A mine ecological carbon sink dynamic monitoring and accounting system based on digital twinning
Patent Information
- Application Number
- CN202611162832.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-03
- Publication Date
- 2026-11-10
AI Technical Summary
Existing mine ecological carbon sink monitoring technologies cannot adapt to the complex nonlinear physical evolution such as subsidence in mining goaf areas and slope deformation of spoil heaps, resulting in spatial distortion of digital twin models, difficulty in eliminating false greening areas, insufficient monitoring resolution, inability to accurately locate the root causes of abnormal carbon sinks, and lack of ecological restoration and regulation strategies.
A dynamic monitoring and accounting system for mine ecological carbon sinks based on digital twins is adopted. Data is collected through a cluster of multi-source heterogeneous sensing nodes to construct an adaptive octree voxel grid three-dimensional digital twin of the mine. Combined with event-triggered reconstruction, causal tracing and decision-making modules driven by carbon sink accounting residuals, dynamic net carbon sink flux calculation and ecological restoration regulation are realized.
It improves the spatiotemporal resolution and anti-interference capability of carbon sink monitoring in complex mining areas, accurately locates the root causes of carbon sink anomalies, generates targeted ecological restoration strategies, and ensures the clarity and security of the spatial boundaries of carbon sink assets.
Smart Images

Figure CN123070461A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of mine ecological monitoring technology, and in particular to a dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins. Background Technology
[0002] With increasing global emphasis on climate change and "dual carbon" targets, the assessment and dynamic monitoring of carbon sequestration capacity in mine ecological restoration areas has become a crucial aspect of ecological environment governance and green mine construction. Mine ecosystems are characterized by complex topography, highly heterogeneous soil pollution, and frequent human-induced disturbances. In recent years, IoT multi-source sensing technology and digital twin technology have been gradually introduced into the field of mine environmental monitoring. By deploying sensor clusters to collect multi-dimensional ecological time-series data and constructing three-dimensional visualization models, basic data support and visualization management methods have been provided for vegetation growth, soil physicochemical property evolution, and carbon flux assessment in mining areas, thereby improving the informatization level of mine ecological governance to a certain extent.
[0003] However, existing mine ecological carbon sequestration monitoring and accounting technologies suffer from several drawbacks in practical applications. These technologies often rely on static update mechanisms with preset cycles or fixed-resolution grids, failing to adapt to complex nonlinear physical evolutions such as subsidence in mining goaf areas and slope deformation at spoil heaps. This leads to spatial distortion in digital twin models and difficulty in effectively eliminating "false green areas" in remote sensing images. Furthermore, the monitoring resolution is severely insufficient in areas with heterogeneous heavy metal pollution and abrupt pH changes. Traditional methods often focus only on the natural physiological and ecological processes of vegetation and soil, neglecting the instantaneous carbon emission disturbances caused by physical and mechanical damage such as blasting and heavy machinery operations. This results in a significant decrease in accounting accuracy and anti-interference capabilities under complex and highly disturbed environments. Existing monitoring systems are often in a "data black box" state, only outputting macroscopic changes in total carbon sequestration, lacking a causal tracing mechanism that integrates groundwater hydrological physical connectivity and data-driven characteristics. This makes it difficult to accurately pinpoint the root causes of local carbon sequestration anomalies and generate targeted ecological restoration and regulation strategies. Therefore, we propose a dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins. Summary of the Invention
[0004] This invention addresses the shortcomings of existing technologies by proposing a dynamic monitoring and accounting system for mine ecological carbon sinks based on digital twins. This system can effectively eliminate interference from common false green areas in mining areas, ensure the authenticity of three-dimensional spatial topology, and significantly improve the spatiotemporal resolution, anti-interference ability, and absolute accuracy of carbon sink monitoring under complex mining areas with strong disturbances. It also solves the problems of ambiguous spatial boundaries, easy tampering, and difficulty in confirming ownership of mine carbon sink assets.
[0005] To achieve the above objectives, the present invention provides a dynamic monitoring and accounting system for mine ecological carbon sinks based on digital twins, specifically including a multi-source heterogeneous sensing node cluster with communication connection, a digital twin reconstruction engine, a carbon sink dynamic accounting module, a causal tracing and decision-making module, and a closed-loop adaptive calibration module.
[0006] The multi-source heterogeneous sensing node cluster is deployed in the mine ecological restoration area to collect multi-dimensional ecological time-series data including hyperspectral images, lidar point clouds, soil physicochemical parameters and micro-meteorological data.
[0007] The digital twin reconstruction is based on the ecological time-series data collected by a multi-source heterogeneous sensing node cluster to construct a 3D digital twin of the mine containing an adaptive octree voxel grid. The digital twin reconstruction engine adopts an event-triggered dynamic reconstruction strategy driven by carbon sink accounting residuals. It triggers the local subdivision or merging reconstruction of the adaptive octree voxel grid according to the local error gradient field fed back by carbon sink accounting. The local error gradient field is the partial derivative vector field of the accounting residual of each voxel grid output by the carbon sink dynamic accounting module with respect to its ecological attribute parameters. Its magnitude reflects the degree of uncertainty of the carbon sink accounting of the grid.
[0008] The carbon sequestration dynamic accounting module runs a vegetation-soil carbon sequestration coupled evolution model that integrates mining area disturbance impulse functions within the reconstructed adaptive octree voxel grid, and outputs the dynamic net carbon sequestration flux and accounting residuals of each voxel grid.
[0009] The causal tracing and decision-making module is used to construct a directed causal graph containing a mine functional zoning mask, calculate the causal contribution of carbon sink anomalies of each grid node based on the counterfactual reasoning algorithm, and generate ecological restoration and regulation strategies.
[0010] The closed-loop adaptive calibration module performs two-layer time-scale data assimilation based on the calculated residuals, and updates the physicochemical property parameters of the three-dimensional digital twin.
[0011] A further technical solution of the present invention: the system forms a three-way cross-coupling mechanism: the first coupling path is that the gradient field of the calculated residual triggers the grid reconstruction of the digital twin reconstruction engine; the second coupling path is that the gradient field of the calculated residual serves as the prior uncertainty input of the causal tracing and decision module; the third coupling path is that the abnormal contribution degree output by the causal tracing and decision module dynamically adjusts the sampling frequency of the multi-source heterogeneous sensing node cluster through an alpha weighted adjustment function.
[0012] The preferred technical solution of this invention is as follows: When constructing a three-dimensional digital twin of a mine, the digital twin reconstruction engine performs nonlinear deformation field registration based on mine physical constraints. Specifically, this includes using a digital elevation model generated from lidar point cloud data as a reference spatial coordinate system, constructing an energy functional that includes smoothness constraint terms, deformation gradient penalty terms, and mine-specific physical prior terms; and using the finite element method to solve for the optimal deformation displacement field of the energy functional, mapping the pixel coordinates of the hyperspectral image to the three-dimensional spatial coordinates of the adaptive octree voxel mesh.
[0013] The mine-specific physical prior terms include a topographic settlement gradient field regularization term and a spoil heap slope angle constraint term. The specific calculation process for the topographic settlement gradient field regularization term is as follows:
[0014] The divergence of the deformation displacement field vector at the current iteration step is calculated to characterize the local volumetric deformation rate, and the settlement rate scalar field at the corresponding spatial location monitored by synthetic aperture radar interferometry is obtained.
[0015] At each node of the adaptive octree voxel grid, the divergence value of the deformation displacement field vector is subtracted from the value at the corresponding position of the settlement rate scalar field to obtain the residual scalar field.
[0016] The residual value of each node in the residual scalar field is squared to obtain the squared residual field;
[0017] The squared residual values of all nodes in the squared residual field are multiplied by the physical volume weight of the corresponding voxel grid and then globally summed. The result of the global summation is the L2 norm spatial integral value of the regularization term of the terrain settlement gradient field, which is used as a soft constraint penalty term to limit the vertical displacement deformation of the voxel grid above the goaf. When the angle between the local normal vector of the voxel grid and the horizontal plane exceeds the soil repose angle threshold after deformation, the slope angle constraint term of the spoil heap applies a penalty potential energy according to the quadratic function of the excess angle.
[0018] The soil angle of repose threshold is a dynamic parameter calibrated through indoor direct shear tests based on the specific soil and rock type of the spoil heap.
[0019] The preferred technical solution of this invention: The logic of the digital twin reconstruction engine triggering local subdivision or merging reconstruction of the adaptive octree voxel mesh is as follows:
[0020] Real-time calculation of the spatial variation coefficient of available heavy metal concentration in soil within the current voxel grid and the spatial pH gradient of acidic mine wastewater;
[0021] When the spatial variation coefficient is greater than the preset pollution heterogeneity threshold, or the pH spatial gradient is greater than the preset acid-base mutation threshold, a local subdivision mechanism is triggered to split the current voxel grid into eight sub-voxel grids; the pollution heterogeneity threshold is determined based on the fractal dimension of the mining area geochemical background survey, and is triggered when the spatial variation coefficient exceeds a preset multiple standard deviation of the background value fractal dimension.
[0022] The subdivided voxel grids inherit the carbon sink accounting history of the parent grids and allocate the initial carbon pool of the parent grids through three-dimensional volume weighting.
[0023] When the joint information entropy of the normalized vegetation index and soil moisture in adjacent sub-voxel grids is less than a preset homogeneity threshold, a grid merging operation is performed to merge multiple sub-voxel grids into a parent voxel grid. The joint information entropy is used to capture the nonlinear coupling relationship between the normalized vegetation index and soil moisture to eliminate false green areas.
[0024] The preferred technical solution of this invention is as follows: the dynamic carbon sequestration accounting module runs a vegetation-soil carbon sequestration coupled evolution model that integrates the mining area disturbance impulse function, and the calculation logic for calculating the dynamic net carbon sequestration flux satisfies the following algorithm formula:
[0025]
[0026] In the formula:
[0027] Let be the dynamic net carbon flux of the i-th voxel grid at time t;
[0028] Total primary productivity of vegetation;
[0029] For the rate of carbon emissions from soil respiration;
[0030] The stress attenuation coefficient;
[0031] The comprehensive environmental stress index;
[0032] For multidimensional volume correction tensors;
[0033] Let be the Dirac impulse function, representing the impulse at... Mining area disturbances that occur constantly;
[0034] For the first Instantaneous carbon release equivalent due to the secondary disturbance event;
[0035] The right side of the equation in the algorithm formula is formed by adding a continuous carbon sink integral term and a discrete perturbation pulse summation term, wherein the continuous carbon sink integral term includes... The first integral expression characterizes physiological and ecological processes; the discrete perturbation pulse summation term contains... The second summation expression characterizes the physical-mechanical destruction process;
[0036] The method for obtaining the multidimensional volume correction tensor is as follows: the effective photosynthetic volume is calculated by inverting the canopy height using multispectral remote sensing data, the root penetration volume is calculated using the ground-penetrating radar echo delay preset at the center frequency, and the two are then multiplied by the Hadamard product.
[0037] The instantaneous carbon release equivalent is obtained by capturing the vibration wave energy generated by blasting or heavy machinery operations through a micro-vibration sensor array deployed in the mining area, and calculating it based on the mapping relationship between vibration wave energy and soil porosity destruction rate.
[0038] The preferred technical solution of this invention: The calibration and acquisition methods for the stress attenuation coefficient and the comprehensive environmental stress index in the vegetation-soil carbon sequestration coupled evolution model are as follows:
[0039] The comprehensive environmental stress index is calculated by integrating the available concentration of heavy metals in the soil, the mineral weathering rate, and the seepage coefficient of acid mine wastewater, using a time-varying objective weight model based on different remediation stages of the mining area. The specific calculation process of the time-varying objective weight model is as follows: obtain the first weight vector in the early stage of remediation and the second weight vector in the later stage of remediation; use the current remediation period as the independent variable and substitute it into the logistic function to calculate the transition coefficient at the current moment; use the transition coefficient to perform a weighted linear combination of the first weight vector and the second weight vector to obtain the time-varying objective weight of each indicator at the current moment; the transition midpoint of the logistic function is located at the midpoint of the transition between the soil physicochemical properties meeting the standards and the vegetation canopy closure in the mining area.
[0040] The calibration method for the stress attenuation coefficient is as follows: Time-series data of net ecosystem exchange measured by the carbon flux monitoring tower in the mining area are obtained, and a Bayesian posterior probability model incorporating the vegetation-soil carbon sink coupled evolution model is constructed. The likelihood function of the Bayesian posterior probability model is composed of the sum of squared residuals between the dynamic net carbon sink flux output by the vegetation-soil carbon sink coupled evolution model and the measured net ecosystem exchange. The hyperparameters of the prior distribution of the Bayesian posterior probability model are determined by the historical blasting vibration frequency and the extreme value of soil bulk density in the mining area. The Metropolis-Hastings algorithm is used as the sampling kernel of the Markov chain Monte Carlo method to iteratively sample the posterior probability distribution of the stress attenuation coefficient. After removing samples from the pre-burning period, the expected value of the posterior distribution during the stationary period is taken as the final calibration value.
[0041] The preferred technical solution of this invention: The calculation logic of the causal tracing and decision-making module for calculating the causal contribution of each grid node to carbon sink anomalies based on the counterfactual reasoning algorithm satisfies the following algorithm formula:
[0042]
[0043] In the formula,
[0044] The causal effect of the input feature intervention of the i-th grid on the total net carbon flux of the system;
[0045] Let i be the input ecological feature vector of the i-th grid.
[0046] The characteristic state after intervention;
[0047] It is a set of structural cause-effect graph variables that contains all the hidden states of the grid.
[0048] It is the set of nodes in a directed causal graph.
[0049] Let j be the set of parent nodes of node j;
[0050] For the joint probability distribution of the observed data;
[0051] The integral term and mathematical expectation symbol in the algorithm formula The resulting difference term is the expected causal effect, which is calculated using the Monte Carlo sampling method with a sampling number of no less than one thousand, and the incoming edges of the parent node of the node being intervened are truncated during the sampling process.
[0052] The edge direction of the directed structural causal graph is determined by the underground water flow direction of the mine and the runoff direction of the spoil heap slope. Furthermore, a mine functional zoning mask matrix is introduced during the graph convolution propagation process, which enables different state transition equations to be used between the spoil heap, tailings dam and reclamation area nodes.
[0053] A preferred technical solution of the present invention: the adjacency matrix of the directed causal graph is constructed using a dual-driven mechanism of physics and data, specifically including:
[0054] Extract the ecological feature vectors of each voxel grid and calculate the data-driven cosine similarity matrix;
[0055] Based on the convection-diffusion equation for groundwater pollution plume diffusion in mines, the solute transport flux between each grid node is solved to construct a physically driven groundwater connectivity matrix.
[0056] The cosine similarity matrix, the groundwater connectivity matrix, and the Gaussian kernel matrix based on spatial distance are fused using Hadamard product to obtain the final asymmetric adjacency matrix of the directed causal graph. The Hadamard product fusion is used to directly set the weight of the corresponding edge after fusion to zero when any element in the matrix is close to zero, thereby achieving hard truncation of physically unreachable or ecologically dissimilar node pairs.
[0057] The preferred technical solution of this invention:
[0058] The detection system also includes a carbon asset anchoring and storage module, configured as follows:
[0059] Extract the three-dimensional spatial coordinates of the adaptive octree voxel grid and the corresponding dynamic net carbon sink flux to generate a voxel-level carbon sink data package containing a spatial anchor fingerprint. The spatial anchor fingerprint is the octree path code of the voxel grid, the three-dimensional centroid coordinates and the concatenated hash of the net carbon sink flux at the current moment. The octree path code uniquely identifies the level position of the voxel in the adaptive octree.
[0060] The spatial anchor fingerprint is used as the unique spatial identifier of the on-chain asset. The voxel-level carbon sink data package is bound one-to-one with the non-fungible token on the blockchain through smart contracts, realizing the triple anchoring of voxel spatial location, carbon sink physical quantity and on-chain digital asset. Any change in any dimension will trigger the automatic marking of the on-chain non-fungible token state.
[0061] When a request to purchase externally verified voluntary emission reductions is received, the smart contract automatically verifies the real-time carbon sink status of the voxel grid corresponding to the spatial anchor fingerprint. If the status is normal, the token transfer is executed and the total carbon sink asset balance of the system is updated.
[0062] The preferred technical solution of this invention: The specific logic of the closed-loop adaptive calibration module performing dual-timescale data assimilation is as follows:
[0063] A first Kalman filter is constructed to update slowly changing soil physicochemical parameters, and a second Kalman filter is constructed to update rapidly changing vegetation growth parameters. The variance of the process noise covariance matrix of the first Kalman filter is smaller than that of the second Kalman filter. The update cycle of the slowly changing parameters is no less than seven days, and the update cycle of the rapidly changing parameters is no more than one day. The observation updates of the two Kalman filters are performed alternately at unequal intervals on the time axis.
[0064] The triggering mechanism of the closed-loop adaptive calibration module adopts a dual-path collaborative triggering: the first path detects the gradual drift of the calculated residual based on the cumulative sum control chart, and the second path detects the abrupt change anomaly of the calculated residual based on the Grubbs test; the outputs of the first path and the second path are logically ORed, and when the first path is triggered, the significance level of the Grubbs test in the second path is automatically increased to enhance the sensitivity to subsequent abrupt change detection.
[0065] The beneficial effects of this invention are as follows:
[0066] 1. This invention overcomes the spatial distortion problem of traditional static digital twin models under the conditions of mine goaf settlement and spoil heap slope deformation; by triggering grid merging with joint information entropy, it effectively eliminates the interference of false green areas commonly found in mining areas, ensuring the authenticity of the three-dimensional spatial topology, and decouples the natural physiological and ecological processes of vegetation from the physical and mechanical destruction processes such as mining blasting and heavy machinery operations, which greatly improves the spatiotemporal resolution, anti-interference ability and absolute accuracy of carbon sink monitoring under the strong disturbance environment of complex mining areas.
[0067] 2. This invention overcomes the limitations of traditional carbon sink monitoring, which only knows the total amount of change but not the root cause of anomalies. It can accurately trace the causal contribution of groundwater pollution plume diffusion or specific operating areas to local carbon sink anomalies, providing a targeted basis for formulating scientific ecological restoration and control strategies for mining areas. Through the triple anchoring of octree path coding, three-dimensional centroid and net carbon sink flux, it completely solves the problems of ambiguous spatial boundaries, easy tampering, and difficulty in confirming ownership of mining carbon sink assets. The automatic verification and transfer mechanism of smart contracts opens up the transformation of ecological value into economic value, ensuring the security and liquidity of carbon sink assets. Attached Figure Description
[0068] Figure 1 This is a schematic diagram of the overall system architecture and three-way cross-coupling mechanism in this invention.
[0069] Figure 2 This is a schematic diagram of the digital twin reconstruction and grid dynamic evolution process in this invention.
[0070] Figure 3 This is a schematic diagram of the dual-track process of carbon sequestration accounting and causal source tracing decision-making in this invention.
[0071] Figure 4 This is a schematic diagram of the closed-loop adaptive calibration and asset storage process in this invention. Detailed Implementation
[0072] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0073] The embodiment provides a dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins, such as Figure 1 As shown in Figure 4, the system includes a multi-source heterogeneous sensing node cluster with communication connections, a digital twin reconstruction engine, a carbon sink dynamic accounting module, a causal tracing and decision-making module, a closed-loop adaptive calibration module, and a carbon sink asset anchoring and storage module. The system forms a three-way cross-coupling mechanism. The first coupling path is that the gradient field of the accounting residual triggers the grid reconstruction of the digital twin reconstruction engine. The second coupling path is that the gradient field of the accounting residual serves as the prior uncertainty input of the causal tracing and decision-making module. The third coupling path is that the abnormal contribution degree output by the causal tracing and decision-making module dynamically adjusts the sampling frequency of the multi-source heterogeneous sensing node cluster through an alpha-weighted adjustment function.
[0074] The multi-source heterogeneous sensing node cluster is deployed in the mine ecological restoration area to collect multi-dimensional ecological time-series data, including hyperspectral images, lidar point clouds, soil physicochemical parameters, and micro-meteorological data.
[0075] The digital twin reconstruction engine is communicatively connected to the multi-source heterogeneous sensing node cluster and is configured to construct a three-dimensional digital twin of the mine containing an adaptive octree voxel grid based on the multi-dimensional ecological time-series data. The digital twin reconstruction engine differs from the static update mechanism with a preset cycle and adopts an event-triggered dynamic reconstruction strategy driven by carbon sink accounting residuals. It triggers the local subdivision or merging reconstruction of the adaptive octree voxel grid according to the local error gradient field fed back by carbon sink accounting. The local error gradient field is the partial derivative vector field of each voxel grid accounting residual output by the carbon sink dynamic accounting module with respect to its ecological attribute parameters. Its magnitude reflects the degree of uncertainty of the carbon sink accounting of the grid.
[0076] When constructing the three-dimensional digital twin of the mine, the digital twin reconstruction engine performs nonlinear deformation field registration based on the physical constraints of the mine. Specifically, this includes: using the digital elevation model generated from lidar point cloud data as the reference spatial coordinate system, constructing an energy functional that includes smoothness constraint terms, deformation gradient penalty terms, and mine-specific physical prior terms; solving the optimal deformation displacement field of the energy functional using the finite element method, and mapping the pixel coordinates of the hyperspectral image to the three-dimensional spatial coordinates of the adaptive octree voxel mesh;
[0077] The mine-specific physical prior terms include a topographic settlement gradient field regularization term and a spoil heap slope angle constraint term. The specific calculation process for the topographic settlement gradient field regularization term is as follows:
[0078] The divergence of the deformation displacement field vector at the current iteration step is calculated to characterize the local volumetric deformation rate, and the settlement rate scalar field at the corresponding spatial location monitored by synthetic aperture radar interferometry is obtained.
[0079] At each node of the adaptive octree voxel grid, the divergence value of the deformation displacement field vector is subtracted from the value at the corresponding position of the settlement rate scalar field to obtain the residual scalar field.
[0080] The residual value of each node in the residual scalar field is squared to obtain the squared residual field;
[0081] The squared residual values of all nodes in the squared residual field are multiplied by the physical volume weight of the corresponding voxel grid and then globally summed. The result of the global summation is the L2 norm spatial integral value of the regularization term of the terrain settlement gradient field, which is used as a soft constraint penalty term to limit the vertical displacement deformation of the voxel grid above the goaf. When the angle between the local normal vector of the voxel grid and the horizontal plane exceeds the soil repose angle threshold after deformation, the slope angle constraint term of the spoil heap applies a penalty potential energy according to the quadratic function of the excess angle.
[0082] The soil angle of repose threshold is a dynamic parameter calibrated through indoor direct shear tests based on the specific soil and rock type of the spoil heap.
[0083] The logic of the digital twin reconstruction engine triggering the local subdivision or merging reconstruction of the adaptive octree voxel grid is to calculate in real time the spatial variation coefficient of the effective concentration of soil heavy metals and the spatial gradient of pH value of acidic mine wastewater within the current voxel grid.
[0084] When the spatial variation coefficient is greater than a preset pollution heterogeneity threshold, or the pH spatial gradient is greater than a preset acid-base mutation threshold, a local subdivision mechanism is triggered, splitting the current voxel grid into eight sub-voxel grids. The pollution heterogeneity threshold is determined based on the fractal dimension of the mining area's geochemical background survey, and is triggered when the spatial variation coefficient exceeds a preset multiple standard deviation of the background value's fractal dimension. The subdivided sub-voxel grids inherit the carbon sink accounting history of the parent grid and allocate the initial carbon pool inventory of the parent grid through three-dimensional volume weighting.
[0085] When the joint information entropy of the normalized vegetation index and soil moisture in adjacent sub-voxel grids is less than a preset homogeneity threshold, a grid merging operation is performed to merge multiple sub-voxel grids into a parent voxel grid. The joint information entropy is used to capture the nonlinear coupling relationship between the normalized vegetation index and soil moisture to eliminate false green areas.
[0086] The dynamic carbon sink accounting module is communicatively connected to the digital twin reconstruction engine. It is configured to run a vegetation-soil carbon sink coupled evolution model incorporating mining area perturbation impulse functions within the reconstructed adaptive octree voxel grid, outputting the dynamic net carbon sink flux and accounting residuals for each voxel grid. The calculation logic for the dynamic net carbon sink flux, obtained by the dynamic carbon sink dynamic accounting module running the vegetation-soil carbon sink coupled evolution model incorporating mining area perturbation impulse functions, satisfies the following algorithm formula:
[0087]
[0088] In the formula: Let be the dynamic net carbon flux of the i-th voxel grid at time t;
[0089] Total primary productivity of vegetation;
[0090] For the rate of carbon emissions from soil respiration;
[0091] The stress attenuation coefficient;
[0092] The comprehensive environmental stress index;
[0093] For multidimensional volume correction tensors;
[0094] Let be the Dirac impulse function, representing the impulse at... Mining area disturbances that occur constantly;
[0095] The instantaneous carbon release equivalent caused by the k-th perturbation event;
[0096] The right side of the equation in the algorithm formula is formed by adding a continuous carbon sink integral term and a discrete perturbation pulse summation term, wherein the continuous carbon sink integral term includes... The first integral expression characterizes physiological and ecological processes; the discrete perturbation pulse summation term contains... The second summation expression characterizes the physical-mechanical destruction process;
[0097] The method for obtaining the multidimensional volume correction tensor is to calculate the effective photosynthetic volume by inverting the canopy height using multispectral remote sensing data, calculate the root penetration volume using the ground-penetrating radar echo delay preset at the center frequency, and perform the Hadamard product on the two.
[0098] The instantaneous carbon release equivalent is obtained by capturing the vibration wave energy generated by blasting or heavy machinery operations through a micro-vibration sensor array deployed in the mining area, and calculating it based on the mapping relationship between vibration wave energy and soil porosity destruction rate.
[0099] The calibration and acquisition methods of the stress attenuation coefficient and the comprehensive environmental stress index in the vegetation-soil carbon sink coupled evolution model are as follows: The comprehensive environmental stress index is calculated by integrating the available concentration of heavy metals in the soil, the mineral weathering rate, and the seepage coefficient of acid mine wastewater, using a time-varying objective weight model based on different remediation stages of the mining area; The specific calculation process of the time-varying objective weight model is as follows: Obtain the first weight vector in the early stage of remediation and the second weight vector in the later stage of remediation in the mining area, substitute the current remediation years as the independent variable into the logistic function to calculate the transition coefficient at the current moment, and use the transition coefficient to perform a weighted linear combination of the first weight vector and the second weight vector to obtain the time-varying objective weight of each indicator at the current moment. The transition midpoint of the logistic function is located at the mid-term node of the alternation between the attainment of soil physicochemical properties and the closure of the vegetation canopy in the mining area;
[0100] The calibration method for the stress attenuation coefficient is as follows: Time-series data of net ecosystem exchange measured by the carbon flux monitoring tower in the mining area are obtained, and a Bayesian posterior probability model incorporating the vegetation-soil carbon sink coupled evolution model is constructed. The likelihood function of the Bayesian posterior probability model is composed of the sum of squared residuals between the dynamic net carbon sink flux output by the vegetation-soil carbon sink coupled evolution model and the measured net ecosystem exchange. The hyperparameters of the prior distribution of the Bayesian posterior probability model are determined by the historical blasting vibration frequency and the extreme value of soil bulk density in the mining area. The Metropolis-Hastings algorithm is used as the sampling kernel of the Markov chain Monte Carlo method to iteratively sample the posterior probability distribution of the stress attenuation coefficient. After removing samples from the pre-burning period, the expected value of the posterior distribution during the stationary period is taken as the final calibration value.
[0101] The causal tracing and decision-making module is used to construct a directed causal graph containing a mine functional zoning mask, calculate the causal contribution of carbon sink anomalies to each grid node based on a counterfactual reasoning algorithm, and generate ecological restoration and regulation strategies. The calculation logic of the causal tracing and decision-making module for calculating the causal contribution of carbon sink anomalies to each grid node based on the counterfactual reasoning algorithm satisfies the following algorithm formula:
[0102]
[0103] In the formula, The causal effect of the input feature intervention of the i-th grid on the total net carbon flux of the system;
[0104] Let i be the input ecological feature vector of the i-th grid.
[0105] The characteristic state after intervention;
[0106] It is a set of structural cause-effect graph variables that contains all the hidden states of the grid.
[0107] It is the set of nodes in a directed causal graph.
[0108] Let j be the set of parent nodes of node j;
[0109] For the joint probability distribution of the observed data;
[0110] The integral term and mathematical expectation symbol in the algorithm formula The resulting difference term is the expected causal effect, which is calculated using the Monte Carlo sampling method with a sampling number of no less than one thousand, and the incoming edges of the parent node of the node being intervened are truncated during the sampling process.
[0111] The edge direction of the directed structural causal graph is determined by the underground water flow direction of the mine and the runoff direction of the spoil heap slope. Furthermore, a mine functional zoning mask matrix is introduced during the graph convolution propagation process, which enables different state transition equations to be used between the spoil heap, tailings dam and reclamation area nodes.
[0112] The adjacency matrix of the directed causal graph is constructed using a dual-driven mechanism of physics and data. Specifically, it includes: extracting the ecological feature vectors of each voxel grid to calculate a data-driven cosine similarity matrix; solving the solute transport flux between grid nodes based on the diffusion-convection-diffusion equation of groundwater pollution plume in mines to construct a physics-driven groundwater connectivity matrix; fusing the cosine similarity matrix, the groundwater connectivity matrix, and the Gaussian kernel matrix based on spatial distance using Hadamard product to obtain the final asymmetric adjacency matrix of the directed causal graph; the Hadamard product fusion is used to directly set the weight of the corresponding edge after fusion to zero when any element in the matrix is close to zero, thereby achieving hard truncation of physically unreachable or ecologically dissimilar node pairs.
[0113] The closed-loop adaptive calibration module performs two-layer time-scale data assimilation based on the calculated residuals to update the physicochemical property parameters of the three-dimensional digital twin. The specific logic of the closed-loop adaptive calibration module performing two-layer time-scale data assimilation is as follows: a first Kalman filter is constructed to update slowly changing soil physicochemical parameters, and a second Kalman filter is constructed to update rapidly changing vegetation growth parameters. The variance of the process noise covariance matrix of the first Kalman filter is smaller than that of the second Kalman filter. The update cycle for the slowly changing parameters is no less than seven days, and the update cycle for the rapidly changing parameters is no more than one day. The observation updates of the two Kalman filters are performed alternately at unequal intervals along the time axis.
[0114] The triggering mechanism of the closed-loop adaptive calibration module adopts a dual-path collaborative triggering: the first path detects the gradual drift of the calculated residual based on the cumulative sum control chart, and the second path detects the abrupt anomaly of the calculated residual based on the Grubbs test; the outputs of the first path and the second path are logically ORed, and when the first path is triggered, the significance level of the Grubbs test in the second path is automatically increased to enhance the sensitivity to subsequent abrupt change detection;
[0115] The carbon sink asset anchoring and notarization module is configured to: extract the three-dimensional spatial coordinates of the adaptive octree voxel grid and the corresponding dynamic net carbon sink flux, generate a voxel-level carbon sink data package containing a spatial anchoring fingerprint, wherein the spatial anchoring fingerprint is the octree path code of the voxel grid, the three-dimensional centroid coordinates, and the concatenated hash of the net carbon sink flux at the current moment, wherein the octree path code uniquely identifies the voxel's hierarchical position in the adaptive octree; use the spatial anchoring fingerprint as a unique spatial identifier for on-chain assets, and bind the voxel-level carbon sink data package to non-fungible tokens on the blockchain one-to-one through a smart contract, thereby achieving triple anchoring of voxel spatial position, carbon sink physical quantity, and on-chain digital assets, and any change in any dimension will trigger automatic marking of the on-chain non-fungible token state;
[0116] When a request to purchase externally verified voluntary emission reductions is received, the smart contract automatically verifies the real-time carbon sink status of the voxel grid corresponding to the spatial anchor fingerprint. If the status is normal, the token transfer is executed and the total carbon sink asset balance of the system is updated.
[0117] In the above embodiments, the construction of a directed causal graph driven by both physics and data, the calculation of counterfactual reasoning and abnormal causal contribution of carbon sink based on random sampling methods, the adaptive adjustment of the perception sampling frequency under a three-way cross-coupling mechanism, the generation of voxel-level spatial anchor fingerprints, and the one-to-one binding of on-chain non-fungible tokens and automated transaction verification based on smart contracts are all included.
[0118] Working principle: In actual use, the system collects hyperspectral, point cloud and physicochemical time series data through a cluster of multi-source heterogeneous sensing nodes. The digital twin reconstruction engine uses the lidar point cloud as a reference to build an energy optimization model that includes smoothness constraints and mine-specific physical priors. It solves the optimal deformation displacement field to map three-dimensional spatial coordinates. The system monitors the pollution heterogeneity and acid-base gradient within the grid in real time. Through an event-triggered mechanism, it dynamically executes the local subdivision or merging of the adaptive octree voxel grid, thereby constructing a high-fidelity three-dimensional digital twin that conforms to the physical evolution law of the mine.
[0119] Within the reconstructed voxel grid, the carbon sequestration dynamic accounting module runs the vegetation-soil carbon sequestration coupled evolution model, collaboratively calculating the continuous carbon sequestration integral term representing physiological and ecological processes and the discrete perturbation impulse summation term representing physical and mechanical damage. Combined with the multidimensional volume correction matrix, it outputs the dynamic net carbon sequestration flux and accounting residuals. The closed-loop adaptive calibration module uses a dual-path collaborative mechanism to detect the residual sequence, triggering dual-timescale data assimilation and updating soil and vegetation parameters in the long and short term respectively. The accounting error gradient field formed in this process not only serves as the input of prior uncertainty but also triggers the dynamic adjustment of grid reconstruction and sensing sampling frequency, forming a tight three-way cross-coupling mechanism.
[0120] Based on accounting data, the causal tracing and decision-making module integrates a data-driven similarity matrix and a physically-driven hydrological connectivity matrix to construct a directed causal graph containing a mine functional zoning mask. The system executes a counterfactual reasoning algorithm using random sampling to accurately calculate the causal contribution of each node to carbon sink anomalies and generate targeted control strategies. The carbon sink asset anchoring and evidence preservation module extracts voxel-level spatial anchoring fingerprints containing octree path codes and binds them one-to-one with on-chain non-fungible tokens via smart contracts, achieving triple anchoring of space, physical quantity, and digital assets. During external subscriptions, the system automatically verifies the status and executes token transfers, completing a closed-loop process from ecological monitoring to asset monetization.
[0121] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.
Claims
1. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins, characterized in that, Specifically, it includes a multi-source heterogeneous sensing node cluster with communication connections, a digital twin reconstruction engine, a carbon sink dynamic accounting module, a causal tracing and decision-making module, and a closed-loop adaptive calibration module. The multi-source heterogeneous sensing node cluster is deployed in the mine ecological restoration area to collect multi-dimensional ecological time-series data including hyperspectral images, lidar point clouds, soil physicochemical parameters and micro-meteorological data. The digital twin reconstruction is based on the ecological time-series data collected by a multi-source heterogeneous sensing node cluster to construct a 3D digital twin of the mine containing an adaptive octree voxel grid. The digital twin reconstruction engine adopts an event-triggered dynamic reconstruction strategy driven by carbon sink accounting residuals. It triggers the local subdivision or merging reconstruction of the adaptive octree voxel grid according to the local error gradient field fed back by carbon sink accounting. The local error gradient field is the partial derivative vector field of the accounting residual of each voxel grid output by the carbon sink dynamic accounting module with respect to its ecological attribute parameters. Its magnitude reflects the degree of uncertainty of the carbon sink accounting of the grid. The carbon sequestration dynamic accounting module runs a vegetation-soil carbon sequestration coupled evolution model that integrates mining area disturbance impulse functions within the reconstructed adaptive octree voxel grid, and outputs the dynamic net carbon sequestration flux and accounting residuals of each voxel grid. The causal tracing and decision-making module is used to construct a directed causal graph containing a mine functional zoning mask, calculate the causal contribution of carbon sink anomalies of each grid node based on the counterfactual reasoning algorithm, and generate ecological restoration and regulation strategies. The closed-loop adaptive calibration module performs two-layer time-scale data assimilation based on the calculated residuals, and updates the physicochemical property parameters of the three-dimensional digital twin.
2. The dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins according to claim 1, characterized in that, The system forms a three-way cross-coupling mechanism: the first coupling path is that the gradient field of the calculated residual triggers the grid reconstruction of the digital twin reconstruction engine; the second coupling path is that the gradient field of the calculated residual serves as the prior uncertainty input for the causal tracing and decision-making module. The third coupling path involves dynamically adjusting the sampling frequency of the multi-source heterogeneous sensing node cluster using an alpha-weighted adjustment function based on the abnormal contribution output of the causal tracing and decision-making module.
3. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins as described in claim 1 or 2, characterized in that, When constructing a 3D digital twin of a mine, the digital twin reconstruction engine performs nonlinear deformation field registration based on mine physical constraints. Specifically, this includes: using a digital elevation model generated from lidar point cloud data as the reference spatial coordinate system, constructing an energy functional that includes smoothness constraint terms, deformation gradient penalty terms, and mine-specific physical prior terms; solving for the optimal deformation displacement field of the energy functional using the finite element method, and mapping the pixel coordinates of the hyperspectral image to the 3D spatial coordinates of the adaptive octree voxel mesh; The mine-specific physical prior terms include a topographic settlement gradient field regularization term and a spoil heap slope angle constraint term. The specific calculation process for the topographic settlement gradient field regularization term is as follows: The divergence of the deformation displacement field vector at the current iteration step is calculated to characterize the local volumetric deformation rate, and the settlement rate scalar field at the corresponding spatial location monitored by synthetic aperture radar interferometry is obtained. At each node of the adaptive octree voxel grid, the divergence value of the deformation displacement field vector is subtracted from the value at the corresponding position of the settlement rate scalar field to obtain the residual scalar field. The residual value of each node in the residual scalar field is squared to obtain the squared residual field; The squared residual values of all nodes in the squared residual field are multiplied by the physical volume weight of the corresponding voxel grid and then globally summed. The result of the global summation is the L2 norm spatial integral value of the regularization term of the terrain settlement gradient field, which is used as a soft constraint penalty term to limit the vertical displacement deformation of the voxel grid above the goaf. When the angle between the local normal vector of the voxel grid and the horizontal plane exceeds the soil repose angle threshold after deformation, the slope angle constraint term of the spoil heap applies a penalty potential energy according to the quadratic function of the excess angle. The soil angle of repose threshold is a dynamic parameter calibrated through indoor direct shear tests based on the specific soil and rock type of the spoil heap.
4. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins as described in claim 1 or 2, characterized in that, The logic for triggering local subdivision or merging reconstruction of the adaptive octree voxel mesh by the digital twin reconstruction engine is as follows: Real-time calculation of the spatial variation coefficient of available heavy metal concentration in soil within the current voxel grid and the spatial pH gradient of acidic mine wastewater; When the spatial variation coefficient is greater than the preset pollution heterogeneity threshold, or the pH spatial gradient is greater than the preset acid-base mutation threshold, a local subdivision mechanism is triggered to split the current voxel grid into eight sub-voxel grids; the pollution heterogeneity threshold is determined based on the fractal dimension of the mining area geochemical background survey, and is triggered when the spatial variation coefficient exceeds a preset multiple standard deviation of the background value fractal dimension. The subdivided voxel grids inherit the carbon sink accounting history of the parent grids and allocate the initial carbon pool of the parent grids through three-dimensional volume weighting. When the joint information entropy of the normalized vegetation index and soil moisture in adjacent sub-voxel grids is less than a preset homogeneity threshold, a grid merging operation is performed to merge multiple sub-voxel grids into a parent voxel grid. The joint information entropy is used to capture the nonlinear coupling relationship between the normalized vegetation index and soil moisture to eliminate false green areas.
5. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins according to claim 4, characterized in that, The dynamic carbon sequestration accounting module runs a vegetation-soil carbon sequestration coupled evolution model that integrates the mining area disturbance impulse function. The calculation logic for dynamic net carbon sequestration flux satisfies the following algorithm formula: ; In the formula: Let be the dynamic net carbon flux of the i-th voxel grid at time t; Total primary productivity of vegetation; For the rate of carbon emissions from soil respiration; The stress attenuation coefficient; The comprehensive environmental stress index; For multidimensional volume correction tensors; Let be the Dirac impulse function, representing the impulse at... Mining area disturbances that occur constantly; The instantaneous carbon release equivalent caused by the k-th perturbation event; The right side of the equation in the algorithm formula is formed by adding a continuous carbon sink integral term and a discrete perturbation pulse summation term, wherein the continuous carbon sink integral term includes... The first integral expression characterizes physiological and ecological processes; the discrete perturbation pulse summation term contains... The second summation expression characterizes the physical-mechanical destruction process; The method for obtaining the multidimensional volume correction tensor is as follows: the effective photosynthetic volume is calculated by inverting the canopy height using multispectral remote sensing data, the root penetration volume is calculated using the ground-penetrating radar echo delay preset at the center frequency, and the two are then multiplied by the Hadamard product. The instantaneous carbon release equivalent is obtained by capturing the vibration wave energy generated by blasting or heavy machinery operations through a micro-vibration sensor array deployed in the mining area, and calculating it based on the mapping relationship between vibration wave energy and soil porosity destruction rate.
6. The dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins according to claim 5, characterized in that, The calibration and acquisition methods for the stress attenuation coefficient and the comprehensive environmental stress index in the vegetation-soil carbon sequestration coupled evolution model are as follows: The comprehensive environmental stress index is calculated by integrating the available concentration of heavy metals in the soil, the mineral weathering rate, and the seepage coefficient of acid mine wastewater, using a time-varying objective weight model based on different remediation stages of the mining area. The specific calculation process of the time-varying objective weight model is as follows: obtain the first weight vector in the early stage of remediation and the second weight vector in the later stage of remediation; use the current remediation period as the independent variable and substitute it into the logistic function to calculate the transition coefficient at the current moment; use the transition coefficient to perform a weighted linear combination of the first weight vector and the second weight vector to obtain the time-varying objective weight of each indicator at the current moment; the transition midpoint of the logistic function is located at the midpoint of the transition between the soil physicochemical properties meeting the standards and the vegetation canopy closure in the mining area. The calibration method for the stress attenuation coefficient is as follows: Time-series data of net ecosystem exchange measured by the carbon flux monitoring tower in the mining area are obtained, and a Bayesian posterior probability model incorporating the vegetation-soil carbon sink coupled evolution model is constructed. The likelihood function of the Bayesian posterior probability model is composed of the sum of squared residuals between the dynamic net carbon sink flux output by the vegetation-soil carbon sink coupled evolution model and the measured net ecosystem exchange. The hyperparameters of the prior distribution of the Bayesian posterior probability model are determined by the historical blasting vibration frequency and the extreme value of soil bulk density in the mining area. The Metropolis-Hastings algorithm is used as the sampling kernel of the Markov chain Monte Carlo method to iteratively sample the posterior probability distribution of the stress attenuation coefficient. After removing samples from the pre-burning period, the expected value of the posterior distribution during the stationary period is taken as the final calibration value.
7. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins as described in claim 6, characterized in that, The calculation logic of the causal attribution and decision-making module for calculating the causal contribution of each grid node to carbon sequestration anomalies based on the counterfactual reasoning algorithm satisfies the following algorithm formula: ; In the formula The causal effect of the input feature intervention of the i-th grid on the total net carbon flux of the system; Let i be the input ecological feature vector of the i-th grid. The characteristic state after intervention; It is a set of structural cause-effect graph variables that contains all the hidden states of the grid. It is the set of nodes in a directed causal graph. Let j be the set of parent nodes of node j; For the joint probability distribution of the observed data; The integral term and mathematical expectation symbol in the algorithm formula The resulting difference term is the expected causal effect, which is calculated using the Monte Carlo sampling method with a sampling number of no less than one thousand, and the incoming edges of the parent node of the node being intervened are truncated during the sampling process. The edge direction of the directed structural causal graph is determined by the underground water flow direction of the mine and the runoff direction of the spoil heap slope. Furthermore, a mine functional zoning mask matrix is introduced during the graph convolution propagation process, which enables different state transition equations to be used between the spoil heap, tailings dam and reclamation area nodes.
8. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins according to claim 7, characterized in that, The adjacency matrix of the directed causal graph is constructed using a dual-driven mechanism of physics and data, specifically including: Extract the ecological feature vectors of each voxel grid and calculate the data-driven cosine similarity matrix; Based on the convection-diffusion equation for groundwater pollution plume diffusion in mines, the solute transport flux between each grid node is solved to construct a physically driven groundwater connectivity matrix. The cosine similarity matrix, the groundwater connectivity matrix, and the Gaussian kernel matrix based on spatial distance are fused using Hadamard product to obtain the final asymmetric adjacency matrix of the directed causal graph. The Hadamard product fusion is used to directly set the weight of the corresponding edge after fusion to zero when any element in the matrix is close to zero, thereby achieving hard truncation of physically unreachable or ecologically dissimilar node pairs.
9. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins as described in claim 1 or 2, characterized in that, It also includes a carbon asset anchoring and certification module, configured as follows: Extract the three-dimensional spatial coordinates of the adaptive octree voxel grid and the corresponding dynamic net carbon sink flux to generate a voxel-level carbon sink data package containing a spatial anchor fingerprint. The spatial anchor fingerprint is the octree path code of the voxel grid, the three-dimensional centroid coordinates and the concatenated hash of the net carbon sink flux at the current moment. The octree path code uniquely identifies the level position of the voxel in the adaptive octree. The spatial anchor fingerprint is used as the unique spatial identifier of the on-chain asset. The voxel-level carbon sink data package is bound one-to-one with the non-fungible token on the blockchain through smart contracts, realizing the triple anchoring of voxel spatial location, carbon sink physical quantity and on-chain digital asset. Any change in any dimension will trigger the automatic marking of the on-chain non-fungible token state. When a request to purchase externally verified voluntary emission reductions is received, the smart contract automatically verifies the real-time carbon sink status of the voxel grid corresponding to the spatial anchor fingerprint. If the status is normal, the token transfer is executed and the total carbon sink asset balance of the system is updated.
10. A dynamic monitoring and accounting system for mine ecological carbon sequestration based on digital twins according to claim 2, characterized in that, The specific logic for the closed-loop adaptive calibration module to perform two-layer time-scale data assimilation is as follows: A first Kalman filter is constructed to update slowly changing soil physicochemical parameters, and a second Kalman filter is constructed to update rapidly changing vegetation growth parameters. The variance of the process noise covariance matrix of the first Kalman filter is smaller than that of the second Kalman filter. The update cycle of the slowly changing parameters is no less than seven days, and the update cycle of the rapidly changing parameters is no more than one day. The observation updates of the two Kalman filters are performed alternately at unequal intervals on the time axis. The triggering mechanism of the closed-loop adaptive calibration module adopts a dual-path collaborative triggering: the first path detects the gradual drift of the calculated residual based on the cumulative sum control chart, and the second path detects the abrupt change anomaly of the calculated residual based on the Grubbs test; the outputs of the first path and the second path are logically ORed, and when the first path is triggered, the significance level of the Grubbs test in the second path is automatically increased to enhance the sensitivity to subsequent abrupt change detection.
