A method and system for analyzing the impact of driver-following behavior and vehicle emissions

Abstract

This invention provides a method and system for analyzing the impact of driver following behavior on vehicle emissions, relating to the technical fields of traffic engineering and environmental science. The method includes: S1, extracting driver characteristic indicators from vehicle trajectory data of a following vehicle convoy, and visualizing the spatiotemporal trajectory of vehicles based on the vehicle trajectory data; S2, measuring the driver's heterogeneity parameters using the vehicle spatiotemporal trajectory map, and statistically determining the number of times the driver follows another vehicle using an improved Newell model combined with a dynamic time warping algorithm; S3, collecting driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters from the vehicle trajectory data of the following vehicle convoy, inputting them into a localized MOVES model, and estimating the emission factors of road traffic pollutants; S4, constructing a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and the emission factors of road traffic pollutants.

Description

Technical Field

[0001] This invention relates to the technical fields of traffic engineering and environmental science, specifically to a method and system for analyzing the impact of driver following behavior on vehicle emissions. Background Technology

[0002] Studies have shown that different driving styles significantly impact pollutant emissions: aggressive driving generates substantial additional carbon emissions compared to gentle, steady driving. Therefore, incorporating driving behavior factors into traffic emissions assessments is crucial for accurately reflecting actual emissions.

[0003] Currently, following other vehicles is a common driving behavior, especially in areas such as urban main roads, intersections, and near overpasses, where vehicles are forced to maintain close following distances and frequently accelerate and decelerate. However, existing technologies still have shortcomings in driving behavior modeling and traffic pollution assessment. Commonly used micro-level following models assume uniform driving responses and fail to adequately characterize the differences in driver following behavior in real-world traffic, resulting in vehicle trajectories that often do not accurately reflect different driving habits. Furthermore, vehicle types, road conditions, and driving habits vary across different areas, making it difficult for uniform models to accurately assess the impact of different driving behaviors on pollutant emissions.

[0004] Based on the above, it is clear that existing technologies lack a deep characterization of the relationship between real driving behavior characteristics and emissions, making it difficult to meet the needs of refined traffic emission management. Furthermore, existing technologies have not yet fully established a quantitative correlation between differences in driving behavior and traffic emissions, resulting in a lack of reliable basis for targeted emission reduction measures. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for analyzing the impact of driver following behavior on vehicle emissions. It proposes a driving behavior recognition method that integrates vehicle trajectory data, an improved Newell model, and a dynamic time warping algorithm. This method can accurately quantify the frequency of driver following behavior and driving stability. Furthermore, a semi-parametric generalized additive model is constructed based on this model, revealing the nonlinear correlation between various driving behavior variables and typical traffic pollutant emission factors. This invention not only significantly improves the precision of driving behavior modeling but also enhances the scientific rigor and operability of emission estimation.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following solution:

[0007] A method for analyzing the impact of driver following behavior on vehicle emissions, the method comprising the following steps:

[0008] S1. Obtain the following vehicle fleet by filtering vehicle trajectory data, extract driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualize the spatiotemporal trajectory of the vehicles based on the vehicle trajectory data.

[0009] S2. The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver's following behavior.

[0010] S3. Based on the vehicle trajectory data of the following vehicle fleet, driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters are collected and input into the localized MOVES model to estimate the emission factors of road traffic pollutants.

[0011] S4. Construct a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

[0012] A further preferred technical solution is that the driver characteristic indicators include maximum speed, minimum speed, average speed, speed standard deviation, maximum acceleration, minimum acceleration, average acceleration, acceleration standard deviation, average positive acceleration, average negative acceleration, positive acceleration standard deviation, and negative acceleration standard deviation.

[0013] A further preferred technical solution is as follows: In S1, the process of visualizing the spatiotemporal trajectory of a vehicle based on vehicle trajectory data is as follows:

[0014] The vehicle's travel time and corresponding spatial location are obtained from the vehicle trajectory data of the following convoy. The vehicle's travel time and spatial location are then plotted into a two-dimensional image to form a vehicle spatiotemporal trajectory map.

[0015] A further preferred technical solution is as follows: In S2, the process of measuring the driver's heterogeneity parameters using the vehicle's spatiotemporal trajectory map is as follows:

[0016] SA1. Measure the density and flow of the following vehicle fleet using tools, and then calculate the actual time delay parameters using the Newell following model and the absolute speed and density of drivers during actual following.

[0017] SA2. The ideal time hysteresis parameters are calculated using the Newell vehicle following model and preset constants.

[0018] SA3 is calculated based on the actual time hysteresis parameter and the ideal time hysteresis parameter. The calculation formula is: ,in, This is the actual time hysteresis parameter. The ideal time hysteresis parameter;

[0019] SA4, according to the formula Calculation yields quantitative indicators ;

[0020] SA5, Standardized Quantitative Indicators This means that the actual following time and following distance are reduced to 1 minute * 1 km.

[0021] A further preferred technical solution is as follows: In S2, the process of statistically obtaining the number of times the driver follows another vehicle by combining the improved Newell model with the dynamic time warping algorithm is as follows:

[0022] SB1. Obtain the time-series trajectory data of vehicles ahead and behind in the current road segment based on the vehicle trajectory data of the following convoy, and perform piecewise linear approximation processing on the time-series trajectory data to obtain the trajectory point set;

[0023] SB2. Construct a cost matrix based on the difference in vehicle speed or spatial location to quantify the matching cost between trajectory points, and add constraints when constructing the cost matrix;

[0024] SB3. Use the dynamic time warping algorithm to backtrack the path to obtain the optimal matching path and obtain the feature points of the subsequent vehicle's response to the preceding vehicle's behavior.

[0025] SB4. Count the total number of matching points based on the optimal matching path, and record it as the number of times the driver follows the vehicle.

[0026] A further preferred technical solution is as follows: In S3, the process of estimating the emission factors of road traffic pollutants is as follows:

[0027] Based on the vehicle trajectory data of the following vehicle fleet, the MOVES model is input with driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters.

[0028] The driving environment parameters include road length and road gradient; the vehicle technical condition parameters include vehicle age distribution and inspection and maintenance system; and the vehicle operating condition parameters include speed and acceleration.

[0029] The driving behavior is mapped to the driving mode defined in the MOVES model by using the operating condition classification method, and the emission factors of road traffic pollutants are output by calling the local emission factor library.

[0030] A further preferred technical solution is as follows: the driver characteristic index is an independent variable used to characterize the driver's following behavior and is related to speed and acceleration; step S4 includes the following steps:

[0031] S41. Divide the emission factor by the number of times you follow another vehicle to get the emissions generated in a single following action.

[0032] S42. Use variance expansion factor to examine standardized quantitative indicators. To detect collinearity with driver characteristic indicators, independent variables with VIF values ​​greater than preset values ​​in the inspection results will be deleted.

[0033] S43. Use the backward elimination method to screen the independent variables;

[0034] S44. Estimate the model parameters of the analytical model using the maximum likelihood estimation method;

[0035] S45. Evaluate the analysis module by calculating the degree of fit between the predicted values ​​of the analysis model and the actual observed values;

[0036] S46. Combining the descriptive statistical results and partial dependency plots of the analytical model, conduct a visual analysis and interpretation of the relationship between the independent variables and emission factors.

[0037] A further preferred technical solution is as follows: In S1, the process of obtaining the following vehicle convoy by filtering vehicle trajectory data is as follows:

[0038] SC1. Based on the vehicle trajectory data, a single vehicle is selected to obtain the first vehicle. The lateral position of the first vehicle in the lane remains unchanged during the observation period.

[0039] SC2. By comparing the longitudinal position of the first vehicle during the observation period, a vehicle sequence is obtained by tracking the vehicles one by one in a specific lane.

[0040] SC3. Determine the order of vehicles in the sequence, filter out the driving trajectories of vehicles in the sequence for a continuous preset time period, and record them as vehicle trajectory data.

[0041] A driver following behavior and vehicle emission impact analysis system, applying the aforementioned driver following behavior and vehicle emission impact analysis method, includes:

[0042] Data processing and analysis module: It obtains the following vehicle fleet by filtering vehicle trajectory data, extracts driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualizes the spatiotemporal trajectory of vehicles based on the vehicle trajectory data.

[0043] Driver following behavior extraction module: The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver following behavior.

[0044] Emission factor calculation module: Based on the vehicle trajectory data collected from the following vehicle fleet, driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters are obtained and input into the localized MOVES model to estimate the emission factors of road traffic pollutants.

[0045] Analysis Model Construction Module: Constructs a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

[0046] The beneficial effects of this invention are:

[0047] This invention provides a method and system for analyzing the impact of driver following behavior on vehicle emissions. To address the problem that existing technologies lack a deep characterization of the relationship between real-world driving behavior characteristics and emissions, making it difficult to meet the needs of refined traffic emission management, this invention proposes an improved driving behavior modeling and emissions assessment technology. It extracts driver characteristic parameters by combining real-world road trajectory data, and then introduces quantitative indicators of driver behavior deviating from the ideal following state into an improved Newell model. By combining it with the dynamic time warping algorithm, the frequency of following vehicles can be accurately identified, achieving a comprehensive characterization of the heterogeneity of driving behavior. Next, a localized MOVES model is used to accurately estimate vehicle emissions. Finally, a semi-parametric generalized additive model is constructed, enabling this invention to reveal the nonlinear correlation between various driving behavior variables and typical traffic pollutant emission factors.

[0048] Based on the above principles, compared with the prior art, the present invention has the following technical advantages and effects:

[0049] (1) Precisely quantify differences in driving behavior: proposed quantitative indicators It can not only reflect the deviation between instantaneous driving response and ideal behavior, but also take into account the cumulative effect over time, effectively capturing the driver's behavioral stability and deviation trend during traffic oscillations, thus improving the accuracy and completeness of driving style recognition;

[0050] (2) Enhance the adaptability and interpretability of the model: The constructed semi-parametric generalized additive model (GAM) comprehensively considers the nonlinear relationship between variables, takes into account the flexibility and interpretability of the model, and effectively reveals the nonlinear response patterns of different velocities and acceleration behaviors to various pollutant emission factors.

[0051] (3) Achieve micro-level emission prediction: By combining the localized MOVES model with natural vehicle trajectory data, it is possible to quantitatively estimate vehicle emission factors at high spatial and temporal resolution, providing a micro-level decision-making basis for ecological driving behavior assessment and traffic emission reduction strategy formulation.

[0052] (4) Improve the scientificity and practicality of traffic emission modeling: The method of this invention realizes a complete chain from original trajectory collection, vehicle following behavior recognition to nonlinear modeling and pollutant correlation analysis. It has good scalability and embeddability and is applicable to multiple application scenarios such as intelligent connected vehicle systems, urban traffic emission monitoring platforms, and vehicle-mounted green driving assistance systems.

[0053] In summary, this invention has achieved breakthroughs in vehicle following behavior recognition accuracy, quantitative modeling of driving behavior differences, and emission response curve fitting capabilities, significantly improving the adaptability of existing traffic emission analysis models to complex real-world driving behaviors and providing key technical support for realizing green and intelligent transportation systems. Attached Figure Description

[0054] Figure 1 This is a schematic diagram illustrating the overall concept of a method for analyzing the impact of driver following behavior on vehicle emissions in Embodiment 1 of the present invention.

[0055] Figure 2 This is a schematic diagram illustrating the specific process of obtaining the number of times a driver follows another vehicle in Embodiment 1 of the present invention. Detailed Implementation

[0056] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0057] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention.

[0058] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0059] Furthermore, for clarity and brevity, descriptions of well-known structures, functions, and configurations may have been omitted. Those skilled in the art will recognize that various changes and modifications can be made to the examples described herein without departing from the spirit and scope of this disclosure.

[0060] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and equipment should be considered part of the specification.

[0061] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0062] The present invention will now be described in detail with reference to the accompanying drawings and embodiments:

[0063] Example 1

[0064] Current technologies still have shortcomings in driving behavior modeling and traffic pollution assessment. On the one hand, commonly used micro-level following models assume uniform driving responses and fail to adequately depict the differences in driver following behavior in real-world traffic, resulting in vehicle trajectories that often do not accurately reflect different driving habits. For example, the existing Newell model, due to its failure to consider the heterogeneity of driver behavior, can lead to significant errors in emissions estimation. On the other hand, existing vehicle emissions estimation mainly relies on general emissions calculation models, which are built based on standardized operating conditions and average driving patterns. If these models are directly applied to actual traffic in specific regions without localized correction, their accuracy is often difficult to guarantee. Different regions have different vehicle types, road conditions, and driving habits, making it difficult for uniform models to accurately assess the impact of different driving behaviors on pollutant emissions.

[0065] As can be seen from the above, existing technologies lack a deep characterization of the relationship between real-world driving behavior and emissions, making it difficult to meet the needs of refined traffic emission management. Furthermore, existing technologies have not yet fully established a quantitative correlation between differences in driving behavior and traffic emissions, resulting in a lack of reliable basis for targeted emission reduction measures.

[0066] Therefore, this invention proposes a method for analyzing the impact of driver following behavior on vehicle emissions. It integrates vehicle trajectory data, an improved Newell model, and a dynamic time warping algorithm for driving behavior identification, accurately quantifying the frequency of driver following behavior and driving stability. Based on this, a semi-parametric generalized additive model is constructed to reveal the nonlinear correlation between various driving behavior variables and typical traffic pollutant emission factors. This invention not only significantly improves the precision of driving behavior modeling but also enhances the scientific rigor and operability of emission estimation. Specifically, this invention is applicable to real-time traffic emission monitoring systems, intelligent driving decision-making systems, and traffic low-carbon strategy evaluation platforms, and has promising application prospects.

[0067] A method for analyzing the impact of driver following behavior on vehicle emissions, the method comprising the following steps:

[0068] S1. Obtain the following vehicle fleet by filtering vehicle trajectory data, extract driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualize the spatiotemporal trajectory of the vehicles based on the vehicle trajectory data.

[0069] S2. The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver's following behavior.

[0070] S3. Based on the vehicle trajectory data of the following vehicle fleet, driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters are collected and input into the localized MOVES model to estimate the emission factors of road traffic pollutants.

[0071] S4. Construct a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

[0072] Based on the above principles, the overall concept of the driver following behavior and vehicle emission impact analysis method proposed in this invention mainly includes a data processing and analysis module, a driver following behavior extraction module, an emission factor calculation module, and an analysis model construction module. Steps S1-S4 can be implemented in these four modules.

[0073] like Figure 1 As shown, in the data processing and analysis module, vehicle trajectory data is filtered to obtain the following vehicle fleet. Based on the vehicle trajectory data of the following vehicle fleet, driver characteristic indicators are extracted. At the same time, the spatiotemporal trajectory of the vehicles is visualized based on the vehicle trajectory data, and driver characteristic indicators can be constructed.

[0074] like Figure 1 As shown, in the driver following behavior extraction module, an improved Newell model is proposed based on the classic Newell following model and driver heterogeneity theory in existing technologies. This improved Newell model can quantify the behavioral differences of drivers during actual following, thus obtaining driver heterogeneity. Then, driver heterogeneity parameters are measured using vehicle spatiotemporal trajectory maps to characterize driver heterogeneity. Simultaneously, the number of times a driver follows another vehicle is statistically obtained by combining the improved Newell model with a dynamic time warping algorithm. It can be seen that this invention measures driver behavior through driver heterogeneity and the number of times a driver follows another vehicle; that is, it quantifies the behavioral differences of drivers during actual following by using heterogeneity parameters and the number of times a driver follows another vehicle.

[0075] like Figure 1 As shown, in the emission factor calculation module, driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters are integrated to construct a localized MOVES model, namely a localized mobile source emission calculation model. The emission factors of road traffic pollutants can be accurately estimated through the localized mobile source emission calculation model.

[0076] like Figure 1 As shown, in the analysis model construction module, an analysis module is established to analyze the impact of driver behavior. In this invention, a semi-parametric generalized additive model is selected to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors, and suggestions for driving strategies can be made.

[0077] A further preferred technical solution is as follows: In S1, the vehicle trajectory data in this invention adopts natural driving trajectory data, enabling the invention to deeply characterize the relationship between real driving behavior characteristics and emissions, thus meeting the requirements of refined traffic emission management and providing a reliable basis for emission reduction measures. The specific content of the vehicle trajectory data is shown in Table 1.

[0078] Table 1. Details of Natural Driving Trajectory Data

[0079]

[0080] In order to filter out following convoys and avoid the combined effects of overtaking or other driving behaviors, this invention utilizes Python software to set filtering conditions to filter the original CSV format vehicle trajectory data. The specific filtering process is as follows:

[0081] SC1. Based on the vehicle trajectory data, a single vehicle is selected to obtain the first vehicle. The lateral position of the first vehicle in the lane remains unchanged during the observation period.

[0082] SC2. By comparing the longitudinal position of the first vehicle during the observation period, a vehicle sequence is obtained by tracking the vehicles one by one in a specific lane.

[0083] SC3. Determine the order of vehicles in the vehicle sequence, filter out the driving trajectories of vehicles in the sequence for a continuous preset time period, and record them as vehicle trajectory data. The continuous preset time period is at least 15 seconds.

[0084] Once the following vehicle fleet combinations that meet the above screening criteria are obtained, twelve speed and acceleration-related parameters, i.e., driver characteristic indicators, can be extracted according to the formula in Table 2 to characterize the driver's following behavior.

[0085] Table 2 Driver Characteristic Indicators

[0086]

[0087] A further preferred technical solution is as follows: In S1, the process of visualizing the spatiotemporal trajectory of a vehicle based on vehicle trajectory data is as follows: The travel time and corresponding spatial position of each vehicle are obtained from the vehicle trajectory data of the following convoy. The travel time and spatial position of each vehicle are then plotted into a two-dimensional image to form a vehicle spatiotemporal trajectory map. Specifically, MATLAB can be used to plot the travel time and spatial position of a vehicle into a two-dimensional image.

[0088] A further preferred technical solution is as follows: In S2, the process of measuring the driver's heterogeneity parameters using the vehicle's spatiotemporal trajectory map is as follows:

[0089] SA1. Measure the density and flow of the following vehicle fleet using tools, and then calculate the actual time delay parameters using the Newell following model and the absolute speed and density of drivers during actual following.

[0090] Specifically, the density and flow of following vehicles can be measured using the Trajectory Explorer tool; then, the formula in the Newell model can be used... Obtain a series of time delays during the actual driving period of the accompanying driver. That is, the actual time hysteresis parameter u is the absolute velocity of the traffic wave, and k is the density of the traffic flow;

[0091] SA2. The ideal time hysteresis parameters are calculated using the Newell vehicle following model and preset constants.

[0092] Specifically, based on the formula in the Newell model Assuming the wave speed u is constant during the ideal following period, and k is extracted using the Edie method, the ideal time hysteresis parameter can be obtained. ;

[0093] SA3 is calculated based on the actual time hysteresis parameter and the ideal time hysteresis parameter. The calculation formula is: ,in, This is the actual time hysteresis parameter. The ideal time hysteresis parameter;

[0094] SA4, according to the formula Calculation yields quantitative indicators ;

[0095] SA5, Standardized Quantitative Indicators This reduces the actual following time and following distance to 1 minute * 1 km; it can be seen that the proposed quantitative indicators... It not only reflects the deviation between instantaneous driving response and ideal behavior, but also takes into account the cumulative effect over time, which can effectively capture the driver's behavioral stability and deviation trend during traffic oscillations, thus improving the accuracy and completeness of driving style recognition.

[0096] A further preferred technical solution is as follows: In S2, the process of statistically obtaining the number of times the driver follows another vehicle by combining the improved Newell model with the dynamic time warping algorithm is as follows:

[0097] SB1. Obtain the time-series trajectory data of vehicles ahead and behind in the current road segment based on the vehicle trajectory data of the following convoy, and perform piecewise linear approximation processing on the time-series trajectory data to obtain the trajectory point set;

[0098] SB2. Construct a cost matrix based on the difference in vehicle speed or spatial location to quantify the matching cost between trajectory points, and add constraints when constructing the cost matrix;

[0099] SB3. Use the dynamic time warping algorithm to backtrack the path to obtain the optimal matching path and obtain the feature points of the subsequent vehicle's response to the preceding vehicle's behavior.

[0100] SB4. Count the total number of matching points based on the optimal matching path, and record it as the number of times the driver follows the vehicle.

[0101] Based on the above principles, the statistics on the number of times a passenger rides on another vehicle will be further elaborated as follows:

[0102] To quantify the emissions impact of each following action, it is necessary to calculate the number of following actions in the observation data. Therefore, this invention is based on the Newell following model, combining the improved Newell model with the dynamic time warping algorithm to statistically obtain the number of following actions by the driver. The specific process is as follows:

[0103] like Figure 2 As shown, firstly, time-series trajectory data of vehicles ahead and behind in the target road segment are acquired, i.e., the trajectory data of vehicles ahead and behind. The trajectory data is then processed by piecewise linear approximation to obtain the feature point set of the trajectory. Then, a cost matrix is ​​constructed. Specifically, in the process of constructing the cost matrix, the cost matrix is ​​calculated based on the speed data, and then the cumulative cost matrix is ​​calculated through time and space constraints. Finally, constraints are added, including boundary constraints, monotonicity constraints, and continuity constraints. Next, the dynamic time warping algorithm is used to backtrack the path to obtain the optimal matching path and find the feature points of the subsequent vehicle's response to the preceding vehicle's behavior to obtain the stimulus-response matching points of the vehicles ahead and behind. Furthermore, the stimulus-response matching points of the vehicles ahead and behind are standardized and converted into the number of following behavior responses per unit time * unit distance (1min * 1km). The purpose is to eliminate the influence of trajectory data with different following durations on the total number of matching points and to perform time standardization processing on the total number of matching points.

[0104] Specifically, the constraints used to quantify the matching cost between trajectory points include the following:

[0105] a. The reaction time should be greater than zero to avoid physical inconsistencies;

[0106] b. Maintain a following distance greater than zero to ensure driving safety;

[0107] c. The matching path must meet the requirements of continuity and monotonicity of the starting boundary and time sequence;

[0108] d. Introduce a non-linear penalty term to prevent non-physical matching results.

[0109] Specifically, the idea behind the dynamic time warping algorithm is as follows:

[0110] Step 1: Input two simplified time series data, namely vehicle trajectory data, which represent the speed or position of the two vehicles as a function of time.

[0111] Step 2: Initialize the cost matrix D as follows:

[0112] Initialize matrix D with dimension (len(sequence 1) + 1) x (len(sequence 2) + 1);

[0113] The third step is to set boundary conditions, typically with the cost of the starting point being 0, as follows:

[0114] D[0][0] = 0;

[0115] Step 4: Fill in the cost matrix, as follows:

[0116] For i from 1 to len (sequence 1):

[0117] For j from 1 to len (sequence 2):

[0118] Calculate the current cost C(i,j) = abs(sequence1[i]-sequence2[j])

[0119] #Calculate the cumulative cost and select the matching path with the minimum cost;

[0120] D[i][j]= C(i,j) + min(D[i-1][j], D[i][j-1], D[i-1][j-1]);

[0121] Step 5: Backtrack to find the optimal matching path, as follows:

[0122] optimal_path = empty list

[0123] i = len(sequence1)

[0124] j = len(sequence2)

[0125] #Tracing back from the finish line to the starting point

[0126] When i>0 and j>0:

[0127] Add (i, j) to optimal_path

[0128] # Backtracking Path: Choosing the direction with the minimum cumulative cost

[0129] If D[i-1][j]<= D[i][j-1] and D[i-1][j]<= D[i-1][j-1]:

[0130] i = i-1

[0131] Otherwise, if D[i][j-1]<= D[i-1][j] and D[i][j-1]<= D[i-1][j-1]:

[0132] j = j-1

[0133] otherwise:

[0134] i = i-1

[0135] j = j-1

[0136] Step 6: Calculate parameters such as driver reaction time and distance to other vehicles, as follows:

[0137] For each matching point (i, j) in the optimal_path:

[0138] Calculate the total number of matching points, sumD

[0139] Returns optimal_path, sumD

[0140] Step 7: Output the optimal matching path and calculate the total number of matching points based on the matching results.

[0141] A further preferred technical solution is as follows: In S3, the process of estimating the emission factors of road traffic pollutants is as follows:

[0142] Based on the vehicle trajectory data of the following vehicle fleet, the MOVES model is input with driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters.

[0143] The driving environment parameters include road length and road gradient; the vehicle technical condition parameters include vehicle age distribution and inspection and maintenance system; and the vehicle operating condition parameters include speed and acceleration.

[0144] The driving behavior is mapped to the driving mode defined in the MOVES model by using the operating condition classification method, and the emission factors of road traffic pollutants are output by calling the local emission factor library.

[0145] The MOVES model refers to the Motor Vehicle Emission Simulator model, which is localized to incorporate actual traffic environment characteristics to estimate the emission factors of road traffic pollutants. Specifically, to adapt to the actual conditions of the study area, the MOVES model requires localized input parameters, as shown in Table 3:

[0146] Table 3 Input Parameters

[0147]

[0148] In summary, based on the collected vehicle trajectory data, the MOVES model is input with driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters. The operating condition classification method is used to map the driver's following behavior to the operating mode defined in the MOVES model. By calling the localized emission factor library, the emission factors of road traffic pollutants are output, which can achieve accurate estimation of the emission impact of specific driving behaviors.

[0149] A further preferred technical solution is as follows: Since the driver characteristic index is an independent variable used to characterize the driver's following behavior and is related to speed and acceleration, step S4 includes the following steps:

[0150] S41. Divide the emission factor by the number of times you follow another vehicle to get the emissions generated in a single following action.

[0151] S42. Use variance expansion factor to examine quantitative indicators. To detect collinearity with driver characteristic indicators, independent variables with VIF values ​​greater than preset values ​​in the inspection results will be deleted.

[0152] S43. Use the backward elimination method to screen the independent variables;

[0153] S44. Estimate the model parameters of the analytical model using the maximum likelihood estimation method;

[0154] S45. Evaluate the analysis module by calculating the degree of fit between the predicted values ​​of the analysis model and the actual observed values;

[0155] S46. Combining the descriptive statistical results and partial dependency plots of the analytical model, conduct a visual analysis and interpretation of the relationship between the independent variables and emission factors.

[0156] Based on the above principles, the analysis of the impact of driver following behavior on vehicle emissions is further elaborated as follows:

[0157] Because most existing models of the relationship between driving behavior and traffic emissions rely on linear models, they are difficult to reveal the potential nonlinear effects between variables and cannot accurately capture the non-constant effects of driving behavior parameters (such as acceleration, speed fluctuations, etc.) on emission factors in different numerical ranges.

[0158] Therefore, this invention proposes to construct a semi-parametric generalized additive model that combines a generalized linear model with a smoothing function technique, which can effectively analyze the complex nonlinear relationship between driving behavior and various pollutant emission factors.

[0159] Specifically, the construction steps are as follows:

[0160] (1) The emissions generated per vehicle following behavior are obtained by dividing the pollutant factor by the number of times the vehicle was followed;

[0161] (2) Examine the independent variables (quantitative indicators) Collinearity among the other twelve independent variables related to velocity and acceleration was specifically examined using the variance inflation factor to check quantitative indicators. To prevent collinearity among driver characteristic indicators, independent variables with VIF values ​​greater than 10 were removed. This was done to ensure that there was no strong collinearity among the independent variables entering the analysis model.

[0162] (3) Variable selection: The backward elimination method is used to screen the independent variables, which can eliminate variables that contribute little to the model and retain variables with high significance to improve the interpretability and simplicity of the model;

[0163] (4) Model solution: The maximum likelihood estimation method is used to estimate the model parameters, which can ensure the stability and effectiveness of the fit;

[0164] (5) Model evaluation: The model's fit can be evaluated by calculating the degree of fit between the model's predicted values ​​and the actual observed values;

[0165] (6) Results analysis: By combining the model’s descriptive statistics and partial dependency graphs, the relationship between variables and target factors can be visualized and explained.

[0166] In summary, this invention provides a method and system for analyzing the impact of driver following behavior on vehicle emissions. To address the problem that existing technologies lack a deep characterization of the relationship between real-world driving behavior characteristics and emissions, making it difficult to meet the needs of refined traffic emission management, this invention proposes an improved driving behavior modeling and emissions assessment technology. This involves extracting driver characteristic parameters by combining real-world road trajectory data, and then introducing quantitative indicators of driver behavior deviating from the ideal following state into the improved Newell model. By combining it with the dynamic time warping algorithm, the frequency of following vehicles can be accurately identified, achieving a comprehensive characterization of the heterogeneity of driving behavior. Next, a localized MOVES model is used to accurately estimate vehicle emissions. Finally, a semi-parametric generalized additive model is constructed, enabling this invention to reveal the nonlinear correlation between various driving behavior variables and typical traffic pollutant emission factors.

[0167] Example 2

[0168] A driver following behavior and vehicle emission impact analysis system, applying the aforementioned driver following behavior and vehicle emission impact analysis method, includes:

[0169] Data processing and analysis module: It obtains the following vehicle fleet by filtering vehicle trajectory data, extracts driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualizes the spatiotemporal trajectory of vehicles based on the vehicle trajectory data.

[0170] Driver following behavior extraction module: The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver following behavior.

[0171] Emission factor calculation module: Based on the vehicle trajectory data collected from the following vehicle fleet, driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters are obtained and input into the localized MOVES model to estimate the emission factors of road traffic pollutants.

[0172] Analysis Model Construction Module: Constructs a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

[0173] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Based on the technical essence of the present invention, any simple modifications, equivalent substitutions, and improvements made to the above embodiments within the spirit and principles of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for analyzing the impact of driver following behavior on vehicle emissions, characterized in that, The method includes the following steps: S1. Obtain the following vehicle fleet by filtering vehicle trajectory data, extract driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualize the spatiotemporal trajectory of the vehicles based on the vehicle trajectory data. S2. The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver's following behavior. In S2, the process of obtaining the driver's heterogeneity parameters using the vehicle's spatiotemporal trajectory map is as follows: SA1. Measure the density and flow of the following vehicle fleet using tools, and then calculate the actual time delay parameters using the Newell following model and the absolute speed and density of drivers during actual following. SA2. The ideal time hysteresis parameters are calculated using the Newell vehicle following model and preset constants. SA3 is calculated based on the actual time hysteresis parameter and the ideal time hysteresis parameter. The calculation formula is: ,in, This is the actual time hysteresis parameter. The ideal time hysteresis parameter; SA4, according to the formula Calculation yields quantitative indicators ; in, This refers to the start time when a vehicle begins to perform corresponding driving actions due to a change in the state of the vehicle in front. This refers to the time during which a vehicle maintains a stable following position. SA5, Standardized Quantitative Indicators This means that the actual following time and following distance are reduced to 1 minute * 1 km; In S2, the process of statistically obtaining the number of times a driver follows another vehicle by combining the improved Newell model with the dynamic time warping algorithm is as follows: SB1. Obtain the time-series trajectory data of vehicles ahead and behind in the current road segment based on the vehicle trajectory data of the following convoy, and perform piecewise linear approximation processing on the time-series trajectory data to obtain the trajectory point set; SB2. Construct a cost matrix based on the difference in vehicle speed or spatial location to quantify the matching cost between trajectory points, and add constraints when constructing the cost matrix; SB3. Use the dynamic time warping algorithm to backtrack the path to obtain the optimal matching path and obtain the feature points of the subsequent vehicle's response to the preceding vehicle's behavior. SB4. Count the total number of matching points based on the optimal matching path, and record it as the number of times the driver follows the vehicle; S3. Based on the vehicle trajectory data of the following vehicle fleet, driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters are collected and input into the localized MOVES model to estimate the emission factors of road traffic pollutants. S4. Construct a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

2. The method for analyzing the impact of driver following behavior on vehicle emissions according to claim 1, characterized in that, The driver characteristic indicators include maximum speed, minimum speed, average speed, speed standard deviation, maximum acceleration, minimum acceleration, average acceleration, acceleration standard deviation, average positive acceleration, average negative acceleration, positive acceleration standard deviation, and negative acceleration standard deviation.

3. The method for analyzing the impact of driver following behavior on vehicle emissions according to claim 1, characterized in that, In S1, the process of visualizing the spatiotemporal trajectory of a vehicle based on vehicle trajectory data is as follows: The vehicle's travel time and corresponding spatial location are obtained from the vehicle trajectory data of the following convoy. The vehicle's travel time and spatial location are then plotted into a two-dimensional image to form a vehicle spatiotemporal trajectory map.

4. The method for analyzing the impact of driver following behavior on vehicle emissions according to claim 1, characterized in that, In S3, the process of estimating the emission factors of road traffic pollutants is as follows: Based on the vehicle trajectory data of the following vehicle fleet, the MOVES model is input with driving environment parameters, vehicle technical condition parameters and vehicle operating condition parameters. The driving environment parameters include road length and road gradient; the vehicle technical condition parameters include vehicle age distribution and inspection and maintenance system; and the vehicle operating condition parameters include speed and acceleration. The driving behavior is mapped to the driving mode defined in the MOVES model by using the operating condition classification method, and the emission factors of road traffic pollutants are output by calling the local emission factor library.

5. The method for analyzing the impact of driver following behavior on vehicle emissions according to claim 1, characterized in that, The driver characteristic index is an independent variable that describes the driver's following behavior and is related to speed and acceleration. S4 includes the following steps: S41. Divide the emission factor by the number of times you follow another vehicle to get the emissions generated in a single following action. S42. Use variance expansion factor to examine standardized quantitative indicators. To detect collinearity with driver characteristic indicators, independent variables with VIF values ​​greater than preset values ​​in the inspection results will be deleted. S43. Use the backward elimination method to screen the independent variables; S44. Estimate the model parameters of the analytical model using the maximum likelihood estimation method; S45. Evaluate the analysis module by calculating the degree of fit between the predicted values ​​of the analysis model and the actual observed values; S46. Combining the descriptive statistical results and partial dependency plots of the analytical model, conduct a visual analysis and interpretation of the relationship between the independent variables and emission factors.

6. The method for analyzing the impact of driver following behavior on vehicle emissions according to claim 1, characterized in that, In S1, the process of obtaining the following vehicle convoy by filtering vehicle trajectory data is as follows: SC1. Based on the vehicle trajectory data, a single vehicle is selected to obtain the first vehicle. The lateral position of the first vehicle in the lane remains unchanged during the observation period. SC2. By comparing the longitudinal position of the first vehicle during the observation period, a vehicle sequence is obtained by tracking the vehicles one by one in the lane. SC3. Determine the order of vehicles in the sequence, filter out the driving trajectories of vehicles in the sequence for a continuous preset time period, and record them as vehicle trajectory data.

7. A system for analyzing the impact of driver following behavior on vehicle emissions, characterized in that, The method for analyzing the impact of driver following behavior on vehicle emissions as described in any one of claims 1-6 includes: Data processing and analysis module: It obtains the following vehicle fleet by filtering vehicle trajectory data, extracts driver characteristic indicators based on the vehicle trajectory data of the following vehicle fleet, and visualizes the spatiotemporal trajectory of vehicles based on the vehicle trajectory data. Driver following behavior extraction module: The heterogeneity parameters of the driver are obtained by measuring the vehicle spatiotemporal trajectory map. The heterogeneity parameters are introduced into the improved Newell model. The number of times the driver follows another vehicle is statistically obtained by combining the improved Newell model with the dynamic time warping algorithm, so as to characterize the heterogeneity of the driver following behavior. Emission factor calculation module: Based on the vehicle trajectory data collected from the following vehicle fleet, driving environment parameters, vehicle technical condition parameters, and vehicle operating condition parameters are obtained and input into the localized MOVES model to estimate the emission factors of road traffic pollutants. Analysis Model Construction Module: Constructs a semi-parametric generalized additive model to analyze the nonlinear correlation between driver following behavior and road traffic pollutant emission factors.

Images