Urban intersection carbon emission evaluation method based on traffic flow parameters

Abstract

This invention discloses a method for assessing carbon emissions at urban intersections based on traffic flow parameters, including the following steps: Step S1: Constructing a typical signalized intersection scenario and setting up a traffic demand model covering different traffic flow states and new energy penetration rates; Step S2: Obtaining traffic flow parameters and carbon emission data through simulation; Step S3: Identifying key influencing factors using correlation analysis; Step S4: Decomposing single-vehicle carbon emissions into zero-flow carbon emissions and delay carbon emissions, establishing a model of the relationship between delay carbon emissions and traffic flow parameters, and an assessment model of total carbon emissions at the intersection; Step S5: Verifying the effectiveness of the model through practical intersection applications. This invention discloses a method for assessing carbon emissions at urban intersections based on traffic flow parameters, selecting the number of stops and average vehicle delay as core parameters to construct a direct quantitative relationship model between traffic flow parameters and carbon emissions, achieving high-precision, refined, and low-cost assessment of carbon emissions at urban intersections.

Description

Technical Field

[0001] This invention belongs to the field of urban traffic low-carbon management technology, specifically relating to a method for assessing carbon emissions at urban intersections based on traffic flow parameters. Background Technology

[0002] Under the dual pressures of addressing climate change and ensuring sustainable urban development, dual carbon targets place profound demands on the low-carbon transformation of urban transportation systems. As a major source of urban energy consumption and carbon emissions, the effectiveness of emissions reduction in the transportation sector directly impacts the achievement of overall goals. Road intersections are bottlenecks in road network capacity and high-incidence areas of frequent vehicle starts and stops, with emission intensity per unit time far exceeding that of continuous flow road sections, making them hotspots for urban carbon footprint. Achieving precise sensing and control of carbon emissions at the intersection scale is a key breakthrough in promoting the evolution of urban transportation low-carbon management towards refinement and intelligence.

[0003] Existing methods for calculating carbon emissions from transportation mainly include the direct acquisition method, which relies on authoritative databases to extract data and is suitable for macro-level analysis; the top-down method, which calculates the total amount based on energy consumption and emission factors and depends on statistical standards; and the bottom-up method, which has relatively high accuracy but requires a large amount of localized vehicle operation data.

[0004] The above methods each have their own emphasis on evaluation scales, but they still have the following key drawbacks: First, the use of static and averaged data at the macro level makes it impossible to distinguish the periodic changes in carbon emissions over time (such as peak hours, off-peak hours, weekdays and holidays, and seasonal differences), resulting in a partial distortion in the portrayal of the dynamic characteristics of carbon emissions.

[0005] Second, the dynamic variable relationship between carbon emissions and traffic flow operation status has not been established. Core traffic flow parameters such as the number of stops and delays have not been included in the variable system, resulting in a "black box" state for the impact mechanism of congestion on carbon emissions. It is impossible to truly reflect the dynamic impact of traffic organization efficiency such as signal coordination, lane management, and left-turn restrictions on carbon emissions, making it difficult to provide direct carbon emission impact references for specific traffic management decisions.

[0006] Third, the spatiotemporal resolution is insufficient, making it difficult to capture the micro-spatiotemporal characteristics of road networks and intersections, and thus difficult to support a refined assessment of carbon emissions from individual intersections or regional road networks.

[0007] Fourth, there is a prominent contradiction between the universality and localization of models, and there is a lack of a unified and easily promoted micro-emission assessment framework. Micro-emission models, represented by COPERT, MOBILE, and MOVES, often require a large amount of data on traffic flow, vehicle composition, and infrastructure geometry. They have complex parameters, are difficult to localize, and rely on time-consuming and labor-intensive simulation modeling, making them difficult to widely apply in urban traffic management.

[0008] Therefore, there is an urgent need for a high-precision and easily applicable method for assessing carbon emissions at urban intersections, with traffic flow parameters as the core. Summary of the Invention

[0009] The main objective of this invention is to provide a method for assessing carbon emissions at urban intersections based on traffic flow parameters. The method selects the number of stops and average vehicle delay as core parameters, constructs a direct quantitative relationship model between traffic flow parameters and carbon emissions, and achieves high-precision, refined, and low-cost assessment of carbon emissions at urban intersections.

[0010] To achieve the above objectives, this invention provides a method for assessing carbon emissions at urban intersections based on traffic flow parameters, comprising the following steps: Step S1: Construct a typical signal-controlled intersection scenario, design channelization schemes and signal timing, and set up a traffic demand model that covers different traffic flow states and new energy penetration rates; Step S2: Obtain traffic flow parameters and carbon emission data through simulation, and verify the simulation flow rate and travel time ratio to ensure coverage of the entire traffic operation status; Step S3: Use correlation analysis to identify key influencing factors and determine average vehicle delay and average number of stops as core parameters affecting carbon emissions; Step S4: Decompose the carbon emissions of a single vehicle into zero-flow carbon emissions and delay carbon emissions, construct a calculation method based on the cumulative emissions of a single vehicle, establish a carbon emission assessment model for delay carbon emissions, average vehicle delay, and average number of stops, and calculate the total carbon emissions of the intersection based on the cumulative carbon emissions of a single vehicle. Step S5: Verify the effectiveness of the model through actual intersection applications.

[0011] As a further preferred technical solution to the above technical solution, for step S1, design signal-controlled intersections of arterial roads, determine the intersection range, channelization scheme and signal timing parameters; set traffic demand gradients covering all states from smooth to congested and different new energy penetration rates, divide low-flow groups and regular groups, fix the flow of trucks and buses, and form multiple traffic demand combinations.

[0012] As a further preferred technical solution to the above technical solution, for step S2, the intersection scenario constructed by running the TessNG simulation software is used to output traffic flow parameters and carbon emission indicators; through simulation flow and travel time ratio verification, it is ensured that the simulation covers five traffic operation states: smooth, basically smooth, light congestion, moderate congestion, and severe congestion.

[0013] As a further preferred technical solution to the above technical solution, the specific implementation of establishing the carbon emission assessment model in step S4 is as follows: Step S4.1: Construct a framework for calculating total carbon emissions. Establish a method for calculating total carbon emissions at intersections based on the cumulative emissions of individual vehicles. Total emissions are the sum of the actual carbon emissions of all carbon-emitting vehicles, categorized by vehicle type and fuel type. Step S4.2: Deconstruct the carbon emission structure of a single vehicle, breaking down the actual carbon emissions of a single vehicle into: Zero-flow carbon emissions: Benchmark emissions for vehicles passing through intersections without delay or stopping; Delayed carbon emissions: Additional emissions caused by congestion, start-stop, acceleration, and deceleration; Step S4.3: Zero-flow carbon emission modeling; Step S4.4: Modeling carbon emissions due to delays. Using average vehicle delay and average number of stops as core input variables, construct a relational model to obtain a carbon emission calculation model for delays. Step S4.5: Synthesis of actual carbon emissions per vehicle, Actual carbon emissions per vehicle = Zero-flow carbon emissions + Delayed carbon emissions; Step S4.6: Summarize the total carbon emissions of the intersection by weighting and summing them according to vehicle type, fuel type, and traffic volume to obtain the total carbon emissions of the intersection.

[0014] As a further preferred technical solution to the above technical solution, for step S5, the effectiveness of the model is verified by comparing the evaluation results of the carbon emission assessment model with the carbon emission results of the TessNG simulation model.

[0015] The beneficial effects of this invention are as follows: 1. Theoretical Innovation: By taking the number of stops and delays as core variables, a direct quantitative relationship between traffic flow status and carbon emissions is established, filling the theoretical gap between macro and micro carbon emission research.

[0016] 2. Improved accuracy: The model takes into account the traffic flow operation status to achieve high-precision assessment of carbon emissions at intersections. It can quantify the changes in carbon emissions before and after the implementation of different traffic control measures (such as signal timing optimization, lane management, left turn restriction, etc.) and dynamically distinguish the time-based emission differences such as peak / off-peak and weekday / holiday.

[0017] 3. Practical and convenient: The method can rely on existing traffic congestion assessment systems (such as urban traffic congestion index platforms, Gaode Maps, etc.), making it feasible to make carbon emission prediction more refined. It also eliminates the need for complex simulation models, greatly reduces the data and calculation threshold, and is suitable for promotion and application in domestic cities.

[0018] 4. Engineering application value: It can directly support decision-making in signal timing optimization, congestion management, and low-carbon planning, providing traffic management departments with actionable emission reduction tools to achieve precise traffic emission reduction policies. Attached Figure Description

[0019] Figure 1This is a flowchart illustrating the present invention.

[0020] Figure 2 This is a schematic diagram of intersection channelization.

[0021] Figure 3 This is a comparison chart of simulated running flow and input flow.

[0022] Figure 4 This is a verification chart of the travel time ratio of the traffic simulation model.

[0023] Figure 5 Analysis of factors affecting carbon emissions at intersections.

[0024] Figure 6 This is a graph showing the relationship between carbon emissions per passenger vehicle delay and the number of stops and average delay per vehicle.

[0025] Figure 7 This is a graph showing the relationship between carbon emissions from single-vehicle delays and the number of stops and vehicle delays.

[0026] Figure 8 This is a basic information diagram of the intersection.

[0027] Figure 9 This is a flowchart of the pre-selected intersection carbon emission assessment process.

[0028] Figure 10 This is a graph showing the actual carbon emissions of passenger cars turning at intersections. Detailed Implementation

[0029] The following description is intended to disclose the present invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art. The basic principles of the invention defined in the following description can be applied to other embodiments, modifications, improvements, equivalents, and other technical solutions that do not depart from the spirit and scope of the invention.

[0030] In the preferred embodiments of the present invention, those skilled in the art should note that TessNG, Spearman, etc., involved in the present invention can be regarded as prior art.

[0031] Preferred embodiment.

[0032] like Figures 1 to 10 As shown, this invention discloses a method for assessing carbon emissions at urban intersections based on traffic flow parameters, characterized by comprising the following steps: Step S1: Construct a typical signal-controlled intersection scenario, design channelization schemes and signal timing, and set up a traffic demand model that covers different traffic flow states and new energy penetration rates; For step S1, design signal-controlled intersections where arterial roads intersect, determine the intersection range, channelization scheme and signal timing parameters; set traffic demand gradients covering all states from smooth to congested and different new energy penetration rates, divide into low-flow groups and regular groups, fix the flow of trucks and buses, and form multiple traffic demand combinations.

[0033] A typical signalized intersection of arterial roads was designed for the experiment, with an intersection area of ​​0.5km × 0.5km. The carbon emissions studied are those emitted by vehicles traveling 0.5km. Specific channelization design is as follows: Figure 2 The signal timing is shown in Table 1.

[0034] Table 1. Intersection Signal Timing

[0035] Note: The yellow flash time between phases is 3 seconds, and the all-red time is 2 seconds.

[0036] The low-flow group starts with no signal control, and the green light time increases second by second until the flow rate reaches 3500 pcu / h.

[0037] A traffic demand model covering various traffic flow states and different proportions of new energy vehicles was established. Calculations showed that the intersection's capacity ranged from 7838 pcu / h (off-peak hours) to 8404 pcu / h (peak hours). The total traffic demand modeling range for this intersection was set to two intervals: 200 pcu / h to 3500 pcu / h (low-flow group) and 3500 pcu / h to 8600 pcu / h (regular group), with each 300 pcu / h increment. The low-flow group focused on the relationship between traffic flow parameters and carbon emissions under zero-flow and unobstructed conditions, without considering the inclusion of new energy vehicles. For the regular group, under each total demand level, the proportion of new energy vehicles started from 0 and increased in increments of 300 pcu / h, with a maximum proportion below 70%, as detailed in Table 2. Regarding traffic composition, the regular group included four main vehicle types: gasoline-powered passenger cars, new energy vehicles, medium-sized trucks, and medium-sized buses (the low-flow group included gasoline-powered passenger cars, medium-sized trucks, and medium-sized buses). Trucks and buses account for a relatively small and fixed proportion of traffic on city center roads, so the truck flow rate was fixed at 80 pcu / h and the bus flow rate at 35 pcu / h across all traffic demand levels. A total of 310 traffic demand profiles were analyzed in the experiment (45 low-flow groups and 265 regular groups).

[0038] Table 2 Total Traffic Flow of Passenger Cars, Trucks, and Buses at the Intersection Table 2 Total Traffic Flow of Passenger Cars, Trucks, and Buses at the Intersection

[0039] Step S2: Obtain traffic flow parameters (flow rate, delay, parking rate, etc.) and carbon emission data (total and per-vehicle carbon emissions) through simulation, and verify the simulation flow rate and travel time ratio to ensure coverage of the entire traffic operation status; For step S2, the intersection scenario constructed using TessNG simulation software is run to output traffic flow parameters and carbon emission indicators; the simulation coverage of five traffic operation states—smooth, basically smooth, lightly congested, moderately congested, and severely congested—is verified through simulated traffic flow and travel time ratio.

[0040] To ensure that the simulated traffic flow matches the input traffic demand and covers traffic conditions from smooth to congested, two metrics are selected for verification: simulated traffic flow and travel time ratio (the ratio of actual travel time to free-flow travel time). The low-flow group has smooth traffic and simple data, so it is not shown. Verification of the regular flow group is as follows: Figure 3 .

[0041] 1. Process Validation: From Figure 3 As can be seen, the simulated traffic flow increases synchronously with the increase in input traffic at the intersection. Later, as the intersection approaches saturation, it eventually stabilizes at around 7455 pcu / h. Before the intersection reaches saturation, the error rate between the simulated hourly traffic flow and the input traffic flow is approximately 5% to 8%, which aligns with the traffic demand gradient design in the experiment.

[0042] 2. Travel Time Ratio Validation: To evaluate the differences in carbon emissions under different traffic flow conditions, the simulation was further validated to cover various traffic flow conditions. Urban road traffic operation conditions are generally divided into five categories: "Smooth: 1 ≤ Travel Time Ratio < 1.3; Basically Smooth: 1.3 ≤ Travel Time Ratio < 1.6; Lightly Congested: 1.6 ≤ Travel Time Ratio < 1.9; Moderately Congested: 1.9 ≤ Travel Time Ratio < 2.2; Severely Congested: Travel Time Ratio ≥ 2.2". Travel time data from 310 simulation models were collected, and the "Travel Time Ratio" index was calculated. For the low-flow group, 1 ≤ Travel Time Ratio < 1.3, and for the regular group, ... Figure 4 As shown in the figure, the traffic simulation model comprehensively covers five traffic operation states, allowing for further analysis of the underlying mechanisms between traffic operation state indicators and CO2 emissions.

[0043] Step S3: Use Spearman correlation analysis to identify key influencing factors and determine average vehicle delay and average number of stops as the core parameters affecting carbon emissions; Carbon emissions at intersections are the result of multiple factors. The primary factor depends on the vehicles themselves, including their type (e.g., gasoline / electric), technological level, and maintenance condition, which determines the baseline emissions per vehicle. Secondly, the core influencing factor stems from the operational state of traffic flow, comprehensively reflected in indicators such as delays, stops, queues, and travel times, and is heavily constrained by intersection geometry and signal control strategies. Frequent starts, stops, idling, acceleration, and deceleration of vehicles following traffic flow significantly increase carbon emissions. Finally, differences in driving behavior and interference from pedestrians and non-motorized vehicles also introduce more uncertainty into intersection carbon emissions, which can be reflected in traffic flow conditions.

[0044] Based on the quantified values ​​of each parameter output from the simulation, the interaction relationships between them are analyzed to identify key indicators affecting carbon emissions at intersections. Considering that the influencing factors may not satisfy the assumptions of linearity and normality (e.g., greater delays lead to more efficient carbon emission growth), this study chooses Spearman correlation analysis, a method robust to outliers, to analyze the monotonic relationships between variables. This method effectively measures the strength and direction of the monotonic relationship between two variables. The analysis results are as follows: Figure 5 .

[0045] The figure above shows that intersection carbon emissions are influenced by multiple coupled factors. First, traffic flow and individual vehicle emission efficiency are the core driving factors, with the flow of fuel-powered passenger cars having the most significant impact on total emissions (correlation coefficient 0.924), and its individual vehicle carbon emissions also showing a strong correlation (0.872), indicating that both factors jointly determine the intensity of intersection carbon emissions. Second, traffic operation status (reflected in average travel time, travel time ratio, average vehicle delay, average queue length, average number of vehicles in the queue, and average number of stops) is strongly correlated with individual vehicle carbon emissions (approximately 0.85) and strongly positively correlated with total CO2 (approximately 0.74), confirming that congestion significantly exacerbates carbon emissions through inefficient operating conditions such as idling and stop-and-go traffic. Furthermore, traffic operation status indicators are highly correlated with each other (>0.99). To avoid parameter multicollinearity leading to model unreliability, "average vehicle delay" and "average number of stops" are selected as key influencing parameters for subsequent modeling studies of individual vehicle carbon emissions. Because average vehicle delay reflects traffic efficiency loss and can represent the state of traffic operation, and the average number of stops is directly related to the start-stop process of fuel consumption, it is more interpretable. Third, the popularization of new energy vehicles has an emission reduction effect, and its proportion is negatively correlated with total CO2 (-0.616), but the emission reduction effect is limited by the absolute number of fuel vehicles. When the proportion of new energy vehicles increases, the remaining fuel vehicles may face a more complex traffic environment, leading to a slight increase in their per-vehicle emission efficiency. In addition, the correlation between truck traffic flow and total CO2 emissions is -0.267 (weak negative correlation), while the correlation between per-truck carbon emissions and total CO2 emissions is 0.434 (moderate positive correlation). Because the proportion of trucks on urban center roads is relatively low, the truck traffic flow value was fixed in the traffic demand model of this study, making the impact of trucks on total emissions less than that of fuel passenger cars. However, the per-truck emissions of trucks are relatively high, and their efficiency improvement still needs to be considered.

[0046] 3. Analysis of the Relationship between Individual Vehicle Carbon Emissions and Traffic Flow Parameters: The carbon emissions generated by a single motor vehicle passing through an intersection can be broken down into two parts: carbon emissions generated under zero-delay conditions, defined as zero-flow carbon emissions, which are directly related to the vehicle's energy type and the intersection's travel distance; and carbon emissions increased under delayed conditions, defined as delayed carbon emissions. This part of the carbon emissions is generated more than under smooth traffic conditions due to increased vehicle travel time and frequent starts and stops, and is directly related to the traffic operation status. Analyzing the results data obtained from the simulation experiment, the relationship between actual carbon emissions, zero-flow carbon emissions, and delayed carbon emissions can be obtained. Taking fuel-powered passenger cars and trucks as examples, the data on the three carbon emission relationships are shown in Table 3.

[0047] Table 3. Carbon Emission Contribution Ratio of Passenger Vehicles

[0048] As shown in the table above, for passenger cars, zero-flow carbon emissions and delay carbon emissions are almost equally distributed under smooth traffic conditions. However, as congestion worsens, the proportion of delay carbon emissions gradually increases, reaching 68.8% under severe congestion, an increase of 20.7 percentage points from smooth to severe congestion. For trucks, delay carbon emissions consistently dominate (78%~86%), and this proportion continues to rise with increasing congestion, increasing by 8 percentage points from smooth to severe congestion. This indicates that truck carbon emissions at intersections are mainly determined by delays. For both trucks and passenger cars, zero-flow carbon emissions are a fixed value (unchanging with traffic conditions), therefore the actual increase in emissions is entirely caused by delay emissions. Reducing delays is the core of emission reduction.

[0049] Further analysis of the relationship between carbon emissions per unit of fuel-powered passenger cars and trucks and their respective traffic flow parameters is shown in Table 4. Figure 6 , Figure 7 .

[0050] Table 4. Indicators and carbon emissions per vehicle for passenger cars and freight vehicles under different traffic flow conditions.

[0051] Data analysis shows that traffic congestion has a significant aggravating effect on vehicle carbon emissions, and this effect exhibits both commonalities and heterogeneities in passenger cars and trucks. As traffic flow deteriorates from "smooth flow" to "severe congestion," both types of vehicles show a positive correlation between average number of stops, average vehicle delay, and carbon emissions. Furthermore, the average number of stops, average vehicle delay, carbon emissions due to delay, and average actual carbon emissions for both types of vehicles all show a monotonically increasing trend. The average number of stops for passenger cars increased from 0.73 to 2.28, the average vehicle delay increased from 42.25 seconds to 88.25 seconds, and the average actual carbon emissions increased from 204.54 g / vehicle to 322.19 g / vehicle. The increase in actual carbon emissions for trucks is even more significant, increasing by 1243.90 g / vehicle during severe congestion compared to smooth flow, approximately 10.6 times the increase for passenger cars (117.65 g / vehicle), indicating that truck emissions are more sensitive to delays. Meanwhile, increased congestion leads to greater dispersion in various indicators and increased inter-individual variability. Under smooth traffic conditions, indicator values ​​are relatively concentrated (e.g., the average delay per vehicle range is approximately 0.24 seconds), while under severe congestion, the average delay per vehicle range exceeds 13 seconds (13.69 seconds for passenger cars, an increase of 56.8 times). The actual carbon emission range also widens accordingly: passenger cars approach 83g / vehicle, while trucks reach as high as 968g / vehicle, with the delay carbon emission range for trucks increasing from 327.81g / vehicle under smooth traffic conditions to 992.68g / vehicle under moderate congestion (an increase of 3.0 times). Furthermore, it is noteworthy that the carbon emission growth for both types of vehicles exhibits a non-linear characteristic: the rate of change in delayed carbon emissions and actual carbon emissions is relatively small before mild congestion, but accelerates after moderate congestion.

[0052] Step S4: Decompose single-vehicle carbon emissions into zero-flow carbon emissions and delay carbon emissions, construct a calculation method based on the cumulative emissions of single vehicles, and establish a carbon emission assessment model for delay carbon emissions, average vehicle delay, and average number of stops. Calculate the total carbon emissions of the intersection based on the cumulative carbon emissions of single vehicles. The specific implementation of establishing the carbon emission assessment model in Step S4 is as follows: Step S4.1: Constructing the framework for calculating total carbon emissions. Establish a method for calculating total carbon emissions at intersections based on the cumulative emissions of individual vehicles. Total emissions are the sum of the actual carbon emissions of all carbon-emitting vehicles, categorized by vehicle type and fuel type. Total carbon emissions are obtained by summing the actual carbon emissions of all carbon-emitting motor vehicles (including gasoline, diesel, natural gas, and liquefied petroleum gas) passing through the intersection. ; in, Based on vehicle type, the study categorizes vehicles into 7 types, taking into account their energy type. ); Based on fuel type, it is divided into 6 major categories according to different energy supply types ( The four types of fuels are gasoline, diesel, pure electric, hybrid, compressed natural gas (CNG), and liquefied natural gas (LNG). For use of fuel The The number of times the vehicle operates. For use of fuel The Actual carbon emissions per vehicle.

[0053] Step S4.2: Deconstruct the carbon emission structure of a single vehicle, breaking down the actual carbon emissions of a single vehicle into: Zero-flow carbon emissions: Benchmark emissions for vehicles passing through intersections without delay or stopping; Delayed carbon emissions: Additional emissions caused by congestion, start-stop, acceleration, and deceleration; It can be broken down into two parts: zero-flow carbon emissions and delayed carbon emissions. ; This refers to the carbon emissions of a vehicle passing through an intersection without stopping under free-flowing conditions. It is directly related to the distance traveled through the intersection and the vehicle's fuel type. Based on fuel type, an emission factor method is used for calibration, expressed as: ; Carbon emission conversion efficiency of fuel energy, measured in grams per unit of fuel. Gasoline: approximately 2.26-2.33 kg CO2 / L; Diesel: approximately 2.66-2.73 kg CO2 / L; CNG: approximately 2.16-2.20 kg CO2 / m³. 3LNG contains approximately 2.75-2.80 kg CO2 / kg. Electric vehicles do not produce carbon emissions during operation, and carbon emissions from the power generation process are not considered here.

[0054] Average travel distance through the intersection, in km.

[0055] :fuel The The overall energy consumption of this type of vehicle is expressed in energy units per 100km. The energy consumption of each vehicle model is shown in Table 5. Table 5. Vehicle Classification and Energy Consumption per 100km

[0056] Step S4.3: Zero-flow carbon emission modeling, i.e. ; Step S4.4: Modeling carbon emissions due to delays. Using average vehicle delay and average number of stops as core input variables, a relational model is constructed to obtain a carbon emission calculation model for delays. Vehicle delay carbon emissions are caused by factors such as traffic control and traffic congestion, with the direct cause being frequent starts, stops, accelerations, and decelerations. This is comprehensively reflected in the Traffic Flow State (TFS). A parametric model is established for carbon emissions due to delays in relation to delay duration and number of stops, expressed as follows: ; Average traffic flow delay time (s); Average number of stops per traffic flow; A model showing the relationship between incremental carbon emissions from single-vehicle delays and their delay time and number of stops; The relationship between carbon emissions from vehicle delays and the number of stops and average vehicle delay often exhibits nonlinearity, heterogeneity, and variable interactions. Furthermore, carbon emissions from delays are continuous variables, making it a regression task. Therefore, the following methods are considered for modeling carbon emissions from delays: Neural networks can fit complex relationships but require large samples and have poor interpretability, making them unsuitable for the medium sample size and interpretability requirements of this study; Random forests and gradient boosting trees, as ensemble learning algorithms, can effectively fit nonlinearities and output feature importance, meeting interpretability requirements; Support vector machines (SVMs) utilize kernel tricks to handle nonlinearity and have strong generalization ability with small samples, making them suitable for this study; Linear regression models are simple and efficient, and are also considered. In addition, to address the heterogeneity under different traffic conditions, K-means clustering is introduced to first identify the inherent structure of the data, and then the optimal regression model (linear regression, tree model, SVR) is selected for each subset to further improve prediction accuracy. Based on the above considerations, five methods—linear regression (LR), random forest (RF), gradient boosting tree (GBR), support vector machine (SVR), and K-means + classification prediction—are selected for modeling comparison.

[0057] (1) Data preprocessing and feature analysis The simulation data were checked and found to be free of missing values, negative values, and anomalies. The number of stops ranged from 0 to 2.4, and the delay ranged from 0 to 96 seconds, both within a reasonable range. "Carbon emissions per truck / passenger vehicle due to delay" was used as the dependent variable, and "average number of stops" and "average delay per vehicle" were used as independent variables.

[0058] (2) Model evaluation results Linear regression (LR), random forest (RF), gradient boosting tree (GBR), support vector machine (SVR), and K-means+ classification prediction were used to analyze the carbon emissions per vehicle delay and the average number of stops and average vehicle delay for passenger cars / trucks. The results are shown in Tables 6, 7, and 8. Table 6 Comparison of Indicators for the First Four Categories of Modeling Methods for Passenger Cars and Freight Trucks Table 6 Comparison of Indicators for the First Four Categories of Modeling Methods for Passenger Cars and Freight Trucks

[0059] The results of the K-means clustering + classification prediction model are as follows: For passenger cars: K=2, the elbow point is obvious, so K=2 is selected. For trucks: K=3, the silhouette coefficient is the highest, reaching 0.5913, and a clear "elbow" inflection point appears, indicating good clustering effect.

[0060] Table 7. K-means+ classification prediction modeling indicators for passenger cars

[0061] Table 8. K-means+ classification prediction modeling indicators for trucks

[0062] After clustering passenger cars and trucks, among various analysis methods, Gradient Boosting Tree (GBR) performed best, indicating that the model has the strongest fitting ability for the relationship between "number of stops - delay - carbon emissions", and is especially suitable for handling the nonlinear characteristics of this type of traffic data.

[0063] (3) Model Comparison Analysis The five methods were comprehensively evaluated based on six dimensions: prediction accuracy, generalization ability, computational efficiency, interpretability, parameter tuning difficulty, and applicable scenarios. Scores ranged from 1 to 5, with 5 being the best, significantly outperforming the other methods in this dimension; 4 being good, performing well but with minor shortcomings; 3 being average, acceptable, but with significant weaknesses; 2 being poor, with obvious defects; and 1 being extremely poor, almost unusable. The scoring results are shown in Table 9.

[0064] Table 9 Comparison of comprehensive scores for five analysis methods

[0065] The analysis of passenger cars and trucks above shows that gradient boosting trees have the highest overall score, with the best "prediction accuracy" and "applicable scenarios," only slightly higher "parameter tuning difficulty," making them the preferred choice for single models. Linear regression excels in "computational efficiency" and "interpretability," but has shortcomings in "prediction accuracy" and "applicable scenarios." KMeans + gradient boosting trees offer the best "prediction accuracy," "generalization ability," and "applicable scenarios" for passenger cars, with slightly lower scores for trucks, but poorer "computational efficiency" and "parameter tuning difficulty," making them suitable for scenarios with extremely high accuracy requirements and acceptable computational costs. From an engineering deployment perspective, gradient boosting trees are preferable for single-model, fast prediction, while K-means + classification prediction can be used for in-depth analysis of data grouping patterns.

[0066] Step S4.5: Synthesis of actual carbon emissions per vehicle, where actual carbon emissions per vehicle = zero-flow carbon emissions + delayed carbon emissions, i.e. ; Step S4.6: Summarize the total carbon emissions of the intersection by weighting and summing them according to vehicle type, fuel type, and traffic volume to obtain the total carbon emissions of the intersection. ; Step S5: Verify the effectiveness of the model through actual intersection application. Verify the effectiveness of the model by comparing the assessment results of the carbon emission assessment model with the carbon emission results of the TessNG simulation model.

[0067] 1. Evaluation results of the established carbon emission assessment model: Based on the carbon emission prediction model, the carbon emissions and total carbon emissions of each vehicle type at the pre-selected intersection were predicted. The steps involved using 300-second intervals (90 minutes for simulation modeling, and 60 minutes for evaluation data), with four intersection entrances, each entrance having three turns (left, straight, and right), and 12 statistical time intervals for each turn, totaling 144 traffic flow data points to be predicted. The experiment used a k-means + classification prediction model for prediction. The specific process is as follows. Figure 9 .

[0068] The average number of stops and average vehicle delay historical data for each turn were standardized using a normalizer and then substituted into the K-means model to obtain the cluster ID. The cluster assignment results for each turn are shown in Table 10.

[0069] Table 10 Clustering Results of Passenger Vehicle Traffic Flow Data at Intersections

[0070] Using the above method, the actual carbon emissions of passenger cars at the intersection are predicted to be 1,178,940g. The carbon emissions of passenger cars at each entrance and each turn, with a time interval of 300 seconds, are as follows: Figure 10 The prediction methods for trucks, buses, and public buses are similar, and the process is omitted. The results are 36151.21g, 73961.11g, and 161322.35g respectively. The total carbon emissions at the intersection are 1450374.66g, or 1450.37kg.

[0071] 2. TessNG simulation modeling and carbon emission results: A simulation model of the intersection was built using TessNG. In the traffic demand analysis, green-plate passenger cars were designated as electric vehicles, with dimensions and other performance parameters identical to blue-plate gasoline-powered passenger cars. Carbon emissions from their energy production were ignored; only the intersection operation process was considered, classifying them as green, emission-free vehicles. Small trucks, medium-sized passenger cars, and medium-sized trucks in the surveyed traffic flow were grouped into the medium-sized truck category (8m long medium-sized vans) in the simulation model, while large vehicles and buses were designated as 12m long passenger cars. The simulation was run to obtain a real-time carbon emission heatmap, and the intersection evaluation indicators and carbon emission results are output as shown in Table 11 below.

[0072] Table 11 Carbon Emissions of Various Vehicle Types at the Intersection

[0073] 3. Comparative Analysis: Table 12 shows the predicted carbon emissions of various vehicle types at the pre-selected intersection and the relative differences between the prediction model and TessNG simulation.

[0074] Table 12. Carbon emission prediction results (g) at intersections using different methods

[0075] The relative differences in the prediction results of the two methods for trucks, buses, public buses, and total emissions are all between 3% and 5%, with a high overall consistency and the differences are within an acceptable range. This demonstrates that constructing a relational model using the number of stops and delays as core parameters achieves a direct quantitative correlation between traffic flow status and carbon emissions. This not only improves the theoretical system of traffic carbon emissions but also provides a precise assessment tool for the low-carbon transformation of urban transportation.

[0076] It is worth mentioning that the technical features such as TessNG and Spearman involved in this patent application should be regarded as prior art. The specific structure, working principle and possible control methods and spatial arrangement of these technical features can be adopted using conventional choices in the field, and should not be regarded as the inventive point of this patent. This patent will not be further elaborated in detail.

[0077] For those skilled in the art, modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the protection scope of this invention.

Claims

1. A method for assessing carbon emissions at urban intersections based on traffic flow parameters, characterized in that, Includes the following steps: Step S1: Construct a typical signal-controlled intersection scenario, design channelization schemes and signal timing, and set up a traffic demand model that covers different traffic flow states and new energy penetration rates; Step S2: Run the constructed intersection scenario using TessNG simulation software to obtain traffic flow parameters and carbon emission data. Verify the simulated traffic flow and travel time ratio to ensure that the simulation covers five traffic operation states: smooth, basically smooth, lightly congested, moderately congested, and severely congested. The simulated traffic flow verification involves comparing the simulated traffic flow with the input traffic flow. The travel time ratio verification involves calculating the ratio of the actual travel time to the free-flow travel time to ensure coverage of the travel time ratio ranges corresponding to the five traffic operation states. Step S3: Spearman correlation analysis is used to identify key influencing factors and determine the average vehicle delay and average number of stops as the core parameters affecting carbon emissions. The selection of Spearman correlation analysis is based on the fact that traffic flow parameters do not satisfy the assumptions of linearity and normality, the nonlinear characteristics that the greater the delay, the higher the carbon emission growth efficiency, and the need for robust monotonic relationship analysis of outliers. Step S4: Decompose single-vehicle carbon emissions into zero-flow carbon emissions and delay carbon emissions, construct a calculation method based on the cumulative emissions of single vehicles, and establish a carbon emission assessment model for delay carbon emissions, average vehicle delay, and average number of stops. Calculate the total carbon emissions of the intersection based on the cumulative carbon emissions of single vehicles. The specific implementation of establishing the carbon emission assessment model is as follows: Step S4.1: Construct a framework for calculating total carbon emissions. Establish a method for calculating total carbon emissions at intersections based on the cumulative emissions of individual vehicles. Total emissions are the sum of the actual carbon emissions of all carbon-emitting vehicles, categorized by vehicle type and fuel type. Step S4.2: Deconstruct the carbon emission structure of a single vehicle, breaking down the actual carbon emissions of a single vehicle into: Zero-flow carbon emissions: Benchmark emissions for vehicles passing through intersections without delay or stopping; Delayed carbon emissions: Additional emissions caused by congestion, start-stop, acceleration, and deceleration; Step S4.3: Zero-flow carbon emission modeling; Zero-flow carbon emissions are directly related to intersection travel distance and vehicle fuel type, and are calibrated using the emission factor method based on fuel type. Step S4.4: Modeling carbon emissions due to delays. Using average vehicle delay and average number of stops as core input variables, K-means clustering is used to first identify the inherent structure of the data, and then gradient boosting tree regression model is called for each subset to obtain the carbon emission calculation model due to delays. Step S4.5: Synthesis of actual carbon emissions per vehicle, Actual carbon emissions per vehicle = Zero-flow carbon emissions + Delayed carbon emissions; Step S4.6: Summarize the total carbon emissions of the intersection by weighting and summing them according to vehicle type, fuel type, and traffic volume to obtain the total carbon emissions of the intersection; Step S5: Verify the effectiveness of the model through actual intersection application. Verify the effectiveness of the model by comparing the assessment results of the carbon emission assessment model with the carbon emission results modeled by TessNG simulation. The verification includes predicting the carbon emissions of each approach and turn at each 300-second interval and performing a relative difference analysis with the TessNG simulation results.

2. The method for assessing carbon emissions at urban intersections based on traffic flow parameters according to claim 1, characterized in that, For step S1, design signal-controlled intersections where arterial roads intersect, determine the intersection range, channelization scheme and signal timing parameters; set traffic demand gradients covering all states from smooth to congested and different new energy penetration rates, divide into low-flow groups and regular groups, fix the flow of trucks and buses, and form multiple traffic demand combinations.

Images