Real-time online traffic simulation system calibration method based on shadow vehicle

Abstract

The invention discloses a shadow vehicle-based real-time online traffic simulation system calibration method, and belongs to the field of intelligent traffic systems, and the method comprises the steps: sampling vehicle driving data, and defining a shadow vehicle; performing space-time coordinate registration on the shadow vehicle; a simulation shadow vehicle is generated for the shadow vehicle after the time-space coordinates are registered through traffic simulation software, and the speed of the simulation shadow vehicle is calculated; carrying out delayed sampling operation on the vehicle driving data of the simulation shadow vehicle, and correcting the simulation process; a simulation period is set, a simulation result is obtained, and calibration of the real-time online traffic simulation system is achieved. According to the method, the data can be more directly and effectively used for rapid calibration of the real-time online traffic simulation system without adopting an optimization algorithm to carry out high-intensity iterative optimization calculation, so that the simulation system can more accurately reproduce and track the actual traffic condition.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation systems, and particularly to a calibration method for a real-time online traffic simulation system based on shadow vehicles. Background Art

[0002] ITS mainly achieves the dynamic balance between traffic travel demand and road network supply capacity through real-time traffic control (including intersection signal control, ramp control, variable speed limit control, etc.) and traffic guidance, etc., so as to effectively reduce traffic congestion and improve travel efficiency. The premise of its effectiveness is that it can reliably reproduce the current traffic conditions, that is, the deviations of the traffic flow, speed, density, travel time, etc. of each road section (a road section refers to a one-way road between two adjacent intersections) output by the simulation system compared with the actual detection data are relatively small. Otherwise, the evaluation results of the traffic control and guidance schemes obtained based on this system lose their practical significance and are difficult to be used to improve the actual traffic conditions. The process of estimating the key input data in the traffic simulation system, such as the dynamic OD traffic matrix and calibrating traffic model (such as vehicle following model, road section speed-density model, etc.) parameters through detection data (traffic flow, speed, density, travel time, etc.) is called the calibration of the traffic simulation system, so that the output of the traffic simulation system is consistent with the actual detection volume of the on-site traffic detector, so that the traffic simulation system can accurately estimate or reproduce the current traffic conditions, thereby providing a reliable initial condition for the traffic simulation deduction and evaluation of loading the predicted dynamic OD traffic matrix.

[0003] Traditionally, traffic simulation systems are mainly calibrated using data such as traffic flow, speed, and occupancy collected by fixed traffic detectors such as video, microwave, geomagnetic, and loop detectors. Since there is a clear mapping relationship (such as the distribution matrix) between the traffic flow detected at the road section and the OD traffic input of the simulation system, the section traffic flow data is more often used to estimate (calibrate) the dynamic OD traffic matrix and is relatively effective; however, there is a complex non-linear mapping relationship between the speed detection volume (including data such as the average speed collected by fixed detectors and the vehicle speeds at different spatial positions collected by mobile detection vehicles) and the dynamic OD traffic and traffic model parameters, and it is difficult to find a clear calculation formula. Therefore, the calibration process usually requires a large number of iterative calculations using optimization algorithms, with low calculation efficiency and not very effective. At present, there is an urgent need to use these data more effectively and directly for the calibration of real-time online traffic simulation systems. Summary of the Invention

[0004] To solve the above technical problems, the present invention provides a calibration method for a real-time online traffic simulation system based on shadow vehicles.

[0005] To achieve the above technology, the specific steps are as follows:

[0006] S1. Sample the vehicle driving data and define the shadow vehicle:

[0007] The method of sampling the vehicle driving data is as follows: obtain the driving state data of the intelligent connected vehicles, running on-vehicle devices, self-driving vehicles of mobile navigation apps, and online car-hailing vehicles in the actual road network at a fixed sampling period T (T = 10S);

[0008] The driving state data includes: vehicle ID (identification number), the moment of sending data, i.e., the physical timestamp t0, longitude and latitude coordinates, the current speed v(t0), the turning direction at the downstream intersection, the current mileage s(t0) (obtained by a navigation system with meter-level accuracy), and the ID (number) of the road section / lane where the vehicle is located;

[0009] Among them, the intelligent connected vehicle in the actual road network is: an intelligent vehicle equipped with an on-board unit OBU that can communicate with a roadside unit RSU in real time;

[0010] According to the ID of the road section / lane where the vehicle is located in the sampled data, convert the longitude and latitude coordinates of all vehicles into plane rectangular coordinates in the traffic simulation system to obtain the specific positions of the vehicles in the simulation road network;

[0011] For each vehicle in the road network that can collect real-time speed and position, if a corresponding virtual vehicle is generated in the simulation system, it will lead to an abnormal increase in the number of vehicles on the road, duplicate calculation of traffic flow, and an incorrect decrease in the simulation vehicle speed due to the increase in traffic density; in the present invention, by making the length of these newly added virtual vehicles 0 and not occupying any road resources, they become shadow vehicles, which will not affect the normal simulation process, and are only used to track the simulation vehicle speeds at different positions, and by comparing with the physical average vehicle speed corresponding to the shadow vehicle, to correct the speeds of the normal simulation vehicles around the shadow vehicle;

[0012] Define the vehicle that satisfies the following formula as a shadow vehicle, and the expression is as follows:

[0013] v(t0) < v f

[0014] v(t0)·T < D end

[0015] Among them, v f is the free flow speed of the road section where the vehicle is located, i.e., the speed limit; D end is the distance of the vehicle from the intersection ahead; the vehicle that satisfies the speed condition and is still in the plane rectangular coordinates in the traffic simulation system within a sampling period T is used as a shadow vehicle;

[0016] S2. Perform spatio-temporal coordinate registration on the shadow vehicle;

[0017] During the traffic simulation and deduction process, the displacement of each vehicle within a complete simulation period is obtained through the road network and the vehicle position is updated. Therefore, the starting moment of the current simulation period of each shadow vehicle, i.e., the simulation timestamp t sim0 is set as the starting point of this period, and the time interval corresponding to the simulation period is [T start , T end . The expression for the simulation step size is as follows:

[0018] T step = T end - T start

[0019] In the formula, T start represents the starting time of the simulation; T end represents the ending time of the simulation; T step represents the simulation step size, and its value range is 1s to 10s, and t sim0 = T start ;

[0020] When the physical timestamp t0 is not equal to T start , then the position of the shadow vehicle at t sim0 is different from its position at t0, and the longitudinal position of the vehicle on the road section needs to be registered. The expression is as follows:

[0021] S new = S old - v(t0) * (t0 - t sim0 )

[0022] In the formula, S old and S new are respectively the longitudinal positions of the vehicle on the road section at the physical timestamp t0 and the simulation timestamp t sim0 ;

[0023] After the registration is completed, the time coordinates (simulation timestamps) of all shadow vehicles are the starting points of a certain simulation period, and the position coordinates are the spatial coordinates corresponding to this time coordinate, that is, the corresponding longitudinal position and the lane where the vehicle is located (lateral position) in the road section;

[0024] If the distance from the shadow vehicle to the stop line ahead is less than the length of the channelized section (60 - 100m, divided according to the road grade), then according to the turning direction of the vehicle at the intersection ahead and the intersection channelization method (i.e., the lane division method for different turning directions, such as 1 straight - right lane, 2 straight - through lanes, and 1 left - turn lane), its corresponding turning lane is determined through the road network and the lane ID where it is located is assigned. When there are multiple lanes with the same turning direction, any one of them can be selected. In other cases, the shadow vehicle can be assigned any lane in the road section, and it does not need to be exactly the same as the lane where the actual vehicle is located.

[0025] S3. Generate a simulated shadow vehicle for the shadow vehicle after registering the spatio-temporal coordinates through traffic simulation software and calculate the speed of the simulated shadow vehicle;

[0026] The traffic simulation software includes: medium or micro traffic simulation software such as DynasTIM, SUMO, and Transmodeler;

[0027] Generating a simulated shadow vehicle means: generating a simulation object in the simulation software with speed, simulation timestamp, longitudinal position, lane where it is located, downstream turning direction, and the length, width, and height of the shadow vehicle are all 0 m;

[0028] Load the generated shadow vehicle into the simulation road network. The loading method is: after the shadow vehicle passes through the registration in S2, input it into the road sections in the road network of the traffic simulation software;

[0029] Calculating the simulated vehicle speed means: calculating the speed v of all simulated vehicles including the shadow vehicle according to a medium (such as a speed-density relationship model, etc.) or a micro traffic model (such as a car-following model, etc.) i , i = 1, 2, …, N, where N is the total number of vehicles in the simulation road network;

[0030] The present invention uses the DynasTIM traffic simulation system. The formula for calculating the speed of simulated vehicles by the medium traffic model is as follows:

[0031]

[0032] In the formula, the subscript i represents the i-th simulated vehicle; the superscript t represents the t-th simulation period, t ∈ T step ; represents the traffic density within the speed influence region (SIR i ) of vehicle i during the simulation period (t - 1); k jam represents the jam density, that is, assuming that all vehicles within SIR i are queuing and driving at the minimum spacing, the traffic density ranges from [0.10, 0.15], and the unit is: standard vehicle / lane / meter; represents the simulation period (t - 1), and the number of vehicles (excluding vehicle i itself) within the SIR i of vehicle i; n represents the number of lanes included in SIR i ; and respectively represent the average driving speeds of vehicle i during the simulation periods t and (t - 1); represents the length of the SIR i of vehicle i during the simulation period (t - 1); resp_time represents the average reaction time of the driver to the traffic conditions ahead, set to 2 s to 3 s; v fDenote the free flow speed, which is set as the speed limit value of the road section; α and β represent the first parameter and the second parameter to be calibrated, and the default value is 1.0.

[0033] S4. Calibrate the simulation process by performing a delayed sampling operation on the vehicle driving data of the simulated shadow vehicles.

[0034] The simulation process includes: the simulated vehicle speed and the total driving state data of the traffic simulation model.

[0035] The delayed sampling is: delaying by one simulation step T step and then sampling, that is, receiving the operating state data of the physical vehicles corresponding to all the shadow vehicles in the simulation system through the communication network, including the vehicle ID, the time of sending the data, i.e., the physical timestamp t1, the longitude and latitude coordinates, the current speed v(t1), the turning direction at the downstream intersection, the current mileage s(t1) (obtained by a navigation system with meter-level accuracy), and the ID (number) of the road section / lane where it is located; according to the actual driving mileage s(t1) and the corresponding physical timestamp t1 of the shadow vehicle i, calculate the actual speed v of the shadow vehicle in the time period [t0, t1] actual =(s(t1)-s(t0)) / (t1 - t0). When the simulated vehicle speed of the shadow vehicle i calculated in S3 is v sim , if the relative error between it and the actual speed is greater than the critical value Δ, then the simulated vehicle speed of the shadow vehicle should be calibrated.

[0036] The expression for the relative error of the actual speed being greater than the critical value Δ is as follows:

[0037]

[0038] In the formula, the critical value Δ is 30%.

[0039] The calibration method is to use the inertial filtering method for calibration, and the expression is as follows:

[0040] v sim_correct =α1v sim +(1 - α1)v actual ;

[0041] delta_v = v sim_correct -v sim ;

[0042] In the formula, v sim_correct represents the calibrated simulated vehicle speed; delta_v represents the difference between the calibrated simulated vehicle speed and the simulated vehicle speed; α1 represents a real number in the interval [0, 0.5].

[0043] If the distance of the shadow vehicle from the stop line ahead is greater than the channelization section length, then on the road section where the shadow vehicle is located, the distance from the position where the shadow vehicle is located is Select the vehicles in all lanes before and after within as the simulation vehicles for speed and displacement correction, and set the set of their ID numbers (identification numbers) as Nc. At the same time, add the ID number of the shadow vehicle to Nc. For the simulation vehicle, its corrected vehicle speed is calculated according to the following formula:

[0044]

[0045] In the formula, represents the corrected simulation vehicle speed when the distance of the shadow vehicle from the stop line ahead is greater than the length of the channelized section; the simulation vehicle speed when the distance of the shadow vehicle from the stop line ahead is greater than the length of the channelized section; when there are M s shadow vehicles on the same road section, the simulation vehicle speed is calculated for each shadow vehicle, with a total of M s times of correction;

[0046] If the corrected simulation vehicle speed when the distance of the shadow vehicle from the stop line ahead is greater than the length of the channelized section is less than the queuing driving speed after the vehicle enters the queue that is then update the traffic capacity of the lane where the shadow vehicle is located according to the following formula:

[0047]

[0048] where c i is the traffic capacity of the lane where the shadow vehicle is located, that is, the maximum number of vehicles that can leave the lane per unit time; is set to 3 m / s;

[0049] In the actual road network, if the physical vehicle corresponding to the shadow vehicle is affected by a traffic event and enters the queuing state, then by calculating c i method, the traffic simulation system can automatically reflect the decrease in road traffic capacity caused by traffic events by adjusting its model parameter c i .

[0050] S5. Set the simulation period, obtain the simulation results, and realize the calibration of the real-time online traffic simulation system;

[0051] The DynasTIM traffic simulation system calculates the displacement s i = v sim(i) * T step of all simulation vehicles within the current simulation period, where v sim(i) is the simulation vehicle speed calculated according to the traffic model; and move all simulation vehicles to new positions according to the displacement s i . When the shadow vehicle leaves the current road section, its life cycle ends and it is deleted from the simulation road network;

[0052] After completing S1 to S5, receive the vehicle driving data of the next sampling period and repeat steps S1 to S5 to continuously generate new shadow vehicles at the sampling period T or update the relevant attributes such as the speed and position of the existing shadow vehicles in the simulation road network, so as to calibrate the real-time online traffic simulation system.

[0053] Advantages of the present invention:

[0054] Regarding the problem that intelligent connected vehicles and navigation apps can provide a large amount of data such as the real-time position, speed, and driving trajectory of vehicles, but currently these data are not effectively used for calibrating real-time online traffic simulation systems. The present invention can more directly and effectively use these data for rapid calibration of real-time online traffic simulation systems without using optimization algorithms (such as genetic algorithms, stochastic gradient descent algorithms, non-linear optimization algorithms, etc.) for high-intensity iterative optimization calculations, enabling the simulation system to more accurately reproduce and track the actual traffic conditions. Description of the Drawings

[0055] Figure 1 is the flowchart of the steps of the present invention;

[0056] Figure 2 is the simulation system diagram of the present invention;

[0057] Figure 3 is the mesoscopic traffic model diagram of the present invention. Detailed Embodiments

[0058] The present invention will be further described in detail below in conjunction with specific embodiments.

[0059] As Figure 1 and Figure 2 shown, a method for calibrating a real-time online traffic simulation system based on shadow vehicles includes the following steps:

[0060] S1. Sample vehicle driving data and define shadow vehicles:

[0061] The method of sampling vehicle driving data is: obtain the driving state data of intelligent connected vehicles, running on-vehicle, self-driving vehicles of mobile phone navigation apps, and in-transit online car-hailing vehicles in the actual road network at a fixed sampling period T (T = 10S);

[0062] The driving state data includes: vehicle ID (identification number), the physical timestamp t0 of the time when the data is sent, longitude and latitude coordinates, the current speed v(t0), the turning direction at the downstream intersection, the current mileage s(t0) (obtained by a navigation system with meter-level accuracy), and the ID (number) of the road section / lane where the vehicle is located;

[0063] Among them, the intelligent connected vehicles in the actual road network are: intelligent vehicles equipped with on-board units (OBUs) that can communicate with roadside units (RSUs) in real time;

[0064] According to the ID of the road section / lane where the vehicle is located sampled, convert the longitude and latitude coordinates of all vehicles into plane rectangular coordinates in the traffic simulation system to obtain the specific positions of the vehicles in the simulated road network;

[0065] For each vehicle in the road network that can collect real-time speed and position, if a corresponding virtual vehicle is generated in the simulation system, it will lead to an abnormal increase in the number of vehicles on the road, duplicate calculation of traffic flow, and an incorrect decrease in the simulated vehicle speed due to the increase in traffic density; in the present invention, by making the length of these newly added virtual vehicles 0 and not occupying any road resources, that is, becoming shadow vehicles, this will not affect the normal simulation process, and is only used to track the simulated vehicle speeds at different positions, and by comparing with the physical average vehicle speed corresponding to the "shadow vehicle", to correct the speeds of the normal simulated vehicles around the shadow vehicle;

[0066] Define the vehicles that meet the following formula as shadow vehicles, and the expression is as follows:

[0067] v(t0) < v f

[0068] v(t0)·T < D end

[0069] Among them, v f is the free flow speed of the road section where the vehicle is located, that is, the speed limit; D end is the distance of the vehicle from the intersection ahead; the vehicle whose speed meets the condition and is still in the plane rectangular coordinates in the traffic simulation system within one sampling period T is used as a shadow vehicle;

[0070] S2. Perform spatio-temporal coordinate registration on the shadow vehicles;

[0071] During the traffic simulation deduction process, the displacement of each vehicle within a complete simulation period is obtained through the road network and the vehicle position is updated. Therefore, it is necessary to set the starting moment of the current simulation period of each shadow vehicle, that is, the simulation timestamp t sim0 as the starting point of this period, and the time interval corresponding to the simulation period is [T start , T end , and the expression of the simulation step size is as follows:

[0072] T step = T end - T start

[0073] In the formula, T start represents the simulation start time; T end represents the simulation end time; Tstep Indicates the simulation step size, with a value range of 1 s to 10 s, and t sim0 = T start ;

[0074] When the physical timestamp t0 is not equal to T start , the position of the shadow vehicle at t sim0 is different from its position at t0, and the longitudinal position of the vehicle on the road section needs to be registered. The expression is as follows:

[0075] S new = S old - v(t0) * (t0 - t sim0 )

[0076] In the formula, S old and S new are the longitudinal positions of the vehicle on the road section at the physical timestamp t0 and the simulation timestamp t sim0 respectively;

[0077] After the registration is completed, the time coordinates (simulation timestamps) of all shadow vehicles are the starting points of a certain simulation period, and the position coordinates are the spatial coordinates corresponding to the time coordinates, that is, the corresponding longitudinal positions and the lanes where they are located (lateral positions) in the road section;

[0078] If the distance from the shadow vehicle to the stop line ahead is less than the length of the channelized section (60 - 100 m, divided according to the road grade), then according to the turning direction of the vehicle at the intersection ahead and the intersection channelization method (that is, the lane division method for different turning directions, such as 1 straight - right lane, 2 straight - through lanes, and 1 left - turn lane), determine its corresponding turning lane through the road network and assign the lane ID where it is located. When there are multiple lanes with the same turning direction, any one of them can be selected. In other cases, the shadow vehicle can be assigned any lane in the road section, and it does not need to be exactly the same as the lane where the actual vehicle is located.

[0079] S3. As Figure 3 shown, generate simulation shadow vehicles for the shadow vehicles with registered spatio - temporal coordinates through traffic simulation software and calculate the speeds of the simulation shadow vehicles;

[0080] The traffic simulation software includes: medium - scale or micro - scale traffic simulation software such as DynasTIM, SUMO, and Transmodeler;

[0081] Generating the simulation shadow vehicle means generating a simulation object in the simulation software with speed, simulation timestamp, longitudinal position, lane where it is located, downstream turning direction, and the length, width, and height of the shadow vehicle are all 0 m;

[0082] Load the generated shadow vehicle into the simulation road network. The loading method is: after the shadow vehicle passes through the registration in S2, input it into the road section in the road network of the traffic simulation software;

[0083] The calculated simulation vehicle speed is: The speed v of all simulation vehicles including shadow vehicles is calculated according to a mesoscopic (such as a speed-density relationship model, etc.) or a microscopic traffic model (such as a car-following model, etc.) i , where i = 1, 2, …, N, and N is the total number of vehicles in the simulation road network;

[0084] The present invention uses the DynasTIM traffic simulation system, and the formula for calculating the speed of simulation vehicles by the mesoscopic traffic model is as follows:

[0085]

[0086] In the formula, the subscript i represents the i-th simulation vehicle; the superscript t represents the t-th simulation period, and t ∈ T step ; represents the traffic density within the speed influence region (SIR i ) of vehicle i during the simulation period (t - 1); k jam represents the jam density, that is, the traffic density when all vehicles within the assumed SIR i are queued and driving at the minimum spacing, and its value range is [0.10, 0.15], and the unit is: standard vehicle / lane / meter; represents the simulation period (t - 1), and the number of vehicles (excluding vehicle i itself) within the SIR i of vehicle i; n represents the number of lanes included in the SIR i ; and respectively represent the average driving speeds of vehicle i during the simulation periods t and (t - 1); represents the length of the SIR i of vehicle i during the simulation period (t - 1); resp_time represents the average reaction time of the driver to the traffic conditions ahead, and is set to 2 s to 3 s; v f represents the free flow speed, and is set to the speed limit value of the road section; α and β represent the first parameter and the second parameter to be calibrated, and the default value is 1.0.

[0087] S4. Correct the simulation process by performing a delayed sampling operation on the vehicle driving data of the simulation shadow vehicle;

[0088] The simulation process includes: the vehicle speed of the simulation and the total driving state data of the traffic simulation model;

[0089] The delayed sampling is: delaying by one simulation step T stepAfter resampling, that is, receiving the operating state data of physical vehicles corresponding to all shadow vehicles in the simulation system through a communication network, including vehicle ID, the time of sending data, i.e., physical timestamp t1, longitude and latitude coordinates, current speed v(t1), turning direction at the downstream intersection, current mileage s(t1) (obtained by a navigation system with meter-level accuracy), and the ID (number) of the road section / lane where the vehicle is located; according to the actual driving mileage s(t1) and the corresponding physical timestamp t1 of shadow vehicle i, calculate the actual speed v of this shadow vehicle within the time period [t0, t1] actual =(s(t1)-s(t0)) / (t1-t0). When the simulated vehicle speed of shadow vehicle i calculated in S3 is v sim , if the relative error between it and the actual speed is greater than the critical value Δ, the simulated vehicle speed of the shadow vehicle should be corrected;

[0090] The expression for the relative error of the actual speed being greater than the critical value Δ is as follows:

[0091]

[0092] In the formula, the critical value Δ is 30%;

[0093] The correction method is to use the inertial filtering method for correction, and the expression is as follows:

[0094] v sim_correct =α1v sim +(1 - α1)v actual ;

[0095] delta_v = v sim_correct -v sim ;

[0096] In the formula, v sim_correct represents the simulated vehicle speed after correction; delta_v represents the difference between the simulated vehicle speed after correction and the simulated vehicle speed; α1 represents a real number in the interval [0, 0.5];

[0097] If the distance of the shadow vehicle from the stop line ahead is greater than the length of the channelized section, then on the road section where this shadow vehicle is located, select all the vehicles in the front and rear lanes within a distance of from the position of the shadow vehicle as the simulated vehicles whose speeds and displacements need to be corrected. Let the set of their ID numbers (identification numbers) be Nc, and at the same time add the ID number of the shadow vehicle to Nc. For the simulated vehicles, their corrected vehicle speeds are calculated according to the following formula:

[0098]

[0099] In the formula, Indicates that the distance of the shadow vehicle from the stop line ahead is greater than the simulated vehicle speed after the channelization section length correction; The simulated vehicle speed when the distance of the shadow vehicle from the stop line ahead is greater than the channelization section length; When there are M s shadow vehicles on the same road section, the simulated vehicle speed is calculated for each shadow vehicle, totaling M s times of correction;

[0100] If the distance of the shadow vehicle from the stop line ahead is greater than the simulated vehicle speed after the channelization section length correction and less than the queuing driving speed after the vehicle enters the queue That is Then update the traffic capacity of the lane where the shadow vehicle is located according to the following formula:

[0101]

[0102] Among them, c i is the traffic capacity of the lane where the shadow vehicle is located, that is, the maximum number of vehicles that can leave the lane per unit time; Is set to 3m / s;

[0103] In the actual road network, if the physical vehicle corresponding to the shadow vehicle is affected by a traffic event and enters the queuing state, then by calculating c i method, the traffic simulation system can automatically reflect the decrease in road traffic capacity caused by traffic events by adjusting its model parameter c i .

[0104] S5. Set the simulation period, obtain the simulation results, and realize the calibration of the real-time online traffic simulation system;

[0105] The DynasTIM traffic simulation system calculates the displacement s of all simulation vehicles within the current simulation period i = v sim(i) *T step , i = 1, 2, …, N, where v sim(i) is the simulated vehicle speed calculated according to the traffic model; And according to the displacement s i Move all simulation vehicles to new positions. When the shadow vehicle leaves the current road section, its life cycle ends and it is deleted from the simulation road network;

[0106] After completing S1~S5, receive the vehicle driving data of the next sampling period and repeat the steps of S1~S5 to continuously generate new shadow vehicles or update the relevant attributes such as the speed and position of the existing shadow vehicles in the simulation road network at the sampling period T, so as to realize the calibration of the real-time online traffic simulation system.

Claims

1. A real-time online traffic simulation system calibration method based on shadow vehicles, characterized in that: The following steps are involved: S1. Sample vehicle driving data and define shadow vehicles; The vehicle driving data includes: vehicle ID, the time when the data is sent, i.e., the physical timestamp t0, latitude and longitude coordinates, current speed, turning direction at the downstream intersection, current mileage s(t0), and the ID of the road section / lane; The shadow vehicle is defined to satisfy the following expression: v(t0) <v f v(t0)·T<D end Among them, v f is the free flow speed of the road section where the vehicle is located; D end is the distance between the vehicle and the intersection ahead; v(t0) is the current speed; T is the fixed sampling period; S2, performing time-space coordinate registration of the shadow vehicle; S3, generating a simulated shadow vehicle for the shadow vehicle after the space-time coordinates are aligned through traffic simulation software and calculating the speed of the simulated shadow vehicle; S4, correcting the simulation process by performing a delayed sampling operation on the vehicle driving data of the simulated shadow vehicle; S5. Set the simulation cycle, obtain the simulation results, and realize the calibration of the real-time online traffic simulation system.

2. The method for calibrating a real-time online traffic simulation system based on shadow vehicles according to claim 1, characterized in that: The expression for performing time-space coordinate registration on the shadow vehicle is as follows: S new =S old -v(t0)*(t0-t sim0 ) In the formula, S old and S new are the vehicle at physical timestamp t0 and simulation timestamp t sim0 The longitudinal position of the road section; Among them, t sim0 =T start , T start Indicates the simulation start time, which satisfies the following expression: T step =T end -T start Where, T start Indicates the simulation start time; T end Indicates the simulation end time; T step Represents the simulation step size.

3. The method for calibrating a real-time online traffic simulation system based on shadow vehicles according to claim 1, characterized in that: The expression for generating a simulated shadow vehicle and calculating the speed of the simulated shadow vehicle by using the traffic simulation software for the shadow vehicle after registering the time-space coordinates is as follows: In the formula, subscript i represents the i-th simulated vehicle; The superscript t represents the tth simulation period, t∈T step ; represents the speed influence area SIR of vehicle i during the simulation period (t-1) i Traffic density within; k jam represents the blocking density; Indicates the simulation period (t-1), in the influence area SIR of vehicle i i The number of vehicles in; n represents SIR i The number of lanes included; and They represent the average speed of vehicle i in simulation period t and (t-1), respectively; Indicates the influence area SIR of vehicle i in the simulation period (t-1) i The length of resp_time represents the average reaction time of the driver to the traffic conditions ahead; v f Indicates free flow speed; α and β indicate the first and second parameters that need to be calibrated, and the default value is 1.0; Traffic simulation software uses DynasTIM traffic simulation system.

4. The method for calibrating a real-time online traffic simulation system based on shadow vehicles according to claim 1, characterized in that: The simulation process is corrected by delaying the sampling operation on the vehicle driving data of the simulated shadow vehicle, and the simulation process includes: simulated vehicle speed and total driving state data of the traffic simulation model; Delayed sampling: delay one simulation step T step Then sample to obtain the vehicle ID, the time when the data is sent, that is, the physical timestamp t1, the longitude and latitude coordinates, the current speed v(t1), the turning direction at the downstream intersection, the current mileage s(t1), and the ID of the road section / lane; according to the actual mileage s(t1) of the shadow vehicle i and the corresponding physical timestamp t1, calculate the actual speed v of the shadow vehicle in the time period [t0, t1] actual =(s(t1)-s(t0)) / (t1-t0); When the simulated speed of shadow vehicle i calculated in S3 is v sim , if the relative error with the actual speed is greater than the critical value Δ, the simulated speed of the shadow vehicle should be corrected; The expression for the relative error of the actual speed being greater than the critical value Δ is as follows: In the formula, the critical value Δ is 30%; The correction method is to use the inertial filtering method, and the expression is as follows: v sim_correct =α1v sim +(1-α1)v actual ; delta_v=v sim_correct -v sim ; In the formula, v sim_correct represents the corrected simulated vehicle speed; delta_v represents the difference between the corrected simulated vehicle speed and the simulated vehicle speed; α1 is represented as a real number in the interval [0,0.5].

Images