A method for identifying high-emission vehicles of nitrogen oxides of diesel vehicles based on emission reduction targets

Abstract

The application relates to the technical field of motor vehicle pollution emission control and emission accounting, and discloses a diesel vehicle nitrogen oxide high-emission vehicle identification method based on emission reduction targets. The method comprises the following steps: obtaining diesel vehicle annual inspection NOx emission concentration data; performing two-dimensional grouping on vehicles according to vehicle types and weight grades, constructing a cumulative mileage-vehicle age statistical model to calculate a benchmark year activity level; introducing an emission stage dimension, performing three-dimensional combination refinement grouping, and calculating total emissions before vehicle fleet treatment; determining a target emission amount; enumerating candidate NOx emission concentration thresholds, calculating total emissions after treatment; screening a feasible threshold set, and selecting an optimal threshold; and identifying vehicles with annual inspection NOx emission concentrations higher than the optimal threshold as high-emission vehicles. The method minimizes the treatment scale under the premise of meeting the emission reduction requirements, improves the emission factor accounting accuracy to the vehicle level, and improves the accuracy of high-emission vehicle identification and treatment efficiency.

Description

Technical Field

[0001] This application relates to the field of motor vehicle pollution emission control and emission accounting technology, and in particular to a method for identifying high-NOx emission diesel vehicles based on emission reduction targets. Background Technology

[0002] Diesel vehicles are a major source of NOx emissions from road traffic, and effectively identifying high-emission vehicles is a prerequisite for implementing precise pollution control and reducing total emissions. Existing methods for identifying high-emission vehicles typically use fixed emission limits as prior thresholds to directly screen vehicles with excessive NOx emission concentrations. This approach has four main shortcomings: First, the threshold setting is decoupled from regional emission reduction targets. Fixed limits do not consider the emission reduction needs of specific regions at specific times, leading to a number of screened vehicles that may far exceed or fall short of the required pollution control scale to achieve the standards. Second, the expression of emission factor differences is insufficient. Using a uniform average emission factor within the same emission stage fails to reflect the actual emission differences between different vehicles, resulting in systematic biases in the calculation results. Third, there is a lack of an integrated mechanism for emission accounting and vehicle identification. The impact of pollution control measures on emission factors cannot be incorporated into the calculation, making it difficult to form a complete technical path of "emission accounting - threshold setting - emission reduction assessment - vehicle identification." Fourth, the efficiency of pollution control lacks optimization. Under the condition of meeting the same emission reduction targets, existing methods lack an optimization mechanism to achieve the predetermined goals with the minimum pollution control scale, leading to excessively high administrative costs or an overly broad scope of pollution control. Therefore, there is an urgent need for a method for identifying high-NOx diesel vehicles based on emission reduction targets, which directly couples regional emission reduction targets with threshold determination, achieves refined expression of vehicle-level emission factors, and achieves emission reduction targets with minimal governance scale. Summary of the Invention

[0003] To address the aforementioned technical issues, this application provides a method for identifying high-NOx-emission diesel vehicles based on emission reduction targets, which aims to improve the accuracy of NOx-emission vehicle identification.

[0004] This application provides a method for identifying high-NOx emission diesel vehicles based on emission reduction targets, the method comprising:

[0005] Step S1: Obtain diesel vehicle annual inspection data from the three-level network system for periodic vehicle inspection, and group the vehicles in two dimensions according to vehicle type and weight class;

[0006] Step S2: Based on the two-dimensional grouping and vehicle age and annual inspection cumulative mileage data, fit the cumulative mileage-vehicle age statistical model for each group to estimate the annual mileage of each vehicle in the base year.

[0007] Step S3: Based on the two-dimensional grouping, introduce the emission stage as the third dimension, and group the vehicles in three dimensions according to the three-dimensional combination of emission stage, vehicle type and weight class to obtain the total emissions of the fleet before treatment.

[0008] Step S4: Set the NOx emission reduction target percentage P% according to the regional environmental management needs, and calculate the target emission amount based on the total emissions of the fleet before treatment;

[0009] Step S5: Determine the lower limit of candidate thresholds based on the feasibility constraints of the treatment scale, locate the key search interval based on the pre-sorting of the expected emission reduction potential of each vehicle and the identification of the inflection point of the cumulative emission reduction curve, enumerate the candidate NOx emission concentration thresholds within the key search interval, and calculate the total emissions of the fleet after treatment.

[0010] Step S6: From the enumerated candidate thresholds, select candidate thresholds that ensure the total emissions after treatment are not higher than the set target emissions to form a feasible threshold set; select the threshold that minimizes the number of suspected high-emission vehicles from the feasible threshold set as the optimal threshold.

[0011] Step S7: Identify vehicles with NOx emission concentrations higher than the optimal threshold during the annual inspection as high-emission vehicles, and output a list of high-emission vehicles including Vehicle Identification Number (VIN), vehicle type, emission stage, NOx emission concentration during the annual inspection, and corresponding emission amount.

[0012] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0013] First, by constructing the response relationship between candidate thresholds and post-treatment emissions, the optimal threshold is determined in reverse by using regional emission reduction targets as hard constraints. This ensures that the identification results accurately support the established emission reduction targets and avoids mismatch between the scale of treatment and emission reduction needs. A "three-dimensional grouping + median ratio mapping" mechanism is adopted, refining the grouping by emission stage, vehicle type, and weight class. A single-vehicle emission factor mapping is established using the median NOx concentration within the same group as a reference benchmark, improving the accuracy of emission factors from the emission stage level to the vehicle level and eliminating the calculation bias caused by a uniform average factor. Then, the threshold with the fewest suspected high-emission vehicles is selected from the feasible threshold set that meets the emission reduction targets. This achieves the established emission reduction effect with minimal necessary treatment, significantly reducing implementation costs and social burden. Based on the pre-sorting of expected emission reduction potential for single vehicles and the identification of inflection points in the cumulative emission reduction curve, the enumeration space is compressed to a highly efficient interval, improving computational efficiency and supporting real-time decision-making for large-scale fleets. Finally, a technical accessibility constraint is introduced to correct the assumption of vehicles with extreme failures. Vehicle stability levels are categorized based on the coefficient of variation from multiple annual inspections. For vehicles with low to medium stability, standard deviation correction or two consecutive instances of exceeding the standard are used to reduce the risk of misjudgment due to detection fluctuations and improve identification reliability. A fleet structure file and seasonal correction coefficient are established. When changes in the fleet structure exceed a threshold, recalculation is triggered, extending the effective application period of the method. Attached Figure Description

[0014] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0015] Figure 1 This is a flowchart illustrating a method for identifying high-NOx emission diesel vehicles based on emission reduction targets, as described in this application.

[0016] Figure 2 The figures represent the percentage of suspected high-emission vehicles, the percentage of emission reduction potential, and the marginal utility curves of emission reduction under different candidate thresholds in this application embodiment. Detailed Implementation

[0017] This application provides a method for identifying high-NOx emission diesel vehicles based on emission reduction targets. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0018] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the method for identifying high-NOx emission diesel vehicles based on emission reduction targets in this application includes:

[0019] Step S1: Obtain diesel vehicle annual inspection data from the three-level network system for periodic vehicle inspection, and group the vehicles in two dimensions according to vehicle type and weight class.

[0020] Obtaining diesel vehicle annual inspection data also includes:

[0021] The annual inspection data for diesel vehicles shall include at least the Vehicle Identification Number (VIN), emission stage, vehicle type, weight class, vehicle age, cumulative mileage for the year, and NOx emission concentration under the deceleration test.

[0022] The acquired data was cleaned, and valid records were retained for fuel type diesel, detection using the loading deceleration method, and NOx emission concentrations that were positive.

[0023] Specifically, diesel vehicle annual inspection data is extracted from the three-level network system for periodic vehicle inspections by city or region. The data fields involved include: inspection report number, inspection time, vehicle identification number (VIN), fuel type, vehicle license plate, license plate color, city, initial registration date, vehicle annual inspection year, emission stage, vehicle type, cumulative mileage of annual inspections, vehicle age, manufacturer, vehicle model, number of times periodic inspections failed, and NOx test results under the loading and deceleration method (i.e., the emission concentration of nitrogen oxides in the exhaust gas of the vehicle under loading and deceleration conditions, in ppm, which is a direct observation of the emission status of a single vehicle).

[0024] Data cleaning shall be performed according to the following rules:

[0025] (a) Retain vehicle records for diesel fuel, and exclude gasoline vehicles, natural gas vehicles, and other fuel types;

[0026] (b) Exclude cases where NOx test results are missing in all annual inspection records of a vehicle. That is, if a vehicle has annual inspection records but has never been tested for NOx, then the vehicle will not be included in the subsequent analysis.

[0027] (c) Remove records with negative or zero NOx test results, because emission concentration is a physical quantity and cannot be negative. A zero value usually indicates a test failure or data entry error.

[0028] (d) Retain the NOx test results of the loading deceleration method; if there are multiple records of a vehicle in the same annual inspection year, take the maximum value of the NOx result as the representative value of that annual inspection. The technical basis for this maximum value rule is that: high emission identification focuses on the highest emission state that the vehicle has actually experienced. A single low value may be caused by accidental factors, while the maximum value can better reflect the vehicle's true emission potential.

[0029] (e) Retain records for vehicles whose major category is passenger car or freight car, and remove vehicles with unclear categories or insufficient sample size, such as tractors and special vehicles;

[0030] (f) Remove records with abnormal cumulative mileage during the annual inspection, including negative mileage, exceeding reasonable limits (such as more than 5 million kilometers) or values ​​that are obviously inconsistent with the vehicle's usage patterns (such as a 10-year-old vehicle with only 1,000 kilometers on the mileage).

[0031] After data cleaning, the vehicle grouping stage begins. A two-dimensional grouping strategy is adopted, where vehicles are divided only by combinations of vehicle type and weight class. Vehicles within the same group share the same vehicle type and weight class. For example, "large passenger buses" form one group, and "heavy-duty trucks" form another. The technical reason for not introducing emission stages as a grouping dimension is that the purpose of this grouping step is to provide a sufficient sample size for fitting the subsequent cumulative mileage-vehicle age statistical model. Emission stages are strongly correlated with vehicle age (older vehicles are mostly in low-emission stages, while newer vehicles are mostly in high-emission stages). Introducing emission stages would lead to overly granular grouping and insufficient sample size in each group, failing to meet the sample size requirements for statistical modeling. Therefore, the two-dimensional grouping scheme is the optimal choice to balance group homogeneity and sample sufficiency. After grouping, the vehicle set within each group is recorded.

[0032] Step S2: Based on the two-dimensional grouping and vehicle age and annual inspection cumulative mileage data, fit the cumulative mileage-vehicle age statistical model to each group to estimate the annual mileage of each vehicle in the base year.

[0033] In step S2, the cumulative mileage-vehicle age statistical model is automatically selected from three types of models, with the one having the smallest root mean square error in the fit being the optimal model. The three types of models include:

[0034] The model employs a monotonically increasing constrained spline model (SCAM) with a Gamma logarithmic connection function; a log-linear model fits the logarithm of cumulative mileage with vehicle age as the independent variable; and a saturated growth nonlinear model, mathematically expressed as: ,in, For the annual inspection cumulative mileage data, For saturation mileage parameters, For the growth rate parameter, This refers to the vehicle's age.

[0035] Specifically, Vehicle Kilometers Traveled (VKT) is a key activity level parameter in emissions accounting, referring to the total mileage a vehicle travels within a base year, measured in kilometers. Since vehicle periodic inspection data is cross-sectional data—meaning it only records cumulative mileage at the time of the annual inspection—it's impossible to directly obtain the annual mileage for any given year. Therefore, statistical modeling methods are needed to establish a functional relationship between cumulative mileage and vehicle age, and then the annual mileage can be indirectly calculated through function differences.

[0036] Data preparation before model fitting includes: for each two-dimensional group, vehicles within the group are segmented by age at six-month (0.5-year) intervals, such as 0-0.5 years, 0.5-1 year, and 1-1.5 years. The median of the cumulative mileage within each age segment is calculated as the representative mileage level for that age segment. The median is chosen instead of the mean because: within the same age segment, the cumulative mileage of different vehicles is discretely distributed due to differences in usage intensity. The extremely high mileage of a few high-intensity vehicles can raise the mean, causing it to deviate from the actual level of most vehicles. The median, however, is not sensitive to extreme values ​​and better represents the typical cumulative mileage status of vehicles in that age segment. Simultaneously, a constraint is set that the number of valid samples in each segment must be no less than 10 vehicles. If the number of valid vehicles in a certain age segment is less than 10, that segment is not included in the model fitting to ensure the reliability of the statistical results.

[0037] Using the median mileage for each vehicle age group as the fitted data point, and with the mean of vehicle age group as the independent variable and the median cumulative mileage as the dependent variable, the following three statistical models were tried in turn, and the model with the smallest root mean square error (RMSE) was selected as the optimal model. RMSE is an indicator that measures the degree of difference between the model's predicted values ​​and the actual observed values. The formula is the square root of the average of the squares of the relative errors of each data point. The smaller the RMSE, the higher the model's fitting accuracy. Before calculating the RMSE, outliers with a relative error of more than 50% were removed. That is, if the absolute value of the relative error of a data point exceeds 50%, it is considered that the point may be affected by abnormal data and is removed before calculating the RMSE to ensure the robustness of the model evaluation.

[0038] The first type of model is the Shape-Constrained Additive Model (SCAM). SCAM is an extension of the Generalized Additive Model (GAM). While GAM establishes the relationship between the response variable and multiple predictor variables through a nonparametric smoothing function, SCAM adds shape constraints to GAM. The shape constraint used in this method is a monotonically increasing constraint, requiring that the cumulative mileage increases monotonically with vehicle age. This constraint aligns with the physical laws of vehicle use: the cumulative mileage cannot decrease with vehicle age. SCAM employs a Gamma distribution family and a log-join function. The Gamma distribution is chosen because the cumulative mileage is positive and right-skewed, making it suitable for modeling positive continuous variables. The log-join function correlates the linear predictor with the logarithmic mean of the response variable, ensuring that the predicted value is always positive. The input data for this model consists of the mean of each vehicle age segment and its corresponding median cumulative mileage; the output is the predicted cumulative mileage for a given vehicle age. Its working principle can be simply summarized as follows: under the constraint of monotonically increasing, a smooth cumulative mileage-vehicle age curve is fitted by spline basis functions. This curve always rises and can flexibly adapt to the nonlinear characteristics in the data.

[0039] The second type of model is the log-linear model. This model uses vehicle age as the independent variable and fits a linear relationship with the logarithm of cumulative mileage, that is, it establishes the following equation: ,in, For the intercept parameter, Let be the slope parameter, and ε be the random error term. Parameter estimation uses ordinary least squares, i.e., finding the value that minimizes the sum of squared residuals. and In prediction, the results are first obtained through linear equations. The predicted value is then converted back to the mileage scale using an exponential transformation. The input data for this model are the mean and median cumulative mileage for each vehicle age group, and the output is the predicted cumulative mileage for a given vehicle age. Its working principle can be simply summarized as follows: assuming a linear relationship between the logarithm of cumulative mileage and vehicle age, a logarithmic transformation converts the non-linear growth into a linear problem, which is then converted back to the linear scale using an exponential transformation. This model is suitable for stages with relatively uniform growth rates.

[0040] The third type of model is the Saturation Growth Nonlinear LeastSquares (Saturation Growth NLS) model. In the mathematical form of this model, the saturation mileage parameter, in kilometers, represents the theoretical upper limit of the vehicle's cumulative mileage, i.e., the asymptote the cumulative mileage approaches as the vehicle age approaches infinity; the growth rate parameter, in 1 / year, represents the speed at which the vehicle's cumulative mileage approaches the saturation value, with a larger value of b indicating faster growth. Parameters a and b are estimated using the Nonlinear Least Squares (NLS) method, an iterative optimization algorithm that minimizes the sum of squared residuals between the model's predicted values ​​and the actual observed values ​​by continuously adjusting the parameter values. The input data for this model are the mean of vehicle age and the median of cumulative mileage, and the output is the predicted cumulative mileage for a given vehicle age. Its working principle can be simply summarized as follows: assuming that the growth of vehicle cumulative mileage follows a pattern of "fast at first, then slow, and eventually stabilizing," similar to biological growth or market saturation, the two parameters control the upper limit of growth and the growth rate, forming a simplified form of an S-shaped growth curve.

[0041] After fitting the three models, their RMSEs on the valid data points are compared. The model with the smallest RMSE is taken as the optimal mileage-vehicle age model for that group, denoted as . Where g represents the group index and age represents the vehicle age. The technical significance of this automatic selection mechanism is that the usage patterns of vehicles in different groups are different. For example, heavy trucks may grow rapidly in the early stage and tend to saturate in the later stage, while small passenger cars may maintain a relatively uniform linear growth in the long term. No single model can be applied to all groups. Data-driven automatic selection can ensure the optimal fitting effect of the model for each group.

[0042] After obtaining the optimal model, the next step is to calculate the annual mileage of each vehicle. For each vehicle v, based on the relationship between its initial registration date and the base year, the vehicle age at the beginning and end of the base year is calculated. The base year refers to the main inspection year covered by the data, usually the most recent complete calendar year. For example, if the data covers the entire inspection record for 2024, then the base year is set to 2024. Vehicle age at the beginning of the base year. The base year-end vehicle age is the number of years from the vehicle's initial registration date to January 1, 2024. This refers to the number of years up to January 1, 2025. The optimal model for the respective group is used to predict the cumulative mileage at these two points in time, and the difference between the two is the vehicle's base year mileage. The technical principle behind this difference calculation is that the cumulative mileage-vehicle age model provides a predicted cumulative mileage value at any given time, and the difference between the cumulative mileage of two adjacent years naturally corresponds to the mileage driven in that year.

[0043] If the calculated annual mileage is less than or equal to zero, it indicates an anomaly in the model prediction. Possible causes include long-term vehicle inactivity, data entry errors, or model extrapolation failure. In this case, the vehicle's VKT is corrected to the median VKT of valid vehicles within the same group to ensure the reasonableness of the activity level data and avoid the distortion of subsequent emissions calculations by outliers. At this point, each vehicle has an estimated annual mileage for the base year, which will serve as a key input parameter for calculating individual vehicle emissions in step S3.

[0044] Step S3: Based on the two-dimensional grouping, introduce the emission stage as the third dimension, and group the vehicles in three dimensions according to the three-dimensional combination of emission stage, vehicle type and weight class to obtain the total emissions of the fleet before treatment.

[0045] In step S3, the two-dimensional grouping is only used for fitting the cumulative mileage-vehicle age statistical model in step S2, and the three-dimensional grouping is only used for emission factor calculation. The two grouping schemes serve different calculation purposes, specifically including:

[0046] In step S2, the grouping method is determined only by the combination of vehicle type and weight class. Vehicles in the same group have the same vehicle type and weight class, and the emission stage is not introduced as a grouping dimension.

[0047] In step S3, the refined grouping further introduces emission stages based on the grouping in step S2. Vehicles in the same group have the same emission stage, vehicle type and weight class, ensuring the consistency of the emission stage of the reference benchmark in the emission factor mapping.

[0048] The total emissions obtained before the fleet's pollution control measures included:

[0049] Using the median NOx emission concentration of all vehicles in each subgroup as the reference benchmark for the overall emission level of the group, a proportional mapping relationship between the annual inspection NOx emission concentration of a single vehicle and the standard emission factor is established, and the annual inspection NOx emission concentration of each vehicle is converted into the single vehicle NOx emission factor; combined with the annual mileage, the baseline annual NOx emission of each vehicle is calculated and summed to generate the total emission of the fleet before treatment.

[0050] The proportional mapping relationship is as follows: In the formula, For vehicles NOx emission factor To refine the grouping The standard NOx emission factor for the corresponding emission stage, For vehicles The annual inspection NOx emission concentration To refine the grouping The median NOx emission concentration of all vehicles inspected annually.

[0051] When a vehicle has multiple annual inspection records The maximum NOx emission concentration from multiple annual inspections is taken; when the number of valid vehicles in a certain subgroup is less than 2, the median is replaced by the mean NOx concentration of valid vehicles in the subgroup.

[0052] Specifically, based on the two-dimensional grouping in step S1, emission stage is introduced as a third dimension. Vehicles are grouped in three dimensions according to the combination of emission stage, vehicle type, and weight class, and this refined grouping is denoted as g. It is necessary to clarify why two different grouping schemes are used: In step S2, only two-dimensional grouping is performed according to vehicle type and weight class because the fitting of the cumulative mileage-vehicle age statistical model requires a sufficient sample size for each vehicle age segment. Emission stage and vehicle age are strongly correlated; older vehicles mostly correspond to low emission stages, and newer vehicles mostly correspond to high emission stages. Introducing emission stage would lead to overly fine grouping and insufficient sample size within each age segment, failing to meet the modeling constraint of at least 10 vehicles per segment. In step S3, three-dimensional grouping is used because the core premise of emission factor calculation is that vehicles within the same group have the same emission stage technology level. The standard emission factors of different emission stages can differ by several times or even an order of magnitude. Therefore, two-dimensional and three-dimensional grouping serve different calculation purposes and are independent of each other. This design is a key arrangement that balances the feasibility of statistical modeling and the accuracy of emission calculation.

[0053] After the three-dimensional grouping is completed, the median NOx emission concentration of all vehicles in each refined group g is calculated. The median, rather than the mean, is used as a reference for the overall emission level of the group. The technical rationale for choosing the median over the mean is that the NOx concentration distribution of in-use diesel vehicles is typically right-skewed, with most vehicles having emission concentrations concentrated in a relatively low range, while a few high-emission vehicles have extremely high concentrations, causing the mean to be significantly higher than the majority of the distribution. Using the mean as a reference would systematically underestimate the emission factors of normal-emission vehicles and the relative deviation of high-emission vehicles, leading to a decrease in the sensitivity of high-emission identification. The median is insensitive to extreme values; when there are abnormally high-emission vehicles within a group, the median remains stable, ensuring that the emission factor mapping benchmark is not affected by a few extreme values, thus guaranteeing the robustness and repeatability of the method.

[0054] A proportional mapping relationship is established between the annual inspection NOx emission concentration of each vehicle and the standard emission factor, converting the annual inspection NOx emission concentration of each vehicle into the vehicle NOx emission factor. The mathematical expression for the proportional mapping relationship is: ,in is the NOx emission factor of vehicle v, expressed in grams per kilometer, representing the mass of NOx emitted per kilometer the vehicle travels. To refine the standard NOx emission factor for the emission stage corresponding to group g, the unit is grams per kilometer. The measured or recommended emission factors for this emission stage are taken from published literature and relevant national technical specifications. For vehicles that lack the same type of measured emission factor, the emission factor of the same emission stage with similar technical characteristics is used as a substitute. The NOx emission concentration of vehicle v during its annual inspection is expressed in ppm, which is one part per million. When a vehicle has multiple annual inspection records, the maximum NOx concentration from each record is taken to reflect the highest emission status that the vehicle actually experienced during the inspection period. The median NOx emission concentration for all vehicles undergoing annual inspections within group g is given, expressed in ppm.

[0055] The technical meaning of the above formula is as follows: if the NOx concentration of a vehicle during its annual inspection is equal to the median of its group, then its emission factor is exactly equal to the standard emission factor for that emission stage; if it is higher than the median, the emission factor is proportionally amplified; if it is lower than the median, it is proportionally reduced. This linear proportional mapping relationship is selected based on two points: First, within the same emission stage, vehicle type, and weight class, engine design parameters are basically consistent. Under the fixed-condition loading and deceleration test, due to the relatively stable engine operating state, the NOx emission concentration and emission rate are approximately linearly related within a certain range. Second, the linear mapping relationship ensures that the calculated emission factor is consistent with the weighted average of the emission factors of all vehicles within the group. When all vehicles within the group are calculated using this formula, the total emission calculation result of the group is consistent in magnitude with the overall estimate based on the standard emission factor, satisfying the internal consistency requirement for emission calculation.

[0056] When the number of valid vehicles in a refined group is less than 2, a statistically representative median cannot be calculated for that group. In this case, the mean NOx concentration of valid vehicles within that group is used instead of the median in the calculation to ensure its feasibility. The technical significance of this sample sufficiency constraint is to avoid situations where a very small number of samples lead to median distortion or inability to calculate, thus ensuring the applicability of the method to all groups.

[0057] After obtaining the NOx emission factor of a single vehicle, the annual mileage calculated in step S2 is used as a reference. Calculate the NOx emissions of vehicle v in the baseline year: The physical meaning of this formula is: the emission factor represents the emissions per unit mileage, and multiplying it by the annual mileage yields the total annual emissions. Summing up the individual emissions of all effective diesel vehicles gives the total emissions of the fleet before pollution control. This total emission figure serves as the calculation basis for setting subsequent emission reduction targets, and also as a comparison benchmark for calculating emissions under each candidate threshold governance scenario.

[0058] Thus, step S3 completes the entire accounting chain from single-vehicle annual inspection concentration data to the total emissions of the fleet. Its core innovation lies in improving the calculation accuracy of emission factors from the emission stage level to the vehicle level through three-dimensional grouping and median ratio mapping. This allows vehicles with different actual emission levels within the same emission stage to obtain differentiated emission factors, rather than using a uniform average value, thereby providing a foundation for accurate simulation of emission changes under subsequent governance scenarios.

[0059] Step S4: Set the NOx emission reduction target percentage P% according to the regional environmental management needs, and calculate the target emission amount based on the total emissions of the fleet before treatment.

[0060] Step S4 further includes:

[0061] The formula for calculating the target emissions is: And the emission reduction target percentage P% is satisfied. ,in, This represents the theoretical maximum emission reduction ratio of the fleet, and the estimated emission reduction potential at the lower limit of the candidate threshold.

[0062] Specifically, a NOx emission reduction target percentage (P%) is set based on regional environmental management needs. This percentage reflects the region's total NOx control requirements within a specific planning period and is determined by the competent authority based on factors such as the regional air pollution prevention and control plan, air quality improvement targets, pollution source apportionment results, and the contribution rate of diesel vehicle emissions. The physical meaning of P% is: the percentage reduction required in the total emissions of the vehicle fleet after treatment compared to the total emissions before treatment. For example, setting P% to 10% means that the total emissions after treatment are required to be reduced by 10% compared to before treatment.

[0063] The total emissions of the fleet before treatment obtained in step S3 Based on the baseline, calculate the target emissions. The calculation formula is: The derivation logic of this formula is as follows: if emissions need to be reduced by P%, then the remaining allowable emission ratio is (1 - P%). Multiplying this ratio by the total emissions before treatment yields the upper limit of allowable emissions after treatment. This calculation transforms the abstract emission reduction ratio into a specific emission constraint value, providing a clear criterion for selecting the optimal threshold.

[0064] The emission reduction target percentage P% must meet achievability constraints. ,in This represents the theoretical maximum emission reduction ratio for the fleet. The physical meaning is: under the current fleet structure and pollution control technology conditions, the maximum emission reduction potential that can be achieved by controlling high-emission vehicles, corresponding to the lower limit of the candidate threshold. The estimated emission reduction potential at that time. Specifically, when the candidate threshold reaches its minimum value. When the proportion of suspected high-emission vehicles included in the treatment reaches the upper limit of the treatment scale (e.g., 33%), the difference between the total emissions after treatment and the total emissions before treatment is the maximum achievable emission reduction. The proportion of this emission reduction to the total emissions before treatment is the maximum achievable emission reduction. .

[0065] The technical necessity of setting reachability constraints lies in: if P% exceeds... This means that regardless of the candidate threshold used, the emissions after fleet management cannot be reduced below the target emissions, rendering the identification method meaningless and resulting in no feasible solution. Therefore, This constitutes the theoretical upper limit for setting emission reduction targets, and the competent authorities need to estimate it before setting the P% target. This ensures that the objective is technically attainable. This constraint mechanism reflects the rigor and practicality of the method design, avoiding decision failures caused by overly ambitious objectives.

[0066] The estimation method is as follows: In step S5, the lower limit of the candidate threshold is... Perform a full remediation scenario simulation and calculate the total emissions after remediation at this threshold. ,but: This estimation utilizes the governance scenario simulation mechanism in step S5, using the emission reduction effect of the maximum governance scale scenario as a theoretical upper limit, providing a quantitative basis for the reasonable setting of emission reduction targets. If the actual set P% is less than the estimated value, it indicates that there is at least one candidate threshold that can make the post-governance emissions meet the target constraints, and the method has a feasible solution; if P% is greater than or equal to the estimated value, the emission reduction target needs to be adjusted or the governance scale constraints need to be relaxed before re-execution.

[0067] Thus, step S4 completes the transformation from regional environmental management needs to specific emission constraints. Its core function is to directly couple the macro-level emission reduction target with the micro-level threshold determination process, so that the subsequent selection of the optimal threshold always takes meeting the regional emission reduction requirements as a hard constraint, avoiding the drawback of existing fixed threshold methods being decoupled from emission reduction targets.

[0068] Step S5: Determine the lower limit of candidate thresholds based on the feasibility constraints of the treatment scale, locate the key search interval based on the pre-sorting of the expected emission reduction potential of each vehicle and the identification of the inflection point of the cumulative emission reduction curve, enumerate candidate NOx emission concentration thresholds within the key search interval, and calculate the total emissions of the fleet after treatment.

[0069] Locating key search intervals includes:

[0070] Calculate the expected emission reduction for each vehicle after treatment. The calculation formula is: ,in, This is the lower limit of the candidate threshold. To estimate the vehicle Annual mileage, based on Sort the vehicles from largest to smallest; accumulate the expected emission reductions for each vehicle in the sorted sequence to construct a curve showing the cumulative emission reduction as the number of vehicles treated increases; calculate the second difference of the curve to identify the inflection point where the marginal emission reduction effect decays, and record the NOx concentration value corresponding to the inflection point as... ,like The lower limit of the candidate threshold Then As the lower limit, with The upper limit of the search interval is defined as [ ], thus the key search interval is determined as [ ]. , If the NOx concentration value corresponding to the inflection point is lower than Then The key search interval is determined with a lower limit of 1500ppm and an upper limit of 1500ppm; candidate thresholds are enumerated only within the determined key search interval with a preset step size.

[0071] Step S5 further includes:

[0072] For each candidate threshold, a differentiated governance scenario is constructed based on the heterogeneity of emission stages. The emission levels of suspected high-emission vehicles in each refined group are simulated and restored to the emission levels of qualified vehicles that match the technical accessibility of the same emission stage. The total emissions of the fleet after governance are calculated, and the number of suspected high-emission vehicles under the corresponding threshold is recorded.

[0073] The calculation of the total emissions of the fleet after treatment includes:

[0074] Different post-treatment emission level assumptions are set for different emission stages; a technical accessibility constraint is introduced: if the NOx emission concentration of a vehicle during its annual inspection exceeds the technical theoretical upper limit of the emission stage to which the vehicle belongs, the vehicle is judged to have a serious mechanical failure, and the post-treatment emission level of the vehicle is directly set to the standard limit of the emission stage to which the vehicle belongs, rather than taking the statistical quantile of qualified vehicles in the same detailed group; according to the differentiation rules and technical accessibility constraints, the post-treatment emission factors of vehicles in each emission stage are calculated separately, and the total post-treatment emissions of the fleet are summed.

[0075] Among these measures, the uncertainty quantification of emission factors is introduced to make confidence corrections for the determination of the number of suspected high-emission vehicles, including:

[0076] For vehicles with multiple annual inspection records, the coefficient of variation of the corresponding NOx emission concentration is calculated. Based on the coefficient of variation, the vehicles are divided into three categories: high stability, medium stability, and low stability. For high stability vehicles, the measured concentration is directly compared with the threshold. For medium stability vehicles, the measured value plus 0.5 times the standard deviation is used as the judgment value. For low stability vehicles, the measured value plus 1.0 times the standard deviation is used as the judgment value, or the vehicle is identified as a high-emission vehicle only if it exceeds the threshold in two consecutive annual inspections.

[0077] Specifically, the lower limit of the candidate threshold is determined based on the feasibility constraints of governance scale. The technical purpose of this constraint is to ensure the operability of the governance plan in actual management. If the threshold is too low, the proportion of vehicles marked as potentially high-emission vehicles in the total fleet will be too high, exceeding the scale that actual maintenance capacity can handle, rendering the governance plan ineffective. This method uses 33% of the total fleet size as the upper limit for the feasibility of the governance scale, and determines the minimum candidate threshold that ensures the proportion of potentially high-emission vehicles does not exceed 33%. lower than The candidate thresholds are not listed due to the large scale of the governance. The 33% threshold is set based on the following: considering the actual constraints of the regional motor vehicle emission management agency's staffing, repair station capacity, and budget, and the fact that an excessively large scale of governance would trigger social resistance and affect the effectiveness of policy implementation. The method for determining the threshold is as follows: Candidate thresholds are tested progressively from low to high, and the proportion of suspected high-emission vehicles corresponding to each threshold is calculated. The first threshold that reduces this proportion to below 33% is the highest-risk threshold. .

[0078] In determining Next, the process moves to the key search area localization stage. The technical motivation for this stage is: if from... Enumerating candidate thresholds across the entire range up to 1500 ppm in 50 ppm increments typically involves about 17 thresholds. For each threshold, the emission factor and total emissions for the entire fleet must be recalculated, resulting in high computational complexity and making it difficult to support the real-time decision-making needs of large-scale fleets. However, by identifying the inflection point of the cumulative emission reduction curve, the search space can be compressed to the critical interval, significantly improving computational efficiency without overlooking optimal solutions.

[0079] The method for locating the key search interval is as follows: First, calculate the expected emission reduction for each vehicle after treatment. The derivation logic of this calculation formula is as follows: The NOx emission concentration for vehicle v during its annual inspection. To refine the grouping within g, with The median NOx concentration of qualified vehicles is defined by the threshold. The difference between the median and the median represents the expected reduction in NOx concentration after treatment. This concentration difference is then multiplied by a proportional mapping coefficient. This involves converting the concentration difference into an emission factor difference, and then multiplying it by the annual mileage. This yields the expected annual emission reductions. The order of vehicles from largest to smallest, that is, prioritizing vehicles that will benefit the most from governance, reflects the optimization idea of ​​"prioritizing the harvesting of low-hanging fruits".

[0080] The expected emission reductions are accumulated for each vehicle in the sorted vehicle sequence, constructing a curve showing the change in cumulative emission reductions as the number of vehicles undergoing treatment increases. The horizontal axis of this curve represents the number (or percentage) of vehicles undergoing treatment, and the vertical axis represents the cumulative emission reductions. The curve typically exhibits a characteristic of initially rising rapidly and then flattening out: initially, the vehicles undergoing treatment are all high-emission vehicles, resulting in large emission reductions per vehicle and a steep curve slope; as the number of vehicles undergoing treatment increases, the emission levels of the vehicles included in the treatment gradually approach normal levels, the emission reductions per vehicle decrease, and the curve slope gradually decreases. This morphological characteristic provides a data foundation for identifying inflection points.

[0081] The second-order difference is an approximation of the second derivative of discrete data, reflecting the rate of change of the curve's slope. On the cumulative emission reduction curve, the first-order difference represents the marginal emission reduction per additional vehicle, while the second-order difference represents the trend of this marginal emission reduction. When the second-order difference changes from a positive value to a negative value or approaches zero, it indicates a reversal in the growth momentum of marginal emission reduction effectiveness, meaning the emission reduction contribution of additional vehicles significantly diminishes; this point is the inflection point. The NOx concentration value corresponding to the inflection point is used as the upper limit of the candidate threshold search interval. Lower limit of candidate threshold As the lower limit of the search interval, the key search interval is thus determined as [ , The technical meaning of this interval is: to the left of the inflection point (i.e., below). Within the threshold range, each tightening of the threshold still results in a significant increase in the number of vehicles requiring emission reduction, representing a high-probability region that includes the optimal threshold; to the right of the inflection point (i.e., above the threshold range), the newly added vehicles requiring emission reduction still have a high marginal emission reduction effect, and this region contains the optimal threshold. Within this range, the marginal emission reduction utility approaches zero, and further increasing the threshold has a negligible contribution to emission reduction; the probability of the optimal threshold falling within this range is extremely low. If the NOx concentration value corresponding to the inflection point is lower than... This indicates that the marginal utility has diminished overall within the feasible governance scale, at which point... A key search interval is defined with a lower limit of 1500 ppm and an upper limit of 1500 ppm to ensure the integrity of the search space. Candidate thresholds are enumerated only within the defined key search interval with a preset step size, which is usually 50 ppm. That is, a candidate threshold is selected every 50 ppm within the interval to form a candidate set.

[0082] For each candidate threshold, a differentiated governance scenario is constructed based on the heterogeneity of emission stages. The emission levels of suspected high-emission vehicles in each refined group are simulated and restored to the emission levels of qualified vehicles that match the technical accessibility of that emission stage. Based on this, the total emissions of the fleet after governance are calculated, and the number of suspected high-emission vehicles under that threshold is recorded.

[0083] Based on the differentiation rules and technical accessibility constraints, the post-treatment emission factors for vehicles at each emission stage are calculated separately, and then summed to obtain the total post-treatment emissions of the fleet. The specific calculation rule is as follows: for each vehicle v within each refined group g, if... (Suspected high-emission vehicles), then the NOx emission concentration after treatment That is, take the differentiated emission level after treatment corresponding to the emission stage of the group (the value after correction for technical accessibility constraints); if (Qualified vehicles) This means that the emission levels of qualified vehicles remain unchanged. Based on the NOx emission concentration after treatment, the emission factors for each vehicle are updated according to the proportional mapping formula described in step S3: Note that the denominator here is still the original median calculated based on all vehicles in step S3, maintaining the consistency of the emission factor mapping benchmark and avoiding benchmark drift caused by threshold changes, thus ensuring comparability between different governance scenarios. Calculate candidate thresholds. The corresponding total emissions of the fleet after treatment: Simultaneously record the candidate threshold. Corresponding number of suspected high-emission vehicles and its proportion of the total fleet .

[0084] In the process of identifying suspected high-emission vehicles, NOx concentration uncertainty quantification is introduced for confidence correction to reduce the risk of misjudgment due to detection errors and fluctuations in vehicle operating conditions. For vehicles with multiple annual inspection records, the coefficient of variation of their NOx emission concentration is calculated. ,in The standard deviation of NOx concentrations during multiple annual inspections of this vehicle. This is the arithmetic mean of NOx concentrations from multiple annual inspections of the vehicle. The coefficient of variation is the ratio of the standard deviation to the mean; it is a dimensionless indicator that reflects the relative fluctuation of vehicle emission levels. A higher coefficient of variation indicates a more unstable emission state. For example, vehicles are classified into three categories based on the coefficient of variation: high-stability vehicles (…). ≤ 0.2) indicates that the vehicle's emissions are stable and the test results are reliable; medium stability vehicles (0.2 < ≤ 0.5 indicates that the vehicle's emissions fluctuate; low-stability vehicles ( A value > 0.5 indicates that the vehicle's emissions are highly unstable, and the representativeness of a single test result is poor. For high-stability vehicles, the measured concentration is directly compared with the threshold, as the test results are highly reliable and require no correction. For medium-stability vehicles, the measured value plus 0.5 times the standard deviation is used as the judgment value. This revision moderately tightens the judgment criteria, reducing the risk of missed judgments due to negative fluctuations; for vehicles with low stability, the measured value plus 1.0 standard deviation is used as the judgment value, i.e., the judgment value. The correction margin is larger, or the vehicle must exceed the threshold in two consecutive annual inspections to be considered a high-emission vehicle. This repeated verification over time eliminates interference from accidental factors. The judgment value after confidence correction is compared with the candidate threshold to determine the set of suspected high-emission vehicles. This mechanism, while ensuring identification sensitivity, significantly reduces the false judgment rate caused by detection uncertainty, and improves the reliability and credibility of high-emission vehicle identification.

[0085] Repeat the above operation for all thresholds in the candidate set to obtain a complete response relationship with the candidate thresholds as independent variables and the total emissions after treatment and the number of vehicles treated as dependent variables.

[0086] For example, Figure 2 For the curves of the proportion of suspected high-emission vehicles, the proportion of emission reduction potential, and the marginal utility of emission reduction under different candidate thresholds, in Figure 2 Taking the 700ppm to 1300ppm range as an example, the response relationship of different candidate thresholds is analyzed: as the candidate threshold increases from 700ppm to 1300ppm, the proportion of suspected high-emission vehicles decreases continuously from about 33% to below about 3%. The reduction in total emissions after treatment decreases accordingly. The marginal utility of emission reduction (the amount of emissions that can be reduced for every 50ppm increase) shows a decreasing trend in this range, indicating that the stricter the threshold, the higher the average emission contribution of the vehicles included in the treatment, but the marginal reduction decreases (e.g., Figure 2 (As shown). Candidate thresholds above 1300 ppm correspond to emissions exceeding the target emissions after treatment, thus failing to meet emission reduction constraints and not being included in the feasible threshold set.

[0087] Step S6: From the enumerated candidate thresholds, select candidate thresholds that ensure the total emissions after treatment are not higher than the set target emissions to form a feasible threshold set; from the feasible threshold set, select the threshold that minimizes the number of suspected high-emission vehicles as the optimal threshold.

[0088] Specifically, from the candidate set In the process, all candidate thresholds that meet the requirement that the total emissions after treatment do not exceed the target emissions are selected to form a feasible threshold set. : .

[0089] The physical meaning is that each threshold in this set, under the corresponding governance scenario, can reduce the NOx emissions of the fleet to a level no higher than the target emissions level; that is, each threshold can independently support the achievement of the regional emission reduction target. This screening process essentially transforms the abstract emission reduction target into a specific threshold feasible region, completing the mapping from "target constraint" to "feasible solution space".

[0090] This typically involves multiple feasible thresholds. Different feasible thresholds correspond to different governance scales: lower thresholds result in more vehicles being included in the governance, leading to greater emission reductions but higher governance costs; higher thresholds result in fewer vehicles being included in the governance, lower administrative costs, but smaller emission reduction margins. Given that emission reduction targets have already been met, further expanding the governance scale does not generate additional policy benefits but only brings additional administrative costs and social burdens. Therefore, the optimal solution should be selected from the feasible thresholds that meet emission reduction constraints, minimizing the number of vehicles requiring governance.

[0091] Therefore, the optimal threshold Defined as: .

[0092] when When multiple thresholds correspond to the same minimum number of vehicles to be regulated, the largest one is taken as the optimal threshold to ensure that the identification boundary for high-emission vehicles is as lenient as possible and to reduce the probability of normal-emission vehicles being mistakenly included in the regulation scope.

[0093] like An empty set indicates that within the current candidate threshold range, no governance scenario can achieve the preset emission reduction target. In this case, the emission reduction target should be verified first. Does it exceed the fleet's theoretical maximum emission reduction ratio? If the emissions are exceeded, the emission reduction target needs to be adjusted and the method re-executed; if the emissions are not exceeded, the enumeration range of candidate thresholds should be expanded (reduced). After relaxing the feasibility constraints on the scale of governance, the enumeration is repeated.

[0094] Step S7: Identify vehicles with NOx emission concentrations higher than the optimal threshold during the annual inspection as high-emission vehicles, and output a list of high-emission vehicles including Vehicle Identification Number (VIN), vehicle type, emission stage, NOx emission concentration during the annual inspection, and corresponding emission amount.

[0095] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0096] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that 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. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for identifying high-NOx emission diesel vehicles based on emission reduction targets, characterized in that, The method includes: Step S1: Obtain diesel vehicle annual inspection data from the three-level network system for periodic vehicle inspection, and group the vehicles in two dimensions according to vehicle type and weight class; Step S2: Based on the two-dimensional grouping and vehicle age and annual inspection cumulative mileage data, fit the cumulative mileage-vehicle age statistical model for each group to estimate the annual mileage of each vehicle in the base year. Step S3: Introduce emission stage as a third dimension based on the two-dimensional grouping. Group vehicles into three-dimensional groups according to the three-dimensional combination of emission stage, vehicle type, and weight class to obtain the total emissions of the fleet before treatment. The two-dimensional grouping is only used for fitting the cumulative mileage-vehicle age statistical model in step S2, while the three-dimensional grouping is only used for emission factor calculation. The two grouping schemes serve different calculation purposes, specifically including: In step S2, the grouping method is determined only by the combination of vehicle type and weight class. Vehicles in the same group have the same vehicle type and weight class, and the emission stage is not introduced as a grouping dimension. In step S3, the refined grouping further introduces emission stages based on the grouping in step S2. Vehicles in the same group have the same emission stage, vehicle type and weight class, ensuring the consistency of the emission stage of the reference benchmark in the emission factor mapping. The total emissions before fleet remediation were obtained, including: The median NOx emission concentration of all vehicles in each subgroup during annual inspection is used as the reference benchmark for the overall emission level of the group. A proportional mapping relationship between the NOx emission concentration of a single vehicle during annual inspection and the standard emission factor is established, and the NOx emission concentration of each vehicle during annual inspection is converted into the NOx emission factor of a single vehicle. Combined with the annual mileage, the baseline annual NOx emission of each vehicle is calculated and summed to generate the total emission of the fleet before treatment. The proportional mapping relationship is as follows: In the formula, For vehicles NOx emission factor To refine the grouping The standard NOx emission factor for the corresponding emission stage, For vehicles The annual inspection NOx emission concentration To refine the grouping The median NOx emission concentration of all vehicles inspected during the annual inspection. When a vehicle has multiple annual inspection records The maximum NOx emission concentration from multiple annual inspections is taken; when the number of valid vehicles in a certain subgroup is less than 2, the mean NOx concentration of the valid vehicles in the subgroup is used instead of the median. Step S4: Set the NOx emission reduction target percentage P% according to the regional environmental management needs, and calculate the target emission amount based on the total emissions of the fleet before treatment; Step S5: Determine the lower limit of candidate thresholds based on the feasibility constraints of the governance scale. Based on the pre-sorting of the expected emission reduction potential of each vehicle and the identification of the inflection point of the cumulative emission reduction curve, locate the key search interval, enumerate candidate NOx emission concentration thresholds within the key search interval, and calculate the total emissions of the fleet after governance. It also includes: for each candidate threshold, construct differentiated governance scenarios according to the heterogeneity of emission stages, simulate and restore the emission levels of suspected high-emission vehicles in each refined group to the emission levels of qualified vehicles that match the technical accessibility of the same emission stage, calculate the total emissions of the fleet after governance, and record the number of suspected high-emission vehicles under the corresponding threshold. Step S6: From the enumerated candidate thresholds, select candidate thresholds that ensure the total emissions after treatment are not higher than the set target emissions to form a feasible threshold set; select the threshold that minimizes the number of suspected high-emission vehicles from the feasible threshold set as the optimal threshold. Step S7: Identify vehicles with NOx emission concentrations higher than the optimal threshold during the annual inspection as high-emission vehicles, and output a list of high-emission vehicles including Vehicle Identification Number (VIN), vehicle type, emission stage, NOx emission concentration during the annual inspection, and corresponding emission amount.

2. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, Obtaining diesel vehicle annual inspection data also includes: The diesel vehicle annual inspection data includes the Vehicle Identification Number (VIN), emission stage, vehicle type, weight class, vehicle age, cumulative mileage during the annual inspection, and NOx emission concentration under the deceleration test. The acquired data is cleaned, and valid records are retained.

3. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, In step S2, the cumulative mileage-vehicle age statistical model is automatically selected from three types of models, with the one having the smallest root mean square error in the fit being the optimal model. The three types of models are as follows: The model employs a monotonically increasing constrained spline model (SCAM) with a Gamma logarithmic connection function; a log-linear model fits the logarithm of cumulative mileage with vehicle age as the independent variable; and a saturated growth nonlinear model, in the form of: ,in, This is the cumulative mileage data for annual vehicle inspection. For saturation mileage parameters, For the growth rate parameter, This refers to the vehicle's age.

4. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, Step S4 also includes: The formula for calculating the target emissions is: And the emission reduction target percentage P% is satisfied. ,in, This represents the theoretical maximum emission reduction ratio of the fleet, and the estimated emission reduction potential at the lower limit of the candidate threshold.

5. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, Locate the key search range, including: Calculate the expected emission reduction for each vehicle after treatment. The calculation formula is: ,in, This is the lower limit of the candidate threshold. To estimate the vehicle Annual mileage, based on Sort from largest to smallest; The expected emission reductions are accumulated for each vehicle in the sorted vehicle sequence, and a curve showing the change of cumulative emission reductions as the number of vehicles treated increases is constructed. Calculate the second difference of the change curve, identify the inflection point where the marginal emission reduction utility diminishes, and denot the NOx concentration value corresponding to the inflection point as... ,like The lower limit of the candidate threshold Then As the lower limit, with The upper limit of the search interval is defined as [ ], thus the key search interval is determined as [ ]. , If the NOx concentration value corresponding to the inflection point is lower than Then The key search interval is determined with a lower limit of 1500ppm and an upper limit of 1500ppm; candidate thresholds are enumerated only within the determined key search interval with a preset step size.

6. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, The total emissions of the fleet after treatment are calculated, including: Different assumptions about post-treatment emission levels are set for different emission stages; Introducing technical accessibility constraints: If the NOx emission concentration of a vehicle during its annual inspection exceeds the technical theoretical upper limit of the emission stage to which the vehicle belongs, the vehicle is judged to have a serious mechanical failure. The emission level after treatment is directly set to the standard limit of the emission stage to which the vehicle belongs, rather than taking the statistical quantile of qualified vehicles in the same detailed group. Based on the differentiation rules and technical accessibility constraints, the post-treatment emission factors of vehicles at each emission stage are calculated separately, and then summed to obtain the total post-treatment emissions of the fleet.

7. The method for identifying high-NOx emission diesel vehicles based on emission reduction targets according to claim 1, characterized in that, The uncertainty quantification of NOx concentration during annual vehicle inspections is introduced to make confidence corrections for determining the number of suspected high-emission vehicles, including: For vehicles with multiple annual inspection records, the coefficient of variation of the corresponding NOx emission concentration is calculated. Based on the coefficient of variation, the vehicles are divided into three categories: high stability, medium stability, and low stability. For high stability vehicles, the measured concentration is directly compared with the threshold. For medium stability vehicles, the measured value plus 0.5 times the standard deviation is used as the judgment value. For low stability vehicles, the measured value plus 1.0 times the standard deviation is used as the judgment value, or the vehicle is identified as a high-emission vehicle only if it exceeds the threshold in two consecutive annual inspections.

Images