and optimizing signal light settings need to be considered to improve the traffic efficiency.

[0155] Step 130: For the following vehicle pair, input the current radar data into the following model for data processing to obtain the predicted speed data of the following vehicle.

[0156] In one embodiment, after obtaining the following vehicle pair, the parameter calibration of the following model can be carried out. At this time, it is necessary to first obtain the predicted speed data of the following vehicle and use the predicted speed data to reflect the accuracy of the current parameters of the following model. In one embodiment, refer to Figure 9 , Figure 9 Is the flowchart of inputting the current radar data into the following model for data processing to obtain the predicted speed data of the following vehicle provided by the embodiment of the present application, which specifically includes the following steps;

[0157] Step 910: Obtain multiple sets of parameter sets.

[0158] In one embodiment, the following model includes preset simulation parameters, and the following model is expressed as:

[0159]

[0160] Among them, a represents the acceleration data of the following vehicle, a' represents the maximum acceleration of the following vehicle, v0 represents the desired speed of the following vehicle, v α represents the current speed of the following vehicle, Δv represents the speed difference between the front and rear vehicles in the car-following vehicle pair, S α represents the distance between the front and rear vehicles, T represents the safe following time distance, s0 represents the stopping distance, β represents the comfortable deceleration, which can be set according to the actual situation and can be 2, α represents the acceleration index, which is used to describe the acceleration tendency of the driver and can be set according to the actual situation and can be 4.

[0161] According to the above car-following model, the safe following time distance and the stopping distance are adjustable parameters. Therefore, in the embodiments of the present application, they are used as preset simulation parameters to generate multiple corresponding parameter sets, and the first parameter value of the safe following time distance and the second parameter value of the stopping distance in different parameter sets are not exactly the same. For example, the value ranges of the safe following time distance and the stopping distance are divided according to a certain step size, and then different parameter sets are combined and generated. Assume that the value range of the safe following time distance is [1 second, 5 seconds], and the value range of the stopping distance is [2 meters, 10 meters]. The value of the safe following time distance can be determined with a step size of 0.5 seconds, and the value of the stopping distance can be determined with a step size of 1 meter. In this way, multiple different parameter sets can be generated, and the first parameter value of the safe following time distance and the second parameter value of the stopping distance in each parameter set are not exactly the same.

[0162] In one embodiment, the value range of the first parameter value can be determined according to the actual parameters of the car-following vehicle pair. Refer to Figure 10 , Figure 10 which is the determination flowchart of the first parameter value provided by the embodiments of the present application, and specifically includes the following steps:

[0163] Step 1010: For the car-following vehicle pair, obtain the speed difference sequence and the following vehicle acceleration sequence corresponding to the front vehicle and the following vehicle within a preset time period.

[0164] In one embodiment, the front vehicle in the car-following vehicle pair is called A Front , and the following vehicle is called A Back . At this time, at each moment within the preset time period, it includes the vehicle speed and vehicle acceleration of the front vehicle and the following vehicle. Calculate the difference between the vehicle speeds of the two within the preset time period to obtain the speed difference corresponding to each moment. The speed difference sequence formed by arranging the speed differences corresponding to each moment in chronological order can represent the change of the speed difference between the front and rear vehicles over time. Then, arrange the vehicle acceleration corresponding to the following vehicle at each moment within the preset time period in chronological order to obtain the following vehicle acceleration sequence, which reflects the acceleration and deceleration conditions of the following vehicle at different moments.

[0165] It can be understood that a vehicle in the same lane can be the leading vehicle in one following vehicle pair and may be the trailing vehicle in another following vehicle pair.

[0166] Step 1020: Calculate the cross-correlation coefficient at each moment according to the speed difference sequence and the trailing vehicle acceleration sequence, and select the moment corresponding to the maximum value of the cross-correlation coefficient as the reaction time corresponding to the trailing vehicle.

[0167] In one embodiment, given the speed difference sequence and the trailing vehicle acceleration sequence, the cross-correlation coefficient at each moment can be calculated according to the cross-correlation function. Here, the cross-correlation coefficient is used to measure the degree of association between the speed difference sequence and the trailing vehicle acceleration sequence at different moments. Among them, the value of the cross-correlation coefficient ranges from -1 to 1. The closer the value is to 1, the stronger the positive correlation between the two sequences; the closer the value is to -1, the stronger the negative correlation between the two sequences; and when the value is close to 0, it indicates that the correlation between the two sequences is weak.

[0168] In one embodiment, the cross-correlation function is a method for describing the degree of correlation between the values of random signals x(t) and y(t + τ) at any two different moments t and t + τ. The cross-correlation function between random signals x(t) and y(t + τ) is expressed as:

[0169]

[0170] Therefore, the cross-correlation coefficient of x(t) and y(t + τ) with a time shift of τ is expressed as:

[0171]

[0172] Referring to the above calculation process of the cross-correlation coefficient, in the embodiment of the present application, when the preset time period is t, the length between two adjacent moments, that is, the time shift, is denoted as τ. At this time, the speed difference sequence of the leading vehicle and the trailing vehicle with a length of t is used as x(t), and the trailing vehicle acceleration sequence is used as y(t). Applying the above cross-correlation analysis method, within the time shift range [-t, t], calculate the cross-correlation coefficient ρ xy (τ) to obtain a sequence of cross-correlation coefficients.

[0173] In one embodiment, the item with the largest cross-correlation coefficient in this sequence represents the strongest correlation between the two sequences at this time. For example, assuming that when the time shift τ = 2 seconds, the cross-correlation coefficient reaches the maximum value, then it is considered that the correlation between the change in the vehicle speed of the leading vehicle and the vehicle acceleration of the trailing vehicle is the strongest at this time.

[0174] During the following process, the driver of the following vehicle will react according to the speed change of the preceding vehicle and adjust their driving behavior (such as accelerating or decelerating). When the correlation between the two is the strongest, this time shift can reflect the time interval from when the driver of the following vehicle perceives the speed change of the preceding vehicle to when they start to react. Therefore, in the embodiments of the present application, the moment corresponding to the maximum value of the cross-correlation coefficient is approximately considered as the time point when the following vehicle reacts to the speed change of the preceding vehicle, and the duration from the start to this moment is the reaction time of the following vehicle.

[0175] Step 1030: Calculate the average value of the reaction times of all following vehicles within a preset time period to obtain the average reaction time, and generate multiple first parameter values based on the average reaction time.

[0176] In one embodiment, for each following vehicle pair within a preset time period, calculate the average value of the reaction time of the following vehicle in the pair, and the average reaction time of all following vehicles at this intersection can be obtained. In addition, outliers need to be removed from the reaction time of the following vehicle. Since the reaction time generally distributes in the range of (0, 4] seconds, values exceeding this range are deleted.

[0177] Next, after having the average reaction time, multiple values can be selected within the range on both sides of this value as the first parameter values.

[0178] Step 920: Substitute the parameter set into the following model to obtain multiple corresponding following simulation models.

[0179] In one embodiment, since there are multiple parameter sets, substituting different parameter sets into the above following model can obtain the following simulation model corresponding to each parameter set.

[0180] Step 930: For each following simulation model, input the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle.

[0181] In one embodiment, for each following simulation model, calculate a predicted speed data corresponding to the following vehicle. Refer to Figure 11 , Figure 11 is the flowchart of inputting the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle provided by the embodiments of the present application, which specifically includes the following steps:

[0182] Step 1110: Obtain the current speed and the maximum acceleration according to the current radar data of the following vehicle.

[0183] In one embodiment, the current radar data of the following vehicle within a preset time period includes the current speed of the following vehicle at each moment and the maximum acceleration within the preset time period.

[0184] Step 1120: Obtain the speed difference between the front vehicle and the rear vehicle and the distance between the front vehicle and the rear vehicle based on the current radar data corresponding to the front vehicle and the rear vehicle respectively.

[0185] In one embodiment, according to the current radar data of the front vehicle and the rear vehicle respectively, the speed difference between the front vehicle and the rear vehicle and the distance between the front vehicle and the rear vehicle at each moment can be obtained.

[0186] Step 1130: Obtain the desired speed of the rear vehicle, calculate the acceleration data of the rear vehicle based on at least the desired speed, the current speed, the maximum acceleration, the speed difference between the front vehicle and the rear vehicle, and the distance between the front vehicle and the rear vehicle, and use the acceleration data as the predicted speed data.

[0187] In one embodiment, for the desired speed of the rear vehicle, it can also be calculated according to the real-time data of the intersection to match the actual road conditions. Refer to Figure 12 , Figure 12 is the flowchart for obtaining the desired speed of the rear vehicle provided by the embodiment of the present application, which specifically includes the following steps:

[0188] Step 1210: Obtain the vehicle speeds of all vehicles within a preset time period, and use the vehicles whose vehicle speeds are less than the second preset speed at least twice as start-stop vehicles.

[0189] In one embodiment, it is necessary to screen the vehicles at the intersection and select the vehicles with start-stop behaviors as start-stop vehicles. For all vehicles within the preset time period, if there is a start-stop process, the vehicle speed will first decrease and then increase. Therefore, a second preset speed V min ,V min is set greater than the minimum vehicle speed within the preset time period. If the vehicle speed is less than the second preset speed at least twice, it can be considered that there is a process of first decreasing and then increasing, and the vehicle is used as a start-stop vehicle.

[0190] Step 1220: For the start-stop vehicles, extend the two speed timestamps corresponding to the second preset speed in the vehicle speed forward and backward to obtain the start-stop time period.

[0191] In one embodiment, for the start-stop vehicles, obtain the vehicle speed at each moment from all the current radar data corresponding to the vehicle, and search forward and backward along the time dimension. At this time, the second preset speed will correspond to two speed timestamps in the vehicle speed. For example, if the second preset speed is set to 5 km / h, the timestamp when the vehicle first appears at the second preset speed is 10:00:00, and the timestamp when the vehicle appears at the second preset speed for the second time is 10:00:05. Therefore, the two speed timestamps are 10:00:00 and 10:00:05.

[0192] At this time, for the first speed timestamp, select the moment when the vehicle speed is first greater than the second preset speed forward as the first timestamp, for example, 9:59:59. For the second speed timestamp, select the moment when the vehicle speed is first greater than the second preset speed backward as the second timestamp, for example, 10:00:06. For the first timestamp and the second timestamp, extend 3 seconds forward and backward, and the obtained time range is the start-stop time period. For example, the start-stop time period is expressed as [9:59:56, 10:00:09].

[0193] Step 1230: Select at least the maximum value of the vehicle acceleration in the current radar data corresponding to the start-stop time period, and use the moment corresponding to the maximum value as the start time, and obtain the start speed of the start-stop vehicle according to the start time.

[0194] In one embodiment, after obtaining the start-stop time period, at least select the maximum value of the vehicle acceleration in the current radar data corresponding to the start-stop time period. At this time, the maximum value is a positive value. Use the moment corresponding to the maximum value as the start time The vehicle speed corresponding to one second after the start time, that is Obtain the corresponding vehicle speed as the start speed of the start-stop vehicle.

[0195] In addition, it is also possible to at least select the minimum value of the vehicle acceleration in the current radar data corresponding to the start-stop time period, and use the moment corresponding to the minimum value as the stop time

[0196] Step 1240: Obtain the expected speed according to the average value of the start speeds of all start-stop vehicles within the preset time period.

[0197] In one embodiment, obtain the average value of the start speeds of all start-stop vehicles within the preset time period, and use this average value as the expected speed of the vehicle behind in this intersection.

[0198] After having the expected speed v0 of the vehicle behind, according to the expected speed v0, the current speed v α 、the maximum acceleration a’, the speed difference Δv between the front and rear vehicles, and the distance S between the front and rear vehicles α etc., substitute them into the above-mentioned following model, which is expressed as:

[0199]

[0200] Calculate the acceleration data a of the vehicle behind, and then use the acceleration data as the predicted speed data.

[0201] It can be understood that due to different parameter sets, even if there are the same expected speed v0, current speed v α 、the maximum acceleration a’, the speed difference Δv between the front and rear vehicles, and the distance S between the front and rear vehicles αetc., and the obtained predicted speed data is also different. In this embodiment, the following method is used to obtain the following vehicle in each following vehicle pair, and the predicted speed data corresponding to each moment under each following simulation model.

[0202] Step 140: Using the method of orthogonal analysis, error statistics is performed according to the current speed data and the predicted speed data corresponding to the following vehicle to obtain an error result, and the simulation parameters in the following model are calibrated based on the error result to obtain the target simulation parameters corresponding to the simulation parameters.

[0203] In one embodiment, with the predicted speed data corresponding to each following simulation model, the method of orthogonal analysis is used to perform error statistics according to the current speed data and the predicted speed data corresponding to the following vehicle to obtain an error result. Among them, the square difference between the vehicle acceleration at each moment and the corresponding predicted speed data can be calculated first, and the average value of the square differences at each moment corresponding to the same simulation following model is calculated as the error result corresponding to the simulation following model.

[0204] Next, the parameter set with the smallest error result is selected as the target parameter set, and then the first parameter value in the target parameter set is used as the target safe following distance, and the second parameter value in the target parameter set is used as the target stopping distance. The target simulation parameters are obtained according to the target safe following distance and the target stopping distance. That is to say, for different simulation following models, the parameter group with the smallest error value is selected as the target parameter set. At this time, the safe following distance T in the target parameter set is used as the target safe following distance, and the stopping distance s0 in the target parameter set is used as the target stopping distance.

[0205] It can be seen that the following model corresponding to the target parameter set can accurately reflect the vehicle following state of the corresponding intersection. Therefore, the target parameter set is used as the calibration result of the following model.

[0206] The extraction of microscopic traffic flow parameters is of great significance for microscopic traffic simulation. It is the basis and core of the entire microscopic calibration module. Microscopic simulation needs to fit the traffic flow operation in reality as much as possible. The microscopic simulation system tries to restore the actual situation from multiple angles such as traffic flow and path. The microscopic traffic flow simulation parameters are also an important part of it. The driving behaviors of vehicle flows vary greatly in different weather or road conditions. For example, the acceleration and deceleration conditions of vehicles are different in rainy days, foggy days and normal sunny days. Another example is that the marking design is complex and the intersection is inhumane, and a straight intersection. The driver's caution when driving through must also be different. Therefore, it is necessary to calibrate the traffic flow operation parameters of each intersection in real time, and the purpose is to ensure that the microscopic traffic simulation can fit the current traffic flow situation in real time.

[0207] In the embodiments of the present application, the extracted traffic flow parameters also include the driver's reaction time, saturation flow rate, etc. These indicators can all reflect the design level of the intersection to a certain extent, such as indicating signs, intersection markings, and channelization design, etc. In the long-term real-time extraction of parameters, not only can the design of the intersection be evaluated, but also if there is a drastic change in the parameters, it can directly indicate that an accident has occurred at the intersection, which helps traffic decision-making.

[0208] In addition, compared with the traditional traffic flow parameter extraction methods, the embodiments of the present application get rid of the dependence on high-precision map information and can be directly applied at any intersection. In addition, considering that the quality of radar perception at intersections in the current industry is generally not high, the orthogonal analysis method is used to limit the parameter range of possible car-following models in advance, effectively reducing the interference of heterogeneous data on the calculation of index results. Therefore, compared with the traditional methods for obtaining traffic flow parameters at urban intersections, the embodiments of the present application have higher applicability and can be directly connected and used at any intersection with radar perception conditions.

[0209] The following describes the traffic flow simulation parameter calibration method of the embodiments of the present application with a specific example.

[0210] First, the radar data in this embodiment is the 5-minute radar trajectory data collected from an actual cross intersection. That is to say, the preset time period is 5 minutes, and the frequency of the radar detecting vehicle information is 1 time per 0.1 second. That is to say, the acquisition time is 0.1 s. At this time, each frame in the radar trajectory is equal to 0.1 second, and each frame corresponds to a set of current radar data.

[0211] Table 1 below is an example of the current radar data in this embodiment. The vehicle acceleration at each moment can be calculated according to the vehicle speeds corresponding to the vehicle's front and rear timestamps. The current radar data in Table 1 includes vehicle ID, yaw angle, collected timestamp, vehicle acceleration, longitude, latitude, vehicle speed, etc.

[0212]

[0213]

[0214] Table 1 below is an example of the current radar data in this embodiment.

[0215] Next, taking the current radar data as lane division data, obtain the reference vehicle coordinates and reference vehicle speed from the lane division data, then cluster according to the reference vehicle coordinates to obtain the lane centerlines of at least one lane, obtain the lane width, expand the lane centerlines according to the lane width to obtain the lane boundary lines corresponding to each lane, and at the same time, select the vehicles with a reference vehicle speed less than the first preset speed as low-speed vehicles, and cluster the reference vehicle coordinates of the low-speed vehicles to obtain the lane stop line.

[0216] Refer to Figure 13 , Figure 13 which is a schematic diagram of lane area division provided by an embodiment of the present application. It can be seen that the radar trajectories show a linear distribution. Based on DBSCAN clustering, the lane boundary lines of each approach lane and the endpoints of the lane boundary lines are extracted. Taking the 5 lanes of the south approach in the figure as an example, the lane area of the intersection is divided. As Figure 12 shown, 12 endpoints of the south approach are obtained and named s1, s2... s 12 (only endpoint s1 is shown in the figure), and then according to the longitude and latitude coordinates of the 12 endpoints, the approximate straight-line formulas l1, l2... l6 of the six lane lines are further calculated as the lane boundary lines from left to right, and the lane stop line l7 and the possible solid line termination line l8 are obtained. Next, for lane division, the area between l1, l2 and l7, l8 can be named S-1, corresponding to the first lane from the left of the south approach. In this way, the lane field attribute is added and classified for each record of the 5min radar trajectory data in turn. For the records not in the lane area, the lane field remains empty.

[0217] Then, the continuous vehicle passing situation of each lane is determined according to the trajectory data. First, based on the lane boundary lines, the vehicles located in the same lane are determined as lane vehicles, and the timestamp when each lane vehicle passes the lane stop line is obtained as the passing timestamp, and the passing timestamps are sorted in ascending order. Taking the vehicles in the S-1 area in Figure 12 as an example, Lon n and Lat n represent the longitude and latitude of the lane vehicle in the nth frame of the lane. y = ax + b is the straight-line formula of the lane stop line. When the longitude and latitude coordinates of the lane vehicle in the nth frame and the (n + 1)th frame meet the following conditions, it is determined that the vehicle passes the lane stop line in the (n + 1)th frame.

[0218] (a * Lon n + b - Lat n )(a * on n+1 + b - Lat n+1 ) < 0

[0219] After obtaining the time series of each lane vehicle passing the lane stop line in the S-1 area according to the above formula, the passing timestamps are selected pairwise in order from front to back. For each group of selected timestamps, the time difference between them is calculated, and the time difference is shown in Table 2 below.

[0220] Vehicle ID Time difference from the vehicle ahead (s) 2008 3.5 2014 3.8 2021 2.5 2028 1.3 2030 1.8 2033 3.8 2038 3.4 2042 4.8 2047 2.6 2051 1.8 2055 4.1 2061 1.3

[0221] Table 2 is an example of the time difference between the front and rear vehicles.

[0222] As shown in Table 2, if the time difference between passing vehicles of 3 or more consecutive vehicles in the table is < 3 seconds, these vehicles are considered to pass continuously to obtain a continuous passing group, which is recorded as a set of sample data. After traversal, the average headway of the continuous passing group is calculated, and then the saturation flow rate corresponding to the lane is obtained based on the average headway. For example, for the S-1 area, the calculated headway = 1.9 seconds, and the saturation flow rate of the lane = 1895 veh / h.

[0223] The expected speed of the following vehicle at this intersection can also be determined according to the start-stop state. For example, in the 5-minute data, a total of 17 vehicle IDs in the S-1 area were found to have two records of vehicle speed below 3 km / h. Based on this, the average starting speed corresponding to these 17 vehicle IDs is 13.9 km / h.

[0224] Next, multiple sets of following vehicle pairs are selected from the vehicles based on the time stamp. Among them, the following vehicle pair includes the leading vehicle and the following vehicle. For the following vehicle pair, the speed difference sequence and the following vehicle acceleration sequence corresponding to the leading vehicle and the following vehicle within a preset time period are obtained. Table 3 below is an example of the speed difference sequence and the following vehicle acceleration sequence in the embodiments of the present application.

[0225]

[0226] Table 3 is an example of the speed difference sequence and the following vehicle acceleration sequence in the embodiments of the present application

[0227] Then, the cross-correlation coefficient at each moment is calculated based on the speed difference sequence and the following vehicle acceleration sequence, and the moment corresponding to the maximum value of the cross-correlation coefficient is selected as the reaction time corresponding to the following vehicle. Refer to Figure 14 , Figure 14 is a schematic diagram of the cross-correlation coefficient provided by the embodiments of the present application. It can be seen from the figure that when the time shift is 16 frames, the cross-correlation coefficient is the largest, that is, the reaction time of the driver of this vehicle is 1.6 seconds. Based on this, the corresponding average reaction time of this intersection is obtained. This average reaction time can be used to indicate the value of the safe following distance parameter in the following model. Dense values can be taken around the average reaction time to improve the calibration accuracy.

[0228] Next, the safe following distance and the stopping distance in the following model are adjustable parameters. Therefore, they are used as preset simulation parameters to generate multiple corresponding parameter sets. The first parameter value of the safe following distance and the second parameter value of the stopping distance in different parameter sets are not exactly the same. Then, the parameter sets are substituted into the following model to obtain multiple corresponding following simulation models.

[0229] Assume that the value range of the safe following distance is [1 second, 3 seconds], and the value range of the stopping distance is [1 meter, 5 meters]. The following multiple sets of parameter matching results are generated according to the above range.

[0230] Number of the parameter set Safe following distance (s) Stopping distance (m) 1 2.5 1.5 2 1 2.5 3 1.5 2.5 4 3 2.5 5 1 1.5 … … …

[0231] Table 3 is an example of multiple parameter sets

[0232] Next, for the following vehicle pairs, the current radar data is input into each simulation following model for data processing to obtain the predicted speed data of the following vehicle. Using the method of orthogonal analysis, error statistics are performed based on the current speed data and the predicted speed data corresponding to the following vehicle to obtain the error results, as shown in Table 4 below. Finally, the simulation parameters in the following model are calibrated based on the error results to obtain the target simulation parameters corresponding to the simulation parameters

[0233] Parameter set number Safe following distance (s) Stopping distance (m) Square difference 1 0.8 2.6 5.13 2 0.8 2.6 10.72 3 1 2.4 4.46 4 1.2 2.3 1.70 5 1.2 2.4 1.82 … … … …

[0234] Table 4 is an example of error results

[0235] Therefore, it can be obtained that the error result is the smallest when the parameter set number is 4. Therefore, the target safe following distance is set to 1.2 s, and the target stopping distance is set to 2.3 m

[0236] The technical solution provided by the embodiments of this application obtains the current radar data of each vehicle within a preset time period at an intersection, obtains the trajectory data and current speed data of the corresponding vehicle based on the current radar data, and obtains the lane data obtained by dividing the lane area of the intersection, where the lane data is obtained based on the current radar data or historical radar data of the intersection, and the lane data at least includes lane stop lines. Then, the passing timestamp when the vehicle passes the lane stop line is determined according to the trajectory data, and multiple sets of following vehicle pairs are selected from the vehicles based on the passing timestamp, where the following vehicle pair includes a leading vehicle and a following vehicle. Next, for the following vehicle pair, the current radar data is input into the following vehicle model for data processing to obtain the predicted speed data of the following vehicle. The following vehicle model includes preset simulation parameters. Finally, using the method of orthogonal analysis, error statistics are performed according to the current speed data and predicted speed data corresponding to the following vehicle to obtain an error result, and the simulation parameters in the following vehicle model are calibrated based on the error result to obtain the target simulation parameters corresponding to the simulation parameters. In the embodiments of this application, based on the clustering characteristics of the radar trajectory at the intersection, the lanes of the intersection are simulated using radar data. This simulation method can be directly applied to intersections in different scenarios without introducing a high-precision map, thereby reducing the simulation cost. Since at intersections, radar perception is easily interfered by the surrounding environment, resulting in a certain proportion of abnormal data in the collected data. Therefore, this embodiment uses timestamps to select multiple sets of following vehicle pairs, converting the object of the simulation data from the complete vehicle data set to the following vehicle pairs, thereby focusing on the following behavior that has a greater impact on intersection traffic, effectively avoiding the adverse effects of low-quality heterogeneous data on the result accuracy, and reducing the requirements for data quality. Compared with the filtering method used in reinforcement learning in related technologies, the orthogonal analysis method used in this embodiment can effectively avoid the problem that the calibrated parameters deviate from the normal range due to poor data quality, and improve the accuracy of the calibration result.

[0237] The embodiments of this application also provide a device for calibrating traffic flow simulation parameters, which can implement the above traffic flow simulation parameter calibration method. Refer to Figure 15 and this device includes:

[0238] A data acquisition module 1510: used to acquire the current radar data of each vehicle within a preset time period at an intersection, obtain the trajectory data and current speed data of the corresponding vehicle based on the current radar data; and acquire the lane data obtained by dividing the lane area of the intersection, where the lane data is obtained based on the current radar data or historical radar data of the intersection, and the lane data at least includes lane stop lines.

[0239] A following vehicle confirmation module 1520: used to determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple sets of following vehicle pairs from the vehicles based on the passing timestamp, where the following vehicle pair includes a leading vehicle and a following vehicle.

[0240] Model prediction module 1530: For a following vehicle pair, input the current radar data into a following model for data processing to obtain predicted speed data of the following vehicle. The following model includes preset simulation parameters.

[0241] Parameter calibration module 1540: Use the method of orthogonal analysis to perform error statistics based on the current speed data and predicted speed data corresponding to the following vehicle to obtain an error result, and calibrate the simulation parameters in the following model based on the error result to obtain target simulation parameters corresponding to the simulation parameters.

[0242] The specific implementation manner of the traffic flow simulation parameter calibration device in this embodiment is basically the same as that of the above traffic flow simulation parameter calibration method, and will not be elaborated here.

[0243] This application embodiment also provides an electronic device, including:

[0244] At least one memory;

[0245] At least one processor;

[0246] At least one program;

[0247] The program is stored in the memory, and the processor executes the at least one program to implement the traffic flow simulation parameter calibration method described above in this application. The electronic device can be any intelligent terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), an in-vehicle computer, etc.

[0248] Please refer to Figure 16 , Figure 16 which schematically shows the hardware structure of an electronic device in another embodiment. The electronic device includes:

[0249] A processor 1601, which can be implemented by using a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, etc., and is used to execute relevant programs to implement the technical solutions provided by this application embodiment;

[0250] The memory 1602 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 1602 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 1602 and are called by the processor 1601 to execute the traffic flow simulation parameter calibration method of the embodiments of this application;

[0251] The input / output interface 1603 is used to implement information input and output;

[0252] The communication interface 1604 is used to implement communication and interaction between this device and other devices. It can achieve communication through wired means (such as USB, network cable, etc.) or through wireless means (such as mobile network, WIFI, Bluetooth, etc.); and

[0253] The bus 1605 transmits information between various components of the device (such as the processor 1601, the memory 1602, the input / output interface 1603, and the communication interface 1604);

[0254] Among them, the processor 1601, the memory 1602, the input / output interface 1603, and the communication interface 1604 achieve communication connections with each other inside the device through the bus 1605.

[0255] The embodiments of this application also provide a storage medium. The storage medium is a storage medium that stores a computer program. When the computer program is executed by a processor, it implements the above-mentioned traffic flow simulation parameter calibration method.

[0256] As a non-transitory storage medium, the memory can be used to store non-transitory software programs and non-transitory computer executable programs. In addition, the memory can include a high-speed random access memory and can also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes a memory remotely set relative to the processor, and these remote memories can be connected to the processor through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0257] The traffic flow simulation parameter calibration method, device, equipment and storage medium proposed in the embodiments of the present application obtain the current radar data of each vehicle within a preset time period at an intersection, obtain the trajectory data and current speed data of the corresponding vehicle according to the current radar data, and obtain the lane data obtained by dividing the lane area at the intersection, where the lane data is obtained according to the current radar data or historical radar data of the intersection, and the lane data at least includes lane stop lines. Then, the passing timestamp when the vehicle passes the lane stop line is determined according to the trajectory data, and multiple groups of following vehicle pairs are selected from the vehicles based on the passing timestamp, where the following vehicle pair includes a leading vehicle and a following vehicle. Next, for the following vehicle pair, the current radar data is input into the following model for data processing to obtain the predicted speed data of the following vehicle, and the following model includes preset simulation parameters. Finally, using the method of orthogonal analysis, error statistics are performed according to the current speed data and predicted speed data corresponding to the following vehicle to obtain an error result, and the simulation parameters in the following model are calibrated based on the error result to obtain the target simulation parameters corresponding to the simulation parameters. In the embodiments of the present application, based on the aggregation characteristics of radar trajectories at intersections, radar data is used to simulate the lanes at intersections. This simulation method can be directly applied to intersections in different scenarios without introducing high-precision maps, thereby reducing the simulation cost. Since at intersections, radar perception is easily interfered by the surrounding environment, resulting in a certain proportion of abnormal data in the collected data. Therefore, in this embodiment, timestamps are used to select multiple groups of following vehicle pairs, and the object of the simulation data is converted from the complete vehicle data set to the following vehicle pairs, so as to focus on the following behavior that has a greater impact on intersection traffic, effectively avoiding the adverse impact of low-quality heterogeneous data on the result accuracy, and reducing the requirement for data quality. Compared with the filtering method used in reinforcement learning in the related art, the orthogonal analysis method used in this embodiment can effectively avoid the problem that the calibrated parameters deviate from the normal range due to poor data quality, and improve the accuracy of the calibration result.

[0258] The embodiments described in the embodiments of the present application are for more clearly explaining the technical solutions of the embodiments of the present application, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0259] Those skilled in the art can understand that the technical solutions shown in the figures do not constitute a limitation to the embodiments of the present application, and may include more or fewer steps than shown in the figures, or combine some steps, or different steps.

[0260] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0261] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in systems and devices, can be implemented as software, firmware, hardware, and appropriate combinations thereof.

[0262] As used in the specification of this application and the above drawings, the terms "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0263] It should be understood that in this application, "at least one (item)" means one or more, and "multiple" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one)" or a similar expression below refers to any combination of these items, including any combination of single item (one) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0264] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0265] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0266] In addition, in each embodiment of the present application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units.

[0267] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store programs.

[0268] The preferred embodiments of the embodiments of the present application have been described above with reference to the drawings, but this does not limit the scope of the rights of the embodiments of the present application. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of the rights of the embodiments of the present application.

Claims

1. A method for calibrating traffic flow simulation parameters, characterized in that Including: Obtain the current radar data of each vehicle within a preset time period at an intersection, and obtain the trajectory data and current speed data of the corresponding vehicle based on the current radar data; And obtain the lane data obtained by dividing the lane area of the intersection. The lane data is obtained based on the current radar data or historical radar data of the intersection, and the lane data at least includes lane stop lines; Determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple groups of following vehicle pairs from the vehicles based on the passing timestamp. The following vehicle pairs include a leading vehicle and a following vehicle; For the following vehicle pairs, input the current radar data into a following model for data processing to obtain the predicted speed data of the following vehicle. The following model includes preset simulation parameters; Using the method of orthogonal analysis, perform error statistics on the current speed data and the predicted speed data corresponding to the following vehicle to obtain an error result, and calibrate the simulation parameters in the following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters.

2. The traffic flow simulation parameter calibration method according to claim 1, wherein Obtaining the lane data obtained by dividing the lane area of the intersection according to the current radar data or historical radar data includes: Taking the current radar data or historical radar data as lane division data, and obtaining the reference vehicle coordinates and reference vehicle speed from the lane division data; Performing clustering according to the reference vehicle coordinates to obtain the lane centerlines of at least one lane; Obtain the lane width, and expand the lane centerlines according to the lane width to obtain the lane boundary lines corresponding to each lane; Select the vehicles with the reference vehicle speed less than the first preset speed as low-speed vehicles, and perform clustering on the reference vehicle coordinates of the low-speed vehicles to obtain the lane stop lines.

3. The traffic flow simulation parameter calibration method according to claim 1, wherein The lane data further includes lane boundary lines. Determining the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and selecting multiple groups of following vehicle pairs from the vehicles based on the passing timestamp includes: Determine the vehicles located in the same lane as lane vehicles based on the lane boundary lines; Obtain the timestamp when each lane vehicle passes the lane stop line as the passing timestamp, and sort the passing timestamps in ascending order; According to two consecutive passing timestamps, obtain the time difference corresponding to the passing timestamps. If the time difference is less than the preset time, determine that the passing identifiers of the corresponding two lane vehicles are consecutive passing identifiers, and determine multiple groups of the following vehicle pairs based on the consecutive passing identifiers.

4. The traffic flow simulation parameter calibration method according to claim 3, wherein, The current speed data includes the vehicle speed. Determining multiple groups of the following vehicle pairs based on the consecutive passing identifiers includes: Select the two lane vehicles corresponding to the consecutive passing identifiers as the initial vehicle pair. The initial vehicle pair includes an initial leading vehicle and an initial following vehicle; Traverse all the initial vehicle pairs, obtain the recording times of the current radar data corresponding to the initial leading vehicle and the initial following vehicle respectively, and / or, at each moment, determine the speed change amounts of the initial leading vehicle and the initial following vehicle respectively according to the corresponding vehicle speeds, and / or, obtain the distance differences between the initial leading vehicle and the initial following vehicle corresponding to each moment according to the trajectory data; If the recording times are all greater than or equal to a preset number of times, and / or, the speed change amounts are all greater than or equal to a preset speed change amount, and / or, the distance differences are all less than a preset distance difference, regard the initial vehicle pair as the following vehicle pair.

5. The traffic flow simulation parameter calibration method according to claim 3, characterized in that The simulation parameters at least include: the safe following time headway and the stopping distance. The inputting the current radar data into the following model for data processing to obtain the predicted speed data of the following vehicle includes: Obtain multiple parameter sets, and the first parameter values of the safe following time headway and the second parameter values of the stopping distance in different parameter sets are not completely the same; Substitute the parameter sets into the following model to obtain multiple corresponding following simulation models; For each following simulation model, input the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle.

6. The traffic flow simulation parameter calibration method according to claim 5, wherein The inputting the current radar data into the following simulation model for data processing to obtain the predicted speed data corresponding to the following vehicle includes: Obtain the current speed and the maximum acceleration according to the current radar data of the following vehicle; Obtain the speed difference between the leading vehicle and the following vehicle and the distance between the leading vehicle and the following vehicle based on the current radar data corresponding to the leading vehicle and the following vehicle respectively; Obtain the desired speed of the following vehicle, and calculate the acceleration data of the following vehicle at least according to the desired speed, the current speed, the maximum acceleration, the speed difference between the leading vehicle and the following vehicle, and the distance between the leading vehicle and the following vehicle, and regard the acceleration data as the predicted speed data.

7. The traffic flow simulation parameter calibration method according to claim 5, wherein The current speed data includes the vehicle speed, and the current speed data also includes the vehicle acceleration. The obtaining the desired speed of the following vehicle includes: Obtain the vehicle speeds of all the vehicles within the preset time period, and regard the vehicles whose vehicle speeds are less than the second preset speed at least twice as start-stop vehicles; For the start-stop vehicles, expand the two speed timestamps corresponding to the second preset speed in the vehicle speeds forward and backward to obtain the start-stop time period; Select at least the maximum value of the vehicle acceleration in the current radar data corresponding to the start-stop time period, regard the moment corresponding to the maximum value as the start moment, and obtain the start speed of the start-stop vehicle according to the start moment, and the maximum value is a positive value; Obtain the desired speed according to the average value of the start speeds of all the start-stop vehicles within the preset time period.

8. The traffic flow simulation parameter calibration method according to claim 5, characterized in that The calibrating the simulation parameters in the following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters includes: Select the parameter set with the smallest error result as the target parameter set; Use the first parameter value in the target parameters as the target safe following distance, and use the second parameter value in the target parameters as the target stopping distance. Obtain the target simulation parameters according to the target safe following distance and the target stopping distance.

9. The traffic flow simulation parameter calibration method according to claim 3, characterized in that, After determining that the passing identifiers of the two vehicles in the corresponding lanes are consecutive passing identifiers, the method further includes: For each lane, traverse the passing identifiers of the vehicles in the lane. If a preset number of consecutive passing identifiers are all the consecutive passing identifiers, regard the corresponding lane vehicles as a consecutive passing group; Calculate the headway corresponding to the consecutive passing group, and calculate the average value of all the headways to obtain the saturation flow rate corresponding to the lane.

10. The traffic flow simulation parameter calibration method according to claim 5, characterized in that, After selecting multiple following vehicle pairs from the vehicles based on the passing timestamps, the method further includes: For the following vehicle pairs, obtain the speed difference sequence and the rear vehicle acceleration sequence corresponding to the leading vehicle and the rear vehicle within the preset time period; Calculate the cross-correlation coefficient at each moment according to the speed difference sequence and the rear vehicle acceleration sequence, and select the moment corresponding to the maximum value of the cross-correlation coefficient as the reaction time corresponding to the rear vehicle; Calculate the average value of the reaction times of all the rear vehicles within the preset time period to obtain the average reaction time, and generate multiple first parameter values according to the average reaction time.

11. A traffic flow simulation parameter calibration device, characterized in that, It includes: Data acquisition module: used to acquire the current radar data of each vehicle within a preset time period at the intersection, and obtain the trajectory data and the current speed data of the corresponding vehicle according to the current radar data; And acquire the lane data obtained by dividing the lane area at the intersection. The lane data is obtained according to the current radar data or historical radar data of the intersection, and the lane data at least includes lane stop lines; Following confirmation module: used to determine the passing timestamp when the vehicle passes the lane stop line according to the trajectory data, and select multiple following vehicle pairs from the vehicles based on the passing timestamp. The following vehicle pairs include a leading vehicle and a rear vehicle; Model prediction module: used for the following vehicle pairs, input the current radar data into the following model for data processing to obtain the predicted speed data of the rear vehicle. The following model includes preset simulation parameters; Parameter calibration module: used to use the method of orthogonal analysis to perform error statistics according to the current speed data and the predicted speed data corresponding to the rear vehicle to obtain an error result, and calibrate the simulation parameters in the following model based on the error result to obtain the target simulation parameters corresponding to the simulation parameters.

12. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the traffic flow simulation parameter calibration method according to any one of claims 1 to 10.

13. A storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the traffic flow simulation parameter calibration method according to any one of claims 1 to 10.

Images