An AI-based carbon asset and carbon verification management method and platform

Abstract

The application provides an AI-based carbon asset and carbon verification management method and platform, belonging to the technical field of carbon emission management. The method comprises: inputting operation activity data of a target object into a preset energy efficiency analysis model to obtain an energy efficiency analysis result; based on the energy efficiency analysis result, performing energy flow modeling and analysis on the operation activity data of the target object to obtain an energy flow analysis result; inputting the energy flow analysis result into a preset energy efficiency balance optimization model to obtain an optimized energy distribution scheme and predicted carbon emission values of each link based on the energy distribution scheme, wherein the energy efficiency balance optimization model is configured to optimize energy distribution with the goal of minimizing total carbon emissions under the production constraint conditions of the target object; comprehensively calculating the predicted carbon emission values of each link to obtain a predicted total carbon emission value of the target object; and generating a corresponding target carbon asset management strategy according to the predicted total carbon emission value, carbon trading market data and carbon sink data.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of carbon emission management, in particular to an AI-based carbon asset and carbon verification management method and platform. BACKGROUND

[0002] With the promotion of global action on climate change, the proposal of carbon peak and carbon neutralization targets, carbon asset management and carbon verification have become key links for enterprises to realize sustainable development.

[0003] At present, carbon verification and management means rely on manual data collection and experience analysis, which is not only low in efficiency and poor in accuracy, but also difficult to fully grasp the complex and changeable carbon trading market dynamics and carbon emission conditions in enterprise operation. In addition, when processing operation activity data, carbon trading market data and carbon sink data, there are problems such as data integration difficulty and insufficient analysis depth, which are difficult to meet the needs of enterprises for energy utilization efficiency optimization and accurate control of carbon emissions.

[0004] Therefore, an AI-based carbon asset and carbon verification management method and platform are needed to meet the needs of enterprises for energy utilization efficiency optimization and accurate control of carbon emissions. SUMMARY

[0005] In order to solve the above technical problems, the application provides an AI-based carbon asset and carbon verification management method and platform.

[0006] The first aspect of the embodiment of the application provides an AI-based carbon asset and carbon verification management method, which comprises:

[0007] Obtaining multi-source data of a target object, wherein the multi-source data comprises operation activity data, carbon trading market data and carbon sink data of the target object;

[0008] Inputting the operation activity data of the target object into a preset energy efficiency analysis model to obtain an energy efficiency analysis result representing energy utilization efficiency of the target object;

[0009] Based on the energy efficiency analysis result, energy flow modeling and analysis are performed on the operation activity data of the target object to obtain an energy flow analysis result describing energy flow paths, conversion efficiency and loss distribution;

[0010] Inputting the energy flow analysis result into a preset energy efficiency balance optimization model to obtain an optimized energy distribution scheme and predicted carbon emission values of each link based on the energy distribution scheme, wherein the energy efficiency balance optimization model is configured to optimize energy distribution with the goal of minimizing total carbon emissions under the production constraint conditions of the target object;

[0011] Comprehensive calculation of the predicted carbon emission values of each link to obtain a predicted value of the target carbon emission total of the target object;

[0012] generate a corresponding target carbon asset management strategy according to the target carbon emission total amount prediction value, the carbon trading market data, and the carbon sink data.

[0013] In a possible implementation form of the first aspect, the AI-based carbon asset and carbon verification management method further includes:

[0014] obtaining an actual carbon emission verification report of the target object in the actual compliance period;

[0015] calculating actual carbon emission total amount data of the target object based on the actual carbon emission verification report;

[0016] performing deviation analysis on the actual carbon emission total amount data and the target carbon emission total amount prediction value to obtain a deviation analysis result;

[0017] iteratively optimizing parameters of the energy efficiency balance optimization model based on the deviation result.

[0018] In a possible implementation form of the first aspect, iteratively optimizing the parameters of the energy efficiency balance optimization model based on the deviation result includes:

[0019] when an absolute deviation value in the deviation result is greater than a first threshold value or a relative deviation percentage is greater than a second preset threshold value, identifying one or more key links with the greatest deviation contribution between the target carbon emission total amount prediction value and the actual carbon emission total amount data;

[0020] obtaining operation activity data related to the key links and environmental parameter data directly affecting carbon emission of the key links in the actual carbon emission verification report;

[0021] taking the operation activity data, the environmental parameter data, and the corresponding actual carbon emission data of the key links as new training samples;

[0022] training the energy efficiency balance optimization model using the new training samples to iteratively optimize the parameters thereof.

[0023] In a possible implementation form of the first aspect, generating a corresponding target carbon asset management strategy according to the target carbon emission total amount prediction value, the carbon trading market data, and the carbon sink data includes:

[0024] querying a preset carbon asset management strategy knowledge base according to the target carbon emission total amount prediction value, the carbon trading market data, and the carbon sink data, and determining a corresponding target carbon asset management strategy based on a preset matching rule; or

[0025] inputting the target carbon emission total amount prediction value, the carbon trading market data and the carbon sink data into a carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object;

[0026] The target carbon asset management strategy comprises one or more of a carbon quota trading scheme, a carbon sink project offset path and an internal emission reduction measure.

[0027] In a further possible implementation form of the first aspect, the inputting the target carbon emission total amount prediction value, the carbon trading market data and the carbon sink data into a preset carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object comprises:

[0028] determining an initial action space vector of a reinforcement learning algorithm corresponding to the target carbon asset management strategy;

[0029] determining a plurality of standard carbon asset management strategies, and determining an action space vector of a reinforcement learning algorithm corresponding to each standard carbon asset management strategy;

[0030] determining a state space vector of a reinforcement learning algorithm based on the target carbon emission total amount prediction value, the carbon trading market data and the carbon sink data;

[0031] determining a value function of the reinforcement learning algorithm based on the action space vector, the state space vector and a reward function;

[0032] determining an optimal action vector, i.e., the target carbon asset management strategy, based on the initial action space vector and a feedback value obtained by solving the value function;

[0033] The reward function is that a positive reward is given when the action space vector is executed and the predicted or actual total compliance cost is reduced and the production constraint is not violated; and a negative reward is given when the action space vector is executed and the predicted or actual production interruption, the total compliance cost is increased or the constraint is violated.

[0034] In a further possible implementation form of the first aspect, the inputting the target carbon emission total amount prediction value, the carbon trading market data and the carbon sink data into a preset carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object further comprises:

[0035] determining a penalty factor according to a risk index of the carbon trading market;

[0036] determining a value function of the reinforcement learning algorithm based on the action space vector, the state space vector, the reward function and the penalty factor;

[0037] Determine an optimal action vector, i.e., the target carbon asset management strategy, based on the initial action space vector and the feedback value obtained by solving the value function.

[0038] In a possible implementation of the first aspect, the determining the penalty factor based on the risk indicator of the carbon trading market comprises:

[0039] Obtain historical carbon price data, real-time market depth data, transaction volume data, and related policy texts in the carbon trading market data.

[0040] Calculate a price volatility rate based on a standard deviation of the historical carbon price data within a preset time window.

[0041] Calculate a liquidity indicator score based on a combination of the market depth and the transaction volume decay rate according to a preset weight formula.

[0042] Extract keywords based on analysis of the related policy texts by applying natural language processing technology, and determine a policy risk level by combining a preset rule or a classification model.

[0043] Input the price volatility rate indicator, the liquidity indicator score, and the policy risk level into a preset risk scoring function to obtain the penalty factor.

[0044] In a possible implementation of the first aspect, the method for calculating the carbon emission prediction value of each link comprises:

[0045] Analyze the energy distribution scheme to obtain energy types and energy consumption of each link.

[0046] Match the carbon emission factor database to obtain the emission factor corresponding to the energy type of each link.

[0047] For the direct emission link in each link, use the emission factor method to calculate the direct carbon emission of the corresponding link.

[0048] For the link in which a chemical reaction exists, use the mass balance method to calculate the process carbon emission of the corresponding link.

[0049] In a possible implementation of the first aspect, the method for calculating the carbon emission prediction value of each link comprises:

[0050] Obtain first carbon emission data in the production process of a supplier used in each link, and obtain second carbon emission data in the product transportation process of each link.

[0051] The direct carbon emission, the process carbon emission, the corresponding first carbon emission data and second carbon emission data of each link are calculated according to a preset accounting standard to obtain a target carbon emission total amount prediction value of the target object.

[0052] In a possible implementation of the first aspect, the AI-based carbon asset and carbon verification management method further includes:

[0053] Based on the target carbon emission total amount prediction value, the carbon emission prediction value of each link and the energy distribution scheme, a carbon emission verification report is generated according to a preset template;

[0054] The carbon emission verification report is input into a verification rule engine for compliance verification, wherein the verification rule engine is preloaded with carbon verification policies of a jurisdiction where the target object is located and ISO 14064 standard clauses;

[0055] If the compliance verification fails, an abnormal data link is located based on abnormal identification positioning data output by the rule engine;

[0056] If the compliance verification passes, the carbon emission verification report is electronically signed, and a carbon emission verification report with a digital signature is output.

[0057] In a second aspect of the embodiments of the present application, an AI-based carbon asset and carbon verification management platform is provided, including:

[0058] A data acquisition module acquires multi-source data of a target object, the multi-source data including: operation activity data of the target object, carbon trading market data and carbon sink data;

[0059] An energy efficiency analysis module is configured to input the operation activity data of the target object into a preset energy efficiency analysis model to obtain an energy efficiency analysis result representing energy utilization efficiency of the target object;

[0060] An energy flow analysis module is configured to perform energy flow modeling and analysis on the operation activity data of the target object based on the energy efficiency analysis result to obtain an energy flow analysis result describing energy flow paths, conversion efficiency and loss distribution;

[0061] An energy efficiency balance optimization module is configured to input the energy flow analysis result into a preset energy efficiency balance optimization model to obtain an optimized energy distribution scheme and a carbon emission prediction value of each link predicted based on the energy distribution scheme, wherein the energy efficiency balance optimization model is configured to optimize energy distribution with the goal of minimizing total carbon emission under the production constraint condition of the target object;

[0062] A target carbon emission total amount prediction module is configured to comprehensively calculate the carbon emission prediction value of each link to obtain a target carbon emission total amount prediction value of the target object.

[0063] a carbon asset management strategy module configured to generate a corresponding target carbon asset management strategy according to the target total carbon emission prediction value, the carbon trading market data, and the carbon sink data.

[0064] In a third aspect, an electronic device is provided, which includes a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the AI-based carbon asset and carbon verification management method when running the computer program.

[0065] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program implements the steps of the AI-based carbon asset and carbon verification management method when executed by a processor.

[0066] The AI-based carbon asset and carbon verification management method and platform provided by the embodiments of the present application have the following advantages: the present application uses an energy efficiency analysis model and an energy flow modeling to deeply analyze the energy utilization condition, from efficiency evaluation to energy flow detail analysis, to comprehensively master the energy use condition of the target object; secondly, through an energy efficiency balance optimization model, energy distribution is optimized under the condition of meeting production constraints, with the objective of minimizing carbon emissions, which can guarantee production and effectively reduce carbon emissions; finally, the target total carbon emission prediction value is obtained by comprehensively predicting the values of each link, and a carbon asset management strategy is generated in combination with carbon trading and carbon sink data, which meets the needs of enterprises for energy utilization efficiency optimization and precise control of carbon emissions. In addition, the method enables enterprises to reasonably plan transactions, use carbon sink resources, reduce costs, and improve carbon asset value in the carbon market. BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 A flowchart of the AI-based carbon asset and carbon verification management method provided by an embodiment of the present application is shown in the figure;

[0068] Figure 2 A structural block diagram of the AI-based carbon asset and carbon verification management platform provided by an embodiment of the present application is shown in the figure;

[0069] Figure 3 A schematic block diagram of the electronic device provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0070] In the following description, for purposes of explanation and not limitation, specific details are set forth such as particular architectures, technologies, techniques, etc. in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

[0071] For the purpose of making the objectives, technical solutions and advantages of the present application clearer, the following will combine the accompanying drawings to make a detailed description. Figures 1-3 The present application is described by specific embodiments.

[0072] Please refer to Figure 1 , Figure 1 A flowchart of an AI-based carbon asset and carbon verification management method provided by an embodiment of the present application is shown in the figure. The method comprises the following steps.

[0073] S101: Obtain multi-source data of a target object, which includes operation activity data, carbon trading market data and carbon sink data of the target object.

[0074] In the embodiment, the target object is an enterprise or organization that needs carbon asset and carbon verification management. In the embodiment, the operation activity data is obtained in real time from each device of the target object through a preset interface of a carbon asset and carbon verification management platform, which provides data support for subsequent carbon asset management. The operation activity data includes running state, power consumption, running time, material consumption data, energy procurement records and the like of the device.

[0075] The devices of the target object in the embodiment include energy management devices and production operation devices, etc. The energy management devices include energy conversion devices and energy regulation devices. The production operation devices include core production devices and auxiliary production devices. The energy conversion devices include motors, boilers, heat exchangers and the like, which convert primary energy into secondary energy required for production. The energy regulation devices include frequency converters and intelligent switches, etc., which are used to optimize energy distribution, improve energy use efficiency and reduce energy waste by adjusting device running power, start-stop time and the like. The core production devices are determined according to the industry attribute of the target object, such as production line devices of manufacturing industry, reaction kettles of chemical industry and the like. The running state and energy consumption of these production devices directly affect the operation activities and carbon emissions of the target object. The auxiliary production devices include air compressors, ventilators and communication devices, etc., which are used to ensure normal operation of the core production devices.

[0076] This embodiment can calculate the carbon emissions of each device using operational activity data. Specifically, the carbon emission calculation formula is typically based on the device's power consumption and operating time, combined with a carbon emission coefficient. Furthermore, the carbon asset and carbon verification management platform acquires carbon trading market data and carbon sink data through blockchain data interfaces and public data scraping technology. The carbon trading market data includes carbon quota price fluctuations, historical trading data, and policy updates, while the carbon sink data includes progress data on forestry carbon sink projects and soil carbon sequestration monitoring data. In addition, after acquiring multi-source data for the target object, this embodiment uses machine learning algorithms to clean, denoise, and normalize the acquired multi-source data, removing outliers and redundant information to provide a high-quality data foundation for subsequent analysis.

[0077] S102: Input the operational activity data of the target object into the preset energy efficiency analysis model to obtain the energy efficiency analysis results characterizing the energy utilization efficiency of the target object.

[0078] In this embodiment, a preset energy efficiency analysis model is obtained by training a long short-term memory network or a convolutional neural network using historical energy efficiency data. This model can obtain energy efficiency analysis results that characterize the energy utilization efficiency of the target object through in-depth mining and feature extraction of operational activity data, and present them in the form of visual charts to intuitively show the energy utilization efficiency status of the target object. The energy efficiency analysis results include indicators such as energy utilization rate of each production link and equipment energy efficiency ratio, as well as comparative analysis with the industry average level.

[0079] S103: Based on the energy efficiency analysis results, perform energy flow modeling and analysis on the operational activity data of the target object to obtain energy flow analysis results describing the energy flow path, conversion efficiency and loss distribution.

[0080] In this embodiment, based on energy efficiency analysis results, AI-driven energy flow modeling technology is used to perform energy flow modeling and analysis on the operational activity data of the target object. Specifically, a graph neural network is used to construct an energy flow network model, abstracting the target object's production system into a network structure containing energy input nodes, conversion nodes, output nodes, and energy flow edges. This embodiment determines the direction, flow rate, and conversion efficiency of energy flow between nodes through the analysis of operational activity data, simulating the energy flow path at different stages and the energy loss distribution during the conversion process; providing accurate and quantitative analytical basis for energy optimization.

[0081] S104: Input the energy flow analysis results into the preset energy efficiency balance optimization model to obtain the optimized energy allocation scheme and the predicted carbon emission values ​​of each link based on the energy allocation scheme. The energy efficiency balance optimization model is configured as follows: under the production constraints of the target object, optimize energy allocation with the goal of minimizing the total carbon emissions; (the production constraints are: material balance constraints between processes; maximum and minimum load rate constraints of equipment; energy supply stability constraints; and total carbon emission threshold constraints).

[0082] In this embodiment, a linear programming model is used to construct an energy efficiency balance optimization model. This model is constrained by the production constraints of the target object, including inter-process material balance constraints (ensuring a balance between material input and output during production), maximum and minimum equipment load rate constraints (ensuring equipment operates within a reasonable range), energy supply stability constraints (maintaining the continuity and stability of energy supply), and total carbon emission threshold constraints (complying with national or industry-mandated carbon emission limits). The energy allocation is optimized with the goal of minimizing total carbon emissions. In this embodiment, energy flow analysis results are input into the energy efficiency balance optimization model. Through iterative calculation and optimization, an optimized energy allocation scheme is obtained. Simultaneously, based on this energy allocation scheme, the predicted carbon emissions for each stage are calculated.

[0083] In one possible embodiment, the energy efficiency balance optimization model is as follows:

[0084] The objective function (minimizing total carbon emissions) is as follows:

[0085]

[0086] in, For the first The allocation of the first stage Energy quantity, Each step Types of energy; For the first The first step uses the first Carbon emission factors of various energy sources.

[0087] Constraints:

[0088] Inter-process material balance constraints:

[0089]

[0090] in, For input process A collection of links; For output process A collection of links; For input stage The material conversion coefficient is determined by the production process; For output stage The material conversion coefficient is determined by the production process; For the first The first input step consumes the first... Energy quantity; For the first The first output stage consumes the first... Energy quantity.

[0091] Equipment load rate constraints:

[0092]

[0093] in, For the first Minimum load rate of equipment in the process; For the first Maximum load rate of equipment in the process; This represents the equipment load factor per unit of energy.

[0094] Energy supply stability constraints:

[0095]

[0096] in, For the first The minimum supply capacity of this energy source; For the first The maximum supply capacity of this energy source.

[0097] Carbon emission threshold constraints:

[0098]

[0099] in, This is the carbon emission cap stipulated by policy.

[0100] Solve the linear programming problem using the simplex method, and output:

[0101] Optimal energy allocation scheme ;

[0102] Predicted carbon emissions for each stage: ;

[0103] Total carbon emissions forecast .

[0104] S105: Integrate the carbon emission prediction values of each link to obtain a target carbon emission total amount prediction value of the target object. In this embodiment, the target carbon emission total amount prediction value of the target object is calculated by weighted summation of the carbon emission prediction values of each link. In this embodiment, different weights can be given according to the contribution of different production links to carbon emission, so as to obtain more accurate calculation results.

[0105] S106: Generate a corresponding target carbon asset management strategy according to the target carbon emission total amount prediction value, carbon trading market data and carbon sink data.

[0106] In this embodiment, based on the target carbon emission total amount prediction value, combined with the carbon quota price trend, transaction cost in the carbon trading market data and the available carbon sink amount, carbon sink transaction price and other information in the carbon sink data, there are multiple strategy determination methods.

[0107] The first method is to use decision analysis methods such as decision tree and Bayesian network to generate a corresponding target carbon asset management strategy based on the target carbon emission total amount prediction value, combined with the carbon quota price trend, transaction cost in the carbon trading market data and the available carbon sink amount, carbon sink transaction price and other information in the carbon sink data.

[0108] The second method is to query a preset carbon asset management strategy knowledge base, which stores a large number of carbon asset management strategy cases based on different industries and different carbon emission scenarios. Based on the preset matching rules, the current data is matched with the cases in the knowledge base to filter out the most suitable target carbon asset management strategy for the target object. The matching rules include: lowest cost, lowest risk, optimal offset ratio, etc. For example, when the lowest cost is used as the matching rule, the strategy that minimizes the carbon management cost is preferred.

[0109] The third method is to input the target carbon emission total amount prediction value, carbon trading market data and carbon sink data into a carbon asset management strategy recommendation model; the model is based on a reinforcement learning algorithm and outputs a target carbon asset management strategy for the target object.

[0110] In this embodiment, the target carbon asset management strategy includes but is not limited to: developing a carbon quota purchase plan when the predicted carbon emission total amount exceeds a threshold; planning the timing of selling carbon quotas when there are excess carbon quotas; and optimizing the allocation of carbon assets by developing or purchasing carbon sink projects to reduce the carbon management cost and carbon emission risk of the enterprise.

[0111] From the above, the present application uses energy efficiency analysis model and energy flow modeling to deeply analyze the energy utilization situation, from efficiency evaluation to energy flow detail analysis, to comprehensively master the energy use of the target object; secondly, through the energy efficiency balance optimization model, the energy distribution is optimized under the condition of meeting the production constraints, with the target of minimizing carbon emissions, which can not only guarantee production, but also effectively reduce carbon emissions, finally the target carbon emission total amount prediction value is obtained by comprehensively predicting the values of each link, and the carbon asset management strategy is generated combined with carbon trading and carbon sink data, which meets the needs of enterprises for energy utilization efficiency optimization and precise control of carbon emissions. In addition, the method enables enterprises to reasonably plan transactions, use carbon sink resources, reduce costs, and improve carbon asset value in the carbon market.

[0112] In an embodiment of the present application, the AI-based carbon asset and carbon verification management method further comprises:

[0113] Obtaining an actual carbon emission verification report of the target object in the actual performance period;

[0114] Based on the actual carbon emission verification report, the actual carbon emission total amount data of the target object is calculated;

[0115] The deviation analysis result is obtained by performing deviation analysis on the actual carbon emission total amount data and the target carbon emission total amount prediction value;

[0116] Based on the deviation result, the parameters of the energy efficiency balance optimization model are iteratively optimized.

[0117] In this embodiment, the actual carbon emission verification report of the target object in the actual performance period is obtained through the blockchain technology, and the actual carbon emission verification report is uploaded to the blockchain by a third-party certification agency. At the same time, the present embodiment can also perform integrity check on the obtained actual carbon emission verification report, check whether the actual carbon emission verification report contains key contents such as enterprise basic information, verification range, verification method, emission source identification, data quality control, etc., and if there is a missing, trigger the early warning mechanism to prompt the relevant personnel to complete.

[0118] In this embodiment, based on the actual carbon emission verification report, the actual carbon emission total amount data of the target object is calculated, specifically, according to the calculation method of different emission sources, combined with the activity data, emission factor and other parameters provided in the actual carbon emission verification report, according to the relevant standards such as "Greenhouse Gas Emission Accounting and Reporting Requirements", the carbon emissions of each emission source is accurately calculated. The carbon emissions of each emission source is summarized by weighted summation, and the carbon offset amount in the actual carbon emission verification report is calculated, and finally the actual carbon emission total amount data of the target object is obtained. The carbon offset amount in the actual carbon emission verification report is offset through carbon sink projects or carbon quota trading.

[0119] In this embodiment, the actual total carbon emission data and the target total carbon emission prediction value can be analyzed from multiple dimensions. At the numerical level, the absolute deviation of the actual total carbon emission data and the target total carbon emission prediction value and the relative deviation of the actual total carbon emission data and the target total carbon emission prediction value are calculated. At the time dimension, the deviation trend of different time periods is compared to analyze whether the deviation has periodic characteristics.

[0120] In this embodiment, the key emission sources that have a greater impact on the total deviation are determined by deeply analyzing the deviation of each emission source. According to the energy efficiency analysis results, the energy flow analysis results, and the actual situation in the production and operation process, the causes of the deviation are analyzed, such as production process parameter fluctuation, energy supply quality change, unreasonable model parameter setting, etc.

[0121] In this embodiment, a parameter optimization rule library based on the deviation analysis results can be established, which includes the mapping relationship between different deviation types and parameter adjustment strategies. For example, when it is found that the actual carbon emission of a certain production link is much higher than the prediction value and is related to unreasonable energy distribution, the adjustment rule of the energy distribution weight parameter of the link is triggered. This embodiment can also use optimization algorithms such as genetic algorithm to iteratively optimize the parameters of the energy efficiency balance optimization model with the goal of reducing the deviation between the actual total carbon emission and the prediction value. In the optimization process, the iteration termination condition is set, which includes reaching the maximum number of iterations, the target function value converging to a certain accuracy range, etc.

[0122] In summary, through the deviation analysis of the actual carbon emission data and the prediction value, a closed loop of prediction-execution-feedback-optimization is formed, which enables the energy efficiency balance optimization model to dynamically adjust the parameters of the energy efficiency balance optimization model according to the actual operation situation, thereby improving the prediction accuracy and the accuracy of carbon emission accounting.

[0123] In an embodiment of the present application, based on the deviation result, the parameters of the energy efficiency balance optimization model are iteratively optimized, including: when the absolute deviation value in the deviation result is greater than a first threshold or the relative deviation percentage is greater than a second preset threshold, one or more key links that contribute most to the deviation between the target total carbon emission prediction value and the actual total carbon emission data are identified;

[0124] Obtain the operation activity data related to the key link and the environmental parameter data directly affecting the carbon emission of the key link in the actual carbon emission verification report;

[0125] The operation activity data, environmental parameter data, and corresponding actual carbon emission data of the key link are used as new training samples;

[0126] The energy efficiency balance optimization model is trained using the new training samples to iteratively optimize its parameters.

[0127] In the present embodiment, the first threshold value and the second preset threshold value are obtained based on statistical analysis of historical carbon emission data of the industry in which the target object is located. Specifically, by collecting the deviation data between the predicted value and the actual value of carbon emissions of a large number of enterprises in the industry, statistical methods such as calculating the mean and standard deviation are used to determine the reasonable first threshold value and the second preset threshold value in combination with the accuracy requirements of carbon emission management in the industry. For example, if the absolute deviation value corresponding to the mean plus twice the standard deviation of the deviation data in the industry is X, the first threshold value is set to a value slightly higher than X; if the 90th percentile of the relative deviation in the industry is Y%, the second preset threshold value is set to Y%.

[0128] In the present embodiment, when the absolute deviation value in the deviation result is greater than the first threshold value or the relative deviation percentage is greater than the second preset threshold value, a sensitivity analysis method is used to calculate the contribution of each production link to the total deviation; the energy consumption, carbon emission factor and other parameters of each link are disturbed, and the change amplitude of the total deviation is observed. The greater the change amplitude, the higher the contribution of the link to the total deviation, so as to identify one or more key links with the greatest contribution to the deviation between the predicted value of the target carbon emission total and the actual carbon emission total data.

[0129] In the present embodiment, a key link identification model based on association rule mining can be constructed to analyze the association between the predicted value of the target carbon emission total and the actual carbon emission total data in each production link and energy use process, and to identify one or more key links with the greatest contribution to the deviation.

[0130] In the present embodiment, from the actual carbon emission verification report, the text content is parsed by natural language processing technology, and combined with the structured data extraction method, the operation activity data related to the key link is accurately obtained, such as equipment running time, energy input and output flow, material ratio, etc. In the present embodiment, environmental parameter data directly affecting the carbon emission of the key link is collected by meteorological monitoring equipment and geographic information system, including temperature, humidity, wind speed and other meteorological conditions, as well as regional power grid energy structure, real-time policy changes in carbon trading market and other external environmental data. The operation activity data, environmental parameter data and corresponding actual carbon emission data of the key link obtained in the present embodiment are preprocessed; wherein the preprocessing includes: using data normalization method to unify different orders of magnitude of data to the same interval; converting non-numeric data into a numerical form that can be processed by the energy efficiency balance optimization model through data encoding technology.

[0131] In summary, the present embodiment focuses on the key link with the greatest contribution when the deviation exceeds the threshold value, improving the efficiency and pertinence of model iteration. Secondly, the operation data and environmental parameters of the key link are used as new training samples, so that the model can more accurately capture the core factors affecting carbon emission and improve the local prediction accuracy.

[0132] In an embodiment of the present application, a corresponding target carbon asset management strategy is generated according to the target total carbon emission prediction value, carbon trading market data and carbon sink data, including:

[0133] According to the target total carbon emission prediction value, carbon trading market data and carbon sink data, a preset carbon asset management strategy knowledge base is queried, and a corresponding target carbon asset management strategy is determined based on a preset matching rule; or,

[0134] The target total carbon emission prediction value, carbon trading market data and carbon sink data are input into a carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object.

[0135] The target carbon asset management strategy includes one or more of a carbon quota trading scheme, a carbon sink project offset path and an internal emission reduction measure.

[0136] In the present embodiment, the carbon asset management strategy knowledge base stores a large number of carbon asset management strategy cases based on different industries and different carbon emission scenarios. Specifically, the carbon asset management strategy knowledge base includes a rule base, a case base and a model base; wherein the rule base stores deterministic knowledge such as carbon trading policies and regulations and industry standards; the case base collects historical carbon asset management success / failure cases, wherein each case includes enterprise characteristics, strategy selection, implementation effect and other information; and the model base includes carbon emission prediction, carbon price fluctuation analysis, carbon sink project evaluation and other models.

[0137] In the present embodiment, the matching rules include cost minimization, risk minimization, optimal offset ratio, etc. Specifically, if the cost minimization rule is selected, the cost under different strategy combinations is first calculated, including carbon quota purchase cost, carbon sink project investment cost, internal emission reduction measure transformation cost, etc. At the same time, according to the potential cost changes brought by the price fluctuations of the carbon trading market, the Monte Carlo simulation method is used to predict the future carbon price, evaluate the cost expectation of each strategy under different market scenarios, and select the target carbon asset management strategy with the lowest cost. If the risk minimization rule is selected, a risk assessment index system is constructed to score the strategy from the dimensions of market risk, policy risk and technical risk, and the strategy with the lowest comprehensive risk score is preferentially selected.

[0138] In this embodiment, a reinforcement learning model based on deep Q network is used to construct a carbon asset management strategy recommendation model. The defined state space includes: target total carbon emission prediction value, carbon trading market data, carbon sink data, and enterprise's own carbon asset reserves, production and operation status, etc. The defined action space includes: carbon quota trading, carbon sink project selection, and internal emission reduction measures implementation, etc. The reward function is designed according to the actual effect after the implementation of the strategy, for example, positive rewards are given to positive effects such as carbon emission reduction, carbon asset income, cost saving amount, etc., and negative rewards are given to negative results such as carbon emission exceeding the standard, transaction loss, etc.

[0139] In this embodiment, the carbon quota trading scheme includes: comparing the target total carbon emission prediction value with the number of carbon quotas held by the enterprise, and formulating a carbon quota trading plan. When the predicted total carbon emission exceeds the held quota, the price trend of the carbon trading market is analyzed, and sufficient quota is purchased at a low carbon price; when there is surplus quota, the market supply and demand relationship and price fluctuation law are combined to plan the selling opportunity, so as to realize the value-added of carbon assets. At the same time, financial tools such as hedging are used to reduce the transaction risk caused by carbon price fluctuations.

[0140] The carbon sink project offset path includes: evaluating the resources and conditions of the enterprise itself, selecting appropriate carbon sink project types such as forestry carbon sink, wetland carbon sink or carbon credits generated by renewable energy projects. For forestry carbon sink projects, the area of afforestation, tree species selection, planting scale and maintenance plan are planned; for renewable energy projects, the feasibility and economy of the project are analyzed, and the specific strategy of investment construction or purchase of carbon credits is determined. In the implementation process, relevant standards and certification processes are strictly followed to ensure that the emission reduction of the carbon sink project is certifiable and tradable.

[0141] Internal emission reduction measures include: formulating internal emission reduction measures from energy management, production process improvement, equipment upgrading, etc. In energy management, an intelligent energy management system is introduced to monitor energy consumption in real time and optimize energy distribution; in production process, low-carbon production technology is developed or adopted to reduce the carbon emission per unit product; for high-energy-consuming equipment, a device replacement plan is developed to replace energy-saving devices, improve energy efficiency, and reduce carbon emissions.

[0142] In this embodiment, the enterprise can flexibly combine the above three strategies according to its actual situation to form the most suitable target carbon asset management strategy to adapt to the management needs of different enterprises.

[0143] In an embodiment of the present application, the target total carbon emission prediction value, carbon trading market data and carbon sink data are input into the preset carbon asset management strategy recommendation model to obtain the target carbon asset management strategy corresponding to the target object, including:

[0144] determining an initial action space vector of the reinforcement learning algorithm corresponding to the target carbon asset management strategy;

[0145] determining a plurality of standard carbon asset management strategies, and determining an action space vector of the reinforcement learning algorithm corresponding to each standard carbon asset management strategy;

[0146] determining a state space vector of the reinforcement learning algorithm based on the predicted total target carbon emission, carbon trading market data and carbon sink data;

[0147] determining a value function of the reinforcement learning algorithm based on the action space vector, the state space vector and the reward function;

[0148] determining an optimal action vector, i.e. the target carbon asset management strategy, based on the initial action space vector and the feedback value obtained by solving the value function;

[0149] The reward function is: when the action space vector is executed, a positive reward is given when the predicted or actual total compliance cost is reduced and the production constraint is not violated; a negative reward is given when the action space vector leads to predicted or actual production interruption, increased total compliance cost or violation of constraints.

[0150] In this embodiment, various types of optional actions of the carbon asset management strategy are encoded to form an initial action space vector. The carbon quota trading, carbon sink project selection, internal emission reduction measure execution and other action types are encoded, and each action type corresponds to a dimension in the vector. For example, if there are n possible action types, the initial action space vector is an n-dimensional vector, and the value at each position in the vector represents the probability or weight of executing the corresponding action. At the beginning, the weights of each action can be set equally or based on historical experience to set the initial preference.

[0151] In this embodiment, a plurality of standard carbon asset management strategies can be determined by analyzing typical successful cases in the industry, and the action space vector corresponding to each standard carbon asset management strategy can be determined by expert knowledge or historical data statistics.

[0152] In this embodiment, the state space vector of the reinforcement learning algorithm is constructed based on the predicted total target carbon emission, carbon trading market data and carbon sink data. The predicted total target carbon emission is normalized and mapped to the interval [0, 1] as a dimension of the state vector; for carbon trading market data, features such as carbon price trend (e.g. price change rate in the last three months), trading volume, market liquidity, etc. are extracted, and dimensionality reduction methods such as principal component analysis are used to compress them into low-dimensional vectors; for carbon sink data, features such as available carbon sink amount, carbon sink price, yield of different types of carbon sink projects, etc. are extracted, and dimensionality reduction is also performed; the processed feature vectors of each part are spliced to form a complete state space vector.

[0153] In this embodiment, based on the initial action space vector, the feedback value obtained by solving the value function is combined to explore the action space using a greedy strategy to find the action vector with the maximum value; during the exploration process, the exploration probability is dynamically adjusted according to the risk preference of the enterprise, and the enterprise with higher risk preference appropriately increases the exploration intensity to discover better strategies; the enterprise with lower risk preference is more inclined to select known better strategies.

[0154] In an embodiment of the present application, the target total carbon emission prediction value, carbon trading market data and carbon sink data are input into a preset carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object, and the method further comprises:

[0155] The penalty factor is determined according to the risk indicators of the carbon trading market;

[0156] The value function of the reinforcement learning algorithm is determined based on the action space vector, the state space vector, the reward function and the penalty factor;

[0157] The optimal action vector, i.e., the target carbon asset management strategy, is determined based on the initial action space vector and the feedback value obtained by solving the value function.

[0158] In this embodiment, a penalty factor calculation model is constructed based on the risk indicators of the carbon trading market; the model is constructed based on multidimensional factors such as market price fluctuation risk, liquidity risk and policy risk; in this embodiment, the dynamic adjustment of the penalty factor can be obtained by quantifying and weighting the risk indicators. The penalty factor is used to additionally punish high-risk strategies in the value function calculation of reinforcement learning, and guide the agent to learn the optimal strategy balancing risk and return.

[0159] In this embodiment, by adding a risk penalty term in the value function, the dynamic risk of the carbon trading market is converted into a quantifiable decision constraint, avoiding decision failure caused by market uncertainty and preventing a sharp increase in compliance costs caused by policy mutations.

[0160] In one embodiment,

[0161] The value function is:

[0162] ,

[0163] The reward function is:

[0164] ,

[0165] wherein, is the state space vector; is the action space vector; A is the action space, including all possible actions; is the total compliance cost change amount, and ; is the baseline total compliance cost, calculated by the compliance cost when no strategy is implemented in historical data; is the production constraint violation identifier; is the reward base, a constant; is the penalty factor; is the strategy risk exposure; is the discount factor, preset [0, 1), is the absolute value of the total compliance cost change.

[0166] wherein F(s, a) is the production constraint violation identifier, defined as:

[0167] ,

[0168] wherein,

[0169]

[0170] wherein, is the absolute value of the carbon trading volume, is the current carbon price; is the price volatility; is the absolute value of the carbon sink purchase volume; is the unit price of the carbon sink; is the policy risk.

[0171] In an embodiment of the present application, the penalty factor is determined according to the risk indicators of the carbon trading market, comprising:

[0172] acquiring historical carbon price data, real-time market depth data, trading volume data and related policy texts in carbon trading market data;

[0173] calculating the price volatility based on the standard deviation of the historical carbon price data within a preset time window;

[0174] calculating the liquidity indicator score according to a preset weight formula based on the combination of market depth and trading volume decay rate;

[0175] extracting keywords based on the application of natural language processing technology to analyze the related policy texts and determining the policy risk level by combining a preset rule or classification model;

[0176] inputting the price volatility indicator, the liquidity indicator score and the policy risk level into a preset risk scoring function to obtain the penalty factor.

[0177] In this embodiment, based on historical carbon price data within a preset time window, the standard deviation of the price sequence is calculated and used as a price volatility indicator. The above indicator is standardized to map to the interval [0, 1] to obtain a quantitative score of price volatility. In this embodiment, market depth data and trading volume data are combined to construct a liquidity evaluation model. Market depth is defined as the total quantity of the current bid-ask spread, and trading volume decay rate is defined as the sensitivity of trading volume to price changes. In this embodiment, natural language processing techniques are applied to analyze relevant policy texts. The steps include: first, text preprocessing, including word segmentation, stop word removal, and morphological restoration; then extracting keywords from policy texts; constructing a policy impact classification model to classify policy texts and determine policy risk levels; for example, high risk level, medium risk level, and low risk level.

[0178] In this embodiment, the price volatility calculation formula is:

[0179]

[0180] wherein, is the preset time window length; is the time point sequence number; is the carbon price in the time window; is the carbon price in the time window; is the average carbon price in the time window.

[0181] The liquidity score calculation formula is:

[0182]

[0183] wherein, is the liquidity score; is the weight of market depth; is the weight of trading volume decay rate; is the market depth; is the trading volume decay rate.

[0184] The policy risk level calculation formula is:

[0185]

[0186] wherein, is the natural language processing model; is the policy document issued by the government / regulatory agency.

[0187] The penalty factor calculation formula is:

[0188] wherein, is the price volatility weight; is the liquidity score weight; Policy risk weight; wherein the three weights can be obtained according to historical experience, or set by the user.

[0189] In an embodiment of the present application, the calculation method of the carbon emission prediction value of each link comprises:

[0190] The energy allocation scheme is analyzed to obtain the energy type and energy consumption of each link.

[0191] Based on the carbon emission factor database, the emission factor corresponding to the energy type of each link is matched.

[0192] For the direct emission link in each link, the direct carbon emission of the corresponding link is calculated by using the emission factor method.

[0193] For the link with chemical reaction in each link, the process carbon emission of the corresponding link is calculated by using the mass balance method.

[0194] In the present embodiment, the optimized energy allocation scheme is structurally analyzed, and the energy type and energy consumption of each production link are extracted, wherein the energy type includes: electricity, natural gas, coal, biomass energy, etc. A data extraction algorithm is used to identify key data fields from the text or table of the energy allocation scheme, and an energy type-consumption mapping table is established.

[0195] In the present embodiment, the carbon emission factor database comprises: emission factor data of different energy types, different production processes and different regions. Specifically, the database is designed using a hierarchical structure, wherein the first layer is classified according to the energy type, the second layer is classified according to the technical level, and the third layer is classified according to the regional characteristics. When the emission factor corresponding to the energy type of each link is matched, a semantic matching algorithm is used to match the energy type obtained by analysis with the emission factor in the carbon emission factor database, and the emission factor data closest to the actual production is preferentially selected. For example, for electricity energy, the corresponding power grid emission factor is selected according to the regional power grid structure; for fossil fuels, the corresponding emission factor is selected according to the fuel quality and combustion equipment type.

[0196] In this embodiment, for the direct emission link, the emission factor method is used to calculate the carbon emission; wherein, the direct emission link includes fuel combustion and waste disposal, etc. For the link with chemical reaction, the mass balance method is used to calculate the process carbon emission; wherein, the link with chemical reaction includes cement production, chemical synthesis, etc. In this embodiment, for the link with chemical reaction, first, the chemical reaction equation is established to determine the material conservation relationship between the raw material input and the product output; then, the carbon element conversion path in the reaction process is analyzed to calculate the carbon element emission. For example, in cement production, according to the limestone decomposition reaction CaCO3→CaO+CO2, combined with the limestone input and the calcium carbonate content, the CO2 emission generated by decomposition is calculated.

[0197] In an embodiment of the present application, the predicted values of carbon emissions of each link are comprehensively calculated to obtain the predicted value of the total target carbon emission of the target object, comprising:

[0198] The first carbon emission data in the production process of the suppliers used in each link is obtained, and the second carbon emission data in the product transportation process of each link is obtained;

[0199] The direct carbon emission, the process carbon emission, the corresponding first carbon emission data and the second carbon emission data of each link are calculated according to the preset accounting standard to obtain the predicted value of the total target carbon emission of the target object.

[0200] In this embodiment, through data interface or questionnaire survey, the first carbon emission data in the production process of the suppliers of raw materials and intermediate products used in each link is obtained. The data content includes the whole process carbon emission information of raw material mining, processing and transportation to the enterprise. At the same time, by using the data provided by the logistics management system and the transportation service provider, the second carbon emission data in the product transportation process of each link is obtained, including transportation mode (road, railway, waterway, aviation), transportation distance, transportation tool energy consumption and other information.

[0201] In this embodiment, the direct carbon emission, the process carbon emission, the corresponding first carbon emission data and the second carbon emission data of each link are integrated and calculated according to the preset accounting standard (for example, ISO14067 product carbon footprint standard, PAS2050 product and service life cycle greenhouse gas emission evaluation specification).

[0202] In an embodiment of the present application, the AI-based carbon asset and carbon verification management method further comprises:

[0203] Based on the predicted value of the total target carbon emission, the predicted value of the carbon emission of each link and the energy allocation scheme, a carbon emission verification report is generated according to a preset template;

[0204] The carbon emission verification report is input into a verification rule engine for compliance checking, wherein the verification rule engine is pre-stored with carbon verification policies of a jurisdiction where the target object is located and ISO14064 standard clauses;

[0205] If the compliance checking fails, an abnormal data link is located based on an exception identifier output by the rule engine;

[0206] If the compliance checking passes, the carbon emission verification report is electronically signed, and a carbon emission verification report with a digital signature is output.

[0207] In the embodiment, based on the target carbon emission total value, the carbon emission values of each link, and the energy distribution scheme, a carbon emission verification report with clear logic and detailed content is automatically generated according to a preset template. The carbon emission verification report includes enterprise carbon emission status, prediction trend, and emission reduction measures. In the embodiment, a visual chart generation technology can be used to convert carbon emission data into various visual forms such as line charts, column charts, pie charts, and Sankey diagrams, to intuitively display information such as carbon emission proportion of each link, energy flow direction, and carbon emission correlation. In the embodiment, a third-party platform can be introduced to obtain the target carbon emission total value, the carbon emission values of each link, and the energy distribution scheme to generate a carbon emission verification report with clear logic and detailed content.

[0208] In the embodiment, the verification rule engine adopts a micro-service architecture, and the jurisdiction policy rules and the ISO14064 standard rules are split into independent rule micro-services. For the jurisdiction policy rules, a dynamic monitoring mechanism is established, and the updated information of policy and regulation websites is crawled in real time through a web crawler technology. Once a policy change is found, a rule updating process of the verification rule engine is automatically triggered, and a semantic matching and rule conversion technology in natural language processing is used to convert the new policy clauses into executable logical expressions.

[0209] In the embodiment, when the verification rule engine outputs an exception identifier, an association analysis algorithm is used to quickly locate the data abnormal link in combination with the logical relationship between carbon emission data and the business process. For example, through tracing the data source, calculation process, and related influencing factors, it is accurately determined whether the abnormality is caused by data entry error, calculation model deviation, or policy understanding deviation.

[0210] In the embodiment, a distributed electronic signature technology based on a blockchain is used, an asymmetric encryption algorithm is used to generate a unique digital signature, and the digital signature information is bound with the carbon emission verification report content to ensure that the content of the carbon emission verification report cannot be tampered with.

[0211] Corresponding to the AI-based carbon asset and carbon verification management method of the above embodiment, Figure 2A structural block diagram of an AI-based carbon asset and carbon verification management platform is provided for an embodiment of the present application. For ease of illustration, only parts related to the embodiments of the present application are shown. Referring to Figure 2 The AI-based carbon asset and carbon verification management platform 20 includes a data acquisition module 21, an energy efficiency analysis module 22, an energy flow analysis module 23, an energy efficiency balance optimization module 24, a target carbon emission total amount prediction module 25, and a carbon asset management strategy module 26.

[0212] The data acquisition module 21 acquires multi-source data of the target object, including operation activity data of the target object, carbon trading market data, and carbon sink data.

[0213] The energy efficiency analysis module 22 is configured to input the operation activity data of the target object into a preset energy efficiency analysis model to obtain an energy efficiency analysis result representing the energy utilization efficiency of the target object.

[0214] The energy flow analysis module 23 is configured to perform energy flow modeling and analysis on the operation activity data of the target object based on the energy efficiency analysis result to obtain an energy flow analysis result describing the energy flow path, conversion efficiency, and loss distribution.

[0215] The energy efficiency balance optimization module 24 is configured to input the energy flow analysis result into a preset energy efficiency balance optimization model to obtain an optimized energy distribution scheme and predicted carbon emission values of each link based on the energy distribution scheme, wherein the energy efficiency balance optimization model is configured to optimize energy distribution with the goal of minimizing total carbon emissions while meeting the production constraints of the target object.

[0216] The target carbon emission total amount prediction module 25 is configured to comprehensively calculate the predicted carbon emission values of each link to obtain a predicted value of the target carbon emission total amount of the target object.

[0217] The carbon asset management strategy module 26 is configured to generate a corresponding target carbon asset management strategy based on the predicted value of the target carbon emission total amount, the carbon trading market data, and the carbon sink data.

[0218] Referring to Figure 3 , Figure 3 A schematic block diagram of an electronic device is provided for an embodiment of the present application. As shown in Figure 3The electronic device 300 in the embodiment shown can include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The above-mentioned processors 301, input devices 302, output devices 303, and memories 304 complete mutual communication through a communication bus 305. The memory 304 is used to store a computer program, and the computer program includes program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. Among them, the processor 301 is configured to invoke the program instructions to execute the functions of each module in the above-mentioned device embodiments, for example Figure 2 The functions of the data acquisition module 21, the energy efficiency analysis module 22, the energy flow analysis module 23, the energy efficiency balance optimization module 24, the target carbon emission total amount prediction module 25, and the carbon asset management strategy module 26 shown are described.

[0219] It should be understood that in the embodiments of the present application, the processor 301 can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0220] The input device 302 can include a touchpad, a fingerprint collection sensor (used to collect fingerprint information and direction information of a user's fingerprint), a microphone, etc., and the output device 303 can include a display (LCD, etc.), a speaker, etc.

[0221] The memory 304 can include read-only memory and random access memory, and provide instructions and data to the processor 301. A part of the memory 304 can also include non-volatile random access memory. For example, the memory 304 can also store device type information.

[0222] In a specific implementation, the processor 301, the input device 302, and the output device 303 described in the embodiments of the present application can execute any embodiment of the AI-based carbon asset and carbon verification management method provided by the embodiments of the present application, and can also execute the implementation manner of the electronic device described in the embodiments of the present application, which will not be described here.

[0223] In another embodiment of the present application, a computer readable storage medium is provided, which stores a computer program. The computer program includes program instructions, which, when executed by a processor, implement all or part of the processes of the above-mentioned embodiment methods. The computer program can also instruct related hardware to complete the implementation. The computer program can be stored in a computer readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0224] The computer readable storage medium can be an internal storage unit of the electronic device of any of the preceding embodiments, such as a hard disk or a memory of the electronic device. The computer readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the computer readable storage medium can include both the internal storage unit and the external storage device of the electronic device. The computer readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0225] Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be realized in electronic hardware, computer software or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been described in general terms in the above description. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0226] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the electronic device and the units described above can refer to the corresponding processes in the above-mentioned method embodiments, which will not be described here.

[0227] In several embodiments provided in the present application, it should be understood that the disclosed electronic device and method can be implemented in other manners. For example, the embodiments of the apparatus described above are merely schematic; the division of the units is merely logical function division; an actual implementation can be divided into different units depending on actual conditions; or a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, or can be in electrical, mechanical or other forms.

[0228] The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments of the present application.

[0229] In addition, each functional unit in the various embodiments of the present application can be integrated in one processing unit, or each unit can exist physically as a separate unit, or two or more units can be integrated in one unit. The integrated unit can be implemented in the form of hardware, or in the form of a software functional unit.

[0230] The above is merely specific embodiments of the present application, but the protection scope of the present application is not limited thereto; any skilled person in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. An AI-based method for carbon asset and carbon verification management, characterized in that, include: Acquire multi-source data of the target object, including: operational activity data of the target object, carbon trading market data, and carbon sink data; The operational activity data of the target object is input into a preset energy efficiency analysis model to obtain energy efficiency analysis results characterizing the energy utilization efficiency of the target object; the preset energy efficiency analysis model is obtained by training a long short-term memory network or a convolutional neural network using historical energy efficiency data. Based on the energy efficiency analysis results, energy flow modeling and analysis are performed on the operational activity data of the target object to obtain energy flow analysis results describing energy flow paths, conversion efficiency, and loss distribution. The energy flow analysis results are input into a preset energy efficiency balance optimization model to obtain an optimized energy allocation scheme and carbon emission prediction values ​​for each stage based on the energy allocation scheme. The energy efficiency balance optimization model is configured to optimize energy allocation with the goal of minimizing total carbon emissions while meeting the production constraints of the target object. By comprehensively calculating the predicted carbon emissions of each stage, the predicted total carbon emissions of the target object are obtained. Based on the target total carbon emissions forecast, the carbon trading market data, and the carbon sink data, a corresponding target carbon asset management strategy is generated. The step of generating a corresponding target carbon asset management strategy based on the predicted target total carbon emissions, the carbon trading market data, and the carbon sink data includes: Based on the predicted target total carbon emissions, the carbon trading market data, and the carbon sink data, a pre-defined carbon asset management strategy knowledge base is queried, and a corresponding target carbon asset management strategy is determined based on pre-defined matching rules; or, The predicted total carbon emissions, the carbon trading market data, and the carbon sink data are input into the carbon asset management strategy recommendation model to obtain the target carbon asset management strategy corresponding to the target object. The target carbon asset management strategy includes one or more of the following: carbon quota trading schemes, carbon sink project offsetting pathways, and internal emission reduction measures.

2. The AI-based carbon asset and carbon verification management method according to claim 1, characterized in that, Also includes: Obtain the actual carbon emission verification report for the target entity during its actual compliance period; Based on the actual carbon emission verification report, the total actual carbon emissions of the target object are calculated. A deviation analysis is performed between the actual total carbon emissions data and the predicted target total carbon emissions to obtain the deviation analysis results. Based on the deviation analysis results, the parameters of the energy efficiency balance optimization model are iteratively optimized.

3. The AI-based carbon asset and carbon verification management method according to claim 2, characterized in that, The iterative optimization of the parameters of the energy efficiency balance optimization model based on the deviation analysis results includes: When the absolute deviation value in the deviation analysis result is greater than the first threshold or the relative deviation percentage is greater than the second preset threshold, one or more key links that contribute the most to the deviation between the predicted value of the target total carbon emissions and the actual total carbon emissions data are identified. Obtain operational activity data related to the key process and environmental parameter data that directly affect the carbon emissions of the key process from the actual carbon emission verification report; The operational activity data, environmental parameter data, and corresponding actual carbon emission data of the key links will be used as new training samples. The energy efficiency balance optimization model is trained using the newly added training samples, and its parameters are iteratively optimized.

4. The AI-based carbon asset and carbon verification management method according to claim 1, characterized in that, The step of inputting the predicted total carbon emissions, the carbon trading market data, and the carbon sink data into a preset carbon asset management strategy recommendation model to obtain a target carbon asset management strategy corresponding to the target object includes: Determine the initial action space vector of the reinforcement learning algorithm corresponding to the target carbon asset management strategy; Multiple standard carbon asset management strategies are identified, and the action space vector of the reinforcement learning algorithm corresponding to each standard carbon asset management strategy is determined. The state space vector of the reinforcement learning algorithm is determined based on the predicted total carbon emissions, the carbon trading market data, and the carbon sink data. The value function of the reinforcement learning algorithm is determined based on the action space vector, the state space vector, and the reward function. The optimal action vector, i.e. the target carbon asset management strategy, is determined based on the initial action space vector and the feedback value obtained by solving the value function. The reward function is as follows: when the execution of the action space vector reduces the predicted or actual total compliance cost and does not violate production constraints, a positive reward is given; when the execution of the action space vector causes the predicted or actual production to be interrupted, the total compliance cost to increase, or the constraints to be violated, a negative reward is given.

5. The AI-based carbon asset and carbon verification management method according to claim 4, characterized in that, Also includes: Penalty factors are determined based on risk indicators in the carbon trading market; The value function of the reinforcement learning algorithm is determined based on the action space vector, state space vector, reward function, and penalty factor. The optimal action vector, i.e. the target carbon asset management strategy, is determined based on the initial action space vector and the feedback value obtained by solving the value function.

6. The AI-based carbon asset and carbon verification management method according to claim 5, characterized in that, The determination of penalty factors based on risk indicators in the carbon trading market includes: Obtain historical carbon price data, real-time market depth data, trading volume data, and relevant policy texts from the carbon trading market data; Price volatility is calculated based on the standard deviation of historical carbon price data within a preset time window. The liquidity index score is calculated by combining market depth and trading volume decay rate according to a preset weighting formula. Based on the application of natural language processing technology, relevant policy texts are analyzed, keywords are extracted, and policy risk levels are determined by combining preset rules or classification models. The price volatility index, liquidity index score, and policy risk level are input into a preset risk scoring function to obtain the penalty factor.

7. The AI-based carbon asset and carbon verification management method according to claim 1, characterized in that, The calculation methods for the predicted carbon emissions of each stage include: The energy allocation scheme is analyzed to obtain the energy type and energy consumption of each link; The emission factors corresponding to the energy types in each stage are obtained by matching the carbon emission factor database. For the direct emission stages in each of the aforementioned stages, the direct carbon emissions of the corresponding stages are calculated using the emission factor method. For each step involving a chemical reaction, the process carbon emissions of the corresponding step are calculated using the mass balance method.

8. The AI-based carbon asset and carbon verification management method according to claim 1, characterized in that, Also includes: Based on the predicted total carbon emissions, the predicted carbon emissions of each stage, and the energy allocation plan, a carbon emission verification report is generated according to a preset template. The carbon emission verification report is input into the verification rule engine for compliance verification. The verification rule engine is pre-set with the carbon verification policy of the jurisdiction where the target object is located and the ISO14064 standard clauses. If the compliance check fails, the abnormal data link will be located based on the anomaly identifier output by the rule engine; If the compliance verification passes, the carbon emission verification report will be electronically signed, and a carbon emission verification report with a digital signature will be output.

9. An AI-based carbon asset and carbon verification management platform, characterized in that, The AI-based carbon asset and carbon verification management platform, applicable to any one of claims 1-8, comprises: The data acquisition module acquires multi-source data of the target object, including: operational activity data of the target object, carbon trading market data, and carbon sink data; The energy efficiency analysis module is used to input the operational activity data of the target object into a preset energy efficiency analysis model to obtain energy efficiency analysis results that characterize the energy utilization efficiency of the target object; the preset energy efficiency analysis model is obtained by training a long short-term memory network or a convolutional neural network using historical energy efficiency data. The energy flow analysis module is used to perform energy flow modeling and analysis on the operational activity data of the target object based on the energy efficiency analysis results, and to obtain energy flow analysis results describing the energy flow path, conversion efficiency and loss distribution. The energy efficiency balance optimization module is used to input the energy flow analysis results into a preset energy efficiency balance optimization model to obtain an optimized energy allocation scheme and carbon emission prediction values ​​for each link based on the energy allocation scheme. The energy efficiency balance optimization model is configured to optimize energy allocation with the goal of minimizing total carbon emissions while meeting the production constraints of the target object. The target total carbon emission prediction module is used to comprehensively calculate the predicted carbon emission values ​​of each stage to obtain the target total carbon emission prediction value of the target object. The carbon asset management strategy module is used to generate a corresponding target carbon asset management strategy based on the target total carbon emission forecast, the carbon trading market data, and the carbon sink data. The step of generating a corresponding target carbon asset management strategy based on the predicted target total carbon emissions, the carbon trading market data, and the carbon sink data includes: Based on the predicted target total carbon emissions, the carbon trading market data, and the carbon sink data, a pre-defined carbon asset management strategy knowledge base is queried, and a corresponding target carbon asset management strategy is determined based on pre-defined matching rules; or, The predicted total carbon emissions, the carbon trading market data, and the carbon sink data are input into the carbon asset management strategy recommendation model to obtain the target carbon asset management strategy corresponding to the target object. The target carbon asset management strategy includes one or more of the following: carbon quota trading schemes, carbon sink project offsetting pathways, and internal emission reduction measures.

Images