Carbon emission management method and system, storage medium and computer program product

Abstract

The invention discloses a carbon emission management method and system, a storage medium and a computer program product, and relates to the technical field of carbon emission management, and the method comprises the steps: obtaining a carbon emission management instruction inputted by a user; a hierarchical digital twinborn sub-model corresponding to the carbon emission management instruction is called in a pre-constructed multi-hierarchical digital twinborn model system, and the multi-hierarchical digital twinborn model system is divided into a plurality of hierarchical digital twinborn sub-models in a mode of combining administrative region division and project division; if the called hierarchical digital twinning sub-model is a project-level digital twinning model, acquiring project energy consumption data, and calculating an annual cycle total carbon emission amount based on the project energy consumption data; calling a pre-constructed map generator to extract a carbon quota parameter; and generating a carbon quota scheme according to the carbon quota parameter and the annual cycle carbon emission total amount. The method adapts to regional policy differences through a multi-level digital twinborn model system, and supports zero-carbon target quantitative management.

Description

Technical Field

[0001] The present application relates to the technical field of carbon emission management, and in particular to a carbon emission management method, system, storage medium, and computer program product. Background Art

[0002] Different regions have introduced differentiated policies for carbon emissions management, with significant regional variations in carbon quota calculation methods, compliance requirements, and emission reduction incentives. However, mainstream carbon emissions management systems often utilize a single architecture, making them difficult to adapt to the dynamic policy adjustments required by multiple administrative levels (province, city, and district). This results in delayed policy parameter extraction and inefficient cross-regional data collaboration, hindering the provision of accurate compliance guidance for businesses. Furthermore, carbon emissions management in established industrial parks faces the pain points of data fragmentation, delayed monitoring, and empirical decision-making. Industrial park equipment energy consumption data, carbon asset information, and policy documents are scattered across disparate systems, lacking a unified data center platform. This makes it difficult to achieve a closed-loop management system encompassing real-time monitoring, trend prediction, and solution optimization. Furthermore, digital tools are lacking to support the full lifecycle management of carbon assets, such as emission-controlled enterprise quotas, CCERs (Certified Emission Reductions), and green certificates. The disconnect between compliance and market transactions makes it difficult to reduce emission reduction costs through carbon asset optimization.

[0003] Therefore, how to adapt to regional policy differences and support the quantitative management of zero-carbon targets has become a technical problem that needs to be urgently solved in this application.

[0004] The above content is only used to assist in understanding the technical solution of this application and does not constitute an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a carbon emission management method, system, storage medium and computer program product, aiming to solve the technical problem of how to adapt to regional policy differences and support the quantitative management of zero-carbon targets.

[0006] To achieve the above objectives, this application proposes a carbon emission management method, which includes: Obtain carbon emission management instructions input by the user; Invoking a hierarchical digital twin sub-model corresponding to the carbon emission management directive in a pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative region division and project division; If the hierarchical digital twin sub-model called is a project-level digital twin model, then the project energy consumption data is collected, and the total annual carbon emissions are calculated based on the project energy consumption data; Call the pre-built graph generator to extract carbon quota parameters; A carbon quota plan is generated according to the carbon quota parameters and the total carbon emissions for the annual period.

[0007] In one embodiment, before the step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, the step further includes: Defining a two-dimensional division rule based on the administrative region level and the project entity attributes, wherein the two-dimensional division rule includes a geographical boundary division rule and a project attribution division rule; Collect remote sensing geographic information and rule text based on the geographic boundary division rules, and collect project point cloud data using drone oblique photography based on the project attribution division rules; Binding the policy text with the remote sensing geographic information to construct a first-level digital twin model, a second-level digital twin model, and a third-level digital twin model respectively; Distributedly deploying federated learning nodes in the first-level digital twin model, the second-level digital twin model, and the third-level digital twin model; A project-level digital twin model is constructed based on the project point cloud data, a lightweight data upload interface is developed, and the federated learning nodes are connected based on the lightweight data upload interface to obtain a multi-level digital twin model system.

[0008] In one embodiment, the step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system includes: Parsing the carbon emission management directive and extracting the level identification parameters; Retrieving the multi-level digital twin model system based on the level identification parameter and locating the level digital twin sub-model; A call request is sent to the hierarchical digital twin sub-model, the hierarchical digital twin sub-model is called, and a policy text corresponding to the hierarchical identification parameter is rendered in a visual interface.

[0009] In one embodiment, if the hierarchical digital twin sub-model called is a project-level digital twin model, the steps of collecting project energy consumption data and calculating the total annual carbon emissions based on the project energy consumption data include: An IoT gateway based on the project-level digital twin model collects project energy consumption data; Unifying the project energy consumption data to a standard time axis, and using Kalman filtering to remove outliers in the project energy consumption data to obtain standardized project energy consumption data; Call the annual energy structure basic data from the third-level digital twin model and obtain the energy structure correction coefficient issued by the first-level digital twin model through the federal node; Multiplying the annual energy structure basic data and the energy structure correction coefficient to obtain a carbon emission factor; Recall historical data for the same period, and predict energy consumption data for the remaining period based on the historical data for the same period and the carbon emission factor; The energy consumption data of the standardized project and the energy consumption data of the remaining period are accumulated to obtain the total carbon emissions for the annual cycle.

[0010] In one embodiment, after the step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, the step further includes: If the hierarchical digital twin sub-model called is a first-level digital twin model, a pre-built graph generator is called to identify the first-level rule text and extract the first instruction parameter set; generating a regional energy structure correction coefficient based on the first instruction parameter set; Based on the federated learning node, obtain the industrial structure feature vector and energy consumption intensity feature vector uploaded from the second-level digital twin model, and generate an administrative division emission reduction heat map based on the industrial structure feature vector and the energy consumption intensity feature vector; If the called hierarchical digital twin sub-model is a second-level digital twin model, calling the graph generator to identify the second-level rule text and extract the second instruction parameter set; Obtain the second-level industrial structure data, overlay the second-level industrial structure data and the second-level geographic information layer to determine the vector boundaries of high-energy-consuming industrial clusters; Call the regional energy structure correction coefficient issued by the first-level digital twin model, split the regional energy structure correction coefficient according to the second-level industrial structure data, and generate an industry-specific energy structure correction coefficient matrix; If the called hierarchical digital twin sub-model is a third-level digital twin model, calling the graph generator to identify the third-level policy text and extract the third instruction parameter set; The project information uploaded by the project-level digital twin model is called, and a visual project list is generated based on the project information.

[0011] In one embodiment, the step of calling a pre-built graph generator to extract carbon quota parameters includes: Calling a pre-built graph generator to traverse the rule text and extract core carbon asset parameters, wherein the rule text includes a first-level rule text, a second-level rule text, and a third-level rule text; The core carbon asset parameters are logically checked and associated with the benchmark values ​​of the industry to which the project belongs to obtain carbon quota parameters.

[0012] In one embodiment, the step of generating a carbon quota scheme based on the carbon quota parameters and the annual carbon emissions includes: Calculate the allowance shortfall based on the total carbon emissions for the annual period; Calculating the quota gap and the carbon quota parameters using a multi-objective optimization algorithm to generate an initial emission reduction plan; The energy structure correction coefficient is synchronized through the federated learning node, and the initial emission reduction plan is modified to obtain a carbon quota plan.

[0013] In addition, to achieve the above-mentioned purpose, the present application also proposes a carbon emission management system, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the computer program is configured to implement the steps of the carbon emission management method as described above.

[0014] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium. A computer program is stored on the storage medium, and when the computer program is executed by the processor, the steps of the carbon emission management method described above are implemented.

[0015] In addition, to achieve the above objectives, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the carbon emission management method as described above.

[0016] One or more technical solutions proposed in this application have at least the following technical effects: Obtain the carbon emission management instruction input by the user; call the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative area division and project division; if the hierarchical digital twin sub-model called is a project-level digital twin model, then collect the project energy consumption data, and calculate the total annual carbon emissions based on the project energy consumption data; call the pre-built map generator to extract the carbon quota parameters; generate a carbon quota plan based on the carbon quota parameters and the total annual carbon emissions. In this application, by constructing a multi-level digital twin model system with two dimensions of "administrative area + project entity", the carbon emission management instruction input by the user can be directly mapped to the sub-model of the corresponding level; when the instruction points to the project-level model, the project energy consumption data is collected immediately and the total annual carbon emissions are calculated. At the same time, the map generator is called to parse the differentiated carbon quota parameters of the region, and an accurate carbon quota plan is automatically generated. Through instructions, calling models, extracting carbon quota parameters, and forming a closed-loop mechanism for carbon quota plans, we can not only achieve dynamic adaptation to policy differences across regions, but also quantify the zero-carbon target into a predictable, adjustable, and tradable quota path, thereby supporting the quantitative management and implementation of the park's zero-carbon target throughout the entire process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 A flowchart of the first embodiment of the carbon emission management method of this application is provided; Figure 2 A flow chart illustrating the second embodiment of the carbon emission management method of this application; Figure 3 A flowchart of the fourth embodiment of the carbon emission management method of this application is provided; Figure 4 This is a schematic diagram of the module structure of the carbon emission management device according to an embodiment of the present application; Figure 5 This is a schematic diagram of the device structure of the hardware operating environment involved in the carbon emission management method in the embodiment of the present application.

[0020] The purpose, features and advantages of this application will be further explained with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0021] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0022] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0023] The main solution of the embodiment of the present application is: obtaining the carbon emission management instructions input by the user; calling the hierarchical digital twin sub-model corresponding to the carbon emission management instructions in a pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative area division and project division; if the called hierarchical digital twin sub-model is a project-level digital twin model, then the project energy consumption data is collected, and the total annual carbon emissions are calculated based on the project energy consumption data; calling the pre-built map generator to extract carbon quota parameters; and generating a carbon quota plan based on the carbon quota parameters and the total annual carbon emissions.

[0024] The present embodiments take into account the fact that different regions have introduced differentiated policies for carbon emissions management, resulting in significant regional variations in carbon quota calculation methods, compliance requirements, and emission reduction incentives. However, existing carbon emissions management systems often employ a single architecture, making them difficult to adapt to the dynamic policy adjustments required by multiple administrative levels (province, city, and district). This results in delayed policy parameter extraction and inefficient cross-regional data collaboration, making it impossible to provide enterprises with accurate compliance guidance. Furthermore, carbon emissions management in established industrial parks faces the pain points of data fragmentation, delayed monitoring, and empirical decision-making. Industrial park equipment energy consumption data, carbon asset information, and policy documents are scattered across disparate systems, lacking a unified data center platform. This makes it difficult to achieve a closed-loop management system encompassing real-time monitoring, trend prediction, and solution optimization. Furthermore, digital tools are lacking to support the full lifecycle management of carbon assets, such as emission-controlled enterprise quotas, CCERs (Certified Emission Reductions), and green certificates. The disconnect between compliance and market transactions makes it difficult to reduce emission reduction costs through carbon asset optimization.

[0025] Therefore, the solution provided by this application, by constructing a multi-level digital twin model system with two dimensions of "administrative region + project entity", enables the carbon emission management instructions input by users to be directly mapped to the sub-model of the corresponding level; when the instruction points to the project-level model, the project energy consumption data is collected in real time and the total annual carbon emissions are calculated. At the same time, the map generator is called to parse the differentiated carbon quota parameters of the region and automatically generate an accurate carbon quota plan. Through the closed-loop mechanism of instructions, calling models, extracting carbon quota parameters, and forming a carbon quota plan, it not only achieves dynamic adaptation to policy differences between regions, but also quantifies the zero-carbon target into a predictable, adjustable, and tradable quota path, thereby supporting the quantitative management and implementation of the zero-carbon target of the entire park process.

[0026] It should be noted that the execution entity of this embodiment may be a computing service device with data processing, network communication, and program execution capabilities, such as a tablet computer, personal computer, mobile phone, etc., or an electronic device capable of implementing the above functions, a carbon emission management system, a carbon emission management platform, etc. The following uses the carbon emission management system as an example to illustrate this embodiment and the following embodiments.

[0027] Based on this, the present application embodiment provides a carbon emission management method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the carbon emission management method of the present application.

[0028] In this embodiment, the carbon emission management method includes steps S10 to S50: Step S10, obtaining a carbon emission management instruction input by a user; Obtaining carbon emission management instructions input by users refers to receiving instruction data from the government, park, enterprise or third-party platform through the human-computer interaction interface, API interface or Internet of Things gateway. The instruction data is carried in the form of JSON (JavaScript Object Notation, JavaScript, object representation), XML (Extensible Markup Language) or MQTT (Message Queuing Telemetry Transport) messages, and includes user identity, target administrative level identifier, project unique code, management type (query / simulation / transaction / performance) and time range fields.

[0029] The purpose of obtaining carbon emission management instructions is to trigger the precise calling of subsequent multi-level digital twin models and closed-loop management of carbon assets, thereby breaking down the macro-zero-carbon goals into executable and quantifiable operational tasks layer by layer, avoiding the errors and time lags caused by traditional manual disassembly.

[0030] In one possible implementation, the carbon emission management system provides a graphical instruction builder, allowing users to automatically generate instructions by dragging administrative area components and project nodes onto the canvas; in another possible implementation, the system opens a RESTful (Representational State Transfer) interface, allowing the enterprise ERP (Enterprise Resource Planning) or energy management platform to directly push instructions.

[0031] Step S20: calling a hierarchical digital twin sub-model corresponding to the carbon emission management instruction in a pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative region division and project division; The multi-level digital twin model system refers to a four-layer tree-like model cluster composed of an orthogonal combination of "administrative region dimension" and "project entity dimension", including the first-level digital twin model, the second-level digital twin model, the third-level digital twin model, and the project-level digital twin model. Among them, the project-level digital twin model is the model with the smallest granularity. Each level of the model has independent spatial resolution, temporal resolution and data granularity, and realizes cross-layer parameter synchronization through federated learning nodes.

[0032] Exemplarily, the multi-level digital twin model system can be a four-layer model cluster divided by the dual dimensions of "administrative region + project entity": provincial digital twin model, municipal digital twin model, district digital twin model, and project-level digital twin model.

[0033] Administrative division rules: The boundaries of provinces, cities and counties are defined based on the latest statistical division codes and urban-rural division codes.

[0034] Project division rules: Use the unique project code in the project registration certificate or environmental impact assessment approval document as an identifier to map the physical project into a digital twin node.

[0035] In addition, it should be noted that the purpose of calling the hierarchical digital twin model is to automatically route user instructions to the model with corresponding policy authority, data granularity and computing power, thereby reducing cross-regional and cross-project adaptation costs.

[0036] In one possible implementation, the carbon emission management system adopts a distributed service bus + microservice registration center architecture, completing calls through hierarchical keyword matching + load balancing. Specifically, after the user instruction is parsed, the hierarchical identification parameters (such as "provincial-power industry") are extracted. The microservice registration center retrieves the service metadata (including address, version, and load status) of the corresponding hierarchical digital twin sub-model based on the parameters. Next, the distributed service bus combines the hierarchical keyword matching algorithm (such as administrative area code + industry label) to select the optimal node from the available service instances and distributes the call request through a weighted polling load balancing strategy. During the call process, the service bus monitors the health status of the nodes in real time, automatically eliminates abnormal instances and adds backup nodes. Finally, after the target sub-model executes the instruction, the rule text and three-dimensional scene data are returned through the service bus, and dynamic rendering is completed on the visual interface.

[0037] In another possible implementation, a carbon emissions management system based on a blockchain smart contract automatically triggers a cold start of the corresponding model image after obtaining a user's digital signature. Specifically, when a user initiates a carbon emissions management command, they first generate a digital signature using an asymmetric encryption algorithm and submit the command content and signature to a blockchain node. After verifying the signature's legitimacy and user permissions, the smart contract automatically matches the hierarchical identification parameters in the command (such as "first-level model") and calls the hash address of the corresponding model image from off-chain storage. The contract then triggers the container orchestration engine to pull the image and allocate computing resources. It then loads policy parameters and federated learning node configurations via a predefined initialization script, completing the model cold start. During the startup process, the node stores metadata such as image checksums and resource usage on-chain in real time to ensure traceability. Once the model is ready, the smart contract returns the call result to the user and triggers subsequent data collection and carbon quota calculation processes through an event notification mechanism, achieving decentralized, automated, and trusted execution throughout the process.

[0038] For example, in a specific implementation, when the instruction carries the "Province A-City B-Chemical Project C" field, the system first activates the digital twin model of Province A to obtain the provincial energy structure correction coefficient, then cascades the activation of the digital twin model of City B to extract the municipal industry benchmark value, and finally wakes up the project-level digital twin model to load the 3D mesh model generated by drone oblique photography.

[0039] Step S30: If the hierarchical digital twin sub-model called is a project-level digital twin model, then the project energy consumption data is collected, and the total annual carbon emissions are calculated based on the project energy consumption data; Project-level digital twin models: These are high-fidelity digital twins with a single construction or operational project as the smallest management unit. Their 3D visualization shell is constructed from point cloud data collected by drones using mesh reconstruction. The model embeds an IoT gateway, energy consumption monitoring terminals, and edge computing nodes. Mesh reconstruction involves converting point cloud data collected by drones using a 3D mesh generation algorithm into a continuous polygonal mesh model.

[0040] Project energy consumption data: cumulative, instantaneous and status data uploaded by smart electricity meters, water meters, gas meters, steam flow meters, photovoltaic inverters and other terminals at preset time intervals.

[0041] Annual carbon emissions: Energy consumption data is converted into a cumulative annual value of CO2 equivalent using the emission factor method or mass balance method under the ISO 14064-1 framework. ISO 14064-1 refers to the greenhouse gas quantification framework published by the International Organization for Standardization.

[0042] In addition, it should be noted that the purpose of collecting project energy consumption data and calculating the total annual carbon emissions is to establish an "auditable, traceable, and predictable" carbon emission benchmark for the project, support the accurate calculation of quota gaps and subsequent carbon asset development, and avoid compliance risks caused by missing data.

[0043] In one possible implementation, the IoT gateway uploads data through an encrypted channel. In another possible implementation, the edge computing node first cleans the local data and then transmits it back to the cloud-based time series database in batches through a lightweight interface.

[0044] For example, in a specific implementation, a project-level digital twin model of a data center uses the Modbus-TCP (Modbus Transmission Control Protocol) protocol to collect power and flow data from a total of 312 measurement points, including UPS (Uninterruptible Power Supply), chillers, and precision air conditioners. After removing abnormal spikes through Kalman filtering, combined with the regional power grid emission factor of 0.6299tCO2 / MWh issued by the provincial model, the total carbon emissions for the 2024 cycle are calculated to be 15,788tCO2.

[0045] Step S40, calling a pre-built graph generator to extract carbon quota parameters; Graph Generator: An automated rule parsing engine built based on natural language processing and knowledge graph technology. Its input is the carbon emission policy texts, quota allocation plans, and industry benchmark value files publicly released at the national, provincial, municipal, and district levels.

[0046] Carbon quota parameters: a five-tuple consisting of "carbon emission benchmark value per unit product, free quota ratio, pre-borrowing ratio, compliance period, and CCER offset upper limit" corresponding to the industry to which the project belongs.

[0047] In addition, it should be noted that the purpose of extracting carbon quota parameters is to convert unstructured policy provisions into computable numerical rules, thereby eliminating the subjective differences in manual interpretation of policies and ensuring that subsequent quota plans can achieve the goal of zero carbon emissions accurately at the lowest cost and with minimal changes.

[0048] In one possible implementation, the graph generator uses the BERT+CRF model to extract entities and relationships, and completes logical verification through the rule engine. BERT (Bidirectional Encoder Representations from Transformers) is a pre-trained language model based on the Transformer architecture that can understand text semantics through bidirectional context; CRF (Conditional Random Field) is a sequence labeling model that can optimize entity boundary predictions based on contextual features. BERT's bidirectional context understanding capability solves the recognition errors of traditional models for ambiguous texts (such as "carbon sinks" and "carbon trading"). CRF optimizes label predictions through sequence dependencies, improving the recognition accuracy of regular texts.

[0049] Specifically, the carbon emission-related policy text input data is segmented and tagged with parts of speech, and then converted into a sequence of word vectors that can be recognized by the BERT model. The BERT model performs bidirectional encoding on the text and outputs a contextual semantic vector for each character. The CRF layer predicts entity boundaries (such as "company name" and "carbon emissions") and relationship types (such as "belongs to" and "emissions") based on the semantic vectors and label transition probabilities. The extracted results are passed to the rule engine, and the entity type matching and relationship logic are verified according to preset rules (such as the "carbon quota" entity must be associated with the "region" attribute, and the "emission relationship" must include a timestamp), and contradictory or invalid triples are filtered out. The entities and relationships that pass the verification are stored in the knowledge graph to form a graph generator.

[0050] Step S50: Generate a carbon quota plan based on the carbon quota parameters and the annual carbon emissions.

[0051] Carbon quota plan: This plan uses the total annual carbon emissions and carbon quota parameters as inputs and is calculated using a multi-objective optimization algorithm (NSGA-III or improved particle swarm optimization) to combine the quota gap, quota trading strategy, CCER development scale, emission reduction technology investment list, and compliance schedule. The quota gap refers to the difference between the total annual carbon emissions and the free quota, which can be positive (gap) or negative (surplus).

[0052] It should be noted that NSGA-III (Non-dominated Sorting Genetic Algorithm III) is a multi-objective optimization algorithm that guides population evolution by introducing uniformly distributed reference points, combines non-dominated sorting with adaptive normalization strategies, and balances the convergence and diversity of solutions. It is suitable for handling high-dimensional optimization problems with more than three objectives.

[0053] The improved particle swarm optimization algorithm is an improvement to the standard particle swarm optimization algorithm. It enhances the global search capability and avoids premature convergence by dynamically adjusting the inertia weight, learning factor or fusing genetic operators, simulated annealing and other strategies, thereby improving the accuracy and efficiency of solving complex optimization problems.

[0054] By using NSGA-III or improved particle swarm optimization for carbon quota scheme generation, multiple objectives such as emission reduction costs and carbon asset returns can be optimized simultaneously.

[0055] In addition, it should be noted that the purpose of generating a carbon quota plan is to transform the macro-zero carbon target into an annual action blueprint that can be executed by the enterprise, and to simultaneously balance compliance, minimization of emission reduction costs and maximization of carbon asset returns, so as to avoid passive high-priced purchase of quotas or waste of excessive emission reductions in the later stage.

[0056] In one possible implementation, the system uses the improved NSGA-III algorithm to generate the Pareto frontier with the goals of "economic cost, emission reduction, and investment payback period", which is then selected by the enterprise independently. In another possible implementation, the system introduces federated learning nodes to synchronize real-time energy structure correction coefficients and dynamically adjust the plan to respond to policy mutations.

[0057] This embodiment provides a carbon emissions management method. By constructing a multi-level digital twin model system with two dimensions of "administrative region + project entity", the carbon emissions management instructions input by the user can be directly mapped to the sub-model of the corresponding level. When the instruction points to the project-level model, the project energy consumption data is immediately collected and the total annual carbon emissions are calculated. At the same time, the map generator is called to analyze the differentiated carbon quota parameters of the region and automatically generate an accurate carbon quota plan. Through the closed-loop mechanism of instructions, calling models, extracting carbon quota parameters, and forming a carbon quota plan, it not only achieves dynamic adaptation to policy differences between regions, but also quantifies the zero-carbon target into a predictable, adjustable, and tradable quota path, thereby supporting the quantitative management and implementation of the zero-carbon target throughout the park.

[0058] In a feasible implementation, step S20 may include steps S21 to S23: Step S21, parsing the carbon emission management instruction and extracting the level identification parameter; A natural language processing engine is used to perform a joint grammatical and semantic analysis on the original message of the carbon emission management instruction to extract the hierarchical representation parameters; the hierarchical identification parameters indicate the string or enumeration value extracted from the instruction, which is used to uniquely identify the four-level model hierarchy of province, city, district, and project.

[0059] In one possible implementation, the system uses regular expressions and dictionary matching to extract hierarchical identification parameters. Specifically, the system first parses the original carbon emission management instruction message, constructs a regular expression rule library based on a preset administrative division keyword dictionary (such as "province," "city," "district," and "project"), and uses a greedy matching algorithm to locate the hierarchical identification field in the instruction (such as "Province A - City B - Project C"). It then uses a dictionary of administrative division codes to verify the legitimacy of the matching results, filtering out invalid or fuzzy matches. Finally, it extracts standardized hierarchical identification parameters (such as provincial codes, city codes, and unique project codes) and outputs them to the model call module.

[0060] In another possible implementation, the system uses the BERT+CRF sequence labeling model to automatically identify and output hierarchical labels. Specifically, the system tokenizes the instruction text and converts it into word vectors, which are then fed into a pre-trained BERT model for bidirectional context encoding to generate character-level feature vectors containing contextual semantics. The CRF layer predicts the hierarchical label for each character based on the feature vectors and the label transition probability matrix (e.g., the transfer weight from "project level" to "provincial level"). The optimal label sequence is decoded using the Viterbi algorithm, and the hierarchical identification parameters corresponding to the continuous labels are extracted. These parameters are then output after verification by the rule engine.

[0061] Step S22: searching the multi-level digital twin model system based on the level identification parameter to locate the level digital twin sub-model; Node matching queries are performed using hierarchical identification parameters. The matching degree between each digital twin sub-model and the hierarchical identification parameters is retrieved within the multi-level digital twin model system. Based on the matching results, metadata for the target sub-model is returned, including model version, calculation accuracy, data update time, and service address. This reduces the global model space to a single target model, avoiding the performance loss associated with a full scan.

[0062] In one possible implementation, the system uses Redis key-value caching and Bloom filters to accelerate positioning. Specifically, the hierarchical identification parameters of the multi-level digital twin model system are preloaded into the Bloom filter. Upon receiving the hierarchical identification parameters, the Bloom filter is first used to quickly determine whether the parameters exist, filtering out invalid queries. For parameters that are determined to exist, the Redis cache is queried to obtain model metadata (service address, version, etc.). If the cache misses, the database is searched and the results are written to Redis to achieve model positioning. By adopting a dual filtering mechanism of Redis key-value caching and Bloom filters, invalid database accesses are reduced, shortening positioning time.

[0063] Step S23: Send a call request to the hierarchical digital twin sub-model, call the hierarchical digital twin sub-model, and render the rule text corresponding to the hierarchical identification parameter in the visual interface.

[0064] The RESTful call message based on HTTPS (HyperText Transfer Protocol Secure) carries the user identity token, model instance ID, time range, and output format parameters, and sends a call request to the hierarchical digital twin sub-model to call the hierarchical digital twin sub-model.

[0065] The visualization interface specifically refers to a three-dimensional digital twin browser that supports spatial roaming, attribute query, and timeline playback. The policy text refers to the carbon emission policy clauses that correspond to the target level and have undergone structured analysis, which are superimposed on the administrative boundaries or project red lines in the three-dimensional scene in the form of highlighted cards.

[0066] In addition, it should be noted that sending call requests and rendering policy texts allows users to intuitively view the correspondence between policy requirements and spatial objects in an immersive three-dimensional scene, significantly lowering the threshold for policy understanding and improving decision-making efficiency.

[0067] In a feasible implementation, step S40 may include steps S41 and S42: Step S41 , calling a pre-built graph generator to traverse the rule text and extract core carbon asset parameters, wherein the rule text includes first-level rule text, second-level rule text and third-level rule text.

[0068] A pre-built graph generator is used to scan the provincial, municipal, and district-level rules, paragraph by paragraph, chapter by chapter, and clause by clause. This process extracts a set of numerical fields directly related to allowances, emission reductions, trading, and compliance from the rules to obtain core carbon asset parameters. For example, these parameters include the industry carbon emission baseline, the proportion of free allowances, the proportion of pre-borrowed allowances, the compliance period, the CCER offset cap, and the green electricity deduction coefficient.

[0069] Step S42: Performing a logic check on the core carbon asset parameters and associating them with the benchmark values ​​of the industry to which the project belongs to obtain carbon quota parameters.

[0070] Logical verification: Automatically verify the legality, rationality, and consistency of core carbon asset parameters based on preset business rules, including numerical range checks, unit consistency checks, and time logic checks.

[0071] Project industry benchmark value: the official published carbon emission baseline per unit product or per unit output value of the project's industry.

[0072] Carbon quota parameters: The final set of available parameters after verification and correlation with industry benchmark values, used for subsequent quota gap calculation and emission reduction plan generation.

[0073] The purpose of performing logical verification and associating benchmark values ​​is to eliminate ambiguity in policy texts, numerical errors, and version conflicts, ensure the uniqueness and authority of parameters, and ensure the accuracy of subsequent carbon quota plans.

[0074] Based on the first embodiment of the present application, the second embodiment of the present application is proposed. In the second embodiment of the present application, the same or similar contents as those of the first embodiment can be referred to the above introduction and will not be repeated hereafter.

[0075] On this basis, please refer to Figure 2 , Figure 2 This is a flow chart of the second embodiment provided by this application. Figure 2 As shown, before step S10, the carbon emission management method further includes steps S01 to S05: Step S01: defining a two-dimensional division rule based on the administrative region level and the project entity attributes, wherein the two-dimensional division rule includes a geographical boundary division rule and a project attribution division rule; Project entity attributes: A collection of attributes such as project name, project unique code, industry category, construction location, legal entity, construction scale, production process, etc. stated in the project approval, environmental impact assessment approval or registration certificate.

[0076] Two-dimensional division rule: The division criteria are obtained by orthogonally combining the geographical dimension of "administrative region" and the engineering dimension of "project entity".

[0077] Geographic boundary demarcation rules: Based on the administrative division boundary vector data that can be identified by remote sensing images, natural features such as river systems and traffic arteries are superimposed as auxiliary constraints to form a closed polygonal boundary.

[0078] Project ownership classification rules: Using the construction site coordinates stated in the project registration certificate as the anchor point, determine which geographical boundary polygon the coordinates fall into, and thus determine the corresponding administrative region to which the project belongs.

[0079] In addition, it should be noted that the purpose of defining two-dimensional division rules is to decompose regional carbon management tasks step by step, so that the subsequent digital twin model has both policy consistency and project feasibility; when policies are adjusted or projects are added or reduced, the system only needs to update the rules to automatically rearrange the model hierarchy, significantly reducing maintenance costs.

[0080] In one possible implementation, the geographic boundary division rule uses an API interface to obtain the latest boundary vector in real time; in another possible implementation, the project attribution division rule automatically completes attribution matching by performing spatial inclusion judgment between the project coordinates and the administrative divisions.

[0081] Step S02: collecting remote sensing geographic information and rule text based on the geographic boundary division rule, and collecting project point cloud data by drone oblique photography based on the project attribution division rule; Remote sensing geographic information: multispectral, hyperspectral, and synthetic aperture radar (SAR) images acquired by satellite or aerial remote sensing platforms and their derived surface cover classification, vegetation index, and building density raster data.

[0082] The rule text refers to the carbon emission management policies, quota allocation plans, and industry energy consumption limit standards publicly released by national, provincial, municipal, and district-level departments or authoritative groups; Drone oblique photography: Using a multi-rotor drone equipped with a five-lens oblique camera, the project site is photographed from multiple angles in a "well"-shaped flight path, obtaining a high-resolution image sequence; Project point cloud data: A set of three-dimensional point coordinates generated by encrypting and densely matching oblique images using the motion recovery structure algorithm, which contains X, Y, Z three-dimensional coordinates and RGB color information.

[0083] Step S03: Bind the rule text with the remote sensing geographic information to construct a first-level digital twin model, a second-level digital twin model, and a third-level digital twin model respectively; The administrative division fields in the rule text are semantically and spatially aligned with the vector boundaries in the remote sensing geographic information. These are then stored in a graph database using a "province-city-district" node relationship to construct the first-level digital twin model, the second-level digital twin model, and the third-level digital twin model, i.e., the province-city-district level digital twin model.

[0084] Among them, the provincial digital twin model refers to a carbon emission grid model based on the entire province, which includes multi-dimensional attributes such as energy structure, industrial structure, traffic flow, and population density; the municipal digital twin model refers to a carbon emission grid model based on the entire city, which inherits the provincial model correction coefficient and superimposes the boundaries of the municipal industrial clusters; the district digital twin model refers to a carbon emission grid model based on the entire district, which further superimposes the vector boundaries of district-level key projects and real-time air quality monitoring station data.

[0085] The purpose of constructing provincial, municipal and district models is to form a computational framework for continuous downscaling. The upper-level model provides boundary conditions and initial fields for the lower-level model, and the lower-level model provides verification and feedback for the upper-level model.

[0086] In one possible implementation, a provincial digital twin model couples WRF-Chem (Weather Research and Forecasting model coupled to Chemistry) with LEAP (Long-range Energy Alternatives Planning system) to simulate energy-atmosphere interactions. Specifically, WRF-Chem, as an online coupled meteorological-chemical model, simulates the regional-scale transport of atmospheric pollutants (such as PM2.5 and ozone), the impact of meteorological conditions (such as wind speed and temperature) on carbon emission dispersion, and aerosol-radiative feedbacks (such as the effect of greenhouse gas concentrations on regional climate). LEAP, a bottom-up energy planning tool, uses scenario analysis to predict the impact of different energy policies (such as coal-fired power generation and increasing the proportion of renewable energy) on provincial carbon emissions and outputs a carbon emission inventory by industry and fuel type. Energy carbon emission data generated by LEAP serves as anthropogenic emission source input for WRF-Chem, driving atmospheric pollution simulations. Meteorological correction parameters output by WRF-Chem are fed back into LEAP to optimize the dynamic adjustment of carbon emission factors.

[0087] The coupling of WRF-Chem and LEAP simulates energy-atmosphere interactions, enabling simulation of regional atmospheric pollutant transport, the impact of meteorological conditions on carbon emission dispersion, and aerosol-radiation feedback. LEAP also predicts the impact of energy policies on provincial carbon emissions and outputs carbon emission inventories by industry and fuel type. LEAP data drives WRF-Chem atmospheric pollution simulations, and the meteorological correction parameters it outputs feed back into LEAP to optimize the dynamic adjustment of carbon emission factors, improving simulation accuracy and policy adaptability.

[0088] Step S04: Distributedly deploying federated learning nodes in the first-level digital twin model, the second-level digital twin model, and the third-level digital twin model; Federated learning node: A lightweight computing agent embedded in the digital twin model, which has local model training, gradient encryption, and parameter upload functions.

[0089] Specifically, federated learning nodes are deployed as Docker containers within the three-level digital twin model, forming a collaborative computing network that spans both horizontal and vertical domains. This allows for the sharing of model parameters without leaving the domain, enabling dynamic coordination of cross-regional policy parameters and energy structure correction coefficients.

[0090] In one possible implementation, nodes use personalized federated learning to retain local characteristic parameters for different administrative levels. Specifically, after initializing the global shared model, nodes at each administrative level split the parameters into a shared layer (such as the basic carbon emission factor) and a local characteristic layer (such as the regional industrial structure coefficient). During local training, the shared layer is frozen, and only the characteristic layer parameters are updated. During the global aggregation phase, only the shared layer parameters are uploaded, and a new shared model is generated after weighted averaging using the FedAvg algorithm. Nodes merge the updated shared layer and local characteristic layer parameters, and adapt to the local data distribution through gradient fine-tuning, achieving "global collaboration + hierarchical personalization" model optimization. The FedAvg algorithm (the model aggregation algorithm in federated learning) generates a global model by weighted averaging the local model parameters of each node.

[0091] Step S05: construct a project-level digital twin model based on the project point cloud data, develop a lightweight data upload interface, connect the federated learning node based on the lightweight data upload interface, and obtain a multi-level digital twin model system.

[0092] A project-level digital twin model refers to a high-fidelity digital twin with the red line of a single project site as its spatial scope and a single device as the smallest management unit. Its geometric shell is reconstructed from the project point cloud data via Mesh, and is internally associated with BIM (Building Information Modeling), IoT (Internet of Things) sensors, and video surveillance streams.

[0093] The lightweight data upload interface is a narrowband transmission interface. The project-level model uploads the locally calculated carbon emission factors and equipment energy efficiency curves to the district-level nodes through the lightweight interface, and at the same time receives the correction coefficients and policy parameters issued by the provincial, municipal, and district-level nodes, thereby completing the closed loop of the multi-level digital twin model system.

[0094] In one possible implementation, the lightweight interface uses CoAP (Constrained Application Protocol) for ultra-low power sensors; in another possible implementation, the interface supports WebRTC (Web Real-Time Communication) point-to-point transmission to achieve real-time linkage between monitoring video and twin models.

[0095] In this embodiment, two-dimensional rules are used to divide management boundaries into four levels: province, city, district, and project. Data collection is completed using remote sensing geographic information, policy texts, and drone oblique photography point cloud data. Policy semantics are bound to spatial boundaries to construct a digital twin model. Distributed federated learning nodes are used to achieve cross-domain parameter collaboration. Finally, a project-level digital twin model is generated based on the project point cloud data and connected to the federated network through a lightweight interface. Without leaving the domain, the four-level carbon emission digital twin system is closed-loop connected, ultimately forming a carbon emission quantification management system that is both compliant with regional policy differences and accurate to the project level. This makes the zero-carbon goal decomposable, simulatable, verifiable, and optimizable throughout the planning, construction, and operation lifecycle. This significantly reduces the cost of model reconstruction and data re-collection caused by policy changes or project additions and subtractions, ultimately supporting governments and enterprises in achieving zero-carbon commitments at the lowest compliance cost.

[0096] Based on the first embodiment and / or the second embodiment of the present application, the third embodiment of the present application is proposed. In the third embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction and will not be repeated hereafter.

[0097] In this embodiment, if the hierarchical digital twin sub-model called is a project-level digital twin model, then step S30 of collecting project energy consumption data and calculating the total annual carbon emissions based on the project energy consumption data may include steps S31 to S36: Step S31, collecting project energy consumption data based on the IoT gateway of the project-level digital twin model; The IoT gateway refers to an industrial-grade edge computing device deployed at the project site that supports multi-protocol communication. The IoT gateway collects multi-dimensional data such as accumulated energy consumption, instantaneous power, flow, and status quantities uploaded in real time by terminals such as smart electricity meters, water meters, gas meters, steam flow meters, and photovoltaic inverters, thereby obtaining project energy consumption data.

[0098] Step S32: Unifying the project energy consumption data to a standard time axis and using Kalman filtering to remove outliers in the project energy consumption data to obtain standardized project energy consumption data; The project energy consumption data is aligned to a standard timeline, which refers to a unified time series based on UTC. A Kalman filter is used to remove outliers from the time-aligned project energy consumption data, ultimately resulting in a high-quality energy consumption series that is time-aligned and outlier-removed, i.e., the standardized project energy consumption data.

[0099] Step S33: call the annual energy structure basic data from the third-level digital twin model, and obtain the energy structure correction coefficient issued by the first-level digital twin model through the federal node; Annual energy structure basic data: annual consumption and proportion by energy type (coal, oil, gas, electricity, heat, renewable energy) stored in the district-level model.

[0100] Energy structure correction coefficient: A dimensionless coefficient calculated by the provincial model based on policy targets, external power transfer, and renewable energy increments, used to correct the national default emission factor.

[0101] In addition, it should be noted that the purpose of calling district-level basic data and obtaining provincial-level correction coefficients is to achieve the same caliber and standard of "region-project" data and avoid the confusion caused by the parallel use of multiple sets of factors.

[0102] In one possible implementation, the correction coefficients are pushed daily via an HTTPS REST interface. This interface is a RESTful interface based on the HTTPS protocol, enabling data exchange between the client and the server via HTTP methods. Specifically, at a fixed time each day (e.g., 2:00 AM), a data packet of the correction coefficients for the day is pushed to each node via the HTTPS REST interface. Upon receiving the data, the node verifies the integrity of the data, replaces the local historical coefficients, and logs the update.

[0103] In another possible implementation, gRPC (gRPC Bidirectional Streaming Real-time Synchronization) is used for real-time synchronization. The node establishes a persistent gRPC connection with the server. The server monitors coefficient changes in real time and pushes updates via streaming. The node immediately applies the updates and provides feedback on the synchronization status, achieving dynamic synchronization of the correction coefficients.

[0104] Step S34, multiplying the annual energy structure basic data and the energy structure correction coefficient to obtain a carbon emission factor; The carbon emission factor refers to the key conversion coefficient for converting project energy consumption data into carbon dioxide equivalent. Multiplication refers to multiplying the consumption of each energy type in the annual energy structure basic data by the corresponding correction coefficient one by one and then weighted summing them to obtain the localized comprehensive emission factor.

[0105] In addition, it should be noted that the purpose of generating localized carbon emission factors is to ensure that the project carbon emission accounting results truly reflect the differences in the local power grid, heat, and fuel structure, thereby significantly improving the accuracy of the annual total carbon emissions.

[0106] Step S35: calling historical data for the same period, and predicting energy consumption data for the remaining period based on the historical data for the same period and the carbon emission factor; Historical data for the same period refers to the energy consumption and carbon emission time series of the same project in the same calendar period during the historical period; the energy consumption data for the remaining period refers to the complete data obtained by using machine learning or statistical models to predict the energy consumption of unfinished periods in the future based on historical laws, production plans, weather forecasts, market orders and other multi-dimensional factors.

[0107] In addition, it should be noted that the purpose of calling historical data and predicting energy consumption for the remaining period is to solve the problem that some projects have not yet ended their annual year and the real-time data is incomplete, so as to achieve a complete estimate of the total carbon emissions in the annual cycle; its effect is to lock in the quota gap in advance while ensuring accuracy, so as to win time for enterprises to trade quotas or deploy emission reduction measures.

[0108] In one possible implementation, the system uses the Prophet time series forecasting model. The Prophet time series forecasting model is a time series forecasting algorithm suitable for medium- and long-term forecasts and is good at processing data containing seasonality, trends, and outliers.

[0109] Specifically, historical carbon emission data (such as monthly / quarterly emissions) is input to automatically identify trend items (long-term growth / decline), seasonal items (annual / quarterly cycles) and holiday effects in the time series; the model generates prediction intervals through Bayesian inference and supports customized trend mutation points (such as policy implementation dates); it outputs a total carbon emission forecast curve for the next 1-3 years and provides a trend decomposition diagram (trend + season + residual) to assist in identifying key influencing cycles.

[0110] In another possible implementation, the system uses an XGBoost regression model, with inputs including production schedules, temperature, holidays, and other features. The XGBoost regression model is an ensemble learning model based on gradient boosting trees. It improves prediction accuracy by optimizing the objective function and regularizing, and supports multiple feature inputs.

[0111] Specifically, production schedules (such as industrial output value), meteorological data (average daily temperature), holidays (whether they are work start days) and historical carbon emission data are integrated; the training / test sets are divided, and parameters such as tree depth and learning rate are optimized through grid search to train the model to fit the nonlinear relationship between features and carbon emission values; real-time feature data (such as daily production schedules and weather forecasts) are input, and short-term (daily / weekly) carbon emission forecast values ​​are output, and key driving factors (such as the impact of high temperature weather on air conditioning energy consumption) are identified by sorting feature importance.

[0112] Step S36: accumulating the standardized project energy consumption data and the remaining period energy consumption data to obtain the total annual carbon emissions.

[0113] The standardized project energy consumption data that has occurred and the predicted energy consumption data for the remaining period are spliced ​​in time series, multiplied by the localized carbon emission factor and summed to obtain the total annual carbon emissions, and the total carbon dioxide equivalent emissions of the project in a complete natural year are determined.

[0114] In this embodiment, the IoT gateway collects multi-source energy consumption data from the project site in real time and losslessly, and then uses Kalman filtering and a standard time axis to complete data cleaning and time series unification to ensure data quality and comparability. Then, through the federated learning node, the regional energy structure correction coefficient issued by the provincial model is integrated with the annual energy structure basic data provided by the district model to dynamically generate a localized carbon emission factor. The energy consumption of the remaining period is predicted with the help of historical data from the same period, and finally the total annual carbon emissions are accumulated to ensure the high accuracy of the nucleic acid results.

[0115] Based on the above embodiments of the present application, a fourth embodiment of the present application is proposed. In the fourth embodiment of the present application, the same or similar contents as those of the above embodiments can be referred to the above introduction and will not be repeated hereafter.

[0116] On this basis, please refer to Figure 3 , Figure 3 This is a flow chart of the fourth embodiment of the present application. Figure 3 As shown, after step S20 of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, steps S231 to S238 are also included: Step S231: If the hierarchical digital twin sub-model called is a first-level digital twin model, a pre-built graph generator is called to identify the first-level rule text and extract the first instruction parameter set; The first instruction parameter set refers to a set of structured fields extracted from provincial policy text that can be used to drive the model. The purpose of identifying provincial policy text and extracting provincial policy instruction parameter sets is to convert unstructured policy language into numerical instructions that can be directly read by the model.

[0117] In one possible implementation, the graph generator adopts a rule template and deep learning fusion strategy. The rule template is first used to quickly lock the chapter, and then the BERT model is used to extract the entities.

[0118] Step S232: generating a regional energy structure correction coefficient based on the first instruction parameter set; The regional energy structure correction coefficient refers to a dimensionless correction coefficient generated based on the provincial administrative region, taking into account factors such as the primary energy consumption structure, electricity inflow and outflow, the proportion of renewable energy, and the emission factor of purchased electricity. It is used to adjust the national default emission factor to a localized factor that conforms to the actual situation of the province, so that the carbon emission results output by the provincial digital twin model reflect both the national unified caliber and local differences, thereby improving the accuracy of quota allocation.

[0119] Step S233: Based on the federated learning node, obtain the industrial structure characteristic vector and the energy consumption intensity characteristic vector uploaded from the second-level digital twin model, and generate an administrative division emission reduction heat map based on the industrial structure characteristic vector and the energy consumption intensity characteristic vector; The industrial structure characteristic vector refers to a high-dimensional vector output by the municipal digital twin model, which reflects the proportion of output value, employment, and energy consumption of various industries in the city; the energy consumption intensity characteristic vector refers to a high-dimensional vector output by the municipal digital twin model, which reflects the energy consumption per unit GDP and per unit product energy consumption of various industries in the city; the administrative division emission reduction heat map refers to a visualization layer with provincial administrative divisions as the base map, grids or cities as units, and color depth to indicate the carbon emission reduction potential or urgency.

[0120] In addition, it should be noted that the purpose of obtaining city-level feature vectors and generating provincial-level emission reduction heat maps is to use federated learning to achieve cross-level knowledge sharing, avoid the original data from going out of domain, and at the same time display the spatial distribution of key emission reduction areas in the province with intuitive graphics.

[0121] Step S234: If the called hierarchical digital twin sub-model is a second-level digital twin model, the graph generator is called to identify the second-level rule text and extract the second instruction parameter set; Identify municipal policy texts and extract municipal policy instruction parameter sets, injecting municipal differentiated control requirements into the model. In one possible implementation, the system directly connects to the municipal government's OA interface to obtain structured policy data, uses a graph generator to identify municipal policy texts, and extracts municipal policy instruction parameter sets.

[0122] Step S235: Obtain the second-level industrial structure data, overlay the second-level industrial structure data with the second-level geographic information layer, and determine the vector boundary of the high-energy-consuming industrial cluster area; The second-level industrial structure data refers to a statistical indicator of industry output value, energy consumption, employment, and taxation in the municipal administrative area; the second-level geographic information layer, that is, the municipal geographic information layer, refers to the vector layer containing administrative divisions, land use, roads, water systems, and ecological red lines; the vector boundary of the high-energy-consuming industrial cluster area refers to the polygonal boundary formed by aggregating industrial plots with energy consumption density exceeding the standard through spatial clustering algorithms.

[0123] Step S236: Call the regional energy structure correction coefficient issued by the first-level digital twin model, split the regional energy structure correction coefficient according to the second-level industrial structure data, and generate an industry-specific energy structure correction coefficient matrix; The provincial regional energy structure correction coefficient is decomposed according to the weight of the municipal industrial structure, resulting in a two-dimensional matrix with industries as rows and energy types as columns. Each element represents the correction coefficient of that industry and energy type relative to the provincial average. This precisely transmits the strength of provincial policies to various industries at the municipal level to enhance quota fairness.

[0124] In one possible implementation, the system uses an entropy weighting method to calculate industrial structure weights. Specifically, the data for industrial structure indicators (such as the proportion of output value of each industry and energy consumption intensity) are first standardized to eliminate dimensional differences. The information entropy value of each indicator is then calculated. A smaller entropy value indicates a higher degree of dispersion and greater information content. The entropy value is then used to derive the variance coefficient (1-entropy value), which is positively correlated with the weight. Finally, the variance coefficient is normalized to obtain the objective weight of each industrial structure indicator, with the total weight being 1.

[0125] Step S237: If the called hierarchical digital twin sub-model is a third-level digital twin model, the graph generator is called to identify the third-level rule text and extract the third instruction parameter set; The system uses crawlers to periodically capture policy updates from the district government portal website, calls a graph generator to identify district-level policy texts, and extracts the third instruction parameter set, namely the district-level policy instruction parameter set.

[0126] It is understandable that the difference between the parameter sets of provincial, municipal and district policy directives lies in the level of policy text from which they are extracted.

[0127] Step S238: Call the project information uploaded by the project-level digital twin model and generate a visual project list based on the project information.

[0128] Project information refers to structured data such as project name, project code, construction status, designed production capacity, real-time energy consumption, real-time carbon emissions, obtained quotas, remaining quotas, and emission reduction progress uploaded in real time by the project-level digital twin model; the visual project list refers to the project list displayed in the form of cards or tables on the front end of the district-level digital twin model, which supports multi-dimensional filtering and sorting by industry, energy consumption, and emission reduction progress.

[0129] In this implementation, the policy directive parameter sets at the provincial, municipal, and district levels, parsed using a graph generator, are progressively converted into regional energy structure correction coefficients, industry-specific correction matrices, and visualized project lists. Federated learning nodes are then used to enable cross-level encrypted data sharing and dynamic generation of spatial heat maps. This enables the transmission of macroeconomic policy objectives to project sites, ensuring the simultaneous quantification of zero-carbon targets across administrative and market entities at all levels. Ultimately, this significantly reduces compliance costs and emission reduction risks associated with policy differences and information lags.

[0130] Based on the above embodiments of the present application, a fifth embodiment of the present application is proposed. In the fifth embodiment of the present application, the same or similar contents as those of the above embodiments can be referred to the above introduction and will not be repeated hereafter.

[0131] In this embodiment, step S50 of generating a carbon quota plan according to the carbon quota parameters and the annual carbon emissions may include steps S51 to S53: Step S51, calculating the quota gap based on the total carbon emissions in the annual period; The quota gap refers to the difference between the total annual carbon emissions and the free allowances approved by the competent authorities. A positive value indicates that the gap requires the purchase of allowances or emission reductions, while a negative value indicates a surplus that can be sold or carried forward. The difference between the total annual carbon emissions and the free allowances is calculated using a simple difference or weighted difference algorithm to quantify the compliance pressure on enterprises and provide clear targets for the development of subsequent emission reduction plans.

[0132] In one possible implementation, the system uses a real-time gap dashboard for daily updates; in another possible implementation, the system uses scenario simulations to demonstrate changes in the gap under different emission reduction efforts.

[0133] Step S52: Calculate the quota gap and the carbon quota parameters using a multi-objective optimization algorithm to generate an initial emission reduction plan; The multi-objective optimization algorithm refers to an improved particle swarm optimization algorithm that simultaneously aims to minimize economic costs, maximize emissions reductions, and minimize payback periods. The initial emission reduction plan refers to a Pareto optimal solution that combines technology-based emission reduction, CCER development, quota trading, and green electricity substitution.

[0134] The purpose of using a multi-objective optimization algorithm is to find the optimal or nearly optimal fulfillment path for the enterprise under multiple constraints; its effect is to achieve the optimal balance of cost-benefit-risk while ensuring compliance.

[0135] In one possible implementation, the system uses the NSGA-III algorithm to generate 50 Pareto solutions for companies to choose from. Specifically, 50 random solutions for objective functions such as carbon emissions, costs, and energy consumption are initialized. Domination levels are divided through non-dominated sorting, and congestion distances are calculated to maintain solution diversity. Reference points are introduced to guide population evolution, and offspring are generated by simulating binary crossover and polynomial mutation. Parent and offspring populations are merged, and the optimal individual is selected based on the correlation between the reference points. After iteration until convergence, 50 evenly distributed Pareto solutions are output, and companies can choose a solution based on their actual needs (such as emission reduction priorities).

[0136] Step S53: Synchronizing the energy structure correction coefficient through the federated learning node to modify the initial emission reduction plan to obtain a carbon quota plan.

[0137] The latest energy structure correction coefficient is encrypted and pushed to the project-level model through the federated learning node. The carbon emission factor is recalculated according to the change of the correction coefficient, the marginal cost of emission reduction measures is re-evaluated and the emission reduction plan is updated. Finally, the final executable plan after federal revision is obtained, which includes a carbon quota plan including a technology list, investment budget, time nodes, trading plan, and compliance list.

[0138] It is understandable that synchronization and modification of federated learning nodes can ensure that the plan responds to the provincial power grid structure, renewable energy output and policy adjustments in real time, achieve dynamic optimization, and avoid compliance risks or economic losses caused by outdated plans.

[0139] In this implementation, the quota gap is first accurately calculated using the total annual carbon emissions and approved allowances, quantifying the abstract zero-carbon target into a directly actionable carbon asset gap value. A multi-objective optimization algorithm is then invoked to generate a Pareto-optimal initial emission reduction plan under multiple constraints. The plan is then dynamically revised using the provincial energy structure correction coefficients synchronized in real time through federated learning nodes, resulting in a final, executable carbon quota plan. This approach ensures corporate compliance while minimizing economic costs and the risk of policy lags, significantly improving carbon asset operational efficiency and the agility of achieving the zero-carbon target.

[0140] This application also provides a carbon emission management device, please refer to Figure 4 , the carbon emission management device includes: An acquisition module 10 is used to acquire a carbon emission management instruction input by a user; A multi-level digital twin model module 20 is configured to call a hierarchical digital twin sub-model corresponding to the carbon emission management instruction in a pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative region division and project division; The carbon quota scheme generation module 30 is used to collect project energy consumption data and calculate the total annual carbon emissions based on the project energy consumption data if the hierarchical digital twin sub-model called is a project-level digital twin model; call a pre-built graph generator to extract carbon quota parameters; and generate a carbon quota scheme based on the carbon quota parameters and the total annual carbon emissions.

[0141] The carbon emission management device provided in this application, employing the carbon emission management method described in the aforementioned embodiments, can address the technical issues surrounding carbon emission management. Compared to the prior art, the carbon emission management device provided in this application achieves the same beneficial effects as the carbon emission management method described in the aforementioned embodiments. Other technical features of the carbon emission management device are the same as those disclosed in the aforementioned embodiments and are not further detailed here.

[0142] The present application provides a carbon emission management device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the carbon emission management method in the above-mentioned embodiment one.

[0143] Reference below Figure 5 , which shows a schematic diagram of the structure of a carbon emission management device suitable for implementing embodiments of the present application. The carbon emission management device in embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The carbon emission management device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0144] like Figure 5As shown, the carbon emission management device may include a processing device 1001 (e.g., a central processing unit, graphics processing unit, etc.), which can perform various appropriate actions and processes based on programs stored in a read-only memory 1002 or programs loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the carbon emission management device. The processing device 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An input / output interface 1006 is also connected to the bus. Typically, the following systems may be connected to the input / output interface 1006: an input device 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; a storage device 1003 including, for example, a magnetic tape or hard disk; and a communication device 1009. Communication device 1009 can allow the carbon emission management device to communicate with other devices wirelessly or wired to exchange data. Although the figure shows a carbon emission management device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have alternatively.

[0145] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a read-only memory 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are performed.

[0146] The carbon emission management device provided in this application, employing the carbon emission management method described in the aforementioned embodiment, can address the technical issues surrounding carbon emission management. Compared to the prior art, the carbon emission management device provided in this application achieves the same beneficial effects as the carbon emission management method described in the aforementioned embodiment. Other technical features of the carbon emission management device are the same as those disclosed in the aforementioned embodiment and are not further elaborated here.

[0147] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0148] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0149] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, a computer program) stored thereon, and the computer-readable program instructions are used to execute the carbon emission management method in the above-mentioned embodiment.

[0150] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0151] The computer-readable storage medium may be included in the carbon emission management device, or may exist independently without being incorporated into the carbon emission management device.

[0152] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the carbon emission management device, the carbon emission management device: obtains the carbon emission management instructions input by the user; calls the hierarchical digital twin sub-model corresponding to the carbon emission management instructions in the pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative area division and project division; if the called hierarchical digital twin sub-model is a project-level digital twin model, the project energy consumption data is collected, and the total annual carbon emissions are calculated based on the project energy consumption data; the pre-built map generator is called to extract carbon quota parameters; and a carbon quota plan is generated based on the carbon quota parameters and the total annual carbon emissions.

[0153] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0154] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0155] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0156] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned carbon emission management method, thereby addressing the technical challenges of carbon emission management. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are similar to those of the carbon emission management method provided in the aforementioned embodiments and are not further elaborated here.

[0157] The present application also provides a computer program product, comprising a computer program, which implements the steps of the above-mentioned carbon emission management method when executed by a processor.

[0158] The computer program product provided in this application can solve the technical problems of carbon emission management. Compared with the existing technology, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the carbon emission management method provided in the above embodiment, and will not be repeated here.

[0159] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A carbon emission management method, characterized in that: The carbon emission management method includes: Obtain carbon emission management instructions input by the user; Invoking a hierarchical digital twin sub-model corresponding to the carbon emission management directive in a pre-built multi-level digital twin model system, wherein the multi-level digital twin model system is divided into multiple hierarchical digital twin sub-models by combining administrative region division and project division; If the hierarchical digital twin sub-model called is a project-level digital twin model, then the project energy consumption data is collected, and the total annual carbon emissions are calculated based on the project energy consumption data; Call the pre-built graph generator to extract carbon quota parameters; A carbon quota plan is generated according to the carbon quota parameters and the total carbon emissions for the annual period.

2. The carbon emission management method according to claim 1, wherein: Before the step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, the method further includes: Defining a two-dimensional division rule based on the administrative region level and the project entity attributes, wherein the two-dimensional division rule includes a geographical boundary division rule and a project attribution division rule; Collect remote sensing geographic information and rule text based on the geographic boundary division rules, and collect project point cloud data using drone oblique photography based on the project attribution division rules; Binding the rule text with the remote sensing geographic information to construct a first-level digital twin model, a second-level digital twin model, and a third-level digital twin model respectively; Distributedly deploying federated learning nodes in the first-level digital twin model, the second-level digital twin model, and the third-level digital twin model; A project-level digital twin model is constructed based on the project point cloud data, a lightweight data upload interface is developed, and the federated learning nodes are connected based on the lightweight data upload interface to obtain a multi-level digital twin model system.

3. The carbon emission management method according to claim 2, wherein: The step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system includes: Parsing the carbon emission management directive and extracting the level identification parameters; Retrieving the multi-level digital twin model system based on the level identification parameter and locating the level digital twin sub-model; A call request is sent to the hierarchical digital twin sub-model, the hierarchical digital twin sub-model is called, and the rule text corresponding to the hierarchical identification parameter is rendered in a visual interface.

4. The carbon emission management method according to claim 1, wherein: If the hierarchical digital twin sub-model called is a project-level digital twin model, the steps of collecting project energy consumption data and calculating the total annual carbon emissions based on the project energy consumption data include: An IoT gateway based on the project-level digital twin model collects project energy consumption data; Unifying the project energy consumption data to a standard time axis, and using Kalman filtering to remove outliers in the project energy consumption data to obtain standardized project energy consumption data; Call the annual energy structure basic data from the third-level digital twin model and obtain the energy structure correction coefficient issued by the first-level digital twin model through the federal node; Multiplying the annual energy structure basic data and the energy structure correction coefficient to obtain a carbon emission factor; Recall historical data for the same period, and predict energy consumption data for the remaining period based on the historical data for the same period and the carbon emission factor; The energy consumption data of the standardized project and the energy consumption data of the remaining period are accumulated to obtain the total carbon emissions for the annual cycle.

5. The carbon emission management method according to claim 1, wherein: After the step of calling the hierarchical digital twin sub-model corresponding to the carbon emission management instruction in the pre-built multi-level digital twin model system, the method further includes: If the hierarchical digital twin sub-model called is a first-level digital twin model, a pre-built graph generator is called to identify the first-level rule text and extract the first instruction parameter set; generating a regional energy structure correction coefficient based on the first instruction parameter set; Based on the federated learning node, obtain the industrial structure feature vector and energy consumption intensity feature vector uploaded from the second-level digital twin model, and generate an administrative division emission reduction heat map based on the industrial structure feature vector and the energy consumption intensity feature vector; If the called hierarchical digital twin sub-model is a second-level digital twin model, calling the graph generator to identify the second-level rule text and extract the second instruction parameter set; Obtain the second-level industrial structure data, overlay the second-level industrial structure data and the second-level geographic information layer to determine the vector boundaries of high-energy-consuming industrial clusters; Call the regional energy structure correction coefficient issued by the first-level digital twin model, split the regional energy structure correction coefficient according to the second-level industrial structure data, and generate an industry-specific energy structure correction coefficient matrix; If the called hierarchical digital twin sub-model is a third-level digital twin model, calling the graph generator to identify the third-level rule text and extract the third instruction parameter set; The project information uploaded by the project-level digital twin model is called, and a visual project list is generated based on the project information.

6. The carbon emission management method according to claim 1, wherein: The step of calling a pre-built graph generator to extract carbon quota parameters includes: Calling a pre-built graph generator to traverse the rule text and extract core carbon asset parameters, wherein the rule text includes a first-level rule text, a second-level rule text, and a third-level rule text; The core carbon asset parameters are logically checked and associated with the benchmark values ​​of the industry to which the project belongs to obtain carbon quota parameters.

7. The carbon emission management method according to claim 1, wherein: The step of generating a carbon quota scheme based on the carbon quota parameters and the annual carbon emissions includes: Calculate the allowance shortfall based on the total carbon emissions for the annual period; Calculating the quota gap and the carbon quota parameters using a multi-objective optimization algorithm to generate an initial emission reduction plan; The energy structure correction coefficient is synchronized through the federated learning node, and the initial emission reduction plan is modified to obtain a carbon quota plan.

8. A carbon emission management system, characterized in that: The system includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the carbon emission management method according to any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the carbon emission management method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the carbon emission management method according to any one of claims 1 to 7 are implemented.

Images