Artificial intelligence-based intelligent carbon emission assessment method and system

Abstract

This invention relates to the field of engineering carbon emission assessment and three-dimensional spatial data processing technology, specifically disclosing an intelligent carbon emission assessment method and system based on artificial intelligence. The method involves collecting a three-dimensional micro-topographic digital elevation model of the target area, normalized vegetation index, geological attribute data, excavation volume, and discrete timestamp spatial sequences of construction machinery. A three-dimensional spatial distribution matrix of soil organic carbon is generated based on the vegetation index, elevation, and soil coefficient. Soil carbon release is calculated based on this matrix and the excavation volume. The longitudinal slope and acceleration are extracted from the discrete timestamp sequence, and the power carbon emissions are obtained through equivalent conversion of traction power and fuel calorific value, merging them into the total carbon emissions. The total carbon emissions are bound to spatial coordinates and reconstructed into a continuous spatial carbon perturbation density field using a Gaussian kernel. Equivalent breadth surfaces exceeding safety thresholds are extracted from the density field to generate constraint layers, which are then fused into a comprehensive carbon constraint matrix and transmitted back to an external optimization system to drive the scheme to avoid high-carbon emission areas.

Description

Technical Field

[0001] This invention relates to the field of engineering carbon emission assessment and three-dimensional spatial data processing technology, specifically to an intelligent carbon emission assessment method and system based on artificial intelligence. Background Technology

[0002] Under the guidance of the "dual carbon" objective, carbon emission assessment of candidate schemes for large-scale power transmission and transformation projects and infrastructure projects has become a core indicator for decision-making. In the carbon emission assessment of power transmission and transformation projects and infrastructure projects, existing assessment methods generally adopt the static engineering quota method, which calculates emissions by multiplying the consumption of building materials by a preset carbon emission factor. This type of method mainly focuses on carbon emissions during the material production and transportation stages, lacking systematic quantitative means for assessing soil carbon release caused by micro-topographical and ecological disturbances and carbon emissions from dynamic energy consumption of machinery during construction.

[0003] The existing technology has the following shortcomings:

[0004] Existing carbon emission assessment methods lack spatial data support for deep soil organic carbon and cannot reconstruct discrete carbon emission points into continuous field quantities that can be used for spatial constraints. As a result, the assessment results can only be used to issue reports and cannot drive the iteration of engineering solutions. The entire assessment and optimization process is in an open-loop state. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent carbon emission assessment method and system based on artificial intelligence to solve the problems mentioned above.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] An AI-based intelligent assessment method for carbon emissions includes the following steps:

[0008] S1: Collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area to form the basic dataset.

[0009] S2: Based on the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, a multivariate nonlinear mapping is performed to invert and generate the three-dimensional spatial distribution matrix of soil organic carbon.

[0010] S3: Based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, an exponential asymptotic recovery model with zero-point anchoring is introduced to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, the instantaneous longitudinal slope and acceleration are extracted, and the instantaneous fuel consumption rate is obtained through the traction power equation and the equivalent conversion of the low heating value of fuel oil, and the power carbon emission is obtained by integration. The soil carbon release and power carbon emission are combined into the total carbon emission.

[0011] S4: Bind the total carbon emissions to their corresponding three-dimensional spatial coordinates, and reconstruct them into a continuous spatial carbon perturbation density field by topological shading through an isotropic Gaussian kernel.

[0012] S5: Extract the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generate the ecologically sensitive constraint layer and the high energy consumption penalty layer, merge them into a comprehensive carbon constraint matrix and transmit it in reverse to the external spatial optimization system to drive candidate solutions to avoid high carbon emission areas.

[0013] As a further aspect of the present invention: the inversion to generate the three-dimensional spatial distribution matrix of soil organic carbon specifically includes:

[0014] The normalized vegetation index, elevation value, and soil type conversion coefficient are assigned to each node of the three-dimensional grid in the target area, and the initial carbon estimate of each node is calculated by the regression coefficient fitted by historical geological borehole samples.

[0015] Extract the measured carbon concentration of each historical geological borehole sample at the corresponding node, and calculate the deviation between the measured carbon concentration of the node and the initial carbon estimate.

[0016] Based on the spatial Euclidean distance between each borehole node and the unsampled grid node, the deviation value is propagated to all unsampled nodes by inverse distance weighting. The propagated deviation value is then used to correct the initial carbon estimate of each unsampled node, thereby obtaining the three-dimensional spatial distribution matrix of soil organic carbon.

[0017] As a further aspect of the present invention: the calculation process for the initial carbon estimate is as follows:

[0018] The normalized vegetation index, elevation value, soil type conversion coefficient and measured carbon concentration of each of the multiple historical geological borehole samples were collected to form a reference sample set.

[0019] For each node to be calculated in the target area, a local weight value is assigned to each reference sample based on the reciprocal of the product of the absolute value of the spatial Euclidean distance and the elevation difference between the node to be calculated and each reference sample.

[0020] Using the reference sample set and the corresponding local weight values, the regression coefficients are fitted node by node using the weighted least squares method to minimize the weighted sum of squared residuals.

[0021] Substitute the normalized vegetation index, elevation value, and soil type conversion coefficient of the node to be calculated into the multiple linear expression composed of the regression coefficients to calculate the initial carbon estimate of the node.

[0022] As a further aspect of the present invention: the introduction of a zero-point anchored exponential asymptotic recovery model coupled with the calculation of soil carbon release specifically includes:

[0023] Based on the three-dimensional spatial distribution matrix, the soil organic carbon concentration at each excavation grid node is extracted. Combined with the three-dimensional excavation volume of the grid and the dry bulk density of the soil, the potential maximum carbon release capacity of the grid is calculated.

[0024] The soil exposure time of each excavation grid is determined based on the spatial sequence of discrete timestamps of construction machinery, and the aerobic oxidation rate constant corresponding to the grid is calculated based on the surface temperature.

[0025] Substituting the exposure duration into the exponential asymptotic recovery equation with the initial exposure time as zero, the release amount asymptotically approaches the potential maximum release carbon storage as the exposure duration increases, thereby obtaining the cumulative soil carbon release of the grid.

[0026] The cumulative soil carbon release of all excavated grids is summed, and the output is the total soil carbon release.

[0027] As a further aspect of the present invention: the merging of soil carbon release and dynamic carbon emissions into total carbon emissions specifically includes:

[0028] By differentiating the three-dimensional coordinates corresponding to each adjacent time stamp in the discrete timestamp spatial sequence, the instantaneous velocity, instantaneous longitudinal slope angle, and instantaneous acceleration at each time stamp are obtained;

[0029] Substitute the equipment's self-weight, rolling resistance coefficient, instantaneous longitudinal slope angle, instantaneous acceleration, and instantaneous velocity into the traction power equation to calculate the instantaneous traction power at each timestamp;

[0030] By incorporating engine thermal efficiency and diesel low heating value, instantaneous traction power is converted into instantaneous fuel consumption rate at each timestamp, and time integration is performed over the entire construction period to obtain power carbon emissions.

[0031] Add the carbon emissions from power plants to the total carbon release from the soil to get the total carbon emissions.

[0032] As a further aspect of the present invention: the reconstruction process of the spatial carbon perturbation density field is as follows:

[0033] Each discrete carbon emission is assigned its corresponding three-dimensional spatial coordinates to generate a carbon emission scatter set.

[0034] For any point to be determined in the target space, calculate the three-dimensional Euclidean distance between the point to be determined and each point in the carbon emission point set.

[0035] Using a preset smoothing bandwidth as a scale, the Gaussian weight values ​​of each scatter point corresponding to the point to be determined are calculated based on each three-dimensional Euclidean distance, and the density value of the point to be determined is initialized as the sum of the products of each Gaussian weight value and the carbon emission of the corresponding scatter point.

[0036] Traverse all points in the target space and output a continuous three-dimensional matrix composed of the density values ​​of each point, as the spatial carbon perturbation density field.

[0037] As a further aspect of the present invention: the step of calculating the Gaussian weight values ​​of each scatter point corresponding to the point to be determined based on each three-dimensional Euclidean distance specifically includes:

[0038] Obtain the three-dimensional Euclidean distance between the point to be determined and each carbon emission point;

[0039] Calculate the ratio of the square of each distance value to the square of the preset smoothing bandwidth value to obtain the ratio corresponding to each scatter point;

[0040] The exponent values ​​corresponding to the negative values ​​of each ratio are calculated using the base of the natural constant, thus obtaining the original Gaussian weights of each scatter point for the point to be calculated.

[0041] The original Gaussian weights of all scattered points to be calculated are summed, and the sum is used as the denominator to normalize each original Gaussian weight, outputting the final Gaussian weight value of each scattered point to be calculated.

[0042] As a further aspect of the present invention: the extraction process of the equivalent bread surface is as follows:

[0043] Traverse the spatial carbon perturbation density field and select all grid nodes with density values ​​greater than the first safety threshold to form a high-density node set.

[0044] Calculate the 3D convex hull of a high-density node set, using the closed surface formed by the triangular facets on the convex hull surface as the initial isosurface;

[0045] For each vertex on the initial isosurface, gradually push outward or shrink inward along the weighted average direction of the normals of the triangular facets where the vertex is located, until the absolute value of the difference between the density value at the vertex and the first safety threshold is less than the preset tolerance.

[0046] Retriangulate all the adjusted vertices and output a closed equivalent bread surface.

[0047] As a further aspect of the present invention: the calculation of the three-dimensional convex hull of the high-density node set, using the closed surface enclosed by each triangular facet of the convex hull surface as the initial isosurface, specifically includes:

[0048] It receives a high-density set of nodes as input and selects the node with the minimum elevation value as the starting point.

[0049] Using the initial base point as the origin, calculate the azimuth and zenith angles of all other nodes in the set relative to the origin, and arrange them in ascending order of azimuth angles to construct an initial convex hull linked list;

[0050] Traverse the sorted node sequence, check the visibility of each new node to the current convex hull boundary edge. If it is visible, delete the boundary edge and add a new edge connecting the new node to the convex hull. Gradually expand outward until all nodes have been processed.

[0051] Based on the triangular patches formed by each side in the final convex hull linked list, output the closed surface formed by all the triangular patches as the initial isosurface.

[0052] An AI-based intelligent carbon emission assessment system includes:

[0053] The basic data acquisition module is used to collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area, forming the basic dataset.

[0054] The soil organic carbon spatial inversion module uses the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, to perform multivariate nonlinear mapping and generate a three-dimensional spatial distribution matrix of soil organic carbon.

[0055] The total carbon emission calculation module, based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, introduces a zero-point anchored exponential asymptotic recovery model to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, it extracts the instantaneous longitudinal slope and acceleration, obtains the instantaneous fuel consumption rate through the traction power equation and the equivalent conversion of the low heating value of fuel, and integrates it to obtain the power carbon emission, and combines the soil carbon release and power carbon emission into the total carbon emission;

[0056] The spatial carbon density field reconstruction module binds the total carbon emissions to their corresponding three-dimensional spatial coordinates and reconstructs them into a continuous spatial carbon perturbation density field by performing topological shading with an isotropic Gaussian kernel.

[0057] The carbon constraint feedback module extracts the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generates an ecologically sensitive constraint layer and a high-energy-consumption penalty layer, merges them into a comprehensive carbon constraint matrix, and transmits it back to the external spatial optimization system to drive candidate solutions to avoid high-carbon emission areas.

[0058] The beneficial effects of this invention are:

[0059] (1) By collecting easily obtainable remote sensing and geological data such as normalized vegetation index, elevation value and soil type conversion coefficient, and combining historical borehole samples for multivariate nonlinear mapping inversion, a three-dimensional spatial distribution matrix of deep soil organic carbon that is missing in conventional geographic information systems can be generated. This provides an executable data input method for calculating soil carbon release under micro-topographic ecological disturbance and solves the problem of quantitative calculation due to the lack of underlying data in the assessment process.

[0060] (2) The power carbon emissions obtained by converting the spatial sequence of discrete timestamps of construction machinery through the traction power equation and the low heating value of fuel oil are combined with the soil carbon release calculated based on the zero-point anchoring index model to form the total carbon emissions. Then, it is further reconstructed into a continuous spatial carbon perturbation density field, and the equivalent breadth surface that exceeds the safety threshold is extracted from it to generate a constraint layer. Finally, it is transmitted back to the external optimization system in the form of a comprehensive carbon constraint matrix, so that the carbon emission assessment results can directly participate in the path iteration of candidate schemes, forming a complete closed loop from data collection to scheme feedback. Attached Figure Description

[0061] The invention will now be further described with reference to the accompanying drawings.

[0062] Figure 1 This is a flowchart of the method of the present invention;

[0063] Figure 2 This is a flowchart illustrating the process of generating a three-dimensional spatial distribution matrix of soil organic carbon in this invention.

[0064] Figure 3 This is a system block diagram of the present invention. Detailed Implementation

[0065] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0066] Please see Figure 1 As shown, this invention is an intelligent carbon emission assessment method based on artificial intelligence, comprising the following steps:

[0067] S1: Collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area to form the basic dataset.

[0068] S2: Based on the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, a multivariate nonlinear mapping is performed to invert and generate the three-dimensional spatial distribution matrix of soil organic carbon.

[0069] S3: Based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, an exponential asymptotic recovery model with zero-point anchoring is introduced to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, the instantaneous longitudinal slope and acceleration are extracted, and the instantaneous fuel consumption rate is obtained through the traction power equation and the equivalent conversion of the low heating value of fuel oil, and the power carbon emission is obtained by integration. The soil carbon release and power carbon emission are combined into the total carbon emission.

[0070] S4: Bind the total carbon emissions to their corresponding three-dimensional spatial coordinates, and reconstruct them into a continuous spatial carbon perturbation density field by topological shading through an isotropic Gaussian kernel.

[0071] S5: Extract the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generate the ecologically sensitive constraint layer and the high energy consumption penalty layer, merge them into a comprehensive carbon constraint matrix and transmit it in reverse to the external spatial optimization system to drive candidate solutions to avoid high carbon emission areas.

[0072] In S1, the basic dataset is constructed by collecting the three-dimensional micro-topographic digital elevation model of the target area, the hyperspectral remote sensing normalized vegetation index, the underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel routes of the construction candidate schemes. Specifically, it includes:

[0073] Aerial scanning of the target area was conducted using a drone equipped with a LiDAR. The forward overlap was set to 70%, the lateral overlap to 60%, and the relative flight altitude to 120 meters. This yielded 3D point cloud data with a density of at least 4 points per square meter. After filtering to remove vegetation and building points, a 3D micro-topography digital elevation model with a grid spacing of 2 meters was generated. Simultaneously, a multispectral camera on the same drone acquired red and near-infrared reflectance images of the target area at a ground resolution of 0.5 meters. The normalized vegetation index (NDI) for each grid cell was obtained by calculating the difference between the near-infrared and red reflectance and dividing it by the sum of the two values.

[0074] Historical records of existing geological exploration boreholes within the target area were collected. Each borehole record included borehole coordinates, soil classification names for each depth, and measured soil organic carbon concentration values ​​determined in the laboratory using the potassium dichromate oxidation method after on-site sampling. Based on the soil classification names, a pre-established geological attribute mapping table was consulted to convert each soil classification into a corresponding dimensionless soil category conversion coefficient. This mapping table was compiled based on regional soil survey data, with a coefficient of 0.9 for sandy clay, 2.3 for peat soil, and other soil types determined by linear interpolation based on organic matter content.

[0075] Extract the length, width, and depth dimensions of each foundation excavation area from the design drawings of the construction candidate schemes, and calculate the three-dimensional excavation volume of each excavation area in cubic meters. For excavation areas with slopes, calculate the actual excavation volume after correcting the excavation boundary according to the slope coefficient given in the design drawings, and summarize them to form a list of three-dimensional excavation volumes corresponding to each grid node.

[0076] A dual-frequency global navigation satellite system receiver is installed on the top of the cab of the construction machinery. It records the machinery's three-dimensional spatial coordinates and corresponding timestamps in real time during the construction period at a sampling frequency of 2 Hz, forming a discrete timestamp spatial sequence. Each record in the sequence includes longitude, latitude, elevation, and the time of acquisition. This sequence is arranged in chronological order, with a time interval of 0.5 seconds between adjacent records, covering the entire process from the construction machinery entering the work area to its withdrawal.

[0077] Please see Figure 2 As shown, in S2, based on the normalized vegetation index, elevation values, and soil type conversion coefficients in the basic dataset, and combined with the regression coefficients fitted from historical geological borehole samples, a multivariate nonlinear mapping is performed to invert and generate the three-dimensional spatial distribution matrix of soil organic carbon, specifically including:

[0078] The 3D micro-topography digital elevation model was divided into a 3D grid with a 2-meter grid spacing. Each grid node was assigned three attribute values: the vegetation index value corresponding to the node was extracted from the normalized vegetation index (NVI) raster, the elevation value of the node was extracted from the digital elevation model, and the soil type conversion coefficient was obtained by consulting the geological attribute mapping table based on the soil type at the node's location. All historical geological borehole samples within the target area were collected. Each sample included the normalized vegetation index, elevation value, soil type conversion coefficient, and measured carbon concentration determined by the potassium dichromate oxidation method in the laboratory for the borehole location, forming a reference sample set.

[0079] For each grid node to be calculated, the product of the absolute values ​​of the spatial Euclidean distance and elevation difference between that node and each borehole sample in the reference sample set is calculated. The reciprocal of this product is taken as the local weight value of that borehole sample for that node. Using the four attribute values ​​(vegetation index, elevation value, soil type conversion coefficient, and measured carbon concentration) of all samples in the reference sample set and the corresponding local weight values ​​of each sample, the weighted least squares method is used to fit the regression coefficients, minimizing the weighted sum of squared residuals between the measured carbon concentration of each sample and the predicted value by the linear combination of regression coefficients. This yields four regression coefficients, corresponding to the intercept term, vegetation index term, elevation value term, and soil type conversion coefficient term, respectively. The vegetation index value, elevation value, and soil type conversion coefficient of the node to be calculated are multiplied by the corresponding regression coefficients and then added to the intercept term to obtain the initial carbon estimate value of that node. This process is repeated for all grid nodes to obtain the initial carbon estimate value for each node.

[0080] The measured carbon concentration of each historical geological borehole sample in the grid node is extracted, and the initial carbon estimate obtained in the previous step is also obtained. The difference between the measured carbon concentration and the initial carbon estimate is calculated as the deviation value of that borehole node. For grid nodes without historical geological borehole samples, the deviation value is initially set to zero.

[0081] For each unsampled grid node, the spatial Euclidean distance between that node and all historical geological borehole nodes is calculated. The deviation value of each borehole node is divided by the square of the spatial Euclidean distance from that borehole node to the unsampled nodes to obtain the contribution weight of that borehole node to the unsampled nodes. The sum of the contribution weights of all borehole nodes is used as the denominator. The deviation value of each borehole node is divided by the square of the spatial Euclidean distance from that borehole node to the unsampled nodes, and the sum of these fractions is used as the numerator. The ratio of the numerator to the denominator is the deviation correction value propagated to the unsampled node. The reciprocal of the square of the spatial Euclidean distance is used as a distance attenuation factor, ensuring that unsampled nodes closer to the boreholes receive a larger deviation correction. The deviation correction value for each unsampled grid node is calculated in this manner.

[0082] The initial carbon estimate for each unsampled grid node is added to its corresponding bias correction value to obtain the corrected soil organic carbon concentration for that node. For grid nodes with historical geological borehole samples, the measured carbon concentration is directly used as the final concentration value for that node. The final concentration values ​​of all grid nodes are arranged in spatial order according to the three-dimensional grid to form a three-dimensional spatial distribution matrix of soil organic carbon. Each element in the matrix corresponds to the organic carbon concentration of a grid node, in kilograms per cubic meter.

[0083] In S3, based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, an exponential asymptotic recovery model with zero-point anchoring is introduced to couple the calculation of soil carbon emissions. Simultaneously, based on the discrete timestamp spatial sequence in the basic dataset, instantaneous longitudinal slope and acceleration are extracted. The instantaneous fuel consumption rate is obtained through the traction power equation and equivalent transformation of the low-grade calorific value of fuel, and then integrated to obtain the power carbon emissions. Soil carbon emissions and power carbon emissions are combined into the total carbon emissions, specifically including:

[0084] Based on the three-dimensional spatial distribution matrix of soil organic carbon, the soil organic carbon concentration value at each excavation grid node is extracted, in kilograms per cubic meter. The total mass of the excavated soil at that node is obtained by multiplying the volume of the three-dimensional excavated entity by the dry bulk density of the soil at that node. The dry bulk density of the soil is preset according to soil type: 1450 kg / m³ for sandy clay, 1100 kg / m³ for peat soil, and other soil types are determined by linear interpolation based on measured values. The potential maximum carbon release capacity of that grid node is obtained by multiplying the soil organic carbon concentration value at that node by the total mass of the excavated soil, in kilograms. For excavation areas including slopes, the actual excavation volume corrected for the slope coefficient is included in the calculation.

[0085] Based on the start and end times of the discrete timestamp spatial sequence of the construction machinery, the soil exposure time corresponding to each excavation grid node is determined. The exposure time is calculated from the moment the grid node is mechanically excavated and exposed to the surface until the end of construction, in days. Daily average surface temperature data of the target area during construction is collected, in degrees Celsius, and the average surface temperature within the exposure time is calculated. Based on the aerobic oxidation kinetics of soil microorganisms, a baseline release rate constant of 0.005 per day is set, representing the daily release ratio of soil organic carbon per unit mass at 20 degrees Celsius. A temperature sensitivity coefficient of 2.0 is set, indicating that the oxidation rate doubles for every 10 degrees Celsius increase in temperature. The difference between the average surface temperature and 20 degrees Celsius, divided by 10, is used as the exponent. The exponent is calculated using the temperature sensitivity coefficient as the base, and then multiplied by the baseline release rate constant to obtain the aerobic oxidation rate constant for that grid node, in reciprocals of the daily value. The exponent calculation is achieved using iterative multiplication, i.e., the temperature sensitivity coefficient is multiplied by the difference divided by 10 a certain number of times.

[0086] Substituting the calculated potential maximum carbon release, aerobic oxidation rate constant, and exposure duration into the exponential asymptotic recovery equation, the cumulative soil carbon release for this grid node is calculated. The equation is in the form: cumulative release equals the potential maximum carbon release multiplied by a factor within parentheses minus the value of the exponential function with the natural constant as its base, where the exponent of the exponential function is the negative aerobic oxidation rate constant multiplied by the exposure duration. The exponential function with the natural constant as its base is approximated using a Taylor series expansion, and the sum of the first eight terms is taken to obtain an approximate value. The mathematical expression of this equation is: ;

[0087] in, This represents the cumulative soil carbon release of a single grid node, expressed in kilograms. This indicates the maximum potential carbon release at this node, expressed in kilograms. This represents the aerobic oxidation rate constant at that node, expressed as the reciprocal of the daily rate. This indicates the duration of soil exposure at this node, in days. This represents the natural constant, with a value of 2.71828. When the exposure duration... When the value equals zero, the exponential function value is 1, the value inside the parentheses is 0, and the cumulative release is 0, satisfying the physical reality that the release is zero immediately after construction ends; as the exposure time increases... As carbon release increases, the exponential function value approaches 0, and the cumulative release approaches the potential maximum carbon storage. By iterating through all excavated grid nodes in this manner and summing the cumulative soil carbon releases at each node, the total soil carbon release is obtained, expressed in kilograms.

[0088] Based on the spatial sequence of discrete timestamps of the construction machinery, the data is processed sequentially according to time. For the 3D coordinates corresponding to the i-th timestamp and the 3D coordinates corresponding to the (i-1)-th timestamp, the difference in horizontal displacement and elevation between them are calculated. The horizontal displacement difference is divided by the 0.5-second time interval to obtain the horizontal velocity at that moment, in meters per second; the elevation difference is divided by the horizontal displacement difference to obtain the longitudinal slope tangent at that point, which is converted into the instantaneous longitudinal slope angle by looking up the arctangent table, in degrees; the velocity difference between two adjacent moments is divided by 0.5 seconds to obtain the instantaneous acceleration, in meters per square second. The instantaneous velocity and acceleration of the first timestamp are set to zero.

[0089] Substitute the calculated instantaneous velocity, instantaneous longitudinal slope angle, and instantaneous acceleration, along with the rated self-weight of the construction machinery (25,000 kg), the surface rolling resistance coefficient (0.10), the gravitational acceleration (9.8 m / s²), and the additional hydraulic power (45,000 watts), into the traction power equation. The traction power equals the equipment's self-weight multiplied by the gravitational acceleration multiplied by the rolling resistance coefficient (in parentheses), the cosine of the longitudinal slope angle, the sine of the longitudinal slope angle (in parentheses), plus the equipment's self-weight multiplied by the instantaneous acceleration, the sum of these three multiplied by the instantaneous velocity, and finally the additional hydraulic power. The cosine and sine values ​​of the longitudinal slope angle are obtained by consulting a trigonometric function table. The calculated instantaneous traction power is in watts; if the result is negative, it is taken as zero to avoid negative power during downhill operations.

[0090] Introducing an engine thermal efficiency of 0.35 and a diesel lower heating value of 3.58 joules per liter (LHC), multiplied by 10^7 joules, the instantaneous traction power at each time point is converted into an instantaneous fuel consumption rate. The conversion method is as follows: instantaneous fuel consumption rate equals instantaneous traction power divided by the product of engine thermal efficiency and diesel lower heating value, where instantaneous traction power is in watts (joules per second) and diesel lower heating value is in joules per liter. The result is the instantaneous fuel consumption rate in liters per second. Multiplying the instantaneous fuel consumption rate corresponding to each time point by a 0.5-second time interval yields the fuel consumption within that time interval, in liters. The fuel consumption across all time intervals within the construction period is summed to obtain the total fuel consumption. Multiplying this total fuel consumption by the diesel carbon emission standard of 2.61 kg CO2 per liter gives the power carbon emission, in kilograms of CO2.

[0091] The total carbon emissions, expressed in kilograms of carbon dioxide, are obtained by adding the aforementioned dynamic carbon emissions to the total soil carbon release. This total carbon emissions serve as input data for subsequent steps in the reconstruction of the spatial carbon perturbation density field. The total soil carbon release represents the carbon loss due to soil exposure in the ecological dimension, while the dynamic carbon emissions represent the carbon emissions corresponding to mechanical construction energy consumption. The combined total carbon emissions comprehensively reflect the total carbon footprint of the entire process of micro-topographic ecological disturbance.

[0092] In S4, the total carbon emissions are bound to their corresponding three-dimensional spatial coordinates, and topological shading is performed using an isotropic Gaussian kernel to reconstruct a continuous spatial carbon perturbation density field, specifically including:

[0093] The calculated total carbon emissions are bound to their spatial locations. Each total carbon emission corresponds to a three-dimensional spatial coordinate, which is taken from the location recorded in the discrete timestamp spatial sequence of the construction machinery at the moment the carbon emission was generated, or from the center coordinates of the excavation grid node. Each pair of coordinates and carbon emissions is combined into a carbon emission scatter point, and all scatter points together constitute a carbon emission scatter point set. The unit of carbon emission is kilograms of carbon dioxide, and the unit of coordinates is meters.

[0094] The target space is defined to cover the entire area of ​​the 3D micro-topography digital elevation model, and extended outward by a preset smoothing bandwidth to ensure the continuity of the density field at the boundary. The preset smoothing bandwidth is 15 meters, determined based on the typical operating radius of construction machinery and the spatial diffusion characteristics of carbon emissions. The target space is divided into a 3D grid with the same 2-meter grid spacing as the digital elevation model, with each grid node serving as a target point. For each target point, the 3D Euclidean distance between the target point and each point in the carbon emission distribution set is calculated. The 3D Euclidean distance is calculated as follows: the squares of the coordinate differences between the target point and the distribution points in the east-west direction, the north-south direction, and the vertical elevation difference are summed, and the square root is taken to obtain the distance value in meters.

[0095] The Gaussian weight value for each scatter point corresponding to the point to be determined is calculated based on each 3D Euclidean distance. The calculation process consists of four steps:

[0096] The first step is to obtain the three-dimensional Euclidean distance between the point to be determined and a certain scattered point, and then square the distance to obtain the squared distance value.

[0097] The second step is to square the preset smoothing bandwidth value of 15 meters to obtain the bandwidth square value, and then divide the distance square value by the bandwidth square value to obtain the ratio corresponding to the scatter point.

[0098] The third step is to take the negative of this ratio, and using the natural constant 2.71828 as the base, calculate the exponential function value with the negative of this ratio as the exponent. This yields the original Gaussian weights of the points to be calculated for this scatter plot. The exponential function is approximated using a Taylor series expansion, and the sum of the first eight terms is used to obtain an approximate value.

[0099] The fourth step is to sum the original Gaussian weights of all the scattered points corresponding to the point to be calculated, obtaining the total weight. Then, for each scattered point, divide its original Gaussian weight by the total weight to obtain the final Gaussian weight value for that scattered point to the point to be calculated. After normalization, the sum of the final Gaussian weight values ​​of all scattered points to the point to be calculated equals 1.

[0100] Multiply the carbon emissions of each scatter point by its final Gaussian weight value for the target point to obtain the carbon emission contribution of that scatter point to the target point. Sum the carbon emission contributions from all scatter points to obtain the initial density value of the target point. This density value is expressed in units equivalent to kilograms of carbon dioxide per cubic meter, reflecting the overall intensity of the influence of surrounding carbon emissions on that location. For each target point in the target space, calculate its initial density value using the above steps.

[0101] The initial density values ​​of all points in the target space are arranged in the row, column, and layer order of the 3D mesh to form a continuous 3D matrix, which is the spatial carbon perturbation density field. Each element in the matrix corresponds to the density value of a mesh node, and the density values ​​between adjacent nodes satisfy the second-order differentiable smoothness property, allowing the extraction of isosurfaces and gradient directions from this density field for constraint boundary generation in subsequent steps. This density field serves as the input data for S5 and participates in the construction of the constraint layer.

[0102] In S5, equivalent breadth surfaces exceeding a preset safety threshold are extracted from the spatial carbon perturbation density field to generate an ecologically sensitive constraint layer and a high-energy-consumption penalty layer. These are then fused into a comprehensive carbon constraint matrix and transmitted back to the external spatial optimization system to drive candidate solutions to avoid high-carbon emission regions. Specifically, this includes:

[0103] A first safety threshold and a preset tolerance are set. The first safety threshold is determined based on the regional ecological sensitivity level, taking 0.5 kg CO2 equivalent per cubic meter for ecological redline areas and 1.2 kg CO2 equivalent per cubic meter for general areas. The preset tolerance is set to one-thousandth of the first safety threshold, used to control the approximation accuracy between the isosurface vertex and the theoretical isosurface. The spatial carbon perturbation density field generated in step S4 is traversed, and the density value of each grid node is checked one by one. If the density value of a node is greater than the first safety threshold, the node is assigned to the high-density node set. All nodes that meet the conditions together form the high-density node set, and the node distribution in this set reflects the spatial range in which the carbon perturbation intensity exceeds the allowable upper limit.

[0104] Calculate the 3D convex hull of a high-density node set, and use the closed surface formed by the triangular facets of the convex hull surface as the initial isosurface. The specific calculation steps are as follows:

[0105] The first step is to receive a high-density set of nodes as input, and select the node with the smallest elevation value as the starting point. If the minimum elevation value corresponds to multiple nodes, select the one with the smallest east-west coordinate.

[0106] The second step involves using the initial base point as the origin and calculating the azimuth and zenith angles of all other nodes in the set relative to this origin. The azimuth angle is the angle between the horizontal direction (pointing from the origin to other nodes) and true north, ranging from 0 to 360 degrees. The zenith angle is the angle between the direction (pointing from the origin to other nodes) and the vertical upward direction, ranging from 0 to 180 degrees. All other nodes are arranged in ascending order of azimuth angle; if the azimuth angles are the same, they are arranged in ascending order of zenith angle. An initial convex hull edge linked list is constructed, containing only the edge between the initial base point and the first node in the arrangement.

[0107] The third step involves processing each new node sequentially according to the order of arrangement. For each edge on the current convex hull boundary, it is determined whether the new node is visible. The visibility determination method is as follows: calculate the dot product of the new node and the outward normal of the triangle containing the boundary edge. If the dot product is greater than zero, the node is considered visible. If the new node is visible to a boundary edge, delete that boundary edge and add two new edges connecting the new node and the two endpoints of that boundary edge to the convex hull. After processing each edge, insert the new node into the convex hull and continue processing the next node until all nodes have been processed.

[0108] The fourth step is to extract all triangular facets formed by three sides from the final convex hull linked list. The three vertices of each triangular facet belong to a high-density node set. All triangular facets together form a closed convex polyhedron surface, which is the initial isosurface.

[0109] The position of each vertex on the initial isosurface is adjusted. For each vertex, all triangular faces containing that vertex are obtained, and the unit normal vector of each triangular face is calculated. The unit normal vector is calculated by taking the cross product of the vectors of the two sides of the triangle to obtain the normal vector, and then dividing the normal vector by its magnitude to obtain the unit normal vector. The unit normal vectors of each triangular face are weighted and averaged according to the interior angle of the vertex in each face to obtain the adjustment direction of the vertex. Along the adjustment direction, the vertex position is moved outward or inward by a step size, which is one-tenth of the preset tolerance. After moving the vertex, the density value at that vertex is recalculated by trilinear interpolation through the spatial carbon perturbation density field. The absolute value of the difference between the new density value and the first safety threshold is compared. If the absolute value is greater than the preset tolerance, the movement continues along the adjustment direction, with the step size decreasing to half of the previous step size each time, until the absolute value of the difference is less than the preset tolerance.

[0110] All adjusted vertices are retriangulated according to the triangular topology of the initial isosurface, i.e., the original vertex connections remain unchanged, only the spatial coordinates of the vertices are updated, forming a new closed surface. This surface is output as the isosurface envelope. The spatial region enclosed by this envelope is the high-carbon perturbation zone. All grid nodes within this region are marked as prohibited traversal nodes of the ecologically sensitive constraint layer, and nodes with density values ​​higher than the second safety threshold are marked as high-cost nodes of the high-energy-consumption penalty layer, where the second safety threshold is 1.5 times the first safety threshold. The two layers are merged into a comprehensive carbon constraint matrix, where each element represents the traversal cost of the corresponding spatial location, with a value ranging from 0 to 1. The cost of nodes within the ecologically sensitive constraint layer is set to 1, the cost of nodes within the high-energy-consumption penalty layer is set to 0.7, and the cost of other nodes is 0. This comprehensive carbon constraint matrix is ​​transmitted to an external spatial optimization system as the cost weight for path search.

[0111] Please see Figure 3 As shown, the AI-based intelligent carbon emission assessment system includes:

[0112] The basic data acquisition module is used to collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area, forming the basic dataset.

[0113] The soil organic carbon spatial inversion module uses the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, to perform multivariate nonlinear mapping and generate a three-dimensional spatial distribution matrix of soil organic carbon.

[0114] The total carbon emission calculation module, based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, introduces a zero-point anchored exponential asymptotic recovery model to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, it extracts the instantaneous longitudinal slope and acceleration, obtains the instantaneous fuel consumption rate through the traction power equation and the equivalent conversion of the low heating value of fuel, and integrates it to obtain the power carbon emission, and combines the soil carbon release and power carbon emission into the total carbon emission;

[0115] The spatial carbon density field reconstruction module binds the total carbon emissions to their corresponding three-dimensional spatial coordinates and reconstructs them into a continuous spatial carbon perturbation density field by performing topological shading with an isotropic Gaussian kernel.

[0116] The carbon constraint feedback module extracts the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generates an ecologically sensitive constraint layer and a high-energy-consumption penalty layer, merges them into a comprehensive carbon constraint matrix, and transmits it back to the external spatial optimization system to drive candidate solutions to avoid high-carbon emission areas.

[0117] The working principle of this invention is as follows: A three-dimensional micro-topography digital elevation model, normalized vegetation index (NDI), geological attribute data, excavated entity volume, and discrete timestamp spatial sequences of construction machinery are collected. Based on the NDI, elevation values, and soil type conversion coefficients, a three-dimensional spatial distribution matrix of soil organic carbon is generated. Soil carbon release is calculated using this matrix and the excavated volume. The longitudinal slope and acceleration are extracted from the discrete timestamp sequence and converted into equivalent traction power and fuel calorific value to obtain the dynamic carbon emissions, which are then combined into the total carbon emissions. The total carbon emissions are bound to spatial coordinates and reconstructed into a continuous spatial carbon perturbation density field using a Gaussian kernel. Equivalent breadth surfaces exceeding safety thresholds are extracted from the density field to generate ecologically sensitive constraints and high-energy-consumption penalty layers. These layers are then fused into a comprehensive carbon constraint matrix and transmitted in reverse to an external optimization system to drive the scheme to avoid high-carbon-emission areas.

[0118] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An intelligent carbon emission assessment method based on artificial intelligence, characterized in that, Includes the following steps: S1: Collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area to form the basic dataset. S2: Based on the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, a multivariate nonlinear mapping is performed to invert and generate the three-dimensional spatial distribution matrix of soil organic carbon. S3: Based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, an exponential asymptotic recovery model with zero-point anchoring is introduced to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, the instantaneous longitudinal slope and acceleration are extracted, and the instantaneous fuel consumption rate is obtained through the traction power equation and the equivalent conversion of the low heating value of fuel oil, and the power carbon emission is obtained by integration. The soil carbon release and power carbon emission are combined into the total carbon emission. S4: Bind the total carbon emissions to their corresponding three-dimensional spatial coordinates, and reconstruct them into a continuous spatial carbon perturbation density field by topological shading through an isotropic Gaussian kernel. S5: Extract the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generate the ecologically sensitive constraint layer and the high energy consumption penalty layer, merge them into a comprehensive carbon constraint matrix and transmit it in reverse to the external spatial optimization system to drive candidate solutions to avoid high carbon emission areas.

2. The intelligent carbon emission assessment method based on artificial intelligence according to claim 1, characterized in that, The inversion generates a three-dimensional spatial distribution matrix of soil organic carbon, specifically including: The normalized vegetation index, elevation value, and soil type conversion coefficient are assigned to each node of the three-dimensional grid in the target area, and the initial carbon estimate of each grid node is calculated by the regression coefficient fitted by historical geological borehole samples. Extract the measured carbon concentration of each historical geological borehole sample at the corresponding grid node, and calculate the deviation between the measured carbon concentration of the grid node and the initial carbon estimate. Based on the spatial Euclidean distance between each borehole node and the unsampled grid node, the deviation value is propagated to all unsampled grid nodes by inverse distance weighting. The propagated deviation value is then used to correct the initial carbon estimate of each unsampled grid node, thereby obtaining the three-dimensional spatial distribution matrix of soil organic carbon.

3. The intelligent carbon emission assessment method based on artificial intelligence according to claim 2, characterized in that, The calculation process for the initial carbon estimate is as follows: The normalized vegetation index, elevation value, soil type conversion coefficient and measured carbon concentration of each of the multiple historical geological borehole samples were collected to form a reference sample set. For each grid node to be calculated in the target area, a local weight value is assigned to each reference sample based on the reciprocal of the product of the absolute value of the spatial Euclidean distance and the elevation difference between the grid node to be calculated and each reference sample. Using the reference sample set and the corresponding local weight values, the regression coefficients are fitted node by node using the weighted least squares method to minimize the weighted sum of squared residuals. Substitute the normalized vegetation index, elevation value, and soil type conversion coefficient of the grid node to be calculated into the multiple linear expression composed of the regression coefficients to calculate the initial carbon estimate of each grid node.

4. The intelligent carbon emission assessment method based on artificial intelligence according to claim 1, characterized in that, The calculation of soil carbon release using the exponential asymptotic recovery model with zero-point anchoring specifically includes: Based on the three-dimensional spatial distribution matrix, the soil organic carbon concentration at each excavation grid node is extracted. Combined with the three-dimensional excavation volume of the grid and the dry bulk density of the soil, the potential maximum carbon release capacity of the grid is calculated. The soil exposure time of each excavation grid is determined based on the spatial sequence of discrete timestamps of construction machinery, and the aerobic oxidation rate constant corresponding to the grid is calculated based on the surface temperature. Substituting the exposure duration into the exponential asymptotic recovery equation with the initial exposure time as zero, the release amount asymptotically approaches the potential maximum release carbon storage as the exposure duration increases, thereby obtaining the cumulative soil carbon release of the grid. The cumulative soil carbon release of all excavated grids is summed, and the output is the total soil carbon release.

5. The intelligent carbon emission assessment method based on artificial intelligence according to claim 1, characterized in that, The process of combining soil carbon emissions and dynamic carbon emissions into total carbon emissions specifically includes: By differentiating the three-dimensional coordinates corresponding to each adjacent time stamp in the discrete timestamp spatial sequence, the instantaneous velocity, instantaneous longitudinal slope angle, and instantaneous acceleration at each time stamp are obtained; Substitute the equipment's self-weight, rolling resistance coefficient, instantaneous longitudinal slope angle, instantaneous acceleration, and instantaneous velocity into the traction power equation to calculate the instantaneous traction power at each timestamp; By incorporating engine thermal efficiency and diesel low heating value, instantaneous traction power is converted into instantaneous fuel consumption rate at each timestamp, and time integration is performed over the entire construction period to obtain power carbon emissions. Add the carbon emissions from power plants to the total carbon release from the soil to get the total carbon emissions.

6. The intelligent carbon emission assessment method based on artificial intelligence according to claim 1, characterized in that, The reconstruction process of the spatial carbon perturbation density field is as follows: Each discrete carbon emission is assigned its corresponding three-dimensional spatial coordinates to generate a carbon emission scatter set. For any point to be determined in the target space, calculate the three-dimensional Euclidean distance between the point to be determined and each point in the carbon emission point set. Using a preset smoothing bandwidth as a scale, the Gaussian weight values ​​of each scatter point corresponding to the point to be determined are calculated based on each three-dimensional Euclidean distance, and the density value of the point to be determined is initialized as the sum of the products of each Gaussian weight value and the carbon emission of the corresponding scatter point. Traverse all points in the target space and output a continuous three-dimensional matrix composed of the density values ​​of each point, as the spatial carbon perturbation density field.

7. The intelligent carbon emission assessment method based on artificial intelligence according to claim 6, characterized in that, The calculation of Gaussian weight values ​​for each scatter point corresponding to the point to be determined based on each three-dimensional Euclidean distance specifically includes: Obtain the three-dimensional Euclidean distance between the point to be determined and each carbon emission point; Calculate the ratio of the square of each distance value to the square of the preset smoothing bandwidth value to obtain the ratio corresponding to each scatter point; The exponent values ​​corresponding to the negative values ​​of each ratio are calculated using the base of the natural constant, thus obtaining the original Gaussian weights of each scatter point for the point to be calculated. The original Gaussian weights of all scattered points to be calculated are summed, and the sum is used as the denominator to normalize each original Gaussian weight, outputting the final Gaussian weight value of each scattered point to be calculated.

8. The intelligent carbon emission assessment method based on artificial intelligence according to claim 1, characterized in that, The extraction process of the equivalent bread surface is as follows: Traverse the spatial carbon perturbation density field and select all grid nodes with density values ​​greater than the first safety threshold to form a high-density node set. Calculate the 3D convex hull of a high-density node set, using the closed surface formed by the triangular facets on the convex hull surface as the initial isosurface; For each vertex on the initial isosurface, move outward or inward gradually along the weighted average direction of the normals of the triangular facets where the vertex is located, until the absolute value of the difference between the density value at the vertex and the first safety threshold is less than the preset tolerance. Retriangulate all the adjusted vertices and output a closed equivalent bread surface.

9. The intelligent carbon emission assessment method based on artificial intelligence according to claim 8, characterized in that, The calculation of the three-dimensional convex hull of the high-density node set uses the closed surface formed by the triangular facets on the convex hull surface as the initial isosurface, specifically including: It receives a high-density set of nodes as input and selects the node with the minimum elevation value as the starting point. Using the initial base point as the origin, calculate the azimuth and zenith angles of all other nodes in the set relative to the origin, and arrange them in ascending order of azimuth angles to construct an initial convex hull linked list; Traverse the sorted node sequence, check the visibility of each new node to the current convex hull boundary edge. If it is visible, delete the boundary edge and add a new edge connecting the new node to the convex hull. Gradually expand outward until all nodes have been processed. Based on the triangular patches formed by each side in the final convex hull linked list, output the closed surface formed by all the triangular patches as the initial isosurface.

10. An intelligent carbon emission assessment system based on artificial intelligence, characterized in that, The method for performing the intelligent carbon emission assessment based on artificial intelligence as described in any one of claims 1-9 includes: The basic data acquisition module is used to collect the three-dimensional micro-topography digital elevation model, hyperspectral remote sensing normalized vegetation index, underlying geological attribute data, and the discrete timestamp spatial sequence of the three-dimensional excavation entity volume and construction machinery travel route of the target area, forming the basic dataset. The soil organic carbon spatial inversion module uses the normalized vegetation index, elevation value and soil type conversion coefficient in the basic dataset, combined with the regression coefficients fitted by historical geological borehole samples, to perform multivariate nonlinear mapping and generate a three-dimensional spatial distribution matrix of soil organic carbon. The total carbon emission calculation module, based on the three-dimensional spatial distribution matrix and the three-dimensional excavation entity volume in the basic dataset, introduces a zero-point anchored exponential asymptotic recovery model to couple the calculation of soil carbon release; at the same time, based on the discrete timestamp spatial sequence in the basic dataset, it extracts the instantaneous longitudinal slope and acceleration, obtains the instantaneous fuel consumption rate through the traction power equation and the equivalent conversion of the low heating value of fuel, and integrates it to obtain the power carbon emission, and combines the soil carbon release and power carbon emission into the total carbon emission; The spatial carbon density field reconstruction module binds the total carbon emissions to their corresponding three-dimensional spatial coordinates and reconstructs them into a continuous spatial carbon perturbation density field by performing topological shading with an isotropic Gaussian kernel. The carbon constraint feedback module extracts the equivalent breadth surface that exceeds the preset safety threshold from the spatial carbon perturbation density field, generates an ecologically sensitive constraint layer and a high-energy-consumption penalty layer, merges them into a comprehensive carbon constraint matrix, and transmits it back to the external spatial optimization system to drive candidate solutions to avoid high-carbon emission areas.

Images