A method for calculating carbon emissions of a vehicle based on sparse trajectory data reconstruction

Abstract

The present application relates to a kind of vehicle carbon emission calculation method based on sparse trajectory data reconstruction, which uses the trajectory data of light fuel vehicle driving on ordinary urban road, analyzes the acceleration distribution characteristics of trajectory segment under different sampling frequency, reconstructs sparse trajectory data second by second according to the acceleration distribution characteristics, and then calculates the emission based on the reconstructed trajectory data through carbon emission estimation model.Compared with prior art, the present application avoids the problem that sparse trajectory data is difficult to input into emission estimation model, and overcomes the problem that existing sparse trajectory interpolation reconstruction method will seriously underestimate emission when applied to carbon emission estimation.The present application is suitable for trajectory data of various sampling frequencies, and reconstructs trajectory second by second based on acceleration distribution interpolation as the core, which can effectively improve the accuracy of carbon emission calculation, provides a new perspective for vehicle carbon emission calculation based on sparse trajectory, and has good innovation.

Description

Technical Field

[0001] This invention relates to the field of traffic emission quantification estimation, and in particular to a method for calculating automobile carbon emissions based on sparse trajectory data reconstruction. Background Technology

[0002] Carbon dioxide in vehicle exhaust is a significant contributor to the greenhouse effect, and the prerequisite for formulating emission control measures is the ability to quantify emissions. Existing microscopic, single-vehicle-level carbon emission estimation models generally require second-by-second data to meet model input requirements. This includes two main categories: second-by-second trajectory data (such as speed and acceleration) or second-by-second vehicle engine internal operating data (such as power, RPM, and air-fuel ratio). In recent years, due to the rapid coverage and widespread adoption of GPS, trajectory data has become ideal for applications in the transportation sector, facilitating carbon emission estimation as it is easier to collect than vehicle internal operating parameters. However, currently used GPS devices do not collect second-by-second data; rather, they upload data at regular time intervals, forming sparse vehicle trajectory records. This presents challenges for inputting into emission estimation models for carbon emission estimation. Therefore, proposing a trajectory reconstruction method that meets the requirements of carbon emission estimation for sparse trajectory data is of great significance.

[0003] To adapt non-second-by-second sparse trajectory data to the input requirements of emission estimation models, an interpolation method is needed to estimate the data for completing unknown trajectory points. Existing sparse trajectory reconstruction methods for emission estimation typically assume that acceleration is fixed or linearly variable. Common techniques include linear interpolation or cubic spline interpolation, which uses stable and smooth interpolation based on the velocities of two known trajectory points with a certain time interval to fill in the velocities of the unknown trajectory points in between. Trajectories reconstructed using this type of interpolation cannot reflect the randomness of vehicle acceleration and deceleration behavior on real roads, especially for vehicles traveling on ordinary urban roads, which are subject to intermittent flow caused by external factors such as traffic signals. Their acceleration and deceleration behavior is more random and unstable. Calculating emissions based on stable trajectory data constructed using existing techniques often leads to a significant underestimation of emissions. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a method for calculating automobile carbon emissions based on sparse trajectory data reconstruction.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for calculating automobile carbon emissions based on sparse trajectory data reconstruction includes the following steps:

[0007] S1. Collect multiple segments of known second-by-second trajectory data;

[0008] S2. Collect the sparse trajectory data to be reconstructed, determine the sampling frequency of the sparse trajectory data, and cut the second-by-second trajectory data based on the sampling frequency of the sparse trajectory data to obtain multiple trajectory segments.

[0009] S3. Classify and bin the trajectory segments to obtain multiple segment categories and segment bins;

[0010] S4. Fit the acceleration probability density distribution function to the trajectory segments in each segment box to obtain the acceleration probability density distribution function in each segment box;

[0011] S5. The sparse trajectory data to be reconstructed is cut according to the sampling frequency to obtain multiple sparse trajectory segments;

[0012] S6. Determine the fragment category and fragment bin of the sparse trajectory fragment;

[0013] S7. For each sparse trajectory segment, retrieve the corresponding acceleration probability density distribution function calculated by the segment box based on the second-by-second trajectory segment, and generate acceleration estimates point by point from the start point to the end point of the sparse trajectory segment.

[0014] S8. Based on the known true value of the starting point velocity and the estimated value of the point-by-point acceleration of the sparse trajectory segment, calculate the estimated value of the point-by-point velocity to obtain the reconstructed trajectory segment.

[0015] S9. Determine whether the data of the reconstructed trajectory segment is valid. If it is valid, return to step S6 to reconstruct the next sparse trajectory segment. Otherwise, return to step S7 until all sparse trajectory segments contained in the entire sparse trajectory data to be reconstructed have been reconstructed.

[0016] S10. Substitute all reconstructed trajectory segments into the carbon emission estimation model to calculate the vehicle carbon emissions determined by the sparse trajectory data.

[0017] Furthermore, step S3 includes the following sub-steps:

[0018] The formula for calculating the standard deviation σ of the velocity for each trajectory segment is as follows:

[0019]

[0020] Where v i Let be the velocity of the i-th point in the trajectory segment. is the average velocity of the trajectory segment, and r is the sampling frequency of the sparse trajectory data;

[0021] Let the velocity of the starting point of the trajectory segment be v1, and the velocity of the ending point be v.r+1 Extract all trajectory segments with a standard deviation less than 1.5 m / s, and calculate the velocity difference v between the beginning and end of each trajectory segment. r+1 -v1 sets the maximum difference between the first and last velocities of all trajectory segments as the upper bound value v. r,max The minimum value is set as the state boundary value v. r,min ;

[0022] For all original trajectory segments, state classification and binning are performed according to the following principles:

[0023] S301, if v r+1 -v1≥v r,max If the current trajectory segment is an acceleration segment, it is defined as the UA state, based on the initial velocity v1 and the velocity difference between the beginning and end v. r+1 The range of -v1 further bins all trajectory segments belonging to the UA state;

[0024] S302, if v r+1 -v1≤v r,min If the current trajectory segment is a deceleration segment, it is defined as state UD, based on the initial velocity v1 and the difference between the first and last velocities v. r+1 The range of -v1 further bins all trajectory segments belonging to the UD state;

[0025] S303, if v r,min <v r+1 -v1<v r,max If the current trajectory segment is relatively stable, it is defined as the SS state, based on the initial velocity v1 and the velocity difference between the beginning and end v. r+1 The range of -v1 further bins all trajectory segments belonging to the SS state.

[0026] Further, in step S4, the probability density distribution function includes normal distribution, exponential distribution, Pareto distribution, Dweibull distribution, T distribution, generalized extreme value distribution, gamma distribution, log-normal distribution, Beta distribution, uniform distribution, and log-gamma distribution.

[0027] Furthermore, in step S4, the root mean square error (RMSE) and the goodness of fit (R²) are used. 2 Determine the fitting effect of the acceleration probability density distribution function and select the acceleration probability density distribution function with the best fitting effect;

[0028] The root mean square error (RMSE) and goodness of fit (R) 2 The calculation formula is:

[0029]

[0030]

[0031] Among them, a i For the true value of acceleration, This is an estimate of the acceleration. The acceleration is the true average value, and m is the sample size.

[0032] Further, step S6 includes the following sub-steps:

[0033] Let the velocity of the starting point of the sparse trajectory segment be v1', and the velocity of the ending point be v. r+1 ';

[0034] S601, if v r+1 '-v1'≥v r,max If the current sparse trajectory segment belongs to state UA, then based on the initial velocity v1' and the velocity difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the UA state in step S301, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0035] S602, if v r+1 '-v1'≤v r,min If the current sparse trajectory segment belongs to state UD, then based on the initial velocity v1' and the difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the UD state in step S302, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0036] S603, if v r,min <v r+1 '-v1' < v r,max If the current sparse trajectory segment belongs to the SS state, then based on the initial velocity v1' and the velocity difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the SS state in step S303, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0037] Furthermore, in step S8, the known data in the sparse trajectory segment are the true velocity value of the starting point, the true velocity value of the ending point, and the acceleration estimate of all points;

[0038] The method for calculating the velocity estimate of each point other than the starting point is as follows: starting with the actual velocity value of the starting point, the velocity estimate of the current point is obtained by sequentially adding the acceleration estimate of the previous point and the acceleration estimate of the current point.

[0039] After the velocity estimates of all points except the starting point are calculated, a reconstructed trajectory segment is obtained. The end point of the reconstructed trajectory segment has a true velocity value and a velocity estimate.

[0040] Furthermore, in step S9, determining whether the data of the reconstructed trajectory segment is valid includes the following steps:

[0041] S901, Discriminate the point-by-point velocity estimate obtained in step S8 Is it within a reasonable range [0, V]? max [Inside, where V] max This refers to the actual speed limit on urban roads;

[0042] S902, Determine the point-by-point acceleration estimate obtained in step S8. Is it within a reasonable range? Inside, among which and The calculation formula is as follows:

[0043]

[0044]

[0045] in, The speed is The minimum possible critical value of acceleration, The speed is The critical value of the maximum possible acceleration;

[0046] S903, Determine the point-by-point acceleration estimate obtained in step S8. Does the difference between the estimated acceleration values ​​of the trajectory points at subsequent intervals of 1s, 2s, and 3s conform to...

[0047] S904. Determine the true velocity value v of the end point of the reconstructed trajectory segment obtained in step S8. r+1 and speed estimates Does the difference meet the requirements? Where V thre This is the difference threshold.

[0048] Further, in step S10, based on the second-by-second velocity and second-by-second acceleration of the reconstructed trajectory segment, the specific power value of each point in the reconstructed trajectory segment is calculated, and the second-by-second velocity, second-by-second acceleration and the corresponding specific power value are substituted into the carbon emission estimation model to calculate the second-by-second carbon emissions of each reconstructed trajectory segment. Finally, the second-by-second carbon emissions of all reconstructed trajectory segments are summed in sequence to obtain the total carbon emissions.

[0049] Furthermore, based on the second-by-second velocity and second-by-second acceleration of the reconstructed trajectory segment, the specific power value of each point in the reconstructed trajectory segment is calculated. The formula for calculating the specific power value is as follows:

[0050] VSP=v(1.1a+0.132)+0.000302v 3

[0051] Where VSP is the specific power, v is the velocity per second, and a is the acceleration per second.

[0052] Furthermore, the second-by-second velocity, second-by-second acceleration, and corresponding specific power values ​​of the reconstructed trajectory segments are substituted into the carbon emission estimation model to calculate the second-by-second carbon emissions of each reconstructed trajectory segment, including the following steps:

[0053] If the speed per second in the reconstructed trajectory segment is less than 1.6 km / h, then that second is considered an idling state. The formula for calculating carbon emissions (CE) is:

[0054] CE = exp(-0.4987)

[0055] If the velocity per second in the reconstructed trajectory segment is greater than or equal to 1.6 km / h and the acceleration per second is less than 0, then that second is a deceleration state, and the formula for calculating carbon emissions (CE) is:

[0056] CE=exp(-0.443+0.001v^2+0.011a^2+0.016VSP+0.001v^2a^2+0.001vVSP+0.013v)

[0057] If the velocity per second in the reconstructed trajectory segment is greater than or equal to 1.6 km / h and the acceleration per second is not less than 0, then that second is considered a non-deceleration state, and the formula for calculating carbon emissions (CE) is:

[0058] CE=exp(0.065+0.142VSP-0.004〖VSP〗^2+0.00003794v^3-0.025v+0.0000899v^3a^3-0.013va^2+0.069a)

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] 1. This invention analyzes the acceleration distribution characteristics of trajectory segments at different sampling frequencies, reconstructs sparse trajectory data second by second based on the acceleration distribution characteristics, and then calculates emissions based on the reconstructed second-by-second trajectory data through a carbon emission estimation model, thus avoiding the problem that the original sparse trajectory data is difficult to input into the emission estimation model for carbon emission calculation.

[0061] 2. This invention is applicable to sparse trajectory data with multiple sampling frequencies, and can be adapted to calculate carbon emissions from GPS trajectory data collected at multiple sampling frequencies under real-world conditions, making it highly applicable.

[0062] 3. This invention uses acceleration distribution interpolation as its core to reconstruct sparse trajectories with sampling frequencies greater than 1 second. Based on the reconstructed trajectory, carbon emissions are calculated, which can effectively improve the accuracy of carbon emission calculation. It provides a new perspective for calculating vehicle carbon emissions based on sparse trajectories and has good innovation. Attached Figure Description

[0063] Figure 1 This is a flowchart of the present invention;

[0064] Figure 2 This is a comparison chart of calculation results obtained using different reconstruction methods when the sampling frequency is 5s in an embodiment of the present invention;

[0065] Figure 3 This is a comparison chart of calculation results obtained using different reconstruction methods when the sampling frequency is 15s in an embodiment of the present invention;

[0066] Figure 4 This is a comparison chart of calculation results obtained using different reconstruction methods when the sampling frequency is 30s in an embodiment of the present invention. Detailed Implementation

[0067] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. These embodiments are based on the technical solution of the present invention and provide detailed implementation methods and specific operating procedures. However, the scope of protection of the present invention is not limited to the following embodiments.

[0068] This invention proposes a method for calculating vehicle carbon emissions based on sparse trajectory data reconstruction. Based on trajectory data of light-duty fuel vehicles collected on ordinary urban roads, the method analyzes the acceleration distribution characteristics of trajectory segments at different sampling frequencies. The sparse trajectory data is then reconstructed second by second according to these acceleration distribution characteristics. Finally, emissions are calculated using a carbon emission estimation model based on the reconstructed second-by-second trajectory data. This method is applicable to trajectory data with various sampling frequencies, including higher sampling frequencies (e.g., 3s) or lower sampling frequencies (e.g., 30s). It reconstructs second-by-second trajectories using acceleration distribution interpolation as the core, making it suitable for the input requirements of emission estimation models and effectively improving the accuracy of carbon emission estimation.

[0069] like Figure 1As shown, the entire method mainly includes nine steps: cutting trajectory segments with known second-by-second data according to the sampling frequency, classifying and binning the trajectory segments with known second-by-second data, obtaining the acceleration probability density distribution function of each bin, cutting sparse trajectory segments with emissions to be estimated according to the sampling frequency, determining the category and bin of the sparse trajectory segments, estimating point-by-point acceleration, estimating point-by-point velocity, determining the validity of the reconstructed trajectory segments, and calculating the total carbon emissions of the entire trajectory sequence.

[0070] The specific steps are as follows:

[0071] S1. Collect multiple segments of known second-by-second trajectory data;

[0072] S2. Collect sparse trajectory data to be reconstructed, determine the sampling frequency of the sparse trajectory data, and cut the second-by-second trajectory data based on the sampling frequency of the sparse trajectory data to obtain multiple trajectory segments.

[0073] S3. Classify and bin the trajectory segments to obtain multiple segment categories and segment bins;

[0074] S4. Fit the acceleration probability density distribution function to the trajectory segments in each segment box to obtain the acceleration probability density distribution function in each segment box;

[0075] S5. Cut the sparse trajectory data to be reconstructed according to the sampling frequency to obtain multiple sparse trajectory segments;

[0076] S6. Determine the fragment category and fragment bin of the sparse trajectory fragment;

[0077] S7. For each sparse trajectory segment, retrieve the corresponding acceleration probability density distribution function calculated by the segment box based on the second-by-second trajectory segment, and generate acceleration estimates point by point from the start point to the end point of the sparse trajectory segment.

[0078] S8. Based on the known true value of the starting point velocity and the estimated value of the point-by-point acceleration of the sparse trajectory segment, calculate the estimated value of the point-by-point velocity to obtain the reconstructed trajectory segment;

[0079] S9. Determine whether the data of the reconstructed trajectory segment is valid. If it is valid, return to step S6 to reconstruct the next sparse trajectory segment. Otherwise, return to step S7 until all sparse trajectory segments contained in the entire sparse trajectory data to be reconstructed have been reconstructed.

[0080] S10. Substitute all reconstructed trajectory segments into the carbon emission estimation model to calculate the vehicle carbon emissions determined by the sparse trajectory data.

[0081] In steps S1 and S2, the known second-by-second trajectory data is a continuous second-by-second trajectory sequence formed by the vehicle's movement. Each trajectory data contains velocity and acceleration parameters. The trajectory segments are cut according to the actual sampling frequency r of the sparse trajectory data to be reconstructed. For example, if the sampling frequency of the sparse trajectory data to be reconstructed is 5s, then 1-2-3-4-5-6s and 6-7-8-9-10-11s constitute a trajectory segment, and so on.

[0082] It is worth noting that the second-by-second trajectory data is a large number of known second-by-second trajectory data sequences collected in advance by multiple experimental vehicles in multiple cities or urban roads, which are used as the basis for data research, making the reconstruction of subsequent sparse trajectory data universal; sparse trajectory data is a trajectory sequence that cannot reach the second level, obtained by GPS on any urban road.

[0083] Step S3 includes the following sub-steps:

[0084] The formula for calculating the standard deviation σ of the velocity for each trajectory segment is as follows:

[0085]

[0086] Where v i Let be the velocity (m / s) of the i-th point in the trajectory segment. Let r be the average velocity of the trajectory segment (m / s), and r be the sampling frequency (s) of the sparse trajectory data;

[0087] Let the velocity of the starting point of the trajectory segment be v1, and the velocity of the ending point be v. r+1 Extract all trajectory segments with a standard deviation less than 1.5 m / s, and calculate the velocity difference v between the beginning and end of each trajectory segment. r+1 -v1 sets the maximum difference between the first and last velocities of all trajectory segments as the upper bound value v. r,max The minimum value is set as the state boundary value v. r,min ;

[0088] For all original trajectory segments, state classification and binning are performed according to the following principles:

[0089] S301, if v r+1 -v1≥v r,max If the current trajectory segment is an acceleration segment, it is defined as the UA state, based on the initial velocity v1 and the velocity difference between the beginning and end v. r+1The range of -v1 further divides all trajectory segments belonging to the UA state into bins; the initial velocity should be divided into 3-4 categories based on the sample data within the bin (e.g., 0-20km / h, 20-40km / h, 40-60km / h), and the difference between the initial and final velocities should be divided into 2-3 categories based on the original sample data (e.g., less than 6km / h, not less than 6km / h). Specifically, if v1 belongs to 20-40km / h, v r+1 -v1 belongs to the same bin if it belongs to a trajectory segment with a speed of not less than 6km / h.

[0090] S302, if v r+1 -v1≤v r,min If the current trajectory segment is a deceleration segment, it is defined as state UD, based on the initial velocity v1 and the difference between the first and last velocities v. r+1 The range of -v1 further bins all trajectory segments belonging to the UD state, using the same binning rules as S301.

[0091] S303, if v r,min <v r+1 -v1<v r,max If the current trajectory segment is relatively stable, it is defined as the SS state, based on the initial velocity v1 and the velocity difference between the beginning and end v. r+1 The range of -v1 further bins all trajectory segments belonging to the SS state, following the same binning rules as S301.

[0092] In step S4, the acceleration probability density distribution function is fitted to the trajectory segments within each segment bin to obtain the acceleration probability density distribution function for each segment bin. The probability density distribution functions include normal distribution, exponential distribution, Pareto distribution, Dweibull distribution, T distribution, generalized extreme value distribution, gamma distribution, log-normal distribution, Beta distribution, uniform distribution, and log-gamma distribution. The root mean square error (RMSE) and goodness-of-fit R² are used. 2 Determine the fitting effect of the acceleration probability density distribution function and select the acceleration probability density distribution function with the best fitting effect;

[0093] Root mean square error (RMSE) and goodness of fit (R²) 2 The calculation formula is:

[0094]

[0095]

[0096] Among them, a i The true value of acceleration (m / s²) 2 ), For acceleration estimates (m / s²) 2 ), The true average value of acceleration (m / s²) 2 ), where m is the sample size.

[0097] For step S5, if the sampling frequency of the sparse trajectory sequence to be reconstructed is r, the trajectory segment is cut according to the sampling frequency. For example, if the sampling frequency is 5s, then 1-2-3-4-5-6s is a trajectory segment. For this segment, only the true velocity values ​​of the 1st and 6th seconds are known, and so on.

[0098] Step S6 includes the following sub-steps:

[0099] Let the velocity of the starting point of the sparse trajectory segment be v1', and the velocity of the ending point be v. r+1 ', based on v1' and v r+1 'Classify sparse trajectory fragments into fragment categories and fragment bins;'

[0100] S601, if v r+1 '-v1'≥v r,max If the current sparse trajectory segment belongs to state UA, then based on the initial velocity v1' and the velocity difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the UA state in step S301, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0101] S602, if v r+1 '-v1'≤v r,min If the current sparse trajectory segment belongs to state UD, then based on the initial velocity v1' and the difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the UD state in step S302, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0102] S603, if v r,min <v r+1 '-v1' < v r,max If the current sparse trajectory segment belongs to the SS state, then based on the initial velocity v1' and the velocity difference between the first and last velocities v... r+1 The range of '-v1' is used to determine which bin the current sparse trajectory segment is located in under the SS state in step S303, and the acceleration probability density distribution function obtained in step S4 for that bin is retrieved.

[0103] In step S8, the known data in the sparse trajectory segment are the true velocity value of the starting point, the true velocity value of the ending point, and the estimated acceleration values ​​of all points;

[0104] The method for calculating the velocity estimate of each point other than the starting point is as follows: starting with the actual velocity value of the starting point, the velocity estimate of the current point is obtained by sequentially adding the acceleration estimate of the previous point and the acceleration estimate of the current point.

[0105] After the velocity estimates of all points except the starting point are calculated, a reconstructed trajectory segment is obtained. The end point of the reconstructed trajectory segment has a true velocity value and a velocity estimate.

[0106] Step S9, determining whether the data of the reconstructed trajectory segment is valid includes the following steps:

[0107] S901, Discriminate the point-by-point velocity estimate obtained in step S8 Is it within a reasonable range [0, V]? max [Inside, where V] max This refers to the actual speed limit on urban roads;

[0108] S902, Determine the point-by-point acceleration estimate obtained in step S8. Is it within a reasonable range? Inside, among which and The calculation formula is as follows:

[0109]

[0110]

[0111] in, The speed is The minimum possible acceleration critical value (m / s) 2 ), The speed is The maximum possible critical value of acceleration (m / s²) 2 );

[0112] S903, Determine the point-by-point acceleration estimate obtained in step S8. Does the difference between the estimated acceleration values ​​of the trajectory points at subsequent intervals of 1s, 2s, and 3s conform to...

[0113] S904. Determine the true velocity value v of the end point of the reconstructed trajectory segment obtained in step S8. r+1 and speed estimates Does the difference meet the requirements? Where V thre The difference threshold is set to 0.1 m / s in this embodiment.

[0114] In step S10, based on the second-by-second velocity and second-by-second acceleration of the reconstructed trajectory segment, the specific power value of each point in the reconstructed trajectory segment is calculated. The formula for calculating the specific power value is as follows:

[0115] VSP=v(1.1a+0.132)+0.000302v 3

[0116] Where VSP is the specific power, v is the velocity per second, and a is the acceleration per second.

[0117] The carbon emission estimation model is calculated by substituting the second-by-second velocity, second-by-second acceleration, and corresponding specific power values ​​of the reconstructed trajectory segments into the carbon emission estimation model, including the following steps:

[0118] If the speed per second in the reconstructed trajectory segment is less than 1.6 km / h, then that second is considered an idling state. The formula for calculating carbon emissions (CE) is:

[0119] CE = exp(-0.4987)

[0120] If the velocity per second in the reconstructed trajectory segment is greater than or equal to 1.6 km / h and the acceleration per second is less than 0, then that second is a deceleration state, and the formula for calculating carbon emissions (CE) is:

[0121] CE=exp(-0.443+0.001v^2+0.011a^2+0.016VSP+0.001v^2a^2+0.001vVSP+0.013v)

[0122] If the velocity per second in the reconstructed trajectory segment is greater than or equal to 1.6 km / h and the acceleration per second is not less than 0, then that second is considered a non-deceleration state, and the formula for calculating carbon emissions (CE) is:

[0123] CE=exp(0.065+0.142VSP-0.004〖VSP〗^2+0.00003794v^3-0.025v+0.0000899v^3a^3-0.013va^2+0.069a)

[0124] Finally, the carbon emissions per second in all reconstructed trajectory segments are summed up to obtain the total carbon emissions.

[0125] like Figure 2 , 3As shown in Figure 4, a specific analytical case of the present invention is presented. This embodiment is based on a 120s trajectory sequence. Under sampling frequencies of 5s, 15s, and 30s, the sparse trajectory is reconstructed into a second-by-second trajectory using the core trajectory reconstruction technology of the present invention and conventional linear interpolation and cubic spline interpolation techniques, respectively. The results are then compared with the actual second-by-second trajectory sequence. It can be seen intuitively that the reconstruction results of the sparse trajectory data of the present invention are more consistent with reality under different sampling frequencies, indicating that the carbon emission calculation method of the present invention has better robustness and higher calculation accuracy.

[0126] Table 1 shows the analysis case data results of this invention. Figure 2 The table comparing the final carbon emission estimates in the case studies shows that Method I is the technology of this invention, Method II is the linear interpolation technology, and Method III is the cubic spline interpolation technology. The emission results demonstrate the feasibility of this technology. This invention has higher accuracy in calculating carbon emissions for sparse trajectory data.

[0127] Table 1. Results of Case Data Analysis

[0128]

[0129] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for reconstructing vehicle carbon emission calculation based on sparse trajectory data, characterized in that, Includes the following steps: S1. Collect multiple segments of known second-by-second trajectory data; S2. Collect the sparse trajectory data to be reconstructed, determine the sampling frequency of the sparse trajectory data, and cut the second-by-second trajectory data based on the sampling frequency of the sparse trajectory data to obtain multiple trajectory segments. S3. Classify and bin the trajectory segments to obtain multiple segment categories and segment bins; S4. Fit the acceleration probability density distribution function to the trajectory segments in each segment box to obtain the acceleration probability density distribution function in each segment box; S5. The sparse trajectory data to be reconstructed is cut according to the sampling frequency to obtain multiple sparse trajectory segments; S6. Determine the fragment category and fragment bin of the sparse trajectory fragment; S7. For each sparse trajectory segment, retrieve the corresponding acceleration probability density distribution function calculated by the segment box based on the second-by-second trajectory segment, and generate acceleration estimates point by point from the start point to the end point of the sparse trajectory segment. S8. Based on the known true value of the starting point velocity and the estimated value of the point-by-point acceleration of the sparse trajectory segment, calculate the estimated value of the point-by-point velocity to obtain the reconstructed trajectory segment. S9. Determine whether the data of the reconstructed trajectory segment is valid. If it is valid, return to step S6 to reconstruct the next sparse trajectory segment. Otherwise, return to step S7 until all sparse trajectory segments contained in the entire sparse trajectory data to be reconstructed have been reconstructed. S10. Substitute all reconstructed trajectory segments into the carbon emission estimation model to calculate the vehicle carbon emissions determined by the sparse trajectory data.

2. The method of claim 1, wherein, Step S3 includes the following sub-steps: calculating a standard deviation of the velocity of each trajectory segment which is given by the formula: wherein is the velocity of the i-th point in the trajectory segment, i is the average velocity of the trajectory segment, is the average velocity of the trajectory segment, is the sampling frequency of the sparse trajectory data; The start trajectory point speed value of the trajectory segment is , the end trajectory point speed value is , all trajectory segments with a standard deviation less than 1.5 m / s are extracted, the first and last speed difference of each trajectory segment is calculated , the maximum value of the first and last speed difference of all trajectory segments is set as the state upper limit value , and the minimum value is set as the state lower limit value ; For all original trajectory segments, state classification and binning are performed according to the following principles: S301、If then the current trajectory segment belongs to the accelerating segment, defined as UA state, all trajectory segments belonging to UA state are further binned according to the range of the difference between the start speed and the end speed ; S302、If , the current trajectory segment belongs to the deceleration segment, defined as the UD state, and all trajectory segments belonging to the UD state are further binned according to the range of the initial speed and the head-tail speed difference ; S303、If , the current trajectory segment belongs to the relatively stable segment, defined as SS state, according to the range of the initial speed and the difference between the head and tail speed , all trajectory segments belonging to the SS state are further binned.

3. The method of claim 1, wherein, In step S4, the probability density distribution function includes normal distribution, exponential distribution, Pareto distribution, Dweibull distribution, T distribution, generalized extreme value distribution, gamma distribution, log-normal distribution, Beta distribution, uniform distribution, and log-gamma distribution.

4. The method of claim 3, wherein, In step S4, the root mean square error RMSE and the goodness of fit The fitting effect of the acceleration probability density distribution function is discriminated, and the acceleration probability density distribution function with the best fitting effect is selected. The root mean square error RMSE and the goodness of fit The formula for calculating the root mean square error RMSE and the goodness of fit wherein is the acceleration true value, is the acceleration estimated value, is the acceleration true average value, is the sample size.

5. The method of claim 2, wherein, Step S6 includes the following sub-steps: The start track point speed value of the sparse track segment is recorded as , and the end track point speed value is recorded as ; S601、If then the current sparse trajectory segment belongs to the UA state, and according to the range of the start speed and the head-tail speed difference , it is determined in which bin of the UA state in step S301 the current sparse trajectory segment is located, and the acceleration probability density distribution function obtained in step S4 for the bin is called. S602、if then the current sparse trajectory segment belongs to the UD state, and according to the range of the start speed and the head-tail speed difference , it is judged which bin the current sparse trajectory segment is located in under the UD state in step S302, and the acceleration probability density distribution function obtained in step S4 for the bin is called. S603、If then the current sparse trajectory segment belongs to the SS state, and depending on the range of the start speed and the head-tail speed difference it is determined in which bin of the SS state in step S303 the current sparse trajectory segment is located, and the acceleration probability density distribution function acquired in step S4 for that bin is retrieved.

6. The method of claim 1, wherein, In step S8, the known data in the sparse trajectory segment are the true velocity value of the starting point, the true velocity value of the ending point, and the estimated acceleration values ​​of all points; The method for calculating the velocity estimate of each point other than the starting point is as follows: starting with the actual velocity value of the starting point, the velocity estimate of the current point is obtained by sequentially adding the acceleration estimate of the previous point and the acceleration estimate of the current point. After the velocity estimates of all points except the starting point are calculated, a reconstructed trajectory segment is obtained. The end point of the reconstructed trajectory segment has a true velocity value and a velocity estimate.

7. The method of claim 6, wherein, Step S9, determining whether the data of the reconstructed trajectory segment is valid includes the following steps: S901, determine the point-by-point speed estimate obtained in step S8 whether in a reasonable interval wherein is an actual urban road speed limit value; S902, discriminating the point-by-point acceleration estimate value obtained in step S8 whether in a reasonable interval wherein and the calculation formula is as follows: in, The speed is The critical value of minimum acceleration at time The speed is The critical value of maximum acceleration at time; S903、determine whether the difference between the point-by-point acceleration estimate obtained in step S8 and the trajectory point acceleration estimate obtained after an interval of 1s, 2s, 3s, etc. whether the difference between the point-by-point acceleration estimate obtained in step S8 and the trajectory point acceleration estimate obtained after an interval of 1s, 2s, 3s, etc. , , ; S904. Determine the true velocity value of the end point of the reconstructed trajectory segment obtained in step S8. and speed estimates Does the difference meet the requirements? ,in This is the difference threshold.

8. The method of claim 1, wherein, In step S10, based on the second-by-second velocity and second-by-second acceleration of the reconstructed trajectory segment, the specific power value of each point in the reconstructed trajectory segment is calculated. The second-by-second velocity, second-by-second acceleration and the corresponding specific power value are substituted into the carbon emission estimation model to calculate the second-by-second carbon emissions of each reconstructed trajectory segment. Finally, the second-by-second carbon emissions of all reconstructed trajectory segments are summed to obtain the total carbon emissions.

9. The method of claim 8, wherein, Based on the reconstructed trajectory segment of the second speed and the second acceleration, the specific power value of each point in the reconstructed trajectory segment is calculated, and the calculation formula of the specific power value is: wherein, VSP P is the power, v V is the velocity per second, a A is the acceleration per second.

10. The method of claim 9, wherein, The second speed, the second acceleration and the corresponding specific power value of the reconstructed trajectory segment are substituted into the carbon emission estimation model to calculate the second carbon emission of each reconstructed trajectory segment, including the following steps: If the second speed in the reconstructed trajectory segment is less than 1.6km / h, the second is the idle state, and the formula for calculating the carbon emission CE is: If the second speed in the reconstructed trajectory segment is greater than or equal to 1.6km / h and the second acceleration is less than 0, the second is the deceleration state, and the formula for calculating the carbon emission CE is: If the second speed in the reconstructed trajectory segment is greater than or equal to 1.6km / h and the second acceleration is not less than 0, the second is the non-deceleration state, and the formula for calculating the carbon emission CE is: 。

Images