Carbon asset full-life-cycle traceability and digital asset management platform based on block chain

Abstract

The invention relates to the technical field of digital asset management, in particular to a block chain-based carbon asset full-life-cycle traceability and digital asset management platform. According to the platform, multiple data modeling and intelligent contract mechanisms are fused, and credible management of carbon assets in the whole process from generation, accounting, registration, transaction to auditing is achieved. The method comprises the following steps: collecting operation data of a forestry carbon sink project, constructing a structured Hash fingerprint, generating index root Hash by adopting a Merkle tree, and calculating a carbon sink amount as a basic data source of carbon assets; accounting the carbon sink credibility based on a Bayesian network model and eliminating abnormal data; carrying out anti-fact intervention analysis by utilizing a causal map, and screening abnormal assets with low causal consistency; on-chain matchmaking transaction of the carbon assets and the users is completed through a clustering and greedy matching mechanism; a life cycle auditing module is combined to evaluate responsibility contribution of key behavior variables to ESG indexes, and quantifiable, traceable and verifiable green asset full life cycle management is achieved.

Description

Technical Field

[0001] The present invention relates to the field of digital asset management technology, and specifically to a blockchain-based carbon asset full life cycle traceability and digital asset management platform. Background Art

[0002] As global climate change becomes increasingly serious, carbon assets, as an important vehicle for measuring the value of greenhouse gas emission reductions, are receiving widespread attention for their ownership, accounting, registration, trading, and supervision. However, the existing carbon asset management system generally has the following problems: On the one hand, the diverse and complex data sources of carbon assets, coupled with a lack of unified standards for data collection and storage, result in a lack of verifiability and credibility in carbon sink accounting, which can easily lead to compliance risks such as "data falsification" or "greenwashing." On the other hand, insufficient transparency in the registration and trading of carbon assets leads to issues such as unclear asset attributes, unclear user preferences, and inefficient transaction matching, severely hindering the efficient circulation of the carbon market and the optimal allocation of resources. Furthermore, the environmental (E), social (S), and governance (G) performance indicators of each stage of the carbon asset lifecycle are difficult to track and attribute consistently, and the lack of behavioral audits and accountability mechanisms for ESG results reduces the sustainability assessment value of the assets.

[0003] Blockchain technology, with its distributed storage, immutability, and traceability, offers a new solution for the full lifecycle management of carbon assets. However, most current blockchain-based carbon asset systems remain at the "on-chain ownership confirmation" stage, lacking deep-level support capabilities such as data credibility control, lifecycle behavior modeling, causal path identification, and ESG attribution auditing. These capabilities make it difficult to meet the urgent need for digital and intelligent management platforms required for the development of a high-standard carbon market.

[0004] To this end, a blockchain-based carbon asset full life cycle traceability and digital asset management platform is proposed. Summary of the Invention

[0005] This invention provides a blockchain-based platform for traceability and digital asset management of carbon assets throughout their lifecycle. By performing on-chain storage, credibility verification, causal analysis, and responsibility attribution of carbon sink data, it enables authentic accounting, compliant registration, efficient matching, and full-process auditing of carbon assets, enhancing the transparency and credibility of the carbon market.

[0006] To achieve the above object, the present invention provides the following technical solutions: The blockchain-based carbon asset full life cycle traceability and digital asset management platform includes: The data on-chain evidence storage module collects the operating data of forestry carbon sequestration projects and constructs a structured hash fingerprint. It generates an index root hash based on the Merkle tree and calculates the carbon sequestration amount of the operating data as the basic data source for carbon assets. The carbon sink accounting module is used to use the carbon sink as the observed variable based on the constructed Bayesian network model, determine the credibility of the input variable through posterior inference, and eliminate carbon asset data with credibility below the preset threshold; The digital registration module is used to register carbon assets on a non-homogeneous chain. It uses the causal relationship map of historical carbon assets to simulate the intervention response of the registered object on the key causal path and calculate the causal consistency disturbance score to screen and eliminate abnormal asset status. The on-chain transaction module is used to perform cluster analysis on carbon asset characteristics and user preferences, build asset pools and interested user groups, complete carbon asset transaction matching through an intra-group matching mechanism, and store transaction records on the chain; The life cycle audit module is used to model counterfactual trajectories based on the life cycle of carbon assets, compare the changes in ESG indicators under different paths, and evaluate the responsibility of key behavioral variables in the actual path for ESG outputs.

[0007] Furthermore, the steps of generating the index root hash based on the Merkle tree include: Perform structured extraction and hash calculation on the operating data of the forestry carbon sequestration project to obtain the corresponding hash fingerprint; Arrange all hash fingerprints in a preset order as leaf nodes of the Merkle tree; The hash values ​​of adjacent leaf nodes are merged in pairs and hashed again, and the hash tree structure is constructed layer by layer. Get the top-level root node hash value as the index root hash of the running data.

[0008] Furthermore, the step of determining the credibility of the input variables through a posteriori inference includes: The operating data is set as the input variable and the carbon sink is set as the observation variable. A Bayesian network structure is constructed to represent the causal dependency of each input variable in continuous time and to establish the conditional probability relationship between the input variable and the observation variable. The Bayesian network is trained using historical operating data to learn a priori probability distribution between input variables and observed variables; After receiving the carbon sink observation value at the current moment, the posterior probability distribution of each input variable under the given observation value condition is reversely calculated based on the Bayesian inference method; The posterior probability distribution is compared with the corresponding prior probability distribution, and the KL divergence is used to evaluate the degree of credibility deviation of the input variable under the current observation conditions.

[0009] Furthermore, the steps of screening abnormal asset status include: Based on historical registered carbon asset data, a causal relationship map between carbon asset variables is constructed, and key path variables in the causal relationship map are identified as a set of intervention variables; For the input data of the carbon assets to be registered, simulated intervention operations are applied sequentially along the key path to generate counterfactual samples; Calculate the response changes of the original data output and the intervention sample on the carbon sink valuation results to obtain the intervention response difference set; Taking all intervention results into consideration, the causal consistency disturbance score of the asset to be registered is calculated, which indicates the degree of consistency of its behavior under the key causal mechanism. If the disturbance score is lower than the preset causal consistency threshold, it will be marked as an abnormal asset state and its on-chain registration will be rejected.

[0010] Furthermore, the calculation process of the causal consistency disturbance score includes: Set the simulated intervention value for each intervention variable on the key causal path, keeping other variables unchanged; Calculate the carbon sequestration response value after intervention based on the trained Bayesian network and compare the difference with the original estimated value; The response deviations corresponding to all intervention variables were normalized and the disturbance score values ​​were calculated.

[0011] Furthermore, the implementation steps of the intra-group matching mechanism include: Construct the characteristic vector of carbon assets and the preference vector of user accounts, and perform cluster analysis on carbon assets and user accounts respectively using the DBSCAN algorithm to form several asset pools and corresponding intended user groups; Based on feature semantic mapping, a matching relationship graph is constructed between the asset pool and the user group. Within the same matching relationship graph, candidate matching pairs are constructed for carbon assets and user accounts, and matching similarity scores are calculated. According to the matching score sorting results, the greedy maximum matching strategy is used to determine the final transaction pair, and the matching results and transaction execution information are stored on the chain.

[0012] Furthermore, the implementation steps of the greedy maximum matching strategy include: Sort candidate transaction pairs in descending order according to the matching similarity score; Traverse the sorted candidate transaction pair set in sequence. For each pair of carbon assets and user accounts, if neither of them has participated in any transaction, mark it as a valid transaction pair. Repeat the above steps until the set of candidate trading pairs is completely traversed, and finally obtain the maximum valid trading pair set without duplication.

[0013] Furthermore, the steps to evaluate the key behavioral variables in the actual path include: Divide the carbon asset life cycle into multiple time series stages, extract behavioral variables and corresponding ESG output indicators for each stage; Build a time series forecasting model based on historical carbon asset data to generate ESG outcome trajectories of actual behavior paths; Perform intervention operations on the target behavioral variables to generate a set of ESG output trajectories under counterfactual paths, forming a set of counterfactual trajectories; Compare the differences in ESG output between the actual path and each counterfactual path to quantify the impact of behavioral variables on ESG indicators; According to the degree of influence, the responsibility contribution ranking of key behavioral variables is constructed to infer the key behavioral variables.

[0014] The beneficial effects of the present invention are: 1. This paper incorporates Bayesian networks and causal inference methods to model carbon sink causal pathways based on historical operational data. This method performs a posteriori credibility assessment on input variables and employs causal interventions to generate counterfactual samples, identifying asset states with abnormal perturbation responses. This mechanism effectively filters out assets with inflated carbon sinks and data anomalies, ensuring that carbon assets registered in the blockchain registry actually achieve emissions reductions, thereby enhancing the rigor and effectiveness of the system's overall carbon accounting.

[0015] 2. Through clustering algorithms and feature semantic mapping mechanisms, asset pools and user groups are constructed to efficiently match carbon assets with interested users. A greedy maximum matching strategy is then employed to optimize transaction matching results and improve transaction efficiency. Furthermore, combined with the lifecycle audit module, counterfactual trajectory modeling is employed to analyze the impact of key behavioral variables on ESG indicators. This provides transparent and quantifiable accountability for companies and regulators, promoting refined and compliant carbon asset management.

[0016] 3. This paper constructs a lifecycle behavioral path model for carbon assets and introduces counterfactual trajectory modeling techniques to simulate the ESG output changes of key behavioral variables under different intervention paths. The method systematically compares the indicator deviations between the actual and counterfactual paths, thereby quantifying the impact of each behavioral variable on the final ESG results. This method can identify the key drivers of ESG performance and support companies in auditing the entire lifecycle of carbon assets. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a schematic diagram of the structure of the blockchain-based carbon asset full life cycle traceability and digital asset management platform provided by the present invention; Figure 2It is an execution flow chart of the blockchain-based carbon asset full life cycle traceability and digital asset management platform. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0019] Example 1 Based on the blockchain carbon asset full life cycle traceability and digital asset management platform, such as Figure 1 As shown, including: The data on-chain evidence storage module collects the operating data of forestry carbon sequestration projects and constructs a structured hash fingerprint. It generates an index root hash based on the Merkle tree and calculates the carbon sequestration amount of the operating data as the basic data source for carbon assets. Furthermore, the steps of generating the index root hash based on the Merkle tree include: Perform structured extraction and hash calculation on the operating data of the forestry carbon sequestration project to obtain the corresponding hash fingerprint; Arrange all hash fingerprints in a preset order as leaf nodes of the Merkle tree; The hash values ​​of adjacent leaf nodes are merged in pairs and hashed again, and the hash tree structure is constructed layer by layer. Get the top-level root node hash value as the index root hash of the running data.

[0020] Specifically, the original operational data collected by the forestry carbon sequestration project within a certain period of time is extracted, including: basic forest information, aboveground biomass data, underground biomass data, litter and dead wood data, soil carbon storage data, and meteorological and environmental data; these operational data are structured in a unified format to ensure data structure consistency. The extracted structured data is hashed using a hash function (such as SHA-256) to obtain a unique hash fingerprint for each subfield. For example, the forest area in the basic forest information is "1200km 2 ", the hash value is H1. If the carbon storage in the tree layer in the aboveground biomass data is "3.6 million tons", the hash value is H2, and so on. Arrange all hash fingerprints in the preset field order and use them as leaf nodes of the Merkle tree in turn. This order is fixed and consistent for all data instances to ensure that the generated results are deterministic and verifiable. The hash values ​​of two adjacent leaf nodes are concatenated and then hashed. For example, , , and so on, to obtain the hash value of the upper-level node. If the number of nodes is odd, the last node can copy itself to participate in the merge to ensure that the number of node pairs is even. Repeat the above hash merge process, recursively calculating upward layer by layer, until a unique top-level root hash node is finally constructed, which is the index root hash of the entire set of field-level data.

[0021] By constructing a hash fingerprint for the operating data of the forestry carbon sink project and using the Merkle tree structure to hash layer by layer to generate the index root hash, it not only achieves the integrity verification and tamper-proof traceability of the original data, but also supports rapid data comparison and local verification, significantly improving the security and traceability of carbon asset basic data.

[0022] It is worth noting that the calculation of carbon sequestration using forestry carbon sequestration project operation data such as basic forest land information, aboveground biomass data, belowground biomass data, litter and dead wood data, and soil carbon storage data is a publicly available technology. The specific calculation method can be implemented according to the local standard DB42 / T 2303-2024 or the IPCC standard, and will not be repeated here.

[0023] The carbon sink accounting module is used to use the carbon sink as the observed variable based on the constructed Bayesian network model, determine the credibility of the input variable through posterior inference, and eliminate carbon asset data with credibility below the preset threshold; Furthermore, the step of determining the credibility of the input variables through a posteriori inference includes: The operating data is set as the input variable and the carbon sink is set as the observation variable. A Bayesian network structure is constructed to represent the causal dependency of each input variable in continuous time and to establish the conditional probability relationship between the input variable and the observation variable. The Bayesian network is trained using historical operating data to learn a priori probability distribution between input variables and observed variables; After receiving the carbon sink observation value at the current moment, the posterior probability distribution of each input variable under the given observation value condition is reversely calculated based on the Bayesian inference method; The posterior probability distribution is compared with the corresponding prior probability distribution, and the KL divergence is used to evaluate the degree of credibility deviation of the input variable under the current observation conditions.

[0024] Specifically, multiple variables from the forestry carbon sink project's operational data are set as input nodes of the Bayesian network model, including basic forestland information, aboveground biomass data, belowground biomass data, litter and dead wood data, soil carbon storage data, and meteorological and environmental data. The carbon sink values ​​calculated by the system are used as observation variables and set as output nodes of the network. A time-series Bayesian network structure is constructed to model the causal dependencies between these input variables over time. Conditional probability relationships between the input variables and the observed variables are also established, forming a joint probability graphical model. The Bayesian network is trained using collected operational data from the forestry carbon sink project. Maximum likelihood estimation is used to learn and solidify the prior probability distribution and conditional probability table for each input variable, forming a stable model structure. In actual operation, after receiving the current carbon sink observation in real time, the system uses Bayesian reasoning to infer the posterior probability distribution of the input variables, calculating the posterior distribution of each input variable given the observed value. For each input variable, the degree of difference between its posterior distribution under the current observation conditions and the historical prior distribution is calculated respectively, and the KL divergence is used as the credibility deviation measurement indicator. If the KL divergence of an input variable is higher than the preset threshold (for example, 0.5), it means that the variable shows abnormal deviation in the current context and its credibility is low.

[0025] By modeling the operating data of forestry carbon sink projects as a multivariate Bayesian network and implementing posterior inference with carbon sink amount as the observed variable, it is possible to not only dynamically capture the causal dependency between input variables and carbon sink results, but also quantify the degree of deviation between current input data and historical patterns with the help of KL divergence, thereby effectively identifying abnormal or distorted data and improving the credibility and audit reliability of carbon asset accounting results.

[0026] The digital registration module is used to register carbon assets on a non-homogeneous chain. It uses the causal relationship map of historical carbon assets to simulate the intervention response of the registered object on the key causal path and calculate the causal consistency disturbance score to screen and eliminate abnormal asset status. Furthermore, the steps of screening abnormal asset status include: Based on historical registered carbon asset data, a causal relationship map between carbon asset variables is constructed, and key path variables in the causal relationship map are identified as a set of intervention variables; For the input data of the carbon assets to be registered, simulated intervention operations are applied sequentially along the key path to generate counterfactual samples; Compare the response changes of the original data output and the intervention sample in carbon sequestration valuation results, and calculate the intervention response difference index; Taking all intervention results into consideration, the causal consistency disturbance score of the asset to be registered is calculated, which indicates the degree of consistency of its behavior under the key causal mechanism. If the disturbance score is lower than the preset causal consistency threshold, it will be marked as an abnormal asset state and its on-chain registration will be rejected.

[0027] Specifically, the platform first extracts multidimensional attribute variables from the historically registered carbon asset database, including basic forest information, aboveground biomass data, underground biomass data, litter and dead wood data, soil carbon storage data, and meteorological and environmental data; it uses a structural learning algorithm (such as a causal structural learning algorithm or a causal relationship discovery algorithm, and this embodiment uses a causal relationship discovery algorithm) to construct a causal graph, identify the causal path relationship between variables, and mark the key path variables. Key path variables refer to a set of causal variables that have a significant impact on output indicators (such as carbon sequestration), which are recorded as intervention variable sets. , for the carbon asset records to be registered, extract their structured input data. For each intervention variable Based on the counterfactual reasoning logic, perturbation operations are applied to it in turn (such as adding / subtracting the standard deviation range, replacing it with a quantile value, etc.), and the influence is propagated through the dependency path in the causal graph to generate counterfactual samples after intervention , while keeping other variables unchanged, for each intervention sample Use the trained Bayesian network model in the platform to predict the corresponding carbon sink output The platform records the original sample’s valuation output as , then the response change after each intervention is defined as: After all intervention operations are completed, the intervention response difference set is obtained The platform calculates the average response fluctuation based on all intervention response differences: ,in This is the causal consistency perturbation score of the asset to be registered along the key causal path. A lower score indicates that its behavior is stable under intervention and consistent with historical causal mechanisms. A higher score indicates significant behavioral deviations, potentially posing risks of input anomalies or data falsification.

[0028] By imposing simulated interventions on key causal paths of registered carbon assets and observing changes in their outputs, we can effectively capture their behavioral consistency under the core carbon sink mechanism, and then quantitatively measure their causal rationality with disturbance scores. Compared with traditional anomaly detection methods, this method has greater causal explanatory power and audit transparency, and improves the authenticity and credibility of carbon asset registration from the source.

[0029] Furthermore, the calculation process of the causal consistency disturbance score includes: Set the simulated intervention value for each intervention variable on the key causal path, keeping other variables unchanged; Calculate the carbon sequestration response value after intervention based on the trained Bayesian network and compare the difference with the original estimated value; The response deviations corresponding to all intervention variables were normalized and the disturbance score values ​​were calculated.

[0030] By applying controlled interventions to the variables on the key causal path one by one and evaluating the changes in carbon sink responses while keeping other variables unchanged, the local sensitivity and behavioral stability of the carbon assets to be registered under the causal logic can be accurately identified; by normalizing and integrating the response deviations of each variable to form a unified disturbance score, it is helpful to achieve quantitative judgment of the causal consistency of carbon assets, thereby effectively screening out potential false carbon assets in the registration process.

[0031] The on-chain transaction module is used to perform cluster analysis on carbon asset characteristics and user preferences, build asset pools and interested user groups, complete carbon asset transaction matching through an intra-group matching mechanism, and store transaction records on the chain; Furthermore, the implementation steps of the intra-group matching mechanism include: Construct the characteristic vector of carbon assets and the preference vector of user accounts, and perform cluster analysis on carbon assets and user accounts respectively using the DBSCAN algorithm to form several asset pools and corresponding intended user groups; Based on feature semantic mapping, a matching relationship graph is constructed between the asset pool and the user group. Within the same matching relationship graph, candidate matching pairs are constructed for carbon assets and user accounts, and matching similarity scores are calculated. According to the matching score sorting results, the greedy maximum matching strategy is used to determine the final transaction pair, and the matching results and transaction execution information are stored on the chain.

[0032] Specifically, the platform first extracts structured features for each carbon asset instance, such as project type (such as tree forest, shrub forest), geographical location, land type, expected carbon sequestration, certification status, life cycle stage, and ESG performance. The above feature vectors are quantified into low-dimensional embedding representations, which are recorded as carbon asset feature vectors. Similarly, user accounts set their trading preference vectors, such as preferred asset types, target regions, acceptable carbon sink ranges, project risk tolerance, ESG preference, etc., which constitute the user preference vector Using DBSCAN density clustering algorithm, all carbon asset feature vector sets are and user preference vector set Perform cluster analysis to obtain several asset clusters As an intent pool, and several user clusters As the intended user group. To achieve accurate matching across clusters, the platform constructs a matching relationship map between asset pools and user groups. The method is as follows: Calculate each pair of asset pools With user groups The semantic average vector of the samples in is used as the cluster center; Use feature semantic mapping models (such as cosine similarity, Euclidean distance, BERT embedding similarity, etc.) to measure the similarity between cluster centers; Filter out those whose matching degree exceeds the preset threshold ( , )Yes, establish a matching relationship map.

[0033] In the matching relationship graph, the platform enumerates each combination of carbon assets and candidate user accounts and calculates their matching similarity scores. , generate a list of all candidate matching pairs and their corresponding scores, and use a greedy maximum matching strategy to generate the final trading pair.

[0034] By constructing feature vectors and clustering analysis of carbon assets and user accounts, combining feature semantic mapping to establish a matching relationship map, and using matching similarity scoring and greedy maximum matching strategy for efficient matching, it is possible to significantly improve the efficiency and matching accuracy of carbon asset trading while ensuring transaction compliance and preference alignment. At the same time, with the help of on-chain evidence storage, the transaction process can be ensured to be open, transparent, auditable and traceable.

[0035] Furthermore, the implementation steps of the greedy maximum matching strategy include: Sort candidate transaction pairs in descending order according to the matching similarity score; Traverse the sorted candidate transaction pair set in sequence. For each pair of carbon assets and user accounts, if neither of them has participated in any transaction, mark it as a valid transaction pair. Repeat the above steps until the set of candidate trading pairs is completely traversed, and finally obtain the maximum valid trading pair set without duplication.

[0036] By sorting candidate trading pairs by similarity scores and sequentially screening the optimal trading pairs that do not conflict with each other, we can quickly build the largest effective matching set while ensuring transaction fairness, significantly improving the overall efficiency and resource utilization of carbon asset trading matching, while avoiding duplicate matching of accounts.

[0037] The life cycle audit module is used to model counterfactual trajectories based on the life cycle of carbon assets, compare the changes in ESG indicators under different paths, and evaluate the responsibility of key behavioral variables in the actual path for ESG outputs.

[0038] Furthermore, the steps to evaluate the key behavioral variables in the actual path include: Divide the carbon asset life cycle into multiple time series stages, extract behavioral variables and corresponding ESG output indicators for each stage; Build a time series forecasting model based on historical carbon asset data to generate ESG outcome trajectories of actual behavior paths; Perform intervention operations on the target behavioral variables to generate a set of ESG output trajectories under counterfactual paths, forming a set of counterfactual trajectories; Compare the differences in ESG output between the actual path and each counterfactual path to quantify the impact of behavioral variables on ESG indicators; According to the degree of influence, the responsibility contribution ranking of key behavioral variables is constructed to infer the key behavioral variables.

[0039] Specifically, the entire life cycle of carbon assets is divided into several key stages, including but not limited to data on-chain storage, carbon sink accounting, digital registration, and on-chain transactions. In each stage, observable behavioral variables are extracted (for example, collecting operational data, eliminating untrustworthy data, eliminating abnormal asset status, etc.), as well as ESG indicators (such as carbon sinks) generated in the corresponding stage. Based on historical carbon asset sample data, a time series prediction model is constructed. , used to fit the behavioral data of each stage under the real path and predict the actual ESG output trajectory ,in Indicates the The behavioral variable vector of the stage, represents the corresponding ESG output vector, Represents a time series prediction network, which can use LSTM, Transformer, or RNN network enhanced by causal inference (this embodiment uses LSTM model). , conduct counterfactual intervention while keeping other variables unchanged ( ), and finally form a behavioral variable sequence. Input the behavioral variable sequence after intervention into the time series prediction model to generate the corresponding counterfactual trajectory. Since there are many ways to intervene in counterfactual behavior, multiple intervention versions are generated by enumeration to form a set of counterfactual trajectories. Comparing the deviations between the actual path and each counterfactual path in terms of ESG output dimensions, the impact of the variables is quantified as follows: ; in, Representing behavioral variables The average degree of disturbance to ESG outputs under all intervention trajectories. Construct a ranking table of responsibility contributions and identify the behavioral variables that have the greatest impact on ESG indicators as key behavioral variables.

[0040] By modeling the carbon asset life cycle as a multi-stage behavioral sequence and introducing counterfactual trajectory generation and comparative analysis methods, we can accurately quantify the causal impact of each behavioral variable on ESG output indicators, effectively identify key behavioral nodes that have substantial contributions to carbon asset performance, thereby providing a scientific basis for responsibility attribution judgment and performance auditing, and improving the transparency and credibility of carbon asset management.

[0041] Example 2 This embodiment provides a blockchain-based carbon asset full life cycle traceability and digital asset management platform, the execution process is as follows: Figure 2 The platform is applied to the carbon asset generation, accounting, registration, trading and audit management of the forestry carbon sink project in Zhen'an County, Shaanxi Province, specifically including the collaborative work process of the following modules: The platform first collects multi-source operational data from the Zhen'an County Forestry Carbon Sequestration Project, including forest area, tree species, age group structure, biomass monitoring data, soil sample information, and meteorological and environmental records. It then constructs a hash fingerprint at the field level. Each field is hashed to create a Merkle tree structure, with the root hash used to uniquely index the data structure. Carbon sequestration data (including above- and below-ground biomass carbon storage) is then calculated as the foundational data for carbon assets, ensuring data authenticity and traceability.

[0042] A Bayesian network model was constructed, using variables such as forest soil carbon storage, tree biomass, temperature, and precipitation as inputs, and carbon sequestration as the observed variable. The network model was trained using historical data to obtain a priori distributions. After receiving the current observations, a posteriori inference was performed to identify and remove data records that did not meet the requirements, thereby improving the credibility of carbon asset accounting. After learning the network structure, the model possessed the following capabilities: After receiving a new round of operational data and carbon sink observations, the credibility relationship between each input variable and the observation is evaluated using a posteriori inference; By comparing the prior probability and posterior probability of the input variable, the KL divergence is calculated as the credible deviation of the variable; If the credibility of a variable under observation conditions is lower than the threshold (such as KL divergence>0.5), the carbon asset data is judged to be abnormal and removed from the subsequent processing flow.

[0043] For data that passes credibility checks, the platform registers the corresponding carbon assets as non-fungible tokens (NFTs). Each NFT contains: a blockchain index hash, asset generation time and location, a data summary, and calculated carbon sinks. To prevent carbon asset counterfeiting or fraud, the platform utilizes a behavioral simulation mechanism based on causal inference to construct a causal graph for carbon assets. During the registration phase, the system automatically performs counterfactual interventions on key variables. If the intervention response along the critical path is extremely inconsistent (i.e., ESG valuation fluctuations are abnormal), the asset is flagged as suspicious through a perturbation scoring mechanism and its on-chain registration is rejected.

[0044] The platform uses the DBSCAN clustering algorithm to construct feature vectors (including expected carbon sequestration volume, geographic location, and land type) for all carbon assets and cluster them into asset pools. Simultaneously, user account preference vectors (such as carbon sequestration targets, target regions, and preferred asset types) are clustered to form groups of interested users. The system constructs a graph structure of asset pools and user groups through feature semantic matching. Within each corresponding graph, it generates matching candidate pairs, calculates similarity scores (such as cosine similarity and weighted Euclidean distance), and sorts them in descending order. A greedy maximum matching strategy is employed: after traversing the sorted candidate pairs, if neither the carbon asset nor the user has participated in other transactions, the pair is considered a valid trading pair, ensuring optimal allocation efficiency. Once a transaction is concluded, the system packages the transaction records (including asset ID, buyer and seller, transaction time, etc.) and stores them on-chain.

[0045] The platform regularly performs ESG audits on registered carbon assets. The process is as follows: Divide the carbon asset life cycle into multiple time series stages, extract behavioral variables and corresponding ESG output indicators for each stage; Build a time series forecasting model based on historical carbon asset data to generate ESG outcome trajectories of actual behavior paths; Perform intervention operations on the target behavioral variables to generate a set of ESG output trajectories under counterfactual paths, forming a set of counterfactual trajectories; Compare the differences in ESG output between the actual path and each counterfactual path to quantify the impact of behavioral variables on ESG indicators; According to the degree of influence, the responsibility contribution ranking of key behavioral variables is constructed to infer the key behavioral variables.

[0046] Through this implementation plan, the system achieves closed-loop management of the entire carbon asset lifecycle, from data collection, accounting, registration, trading, and auditing. Blockchain technology ensures data traceability and immutability. Combined with causal inference and clustering algorithms, it significantly improves the authenticity identification of carbon assets and market circulation efficiency, demonstrating its potential in promoting green finance.

[0047] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. Based on the blockchain carbon asset full life cycle traceability and digital asset management platform, it is characterized by: include: The data on-chain evidence storage module collects the operating data of forestry carbon sequestration projects and constructs a structured hash fingerprint. It generates an index root hash based on the Merkle tree and calculates the carbon sequestration amount of the operating data as the basic data source for carbon assets. The carbon sink accounting module is used to use the carbon sink as the observed variable based on the constructed Bayesian network model, determine the credibility of the input variable through posterior inference, and eliminate carbon asset data with credibility below the preset threshold; The digital registration module is used to register carbon assets on a non-homogeneous chain. It uses the causal relationship map of historical carbon assets to simulate the intervention response of the registered object on the key causal path and calculate the causal consistency disturbance score to screen and eliminate abnormal asset status. The on-chain transaction module is used to perform cluster analysis on carbon asset characteristics and user preferences, build asset pools and interested user groups, complete carbon asset transaction matching through an intra-group matching mechanism, and store transaction records on the chain; The life cycle audit module is used to model counterfactual trajectories based on the life cycle of carbon assets, compare the changes in ESG indicators under different paths, and evaluate the responsibility of key behavioral variables in the actual path for ESG outputs.

2. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 1 is characterized in that: The steps to generate the index root hash based on the Merkle tree include: Perform structured extraction and hash calculation on the operating data of the forestry carbon sequestration project to obtain the corresponding hash fingerprint; Arrange all hash fingerprints in a preset order as leaf nodes of the Merkle tree; The hash values ​​of adjacent leaf nodes are merged in pairs and hashed again, and the hash tree structure is constructed layer by layer. Get the top-level root node hash value as the index root hash of the running data.

3. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 1 is characterized in that: The steps to determine the credibility of input variables through a posteriori inference include: The operating data is set as the input variable and the carbon sink is set as the observation variable. A Bayesian network structure is constructed to represent the causal dependency of each input variable in continuous time and to establish the conditional probability relationship between the input variable and the observation variable. The Bayesian network is trained using historical operating data to learn a priori probability distribution between input variables and observed variables; After receiving the carbon sink observation value at the current moment, the posterior probability distribution of each input variable under the given observation value condition is reversely calculated based on the Bayesian inference method; The posterior probability distribution is compared with the corresponding prior probability distribution, and the KL divergence is used to evaluate the degree of credibility deviation of the input variable under the current observation conditions.

4. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 1 is characterized in that: The steps to screen for abnormal asset status include: Based on historical registered carbon asset data, a causal relationship map between carbon asset variables is constructed, and key path variables in the causal relationship map are identified as a set of intervention variables; For the input data of the carbon assets to be registered, simulated intervention operations are applied sequentially along the key path to generate counterfactual samples; Calculate the response changes of the original data output and the intervention sample on the carbon sink valuation results to obtain the intervention response difference set; Taking all intervention results into consideration, the causal consistency disturbance score of the asset to be registered is calculated, which indicates the degree of consistency of its behavior under the key causal mechanism. If the disturbance score is lower than the preset causal consistency threshold, it will be marked as an abnormal asset state and its on-chain registration will be rejected.

5. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 4 is characterized in that: The calculation process of the causal consistency perturbation score includes: Set the simulated intervention value for each intervention variable on the key causal path, keeping other variables unchanged; Calculate the carbon sequestration response value after intervention based on the trained Bayesian network and compare the difference with the original estimated value; The response deviations corresponding to all intervention variables were normalized and the disturbance score values ​​were calculated.

6. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 1 is characterized in that: The implementation steps of the intra-group matching mechanism include: Construct the characteristic vector of carbon assets and the preference vector of user accounts, and perform cluster analysis on carbon assets and user accounts respectively using the DBSCAN algorithm to form several asset pools and corresponding intended user groups; Based on feature semantic mapping, a matching relationship graph is constructed between the asset pool and the user group. Within the same matching relationship graph, candidate matching pairs are constructed for carbon assets and user accounts, and matching similarity scores are calculated. According to the matching score sorting results, the greedy maximum matching strategy is used to determine the final transaction pair, and the matching results and transaction execution information are stored on the chain.

7. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 6 is characterized in that: The implementation steps of the greedy maximum matching strategy include: Sort candidate transaction pairs in descending order according to the matching similarity score; Traverse the sorted candidate transaction pair set in sequence. For each pair of carbon assets and user accounts, if neither of them has participated in any transaction, mark it as a valid transaction pair. Repeat the above steps until the set of candidate trading pairs is completely traversed, and finally obtain the maximum valid trading pair set without duplication.

8. The blockchain-based carbon asset full life cycle traceability and digital asset management platform according to claim 1 is characterized in that: The steps to assess key behavioral variables in the actual path include: Divide the carbon asset life cycle into multiple time series stages, extract behavioral variables and corresponding ESG output indicators for each stage; Build a time series forecasting model based on historical carbon asset data to generate ESG outcome trajectories of actual behavior paths; Perform intervention operations on the target behavioral variables to generate a set of ESG output trajectories under counterfactual paths, forming a set of counterfactual trajectories; Compare the differences in ESG output between the actual path and each counterfactual path to quantify the impact of behavioral variables on ESG indicators; According to the degree of influence, the responsibility contribution ranking of key behavioral variables is constructed to infer the key behavioral variables.

Images