A road heavy truck NOx emission calculation method based on an improved hidden Markov model X Emission calculation method

Abstract

The application relates to the field of urban road traffic emission evaluation, in particular to a road heavy truck NO X emission calculation method based on an improved hidden Markov model. The method comprises the following steps: S1. Road internal and external track identification and in-road track extraction; S2. Track segmentation based on baseline offset; S3. Track key point extraction based on offset distance statistics; S4. Candidate road network reduction based on micro-area segmentation; S5. Track matching based on a hidden Markov model; S6. Road-level vehicle operating condition calculation; S7. Road-level vehicle NO X emission calculation and summarization. The application optimizes the calculation process of a traditional hidden Markov model, ensures the track matching accuracy, and significantly reduces the calculation complexity. Compared with the prior art, the application can effectively solve the problem of low road-level emission calculation efficiency under the condition of low sampling frequency GPS data, and makes the road traffic pollution emission simulation process more efficient.

Description

Technical Field

[0001] This invention relates to the field of urban road traffic emissions assessment, specifically to a method for assessing the NO emissions of heavy-duty trucks on roads based on an improved hidden Markov model. X Emissions calculation method. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of logistics and transportation demands, the proportion of heavy-duty diesel trucks in urban road traffic systems is constantly increasing. Due to the characteristics of heavy-duty diesel trucks—large vehicle weight, high engine power, and high emission intensity—nitrogen oxides emitted by them have become a significant component of urban traffic-related air pollution, significantly impacting urban air quality improvement and public health. Therefore, conducting city-scale NOx emissions testing for heavy-duty diesel trucks is crucial. X A detailed assessment of emissions is of great significance for decision-making regarding traffic pollution prevention and management.

[0003] Existing road traffic emissions assessment methods mainly include macro-level emissions inventory methods based on statistical data and micro-level emissions estimation methods based on vehicle trajectory data. Macro-level emissions inventory methods typically rely on traffic survey data, road classifications, and empirical emissions factors to estimate emissions at the regional or road-level level, making it difficult to reflect differences in vehicle operating conditions across different road sections and resulting in low spatial resolution. In contrast, emissions assessment methods based on GPS trajectory data can characterize the actual operating conditions of vehicles, providing a data foundation for emissions estimation at the road level and even more refined scales.

[0004] In road-level emissions assessment based on GPS trajectory data, rapid and accurate matching of trajectories with road networks is a crucial prerequisite. Currently, Hidden Markov Models (HMMs) are widely used for vehicle trajectory map matching because they can comprehensively consider observation errors and road topology. However, traditional HMM map matching methods often face problems such as high computational complexity, large candidate road network size, and low matching efficiency when processing large-scale vehicle trajectory data at the city scale, easily leading to unstable matching or insufficient computational efficiency.

[0005] Furthermore, existing methods for calculating city-level road-level emission distribution based on large-scale GPS trajectory data still fall short in terms of computational efficiency, making it difficult to meet the actual needs of refined traffic emission assessment and dynamic management.

[0006] Therefore, there is an urgent need for a method that can improve the efficiency and stability of trajectory road matching under low-sampling-frequency GPS trajectory data of heavy-duty diesel trucks, and achieve city-scale road-level NO X Methods for rapid and detailed emission calculations to support urban traffic pollution emission assessment and management decisions. Summary of the Invention

[0007] The technical problem to be solved by this invention is to provide a road heavy-duty truck NO model based on an improved hidden Markov model. X The emission calculation method can quickly and accurately generate NOx emissions for heavy-duty diesel trucks on urban roads based on GPS trajectory data streams. X Emissions distribution.

[0008] Technical solution

[0009] A road heavy-duty truck NO based on an improved hidden Markov model X The emission calculation method includes the following steps:

[0010] S1. Based on the original GPS trajectory data of heavy-duty diesel trucks, the effective on-road trajectory is extracted through on-road and off-road trajectory identification;

[0011] S2. Based on the baseline offset criterion, the effective on-road trajectory is segmented so that the headings of continuous GPS points within the sub-trajectory are consistent with the baseline.

[0012] S3. Extract key motion nodes for each sub-trajectory to achieve adaptive simplification of GPS trajectory data;

[0013] S4. Based on the micro-region stitching method, the candidate road network is reduced to construct a simplified road network for the Hidden Markov Model;

[0014] S5. Based on the Hidden Markov Model, the simplified trajectory and simplified road network are matched to reconstruct the vehicle's driving path;

[0015] S6. Calculate the vehicle's operating conditions at the road level using the trajectory matching results;

[0016] S7. Calculate the road-level emissions of a single vehicle by combining vehicle attributes and road-level operating conditions, and aggregate them to form the city-wide road-level NOx emissions. X Emissions distribution.

[0017] In step S1, the original GPS trajectory sequence O = { for heavy-duty diesel trucks in the city is collected. o 1, o 2,…, o T For each GPS trajectory point in the trajectory sequence, calculate its relationship with the road network. The minimum spatial distance between them. When the minimum distance is less than a preset distance threshold. γ If the condition is met, the trajectory point is determined to be an on-path trajectory point; otherwise, it is determined to be an off-path trajectory point. That is, the on-path trajectory sequence should satisfy:

[0018] (15)

[0019] in, Represents GPS track points o i With road network The minimum spatial distance between them; γ The distance threshold used to distinguish between points on the road trajectory and points away from the road trajectory. is the length of the trajectory point sequence.

[0020] Given the low sampling frequency of GPS data for heavy-duty diesel trucks, this invention does not employ an anomaly detection method based on abrupt changes in heading angle. When sporadic off-road trajectory points appear in a continuous on-road trajectory, i.e., if the time interval between such a trajectory point and its adjacent on-road trajectory points does not exceed a preset time threshold, it is considered a location anomaly caused by GPS positioning error and is removed, thereby obtaining a valid on-road trajectory sequence. .

[0021] In step S2, the effective on-the-road trajectory sequence obtained in step S1 is used. Using the third trajectory point in the trajectory sequence as input, the trajectory segmentation points are identified by traversing the sequence. The specific traversal process is as follows: starting from the starting point of the current sub-trajectory sequence... and the finish line The line connecting the initial sub-trajectory sequence (starting from the origin and ending at the third trajectory point of the input trajectory sequence) serves as the baseline. Calculate the vertical distance from each trajectory point in the sub-trajectory to the baseline. When the offset distance of any trajectory point from the baseline exceeds a preset threshold... δ When the time comes, the trajectory point is identified as a new segment point and updated as the new sub-trajectory start point to continue traversal. The trajectory start point, end point, and segment points identified during the traversal together constitute the sub-trajectory boundary point set. That is, the boundary points should satisfy:

[0022] (16)

[0023] in, δ This is a preset distance threshold; and These represent the starting and ending points of the sub-trajectory, respectively. Indicates the starting point and the end point The baseline formed; Represents trajectory points o i With baseline The baseline offset distance between the two points is defined as the distance between the trajectory points. o i To the baseline The vertical distance.

[0024] In step S3, for each sub-trajectory segment obtained in step S2, starting from the point of that sub-trajectory segment... and the end point As the endpoints of the baseline, the direction of the baseline is determined. For intermediate trajectory points within a sub-trajectory segment, their offset angle relative to the baseline is calculated, and combined with the geographical distance between adjacent trajectory points, the offset distance of the trajectory point is obtained. The calculation formula is as follows:

[0025] (17)

[0026] in, Represents trajectory points Relative to the baseline The offset distance; Indicates the geographical distance between adjacent GPS observation points; Indicates baseline Direction and trajectory points to The angular deviation between the directions of motion is calculated. For each sub-track segment, the average offset distance of all its track points is calculated. When the offset distance of a track point is greater than the average offset distance of the corresponding sub-track segment, that track point is identified as a critical motion point. The start and end points of each sub-track segment and the identified critical motion points are retained to form a simplified on-road trajectory sequence. Each trajectory point satisfies:

[0027] (18)

[0028] in, This represents the average offset distance of all trajectory points within the sub-trajectory defined by the baseline.

[0029] In step S4, for each sub-trajectory segment obtained in step S2, a micro-region corresponding to that sub-trajectory segment is constructed to define the range of candidate roads the vehicle may travel on. The width of the micro-region is set perpendicular to the sub-trajectory direction, and its size is determined by the maximum possible lateral displacement between adjacent trajectory points. The lateral displacement is determined by the vehicle's maximum speed, the GPS sampling time interval, and preset extended parameters. ε Common constraints, i.e.

[0030] (19)

[0031] in, Indicates the relationship with the first i The width of the micro-region corresponding to each sub-trajectory; Indicates the first i The maximum speed between adjacent trajectory points within a sub-trajectory; Δ t Indicates the sampling time interval for GPS trajectory data; ε This represents the extended margin introduced to ensure that there is a feasible path between adjacent trajectory points.

[0032] The length of the micro-region is set along the direction of the sub-trajectory, and its size is determined by the distance between the start and end points of the sub-trajectory segment, combined with the expansion parameter. ε Determined, that is:

[0033] (20)

[0034] in, Indicates the relationship with the first i The length of the micro-region corresponding to each sub-trajectory; This indicates that along the direction of the sub-trajectory, from the starting point of the sub-trajectory... and the end point The distance between the two endpoints.

[0035] Based on this, directional constraints are applied to the initial road network, ensuring that the angle between the direction of a road segment and the direction of a sub-trajectory segment does not exceed a preset maximum allowable angle threshold. The road segments are then analyzed. By merging the candidate road sets corresponding to each sub-trajectory segment, a reduced candidate road network is obtained. ,Right now:

[0036] (twenty one)

[0037] in, Represents the initial road network; e Representing the simplified road network The road section in the middle; Indicates the relationship with the first i The micro-regions corresponding to each sub-trajectory; N Indicates the total number of sub-trajectories; Indicates the direction of road segment e relative to the sub-trajectory baseline. l m,n angular deviation between directions; This indicates the maximum permissible angular deviation threshold.

[0038] In step S5, the simplified trajectory sequence obtained in step S3 is... and the candidate road network obtained in step S4 Construct a Hidden Markov Model (using existing techniques), which includes an initial state probability vector. State transition probability matrix and observation probability matrix The Viterbi algorithm is used to solve the hidden Markov model to obtain the hidden state sequence that maximizes the posterior probability. This hidden state sequence corresponds to the optimal matching path of the vehicle on the road network. ,Right now:

[0039] (twenty two)

[0040] In step S6, the complete on-road trajectory sequence obtained in step S1 is used. and the optimal matching path obtained in step S5 To obtain the projection sequence of the trajectory sequence on the matching path, i.e.:

[0041] (twenty three)

[0042] in, This represents the projected trajectory sequence, where Proj(·) indicates orthogonally projecting the trajectory points onto the matching road path. The projection operator on the road. After trajectory projection is completed, different road segments may correspond to different numbers of projected trajectory points. For any road segment... Define the corresponding set of projection point indices:

[0043] (twenty four)

[0044] in, Indicates road section The projection point index, Represents the projected trajectory points. Indicates location on road section The set of projection point indices on the , whose cardinality is denoted as . , M It is the length of the optimal matching path sequence.

[0045] When the road section When there are at least two projection points on the road segment, idling behavior is identified; when there are fewer than two projection points on the road segment, the operating conditions of the road segment are estimated based on the boundary projection points of adjacent road segments.

[0046] For road segments with fewer than two projection points, select the start and end boundary points for operating condition calculations:

[0047] (25)

[0048] in, This represents the last projection point of the preceding adjacent road segment. This represents the first projection point of the next adjacent road segment.

[0049] For road segments with at least two projection points, the idling trajectory subsequence is identified based on the spatial distance between adjacent projection points. Specifically, when the distance between adjacent projection points is less than a preset distance threshold, the corresponding trajectory segment is determined to be in an idling state; otherwise, it is determined to be in a normal driving state. The trajectory sequence can be identified as follows:

[0050] (26)

[0051] in, and These represent the idling subsequence and the driving subsequence identified on the road segment, respectively. Indicates the spatial distance between adjacent projected trajectory points; This is the preset distance threshold.

[0052] Calculate the segmented average driving speed sequence of road segments :

[0053] (27)

[0054] in, D (·) and Δ t (·) represent the distance traveled along the matched road trajectory and the corresponding time interval, respectively; This indicates the idle speed value.

[0055] In step S7, the segmented average driving speed sequence obtained in step S6 is used. In addition, by combining vehicle attribute information and speed-related emission factors of corresponding emission standards, the emissions of the road segment are calculated:

[0056] hour:

[0057]

[0058] hour:

[0059] (28)

[0060] in, Indicates the emissions of a road segment. Indicates in emission standards x (Including but not limited to speed-related emission factors under China III, China IV, China V, and China VI emission standards.) It should be noted that when the continuous idling duration exceeds a preset time threshold... When the engine is off, its emissions are set to zero.

[0061] The emissions corresponding to the trajectories of all vehicles passing through the same road segment within the same time interval are summed to obtain the total emissions of that road segment within the time interval; further, the emissions of all road segments within the city area are summarized to form the city-scale road-level vehicle emission distribution results.

[0062] The method of this invention has the following beneficial effects:

[0063] Significantly improves trajectory matching and emission calculation efficiency: This invention improves the traditional Hidden Markov Model by extracting key trajectory points and reducing road network micro-regions. While ensuring trajectory matching accuracy, it effectively reduces computational complexity and improves processing efficiency under large-scale GPS trajectory data conditions, thereby supporting city-scale road-level NO X Rapid calculation of emission distribution.

[0064] Enhancing the applicability of low-sampling-frequency trajectory data: In view of the characteristics of low sampling frequency and sparse trajectory points of GPS data for heavy-duty diesel trucks, this invention adopts a differentiated processing strategy of simplified trajectory and original on-road trajectory in the trajectory matching and operating condition calculation processes, thereby improving the stability of road-level operating condition identification and the reliability of emission estimation.

[0065] Enhancing the decision support capabilities for urban traffic pollution assessment and management: This invention can rapidly generate road-level NO data with high spatial resolution. X Emission distribution provides reliable data support for traffic pollution source identification, key road section control, traffic organization optimization, and environmental impact assessment, which helps to serve the coordinated governance of urban traffic and environment. Attached Figure Description

[0066] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0067] Figure 1 This is a flowchart of the method of the present invention;

[0068] Figure 2 A simplified comparison diagram of the trajectory sequences before and after implementation in this embodiment of the invention;

[0069] Figure 3 This is a simplified comparison diagram of the road network before and after implementation of the present invention.

[0070] Figure 4 For the embodiment of the present invention, the hourly urban road level heavy-duty diesel truck NO X Emissions distribution diagram;

[0071] Figure 5 This invention relates to the road NO obtained based on the traditional Hidden Markov algorithm. X Emissions distribution comparison chart. Detailed Implementation

[0072] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0073] like Figure 1 As shown, a road heavy-duty truck NO based on an improved hidden Markov model X The emission calculation method includes the following steps:

[0074] S1. Based on the original GPS trajectory data of heavy-duty diesel trucks, the effective on-road trajectory is extracted through on-road and off-road trajectory identification;

[0075] S2. The on-road trajectory is segmented based on the baseline offset criterion to ensure that the headings of continuous GPS points within the sub-trajectory are consistent with the baseline.

[0076] S3. Extract key motion nodes for each sub-trajectory to achieve adaptive simplification of GPS trajectory data;

[0077] S4. Based on the micro-region stitching method, the candidate road network is reduced to construct a simplified road network for the Hidden Markov Model;

[0078] S5. Based on the Hidden Markov Model, the simplified trajectory and simplified road network are matched to reconstruct the vehicle's driving path;

[0079] S6. Calculate the vehicle's operating conditions at the road level using the trajectory matching results;

[0080] S7. Calculate the road-level emissions of a single vehicle by combining vehicle attributes and road-level operating conditions, and aggregate them to form the city-wide road-level NOx emissions. X Emissions distribution.

[0081] This invention uses a one-hour GPS trajectory data stream of heavy-duty diesel trucks collected in Shanghai on a certain day as an example. It relies on the road heavy-duty truck NO data stream based on the improved Hidden Markov Model proposed in this invention. X Emissions calculation methods to quickly generate city road-level NOx levels that can be used for environmental assessments and policy guidance. X The specific implementation steps for emission distribution are as follows:

[0082] S1. Road Trajectory Identification and On-Road Trajectory Extraction: Using hourly-scale GPS trajectory data of heavy-duty diesel trucks in Shanghai as input, the minimum spatial distance between trajectory points and the road network is calculated, and it is determined whether this distance is less than a preset distance threshold γ (100 m in this embodiment) to distinguish on-road trajectory points from off-road trajectory points. For road edge trajectory points that occasionally appear in continuous on-road trajectories, if the time interval between the trajectory point and its adjacent on-road trajectory points does not exceed a preset time threshold (1 min in this embodiment), it is identified as a location anomaly caused by GPS positioning error and is removed, thereby obtaining a refined on-road trajectory sequence, as shown in Table 1, where POINT() represents latitude and longitude coordinates.

[0083] Table 1. Refined vehicle trajectory sequences (partial list)

[0084]

[0085] S2. Trajectory Segmentation Based on Baseline Offset: Using the on-road trajectory sequence obtained in step S1 as input, traverse starting from the third trajectory point to identify trajectory segmentation points. During the traversal, the starting point of the current sub-trajectory is used as the reference point. o m and the finish line o n The line formed serves as a baseline. l m,n Calculate the vertical distance from each trajectory point in the sub-trajectory to the baseline. When the offset distance of any trajectory point from the baseline exceeds a preset threshold... δ When the time (10 m in this embodiment) is reached, the trajectory point is determined as a new segmentation point. The trajectory start point, the trajectory end point, and the segmentation points identified during the traversal process together constitute the sub-trajectory boundary point set P. seg As shown in Table 2.

[0086] Table 2. Set of boundary points for sub-trajectories (partial list)

[0087]

[0088] S3. Trajectory key point extraction based on offset distance statistics: For each sub-trajectory segment obtained in step S2, extract key points from the starting point of that sub-trajectory segment. o m and the end point o nThe baseline endpoints are used to determine the baseline direction. For intermediate trajectory points within a sub-trajectory segment, their offset angle relative to the baseline is calculated, and combined with the geographical distance between adjacent trajectory points, the offset distance of the trajectory point is obtained. For each sub-trajectory segment, the average offset distance of all trajectory points is calculated. When the offset distance of a trajectory point is greater than the average offset distance of the corresponding sub-trajectory segment, that trajectory point is identified as a critical motion point. The start and end points of each sub-trajectory segment and the identified critical motion points are retained to form a simplified on-road trajectory sequence, as shown in Table 3. Comparison of trajectory sequences before and after simplification: Figure 2 As shown in the figure, the horizontal axis represents longitude and the vertical axis represents latitude.

[0089] Table 3 Simplified on-road trajectory sequences (partial list)

[0090]

[0091] S4. Candidate Road Network Reduction Based on Micro-Region Segmentation: For each sub-trajectory segment obtained in step S2, a micro-region corresponding to that sub-trajectory segment is constructed to limit the range of candidate roads that the vehicle may travel on. The width of the micro-region is set perpendicular to the sub-trajectory direction, and its size is determined by the maximum possible lateral displacement between adjacent trajectory points. This lateral displacement is determined by the vehicle's maximum speed, the GPS sampling time interval, and preset extended parameters. ε Common constraints. The length of the micro-region is set along the sub-trajectory direction, and its size is determined by the distance between the start and end points of the sub-trajectory segment and the expansion parameter ε. Based on this, directional constraints are applied to the road network, ensuring that the angle between the road segment direction and the sub-trajectory segment direction does not exceed a preset maximum allowable angle threshold. (In this embodiment, a road segment of 90° is used.) By merging the candidate road sets corresponding to each sub-trajectory segment, a reduced candidate road network is obtained. ,like Figure 3 As shown in the figure, the horizontal axis represents longitude and the vertical axis represents latitude.

[0092] S5. Trajectory matching based on Hidden Markov Model: For the simplified trajectory sequence obtained in step S3 and the candidate road network obtained in step S4, a Hidden Markov Model is constructed, including the initial state probability vector. State transition probability Matrix and observation probability matrix The Viterbi algorithm is used to solve the hidden Markov model to obtain the hidden state sequence that maximizes the posterior probability. This hidden state sequence corresponds to the optimal matching path of the vehicle on the road network. As shown in Table 4, LINESTRING() represents the path line shape.

[0093] Table 4. Optimal matching path sequence (partial list)

[0094]

[0095] S6. Road-level vehicle operating condition calculation: Using the complete on-road trajectory sequence obtained in step S1 and the optimal matching path obtained in step S5, the projection sequence corresponding to the trajectory sequence on the matching path is obtained. Then, the operating condition of the road segment is calculated based on the number of projection points on each matching road segment. Specifically: when the number of projection points on the road segment is not less than two, idling behavior identification is performed on the road segment; when the number of projection points on the road segment is less than two, the operating condition of the road segment is estimated based on the boundary projection points of adjacent road segments. For road segments with fewer than two projection points, the start and end boundary points for operating condition calculation are selected:

[0096] (29)

[0097] For road segments with at least two projection points, the idling trajectory subsequence is identified based on the spatial distance between adjacent projection points. Specifically, when the distance between adjacent projection points is less than a preset distance threshold (50 m in this embodiment), the corresponding trajectory segment is determined to be in an idling state; otherwise, it is determined to be in a normal driving state. The trajectory sequence can be identified as follows:

[0098] (30)

[0099] Based on the above settings, the segmented average driving speed sequence of road segments can be calculated. :

[0100] (31)

[0101] S7. Road-class vehicles NO X Emissions calculation and summarization, obtaining city-wide road-level NO. X Emissions distribution: using the segmented average driving speed sequence obtained in step S6 In addition, by combining vehicle attribute information and speed-related emission factors of corresponding emission standards, the emissions of the road segment are calculated:

[0102] hour:

[0103]

[0104] hour:

[0105] (32)

[0106] It should be noted that when the continuous idling duration exceeds the preset time threshold... When the time (12 minutes in this embodiment) is taken, it is determined that the engine is in a shutdown state, and the corresponding emission is set to zero.

[0107] The emissions corresponding to the trajectories of all vehicles passing through the same road segment within the same time interval are summed to obtain the total emissions for that road segment within the time interval. Furthermore, the emissions from all road segments within the city area are aggregated to form a city-scale road-level vehicle emission distribution result, such as... Figure 4 As shown in the figure, the horizontal axis represents longitude and the vertical axis represents latitude.

[0108] Meanwhile, under the same computer equipment conditions, the traditional Hidden Markov Map (HMM) algorithm (Newson, P., & Krumm, J., 2009, Hidden Markov map matching through noise and sparseness, ACMSIGSPATIAL) was used to calculate the city-scale road-level vehicle emission distribution for the corresponding hour, and the calculation time of the two methods was statistically analyzed. The results show that the calculation time for obtaining the current city-scale road-level vehicle emission distribution based on GPS trajectory data using the present invention is 26 minutes, while the calculation time using the traditional HMM algorithm is 2 hours and 6 minutes, representing an efficiency improvement of approximately 79%, or about 4.8 times. Furthermore, as shown in Figure 5, compared with the city-scale road-level vehicle emission distribution results obtained by the traditional HMM algorithm, the present invention significantly improves calculation efficiency while maintaining high calculation accuracy.

[0109] The above description is merely a description of preferred embodiments of this application and is not intended to limit the scope of this application in any way. Any changes or modifications made by those skilled in the art based on the above-disclosed technical content should be considered as equivalent and valid embodiments and fall within the scope of protection of the technical solution of this application.

Claims

1. A road heavy-duty truck NO based on an improved hidden Markov model X The emission calculation method is characterized by, Includes the following steps: S1. Based on the original GPS trajectory data of heavy-duty diesel trucks, extract the valid on-road trajectories through on-road and off-road trajectory identification. ; S2. Based on the baseline offset criterion, the effective on-road trajectory is segmented so that the heading of continuous GPS points within the sub-trajectory is consistent with the baseline direction; S3. Extract key motion nodes for each sub-trajectory to achieve adaptive simplification of GPS trajectory data; S4. Based on the micro-region stitching method, the candidate road network is reduced to construct a simplified road network for the Hidden Markov Model; S5. Based on the Hidden Markov Model, the simplified trajectory and simplified road network are matched to reconstruct the vehicle's driving path; S6. Calculate the vehicle's operating conditions at the road level using the trajectory matching results; S7. Calculate the road-level emissions of a single vehicle by combining vehicle attributes and road-level operating conditions, and aggregate them to form the city-wide road-level NOx emissions. X Emissions distribution; Step S4 specifically involves, For each sub-trajectory segment obtained in step S2, a micro-region corresponding to that sub-trajectory segment is constructed to define the candidate road range for vehicle travel; wherein, the width of the micro-region is set perpendicular to the sub-trajectory direction, and its size is determined by the maximum lateral displacement between adjacent trajectory points; the lateral displacement is determined by the vehicle's maximum driving speed, the GPS sampling time interval, and preset extended parameters. ε Common constraints, i.e. in, Indicates the relationship with the first i The width of the micro-region corresponding to each sub-trajectory; Indicates the first i The maximum speed between adjacent trajectory points within a sub-trajectory; Δ t Indicates the sampling time interval for GPS trajectory data; ε This represents the extended margin introduced to ensure that there is a feasible path between adjacent trajectory points; The length of the micro-region is set along the direction of the sub-trajectory, and its size is determined by the distance between the start and end points of the sub-trajectory segment, combined with the expansion parameter. ε Determined, that is: in, Indicates the relationship with the first i The length of the micro-region corresponding to each sub-trajectory; This indicates that along the direction of the sub-trajectory, from the starting point of the sub-trajectory... and the end point The distance between the two endpoints; Based on this, directional constraints are applied to the initial road network, ensuring that the angle between the direction of a road segment and the direction of a sub-trajectory segment does not exceed a preset maximum allowable angle threshold. The road segments are analyzed; by merging the candidate road sets corresponding to each sub-trajectory segment, a reduced candidate road network is obtained. ,Right now: in, Represents the initial road network; e Representing the simplified road network The road section in the middle; Indicates the relationship with the first i The micro-regions corresponding to each sub-trajectory; N Indicates the total number of sub-trajectories; Indicates the direction of road segment e relative to the sub-trajectory baseline. l m,n angular deviation between directions; This indicates the maximum permissible angular deviation threshold. Step S6 specifically involves: The effective on-road trajectory sequence obtained in step S1 And the optimal matching path S obtained in step S5 * To obtain the projection sequence of the trajectory sequence on the matching path, i.e.: in, This represents the projected trajectory sequence, where Proj(·) indicates orthogonally projecting the trajectory points onto the matching road path. Projection operator on; For any road segment Define the corresponding set of projection point indices: in, Indicates road section The projection point index, Represents the projected trajectory points. Indicates location on road section The set of projection point indices on the , whose cardinality is denoted as . , M It is the length of the optimal matching path sequence; When the road section When there are at least two projection points on the road segment, idling behavior identification is performed on the road segment; when there are fewer than two projection points on the road segment, the operating conditions of the road segment are estimated based on the boundary projection points of adjacent road segments. For road segments with fewer than two projection points, select the start and end boundary points for operating condition calculations: in, This represents the last projection point of the preceding adjacent road segment. This represents the first projection point of the next adjacent road segment; For road segments with at least two projection points, the idling trajectory subsequence is identified based on the spatial distance between adjacent projection points: when the distance between adjacent projection points is less than a preset distance threshold, the corresponding trajectory segment is determined to be in an idling state; otherwise, it is determined to be in a normal driving state, and the trajectory sequence is identified as follows: in, and These represent the idling subsequence and the driving subsequence identified on the road segment, respectively. Indicates the spatial distance between adjacent projected trajectory points; This is a preset distance threshold; Calculate the segmented average driving speed sequence of road segments : in, D (·) and Δ t (·) represent the distance traveled along the matched road trajectory and the corresponding time interval, respectively; This indicates the idle speed value.

2. The road heavy-duty truck NO based on the improved hidden Markov model as described in claim 1 X The emission calculation method is characterized by, Step S1 specifically involves: Collect raw GPS trajectory sequences O = { o 1, o 2,…, o T For each GPS trajectory point in the trajectory sequence, calculate its relationship with the road network. The minimum spatial distance between them; when the minimum distance is less than a preset distance threshold γ If the trajectory point is on the road, it is determined as an on-road trajectory point; otherwise, it is determined as an off-road trajectory point. That is, the path trajectory sequence should satisfy: in, Represents GPS track points o i With road network The minimum spatial distance between them; γ The distance threshold used to distinguish between points on the road trajectory and points away from the road trajectory. The length of the trajectory point sequence; By eliminating location anomalies caused by GPS positioning errors, a valid on-road trajectory sequence can be obtained. .

3. The road heavy-duty truck NO based on the improved hidden Markov model as described in claim 2 X The emission calculation method is characterized by, If the time interval between a trajectory point and its adjacent trajectory points in a continuous on-road trajectory does not exceed a preset time threshold, it is considered a location anomaly caused by GPS positioning error.

4. The road heavy-duty truck NO based on the improved hidden Markov model as described in claim 1 X The emission calculation method is characterized by, Step S2 specifically involves: The effective on-road trajectory sequence obtained in step S1 As input, starting from the third trajectory point in the trajectory sequence, traverse and identify trajectory segment points; the traversal process is as follows: Starting from the current sub-trajectory sequence and the finish line The line formed serves as a baseline. Calculate the vertical distance from each trajectory point in the sub-trajectory to the baseline; when the offset distance of any trajectory point from the baseline exceeds a preset threshold... δ When the time comes, the trajectory point is identified as a new segment point and updated as the new sub-trajectory start point to continue traversal; the trajectory start point, end point, and segment points identified during the traversal together constitute the sub-trajectory boundary point set. That is, the boundary points should satisfy: in, δ This is a preset distance threshold; and These represent the starting and ending points of the sub-trajectory, respectively. Indicates the starting point and the end point The baseline formed; Represents trajectory points o i With baseline The baseline offset distance between the two points is defined as the distance between the trajectory points. o i To the baseline The vertical distance.

5. The road heavy-duty truck NO based on the improved hidden Markov model as described in claim 1 X The emission calculation method is characterized by, Step S3 specifically involves: For each sub-trajectory segment obtained in step S2, starting from the point of that sub-trajectory segment... and the end point As the endpoints of the baseline, the direction of the baseline is determined; for intermediate trajectory points within a sub-trajectory segment, their offset angle relative to the baseline is calculated, and combined with the geographical distance between adjacent trajectory points, the offset distance of the trajectory point is obtained. The calculation formula is as follows: in, Represents trajectory points Relative to the baseline The offset distance; Indicates the geographical distance between adjacent GPS observation points; Indicates baseline Direction and trajectory points to The angular deviation between the directions of motion; For each sub-trajectory segment, calculate the average offset distance of all trajectory points; when the offset distance of a trajectory point is greater than the average offset distance of the corresponding sub-trajectory segment, the trajectory point is identified as a critical motion point; retain the start and end points of each sub-trajectory segment and the identified critical motion points to form a simplified on-road trajectory sequence. Each trajectory point satisfies: in, This represents the average offset distance of all trajectory points within the sub-trajectory defined by the baseline.

6. The road heavy-duty truck NO based on the improved hidden Markov model according to claim 1 X The emission calculation method is characterized by, Step S5 specifically involves: Based on the simplified trajectory sequence obtained in step S3 and the candidate road network obtained in step S4 Construct a Hidden Markov Model, which includes an initial state probability vector. State transition probability matrix and observation probability matrix The Viterbi algorithm is used to solve the hidden Markov model to obtain the hidden state sequence that maximizes the posterior probability. This hidden state sequence corresponds to the optimal matching path of the vehicle on the road network. ,Right now: 。 7. The road heavy-duty truck NO based on the improved hidden Markov model according to claim 1 X The emission calculation method is characterized by, Step S7 specifically involves, The segmented average driving speed sequence obtained in step S6 In addition, by combining vehicle attribute information and speed-related emission factors of corresponding emission standards, the emissions of the road segment are calculated: hour: hour: in, Indicates the emissions of a road segment. Indicates in emission standards x Speed-related emission factors; when the continuous idling duration exceeds a preset time threshold. When the engine is off, its corresponding emissions are set to zero. The emissions corresponding to the trajectories of all vehicles passing through the same road segment within the same time interval are summed to obtain the total emissions of that road segment within the time interval; the emissions of all road segments within the city area are summarized to form the city-scale road-level vehicle emission distribution results.

Images