Multi-level dynamic enterprise product carbon emission calculation method and device

Abstract

The invention provides a multi-level dynamic enterprise product carbon emission calculation method and device, and relates to the technical field of data processing. The method comprises the following steps: acquiring carbon emission related data of an enterprise; preprocessing the carbon emission related data of the enterprise to obtain a preprocessed data set; based on the preprocessed data set, utilizing a multi-level carbon emission factor dynamic optimization model to obtain a dynamic carbon emission factor; the total enterprise carbon emission amount is calculated based on the dynamic carbon emission factors; determining a product carbon emission factor by using a product carbon emission regression model based on the total enterprise carbon emission; the product carbon emission factor represents the unit carbon emission of the product of the enterprise. According to the method, different enterprises can be flexibly adapted, available data can be fully utilized, the dynamic change of a regional energy structure can be reflected, and an enterprise optimal carbon emission calculation scheme is formulated, so that the economical efficiency and scientificity of the enterprise carbon emission scheme are ensured, the data calculation error is effectively reduced, and the calculation accuracy is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a multi-level dynamic method, apparatus and equipment for calculating enterprise product carbon emissions. Background Technology

[0002] The National Carbon Emission Monitoring and Analysis Service Platform possesses the capability to calculate national, regional, and industry-specific carbon emission data monthly, with its overall technology reaching international leading levels. However, current research and application in the two key dimensions of regional dynamic carbon emission factors and enterprise-level carbon emissions are still lacking. Calculating regional dynamic carbon emission factors aims to accurately calculate indirect carbon emissions at the consumption end and reveal the temporal changes in the carbon emission intensity of power grids in different regions, thereby providing core data support for local governments to formulate scientific energy transition paths. Enterprise-level carbon emissions, on the other hand, quantifies, monitors, and analyzes the carbon emissions of enterprises to help them formulate emission reduction strategies and monitor their implementation. Currently, research on carbon emission measurement methods for enterprise products is particularly lacking.

[0003] For example, patent application number 202211118933.1, entitled "Optimization Method for Energy Consumption Monitoring Based on Enterprise Carbon Metering," uses a dedicated data acquisition module to sample and analyze various energy data. While this method improves the accuracy of monitoring period determination, it is highly dependent on customized hardware, complex to implement, and costly, making it difficult to widely promote in enterprises of different sizes and types. There are three main methods for calculating carbon emissions in related technologies: actual measurement, material balance, and emission coefficient methods. Actual measurement is costly, technically challenging, and has limited representativeness; material balance relies on comprehensive and accurate material data, which is difficult to obtain, and calculation errors are easily amplified; emission coefficient methods typically use static, universal emission coefficients, which cannot truly reflect the dynamic changes and actual situation of a specific region or enterprise's production process, leading to systematic biases in the calculation results.

[0004] Therefore, there is an urgent need to study a product-level carbon emission calculation method that can flexibly adapt to different enterprises, make full use of available data, and reflect the dynamic changes in regional energy structure. Summary of the Invention

[0005] This invention provides a multi-level dynamic method, apparatus, and equipment for calculating enterprise product carbon emissions, which can flexibly adapt to different enterprises, make full use of available data, and reflect the dynamic changes in regional energy structure.

[0006] In a first aspect, embodiments of the present invention provide a multi-level dynamic method for calculating enterprise product carbon emissions, including: Obtain carbon emission data from enterprises; The carbon emission-related data of enterprises are preprocessed to obtain a preprocessed dataset. Based on the preprocessed dataset, dynamic carbon emission factors are obtained using a multi-level carbon emission factor dynamic optimization model. Calculate total corporate carbon emissions based on dynamic carbon emission factors; Based on the total carbon emissions of enterprises, the carbon emission factor of products is determined using a product carbon emission regression model; the product carbon emission factor represents the unit carbon emission of an enterprise's products.

[0007] In one possible implementation of the first aspect, a dynamic carbon emission factor is obtained based on the preprocessed dataset using a multi-level carbon emission factor dynamic optimization model, including: Calculate the provincial average carbon emission factor based on the preprocessed dataset; Based on the provincial average carbon emission factor, and combined with the electricity consumption and green electricity data of each region in the preprocessed dataset, a multi-level dynamic optimization model of carbon emission factor at the city-district-county-transformer level is constructed. The dynamic carbon emission factors of various cities, districts, counties, and sub-districts at different times were calculated based on a multi-level dynamic optimization model of carbon emission factors.

[0008] In one possible implementation of the first aspect, the provincial average carbon emission factor of thermal power is calculated based on the preprocessed dataset and in conjunction with the first formula. The first formula includes:

[0009]

[0010] in, for p Average carbon dioxide emission factor of provincial power generation; for p Average carbon dioxide emission factor of provincial thermal power plants; for p Direct carbon dioxide emissions from power generation in the province; To p Saves net power output n Average carbon dioxide emission factor of provincial power generation; for n Province p Saves net power output; To p Save net export electricity k National average carbon dioxide emission factor for power generation; for k Guo Xiang p Electricity exported from the province; For regional power grid r The average carbon dioxide emission factor; For regional power grid r Towardsp Saves net power output; for p Total annual electricity consumption of the province; for p Total annual renewable energy power generation in the province; p Target province; n To p Other provinces that have reduced net electricity transmission; k To p Countries that reduce net electricity exports; r for p The regional power grid where the province is located.

[0011] In one possible implementation of the first aspect, calculating the total carbon emissions of an enterprise based on a dynamic carbon emission factor includes: Based on the enterprise's power supply attributes, the enterprise will be matched to the corresponding dynamic carbon emission factor level; Based on the enterprise's electricity consumption data and the corresponding dynamic carbon emission factor level, the indirect carbon emissions generated by the enterprise's electricity consumption are calculated. The dynamic carbon emission factor level includes districts / counties and transformer substations. For enterprises connected to public transformer substations, the dynamic carbon emission factor of the substation to which the enterprise belongs is used for calculation. For enterprises connected to dedicated transformer substations or belonging to dedicated line users, the dynamic carbon emission factor of the district / county to which the enterprise belongs is used for calculation. Obtain enterprise energy consumption data; Calculate the company's direct carbon emissions based on its energy consumption data; The total carbon emissions of an enterprise are determined by combining indirect and direct carbon emissions.

[0012] In one possible implementation of the first aspect, the product carbon emission factor is determined using a product carbon emission regression model based on the enterprise's total carbon emissions, including: A product carbon emission regression model is constructed using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables. The product carbon emission regression model introduces a physical range constraint on the product carbon emission coefficient, and takes minimizing the error between the predicted total carbon emissions of enterprises and the actual total carbon emissions of enterprises as the optimization objective. The product carbon emission regression model is solved by a gradient-based optimization algorithm, and the product carbon emission factor is output.

[0013] In one possible implementation of the first aspect, the gradient-based optimization algorithm is the Lagrange multiplier method or the sequential quadratic programming algorithm.

[0014] In one possible implementation of the first aspect, a product carbon emission regression model is constructed using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as the dependent variable, including: Using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables, a product carbon emission regression model is constructed in conjunction with the second formula. The second formula includes:

[0015] in, Y Total carbon emissions of enterprises; For the company's product output; It is the first p The unit carbon emission factor of a product type, that is, the carbon emission amount corresponding to each unit of production of this type of product; It is a collection of other factors that affect carbon emissions; It is the first O The coefficients of other influencing factors characterize the degree of their impact on carbon emissions; It is a random error term.

[0016] In one possible implementation of the first aspect, the carbon emission-related data of the enterprise is preprocessed to obtain a preprocessed dataset, including: The OneClassSVM algorithm is used to identify outliers in the carbon emission data of enterprises, remove the identified outliers, and mark them as missing values. The missing values ​​are filled using the K-nearest neighbor classification algorithm to form the preprocessed dataset.

[0017] In one possible implementation of the first aspect, identifying outliers in a company's carbon emission-related data based on the OneClassSVM algorithm includes: Construct a numerical data matrix of electricity consumption based on the company's carbon emission data; Based on the numerical data matrix of electricity, a hypersphere model is established using the OneClassSVM algorithm; Based on the hypersphere model, a nonlinear objective function is constructed; The optimized hypersphere model is obtained by optimizing the hypersphere model based on the nonlinear objective function; Calculate the projection distance from each data point in the numerical data matrix of electrical energy to the center of the hypersphere in the optimized hypersphere model; Based on distance and a set anomaly detection threshold, outliers are identified.

[0018] Secondly, embodiments of the present invention provide a multi-level dynamic enterprise product carbon emission calculation device, comprising: The data acquisition module is used to acquire carbon emission-related data from enterprises. The data preprocessing module is used to preprocess the carbon emission-related data of enterprises to obtain preprocessed datasets; The dynamic carbon emission factor calculation module is used to obtain the dynamic carbon emission factor based on the preprocessed dataset using a multi-level dynamic optimization model of carbon emission factors. The total carbon emissions calculation module is used to calculate the total carbon emissions of an enterprise based on dynamic carbon emission factors. The product carbon emission factor calculation module is used to determine the product carbon emission factor based on the total carbon emissions of an enterprise using a product carbon emission regression model; the product carbon emission factor represents the unit carbon emissions of an enterprise's products.

[0019] In this invention, a multi-level carbon metering system for enterprises is established. The multi-level system can be in the form of "city-district-regional," enabling accurate calculation of regional carbon emissions and carbon emissions from high-energy-consuming enterprises, and can be flexibly adapted to different enterprises. This invention mainly relies on existing electricity consumption information collection systems and publicly available data, eliminating the need for expensive dedicated monitoring hardware or frequent manual on-site measurements. Calculations are automatically completed through data preprocessing and algorithm models, effectively overcoming the drawbacks of high cost and difficulty in obtaining data for material balance algorithms in actual measurement methods. Utilizing a multi-level dynamic optimization model for carbon emission factors, dynamic carbon emission factor sequences with frequencies up to 15 minutes can be obtained, finely depicting the differences in electricity carbon emission intensity in different regions and time periods, reflecting the dynamic changes in regional energy structure, thereby significantly improving the spatiotemporal resolution and accuracy of indirect carbon emission accounting for enterprises.

[0020] Traditional methods typically stop at calculating a company's total carbon emissions, failing to scientifically and rationally allocate total emissions to specific products. This invention introduces a regression model with physical constraints, using the company's total carbon emissions as a benchmark to inversely solve for the unit carbon emission factor of each product. This breaks through the key technical bottleneck of transitioning from macro-level accounting at the enterprise level to micro-level traceability at the product level, providing a direct and reliable data foundation for enterprises to manage their product carbon footprint, implement green design, and address carbon barriers in international trade. Attached Figure Description

[0021] Figure 1 This is an application scenario diagram of the multi-level dynamic enterprise product carbon emission calculation method provided in the embodiments of the present invention; Figure 2 This is a flowchart illustrating the implementation of the multi-level dynamic enterprise product carbon emission calculation method provided in this embodiment of the invention. Figure 3 This is a flowchart illustrating the implementation of step 102 provided in an embodiment of the present invention; Figure 4 This is a flowchart illustrating the implementation of step 103 provided in an embodiment of the present invention; Figure 5 This is a flowchart of the multi-level carbon emission calculation process provided in the embodiments of the present invention; Figure 6This is a schematic diagram of the structure of the multi-level dynamic enterprise product carbon emission calculation device provided in the embodiment of the present invention; Figure 7 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0022] To better understand this solution, we will analyze the key pain points in the current carbon accounting field in conjunction with the above background technology.

[0023] In calculating regional dynamic carbon emission factors, there are widespread problems such as complex and varied carbon emission factors, low update frequency, and overly broad calculation scopes. This directly leads to poor accuracy and reliability of regional and enterprise-level carbon emission factors. Regarding carbon emissions from high-energy-consuming enterprises, although some domestic companies have established carbon asset management offices for carbon inventory, their own carbon inventory results often have significant deviations due to inaccurate basic data, unscientific calculation methods, frequent neglect of implicit carbon emissions, and a lack of professional personnel. Furthermore, the carbon inventory methods and data channels available in the market are relatively limited, lacking diversified auxiliary verification tools, further posing a serious challenge to the reliability of carbon inventory results.

[0024] To address the aforementioned systemic challenges, this invention provides an innovative technical solution. The embodiments of this invention will be described in detail below with reference to the accompanying drawings.

[0025] Figure 1 This diagram illustrates an application scenario of the multi-level dynamic enterprise product carbon emission calculation method provided in this embodiment of the invention. Figure 1 As shown, the calculation method operates within a multi-level dynamic enterprise product carbon emission calculation device 100. This device 100 acquires carbon emission-related data from the electricity consumption information collection system, marketing business system, and middleware platform, calculates and predicts dynamic carbon emission factors and product carbon emission factors, and sends the prediction results to the control center 200. Based on the prediction results, the control center 200 provides scientific decision support for managers and decision-makers.

[0026] See Figure 2 The document illustrates a flowchart of the implementation of the multi-level dynamic enterprise product carbon emission calculation method provided in this embodiment of the invention, detailed below: Step 101: Obtain the company's carbon emission data.

[0027] For example, carbon emission-related data includes internal and external data. Internal data is obtained through integration with and extraction from the company's existing information systems, such as existing electricity consumption data collection systems, marketing business systems, and data centers that extract user profile information, user electricity consumption readings, and power generation user readings.

[0028] User profile information includes user ID, name, capacity, metering point ID, electricity consumption category, and power supply area attributes (e.g., public or private transformer), used to identify the enterprise and its electricity consumption attributes. User electricity consumption and power generation user data include user code, comprehensive multiplier, and electricity consumption values ​​at each point. These user and power generation user data are combined to form high-frequency time-series data of power consumption and generation: acquiring daily electricity consumption and power generation values ​​at 96 points (15-minute intervals) for enterprise users and distributed generation users. For cases where hourly data cannot be directly obtained, calculations are performed using forward / reverse active power indicator codes and power factors to form continuous, high-time-resolution energy consumption profile data.

[0029] The formula for calculating hourly power consumption / generation is as follows:

[0030] in, Indicates user number t The power generation reading at that time For the first t Current active power readings in both forward and reverse directions. For the first t +1 indicates positive / reverse active power values; The power factor.

[0031] External data refers to relevant enterprise data, which can include external environmental and operational data related to enterprise carbon emission accounting collected through various channels, such as regional power grid and energy data, enterprise operation and product data, and policy and market data. Regional power grid and energy data includes collected data on total electricity consumption, power generation structure, renewable energy (green electricity) generation, net inter-provincial / inter-regional electricity imports, and their emission factors for the target region (e.g., Hebei Southern Power Grid), providing a foundation for constructing dynamic regional carbon emission factors. Enterprise operation and product data includes data on key high-energy-consuming enterprises, such as monthly product types, detailed output, types and quantities of fossil fuels consumed, and carbon emission allowances held, used to support enterprise-level total emission accounting and product-level carbon emission allocation. Policy and market data includes data obtained from official platforms of government departments, international organizations, and exchanges, tracking the latest carbon trading policies, carbon tariff rules (such as the EU CBAM mechanism), and official carbon emission factors for various energy sources and materials, ensuring that the accounting methods comply with the latest regulations and market requirements.

[0032] This step involves the systematic acquisition and alignment of internal and external data to build a complete dataset covering the macroeconomic policy environment, the mesoeconomic regional power grid, and the microeconomic enterprise operations and product information, providing reliable data input for subsequent accurate calculations.

[0033] Step 102: Preprocess the carbon emission-related data of the enterprise to obtain the preprocessed dataset.

[0034] For example, outliers in a company’s carbon emission data can be identified, removed, and marked as missing values. The missing values ​​can then be filled in to obtain a preprocessed dataset.

[0035] Step 103: Based on the preprocessed dataset, obtain the dynamic carbon emission factor using a multi-level carbon emission factor dynamic optimization model.

[0036] For example, this step aims to calculate dynamic carbon emission factors with high spatiotemporal resolution. By constructing a multi-level factor optimization model at the city-district-transformer level and deducting regional green electricity consumption, a dynamic factor sequence that can reflect the grid carbon intensity in different regions and time periods in real time is generated, thereby improving the spatiotemporal resolution and accuracy of carbon emission measurement and achieving a clear division of carbon emission responsibility for traded electricity.

[0037] Step 104: Calculate the total carbon emissions of the enterprise based on the dynamic carbon emission factor.

[0038] For example, this step applies dynamic carbon emission factors to specific enterprises. By matching the enterprise's power supply attributes (public transformer / dedicated transformer) to select the corresponding level of dynamic factors, and combining the enterprise's electricity consumption data and fossil energy consumption data, indirect and direct carbon emissions are calculated separately, and the total carbon emissions of the enterprise are obtained after summarizing.

[0039] Step 105: Based on the total carbon emissions of the enterprise, determine the carbon emission factor of the product using the product carbon emission regression model.

[0040] Among them, the product carbon emission factor represents the carbon emission per unit of a company's products.

[0041] For example, this step aims to scientifically allocate a company's total carbon emissions to specific products. By constructing a constrained regression model with product output as the independent variable and total corporate carbon emissions as the dependent variable, the carbon emission factor per unit output of various products is solved, thereby achieving accurate traceability of product carbon footprint.

[0042] By following the steps above, the carbon emissions of a company's products can be accurately calculated, and the optimal carbon emission calculation plan can be formulated accordingly. This ensures the economic efficiency and scientific validity of the company's carbon emission plan, effectively reduces data calculation errors, and improves calculation accuracy.

[0043] This embodiment establishes a multi-level carbon metering system for enterprises, where the multi-level system can be in the form of "city-district-regional area," enabling accurate calculation of regional carbon emissions and carbon emissions of high-energy-consuming enterprises, and can be flexibly adapted to different enterprises. This invention mainly relies on existing electricity consumption information collection systems and publicly available data, without relying on expensive dedicated monitoring hardware or frequent manual on-site measurements. Through data preprocessing and automatic calculation using algorithm models, it effectively overcomes the shortcomings of high cost of actual measurement methods and difficulty in obtaining data for material balance algorithms. Utilizing a multi-level dynamic optimization model of carbon emission factors, it can obtain dynamic carbon emission factor sequences with a frequency of up to 15 minutes, finely characterizing the differences in electricity carbon emission intensity in different regions and time periods, reflecting the dynamic changes in regional energy structure, thereby significantly improving the spatiotemporal resolution and accuracy of enterprise indirect carbon emission accounting.

[0044] In one embodiment, a specific preprocessing method is described. The OneClassSVM algorithm is used to identify and detect outliers. The threshold for judging outliers in OneClassSVM is adjusted according to business logic rules, thereby achieving anomaly identification in numerical electricity data. A K-nearest neighbor (KNN) classification algorithm is used to process missing values ​​in the electricity data. The number of neighboring samples used to fill in missing values ​​in the KNN algorithm is specified. First, outliers are identified and removed, and these values ​​are marked as gaps to be filled. Different processing strategies are applied to fill the gaps based on their distribution. See [link to documentation]. Figure 3 Step 102 includes: Step 1021: Based on the OneClassSVM algorithm, identify outliers in the enterprise's carbon emission-related data, remove the identified outliers, and mark them as missing values.

[0045] Specifically, a numerical data matrix of electricity consumption is constructed based on the carbon emission data of enterprises.

[0046] Based on the numerical data matrix of electricity consumption, a hypersphere model is constructed using the OneClassSVM algorithm. This hypersphere model attempts to find a minimal hypersphere that can contain normal data using the OneClassSVM algorithm.

[0047] Based on the hypersphere model, a nonlinear objective function is constructed. A kernel function (Gaussian kernel) is added to the nonlinear objective function to handle the nonlinear problem. The nonlinear objective function can be expressed as:

[0048] The following constraints must be met simultaneously:

[0049] in, This represents the mapping of the normal vectors of the hypersphere in the feature space. The intercept of the hypersphere defines the position of the boundary of the normal data region, and its value determines the size of the hypersphere. As slack variables, they are for the first... i The non-negative variables introduced for each sample are used to compensate for data points that cannot strictly meet the constraints. The pre-defined anomaly rate parameter controls the proportion of training data that becomes support vectors, and its value ranges from (0,1]. It is a very small positive value, used to allow some normal points to be at a certain distance from the boundary of the hypersphere.

[0050] The hypersphere model is optimized based on a nonlinear objective function to obtain the optimized hypersphere model. This involves applying the objective function and constraints defined in the previous step, and training a OneClassSVM model using an optimization algorithm (such as the SMO algorithm, which decomposes the original large optimization problem into multiple smaller quadratic programming subproblems and then solves these subproblems alternately to approximate the global optimum). During training, the model automatically finds the optimal normal vector. w ,intercept ρ and slack variables ζ This ensures that the objective function is minimized while also satisfying the constraints.

[0051] Calculate the distance from each data point in the numerical data matrix of electrical quantities to the center of the hypersphere in the optimized hypersphere model. The negative of this distance is the anomaly score, which is then used to identify outliers.

[0052] After training, the sample points are calculated in the OneClassSVM model. Projected distance to the center of the hypersphere constructed by the model and the corresponding abnormal scores The mathematical representation of .

[0053] Projection distance Calculation formula:

[0054] in, It is the intercept of the hypersphere obtained through training. It defines the normal vector of the hypersphere in the feature space. Sample points The vector mapped to a high-dimensional feature space.

[0055] Abnormal scores The calculation formula is as follows:

[0056] Outliers are identified based on distance and a set anomaly threshold. Specifically, a suitable anomaly threshold is determined using statistical methods such as cross-validation and ROC curve analysis, based on business requirements and data distribution characteristics. For a single data point... x i If the corresponding abnormal score s i Greater than the set threshold t ,Right now s i > t If a point is outside the normal data area, it is considered an outlier; otherwise, it is considered normal data. The set of outliers is represented as follows: .

[0057] Furthermore, by employing Lagrange duality to solve the problem, new data points can be identified. z Whether it is included, if z The distance to the center is less than or equal to the radius. r If the value is within a certain range, it is not an outlier; if it is outside the hypersphere, it is an outlier. All marked outliers will be removed in subsequent data preprocessing steps and marked as missing values.

[0058] When processing missing values ​​in the electricity data, a K-nearest neighbor (KNN) classification algorithm was used. This method first identifies and removes outliers, marking them as gaps to be filled. Then, different processing strategies are adopted based on the distribution of the missing values: when the number of missing values ​​is large, related records are deleted; when the number of missing values ​​is small, the average electricity consumption of enterprises of similar size, during the same period, and in the same region is used to fill the gaps. Step 1022 involves filling the gaps using the KNN classification algorithm, resulting in the preprocessed dataset.

[0059] Furthermore, to more accurately determine the K value, i.e., the number of neighboring samples used to fill in missing values ​​in the KNN algorithm, this method introduces the concept of a business rule threshold. This threshold is set based on the correlation of user electricity consumption data within the same industry, of similar scale, and under the same spatial and temporal context.

[0060] Specifically, determine the K-value of the company's carbon emission-related data.

[0061]

[0062] in, Indicates the location of the missing value. Indicates and Data points in similar locations, Similarity The function is used to evaluate these data points and Numerical similarity. Select the step size with the highest similarity. NThis is used as the K value in the KNN algorithm.

[0063] Calculate the Euclidean distance between each missing value and the non-missing values ​​in the firm's carbon emission data. For each missing value... Calculate its relationship with other data points in the dataset. Euclidean distance :

[0064] Determine and missing values The set of K data points with the smallest Euclidean distance :

[0065] Calculate imputation values ​​based on a set of K data points and the value of K. For each missing value... According to its nearest neighbor set Statistical properties calculation of fill value For example, if the mean is used as the basis for filling, then:

[0066] The fill values ​​are used to fill in the missing values ​​in the carbon emission data of the enterprises. Then the calculated fill values ​​are... Fill in the missing values ​​from the original dataset The missing values ​​are filled in at the specified location to obtain the preprocessed dataset.

[0067] This embodiment employs the OneClassSVM algorithm and dynamically adjusts the judgment threshold based on business rules, overcoming the poor adaptability of traditional fixed threshold methods. This method effectively identifies hidden anomalies caused by equipment failures, meter reading errors, or atypical electricity consumption behavior, ensuring the authenticity and consistency of basic electricity data and reducing calculation errors at the source. The K-nearest neighbor algorithm is used to handle gaps resulting from outlier removal, employing a differentiated filling strategy based on the distribution of the gaps. This method utilizes the similarity of the data itself for filling, which, compared to simple mean or median filling, better preserves the original statistical characteristics and structure of the dataset, providing more continuous and reliable data input for subsequent models.

[0068] In one embodiment, considering that the current carbon emission factor is single and fixed, and does not consider the impact of green electricity on the carbon emission factor, while the actual carbon emission factor should dynamically change with time and space, this invention addresses the problems of large calculation deviations for fixed carbon emission factors and the continued investment required for carbon meter installation. It constructs a multi-level dynamic optimization model of carbon emission factors at the "city-district-region" level, combining green electricity consumption deductions. Utilizing the dynamically calculated regional thermal power average carbon emission factor, it achieves high spatiotemporal resolution carbon emission calculations every 15 minutes, improving the spatiotemporal resolution and accuracy of carbon emission metering, and clearly defining the responsibility for carbon emissions from traded electricity. See also... Figure 4 and Figure 5 Step 103 includes: Step 1031: Calculate the provincial average carbon emission factor based on the preprocessed dataset.

[0069] Based on the preprocessed dataset, the provincial average carbon emission factor is calculated using the first formula.

[0070] The first formula includes:

[0071]

[0072] in, for p The provincial average carbon dioxide emission factor for electricity, expressed in kgCO2 / kWh; for p Average carbon dioxide emission factor of provincial thermal power plants, in kgCO2 / kWh; for p The direct carbon dioxide emissions from power generation in the province, expressed in tCO2; To p Saves net power output n The provincial average carbon dioxide emission factor for electricity, expressed in kgCO2 / kWh; for n The net electricity transmitted from province p to province p, in MWh; To p Save net export electricity k The national average carbon dioxide emission factor for power generation, expressed in kgCO2 / kWh; for k Guo Xiang p The province's net electricity exports, in MWh; For regional power grid r The average carbon dioxide emission factor, expressed in kgCO2 / kWh; For regional power grid r Towards pThe net electricity delivered by the province is expressed in MWh. for p Total annual electricity consumption of the province, in MWh; for p Total annual power generation of the province, in MWh; R To determine the regional renewable energy penetration rate, then for p The province's total annual renewable energy power generation, in MWh; p Target province; n To p Other provinces that have reduced net electricity transmission; k To p Countries that reduce net electricity exports; r for p The regional power grid where the province is located.

[0073] This step, based on preprocessed provincial energy data, deducts non-fossil energy power generation and accurately calculates the ratio of total carbon emissions from provincial thermal power generation and net imported thermal power to thermal power supply, thereby obtaining a provincial average factor characterizing carbon emissions per unit of thermal power generation. This factor isolates the influence of green energy and provides a unified benchmark for thermal power carbon emission intensity for lower-level regions, ensuring consistency at the source in the multi-layered calculation system.

[0074] Step 1032: Based on the provincial average carbon emission factor, and combined with the electricity consumption and green electricity data of each region in the preprocessed dataset, construct a multi-level dynamic optimization model of carbon emission factors at the city-district-county-transformer level.

[0075] Step 1032, following Step 1031, involves calculating regional-level carbon emission factors. The construction of a multi-level dynamic optimization model for carbon emission factors at the city / district / county / regional level is based on the provincial average carbon emission factor. This model uses data on electricity consumption, purchased electricity, electricity generation, and distributed photovoltaic power generation in the target province's regional power grid to calculate the proportion of green electricity in total electricity consumption for each region, and to calculate the dynamic regional (province, city, district, regional) average carbon emission factor.

[0076] The average carbon emission factor for prefecture-level cities is expressed as follows:

[0077] The average carbon emission factor for districts and counties is expressed as follows:

[0078] The average carbon emission factor for the distribution area is expressed as:

[0079] in, for mAverage carbon dioxide emission factor of electricity in prefecture-level cities, in kgCO2 / kWh; for m Total annual electricity consumption of prefecture-level cities, in kWh; for m Total annual renewable energy power generation in prefecture-level cities, in kWh; for d Average carbon dioxide emission factor of electricity in districts and counties, in kgCO2 / kWh; for d Total annual electricity consumption of districts and counties, in kWh; for d Total annual renewable energy power generation in districts and counties, in kWh; for t Average carbon dioxide emission factor of electricity in the power distribution area, in kgCO2 / kWh; for t Total annual electricity consumption of the distribution area, in kWh; for t Total annual renewable energy power generation in the substation area, in kWh; for p Average carbon dioxide emission factor of provincial thermal power plants, in kgCO2 / kWh; m for p Provinces and cities; d for p Provinces, districts, and counties; t for p Taiwan Province.

[0080] This step uses the provincial average carbon emission factor of thermal power as a benchmark, and combines it with the progressively detailed electricity consumption and distributed green power generation data of various cities, counties, and transformer substations from the preprocessed data to construct a top-down, multi-level optimization model. The core of the model is to use the provincial thermal power factor as the basic carbon intensity, and dynamically adjust it according to the proportion of green power consumption in each region, thus forming a carbon emission factor calculation framework that can be progressively transmitted downwards and reflects the differences in green power consumption between regions. This model achieves spatial decoupling and dynamic correlation from macro-level provincial factors to micro-level transformer substation factors.

[0081] Step 1033: Calculate the dynamic carbon emission factors of various cities, districts, counties and sub-districts at different times based on the multi-level carbon emission factor dynamic optimization model.

[0082] Based on the aforementioned multi-level dynamic optimization model for carbon emission factors, real-time or near-real-time electricity consumption and green energy data for each region are input. The dynamic carbon emission factors for cities, districts, and substations are calculated level by level and time point by time (e.g., every 15 minutes). The calculation process fully considers the spatiotemporal fluctuations in green energy penetration rates at each level of region within a specific time period, ultimately outputting a set of regional carbon emission factor sequences with high spatiotemporal resolution. These dynamic carbon emissions can accurately characterize the differences in carbon emission intensity caused by changes in electricity consumption structure and intermittent green energy output in different regions and at different times, providing crucial input for subsequent refined accounting of enterprise-level carbon emissions.

[0083] In one embodiment, the process of calculating the total carbon emissions of an enterprise based on the dynamic carbon emission factor obtained in step 103 is specifically described. Step 104 then includes: First, based on the enterprise's power supply attributes, the enterprise is matched to the corresponding dynamic carbon emission factor level. Based on the enterprise's electricity consumption data and the corresponding dynamic carbon emission factor level, the indirect carbon emissions generated by the enterprise's electricity consumption are calculated. Specifically, for enterprises connected to public transformer substations, the dynamic carbon emission factor of the substation to which the enterprise belongs is used for calculation; for enterprises connected to dedicated transformer substations or belonging to dedicated line users, the dynamic carbon emission factor of the district / county to which the enterprise belongs is used for calculation.

[0084] It should be noted that the dynamic carbon emission factor hierarchy includes provincial, municipal, district / county, and substation levels. The provincial dynamic carbon emission factor, i.e., the provincial average carbon emission factor for thermal power, is the physical benchmark for the entire calculation system. It deducts all renewable energy generation within the province and represents the average carbon emissions generated per kilowatt-hour of thermal power generated in that province. Because the distribution of green electricity is extremely uneven, and power generation data is usually only complete and reliable at the provincial and municipal levels, the provincial factor provides a stable and authoritative starting point for carbon concentration. The dynamic factors for all lower-level regions (municipalities, districts / counties, and substations) are derived from this provincial thermal power benchmark after deductions and dilution based on the local green electricity consumption ratio. This ensures the uniformity and comparability of the carbon emission accounting benchmark across the province. The municipal dynamic carbon emission factor is a core calculation layer and optimization node in the model. The calculation of the district / county dynamic factor is based on the decomposition of total electricity consumption and green electricity data at the municipal level. The accurate calculation of the municipal factor is a prerequisite for ensuring the accuracy of the factors for its subordinate districts / counties.

[0085] Because the dynamic carbon emission factors at the provincial and municipal levels are averages over a large area, they cannot depict the significant differences in grid carbon intensity between different cities within a province, or even between different areas within the same city (such as industrial zones and residential areas, and the presence or absence of distributed photovoltaic power). Therefore, the core function of provincial and municipal factors is to produce high-precision district and substation-level factors for final enterprise accounting.

[0086] For example, to query whether a company belongs to a public transformer substation or a private transformer substation, the calculation is based on the substation's dynamic average carbon emission factor:

[0087] For users belonging to dedicated transformer substations or dedicated line users, the carbon emission factor is calculated based on the district / county dynamic average carbon emission factor:

[0088] in, for c Total carbon emissions from electricity consumption by an enterprise, which is the indirect carbon emissions generated by the enterprise's electricity consumption, is expressed in kgCO2. for i time t Average carbon dioxide emission factor of electricity in the power distribution area, in kgCO2 / kWh; for i time d Average carbon dioxide emission factor of electricity in districts and counties, in kgCO2 / kWh; for i time c Total electricity consumption of the enterprise, in kWh.

[0089] Then, obtain the company's energy consumption data; based on the company's energy consumption data, calculate the company's direct carbon emissions; and combine the indirect carbon emissions with the direct carbon emissions to determine the company's total carbon emissions.

[0090] For example, a company's energy consumption data may include the consumption of fossil fuels, the carbon emission factor of fossil fuels, etc. Of course, the company's energy consumption data has generally been obtained in advance in step 101, but for the sake of completeness, the company's energy consumption data is listed here again.

[0091] Among them, total corporate carbon emissions = carbon emissions from energy use by the enterprise + carbon emissions from electricity use by the enterprise.

[0092] +

[0093] in, for c Total carbon emissions of the enterprise, expressed in kgCO2; fossil fuels m Consumption volume; fossil fuels m The carbon emission factor is expressed in kgCO2 / kWh. for c Total carbon emissions from electricity generated by the enterprise, expressed in kgCO2.

[0094] This embodiment addresses the issue that traditional methods typically use a uniform average factor across all enterprises, neglecting the fundamental differences in grid connection points and actual electricity consumption structures between public transformer users (shared distribution areas) and dedicated transformer / line users (independent access). Instead, it matches a dynamic factor (distribution area level or district / county level) that best reflects the actual carbon intensity of each enterprise's electricity consumption, based on their actual power supply attributes. This eliminates systematic biases in the calculation mechanism, making the results more realistic and reliable. Furthermore, for numerous SMEs connected to public transformer distribution areas, the highest-resolution distribution area-level dynamic factor captures the micro-fluctuations in carbon emission intensity caused by green energy consumption and load changes within their local distribution network, significantly improving calculation accuracy. For dedicated transformer / line users with high electricity consumption and direct access to higher voltage levels, a relatively stable yet still dynamic district / county level factor is used. This ensures calculation feasibility while reflecting changes in the overall energy structure of their region, balancing accuracy and practicality. This invention calculates and integrates indirect carbon emissions from electricity consumption and direct carbon emissions from fossil fuel consumption to generate a company's total carbon emissions, thus constructing a complete and transparent corporate carbon footprint inventory. This follows internationally accepted accounting standards, resulting in a well-structured and source-identifiable carbon inventory. This not only meets compliance reporting requirements but also provides companies with a clear diagnostic basis for identifying key emission reduction priorities.

[0095] In one embodiment, considering the principle of minimizing the deviation between the product carbon emission factor and the official energy emission factor, a range constraint relationship for the product carbon emission coefficient is introduced, innovatively constructing a product carbon emission coefficient constraint to achieve accurate assessment of the product carbon emission factor. Step 105 includes: First, a product carbon emission regression model is constructed using the output data of various products of the enterprise as the independent variable and the total carbon emissions of the enterprise as the dependent variable.

[0096] For example, data on various types of products and their output from companies in the same industry are collected, and the goods are classified and organized according to the commodity list corresponding to the EU Carbon Border Adjustment Mechanism Comprehensive Tariff Catalogue (CN) code. The output of each type of product under each company's product is used as the independent variable, and the company's carbon emissions are used as the dependent variable to establish a constraint coefficient regression model to obtain the carbon emission coefficient of each type of product.

[0097] We obtain the company's carbon emissions and product output, and use a constrained regression model to establish the product carbon emission coefficient. The dependent variable is the company's total carbon emissions (carbon_emission), and the independent variable is the company's product output (product_output).

[0098] Constrained regression models can take into account the relationship between carbon emissions and product output, and constrain the model to ensure that the prediction results meet specific limitations. Let the total carbon emissions of the enterprise be... The company's product output is Other factors affecting carbon emissions are: ,here Index representing product type, An index indicating other influencing factors.

[0099] Using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables, a product carbon emission regression model is constructed by combining the second formula.

[0100] The second formula includes:

[0101] in, Y Total carbon emissions of enterprises; For the company's product output; It is the first P The unit carbon emission factor of a product type, that is, the carbon emission amount corresponding to each unit of production of this type of product; It is a collection of other factors that affect carbon emissions; It is the first O The coefficients of other influencing factors characterize the degree of their impact on carbon emissions; It is a random error term that follows a certain probability distribution, usually assumed to be an independent and identically distributed normal random variable.

[0102] Then, the physical range constraint of the product carbon emission coefficient is introduced into the product carbon emission regression model, and the optimization objective is to minimize the error between the predicted total carbon emission of enterprises and the actual total carbon emission of enterprises. The product carbon emission regression model is solved by a gradient-based optimization algorithm, and the product carbon emission factor is output.

[0103] Regarding the carbon emission coefficient of the product The following constraints are set (based on the carbon emission coefficient reference range provided in official documents or other theoretical and practical experience, reasonable boundary constraints are set for the product's carbon emission coefficient):

[0104] in, and The first The lower and upper limits of the carbon emission coefficient for this type of product are set according to official documents or other normative documents.

[0105] For example, by employing an appropriate optimization algorithm, the parameters of the constrained regression model can be solved while satisfying the constraints. α and βGradient-based optimization algorithms, such as the Lagrangian Multiplier Method (LMM) or Sequential Quadratic Programming (SQP), can handle nonlinear regression models with complex constraints.

[0106] Introducing Lagrange multipliers and Construct the Lagrange function:

[0107] The objective becomes finding the minimum value of the Lagrange function, which requires solving for the following: Find the partial derivatives and set them equal to 0 to obtain the gradient equations:

[0108]

[0109]

[0110]

[0111] Then, the above system of equations can be solved iteratively using gradient descent, Newton's method, or other optimization algorithms. The parameter values ​​are updated in each iteration.

[0112]

[0113]

[0114]

[0115] in, η It's the learning rate. t It represents the number of iterations.

[0116] The final solution involves finding the optimal parameter values ​​and estimating the carbon emission coefficient. The solution process requires iterative steps until stopping criteria are met, such as the gradient approaching zero, the objective function value changing sufficiently, or reaching the maximum number of iterations. The final result is... α This is the carbon emission coefficient of the product that we want to estimate.

[0117] This embodiment constructs a regression model targeting a company's total carbon emissions. Requiring only two relatively readily available data points—total emissions and product output—it can inversely calculate the average carbon emission factor for various products, significantly reducing the accounting threshold and cost. This is particularly suitable for SMEs with complex processes and weak data foundations. Industry common sense, official guidelines, or theoretical extreme values ​​are embedded as upper and lower limits for the product carbon emission coefficient in the regression process. This key design ensures that the calculated coefficients fall within a reasonable physical range, avoiding mathematically feasible but practically absurd results (such as negative carbon emission coefficients), making the model output both statistically optimal and engineering credible. The final output, the "product carbon emission factor," directly quantifies the carbon emissions implied by producing one unit of product. This enables companies to accurately compare the carbon efficiency of different products, identify high-carbon emission product lines, and thus optimize processes, adjust product structures, or implement green design in a targeted manner.

[0118] The multi-level dynamic enterprise product carbon emission calculation method proposed in this application establishes a multi-level carbon metering system covering regions, distribution areas, and high-energy-consuming enterprises. This system has advantages such as more accurate spatiotemporal characterization, low implementation cost, and low technical difficulty, and can achieve accurate measurement of regional and enterprise carbon emissions.

[0119] In terms of time precision, traditional methods typically use static emission factors calculated on an annual basis, which cannot reflect time-period differences. In contrast, this scheme relies on high-frequency sampling data (such as every 15 minutes) to precisely track fluctuations in electricity consumption and generation at different times, enabling the carbon emission factor to dynamically align with actual power dispatch and consumption, thereby significantly improving the timeliness and accuracy of the calculation.

[0120] In terms of spatial precision, traditional methods often only focus on overall emissions across macro-regions. This approach, however, refines the calculation layer by layer to the city, county, and transformer substation levels based on grid topology and data availability, independently calculating the carbon emission factors for each micro-region. This accurately reflects the differences in energy structure and green electricity consumption across different regions, providing reliable data support for differentiated and precise energy conservation and carbon reduction decisions.

[0121] In summary, this scheme achieves a refined description of carbon emissions from macro to micro and from static to dynamic by constructing a carbon measurement system that combines a "multi-level spatial architecture" with "high-frequency dynamic updates." While improving the accuracy of accounting, it maintains a low implementation threshold and has good applicability for promotion.

[0122] Taking a certain manufacturing enterprise A in a certain province as an example, data such as the enterprise's product output, electricity consumption, electricity consumption and green electricity generation in the district / county where enterprise A is located in 2024 are collected. Given the product output data in 2024, the output of products C, D and E are 56.6749, 38.4662 and 65,900 units respectively.

[0123] Based on the formula for calculating the regional dynamic carbon emission factor that takes into account green electricity, the dynamic carbon emission factors of 96 points in B district and county in 2024 were obtained. Based on the formula for calculating corporate carbon emissions based on the dynamic carbon emission factor, the indirect carbon emissions of enterprises based on the dynamic carbon emission factor were calculated to be 8913.89 tons.

[0124] Based on the enterprise's electricity consumption data and the carbon emission calculation formula, the carbon emission of product C can be calculated, and the monthly unit carbon emission of product C is shown in Table 1. Using monthly output as the independent variable and monthly carbon emission as the dependent variable, a regression equation is established to obtain the annual carbon emission factor of the instrument transformer as 0.4103 tons (CO2) / 10,000 units.

[0125] Table 1 Monthly carbon emissions per unit of product C

[0126] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0127] The following are device embodiments of the present invention. For details not described in detail, please refer to the corresponding method embodiments described above.

[0128] Figure 6 The diagram shows a multi-level dynamic enterprise product carbon emission calculation device according to an embodiment of the present invention. For ease of explanation, only the parts related to the embodiment of the present invention are shown, and are described in detail below: like Figure 6 As shown, a multi-level dynamic enterprise product carbon emission calculation device 4 includes: a data acquisition module 401, a data preprocessing module 402, a dynamic carbon emission factor calculation module 403, a total carbon emission calculation module 404, and a product carbon emission factor calculation module 405.

[0129] Data acquisition module 401 is used to acquire carbon emission-related data of enterprises; Data preprocessing module 402 is used to preprocess the carbon emission-related data of enterprises to obtain preprocessed datasets; The dynamic carbon emission factor calculation module 403 is used to obtain the dynamic carbon emission factor based on the preprocessed dataset using a multi-level dynamic optimization model of carbon emission factors. The total carbon emissions calculation module 404 is used to calculate the total carbon emissions of an enterprise based on dynamic carbon emission factors. The product carbon emission factor calculation module 405 is used to determine the product carbon emission factor based on the total carbon emissions of an enterprise using a product carbon emission regression model; the product carbon emission factor represents the unit carbon emissions of an enterprise's products.

[0130] For example, the dynamic carbon emission factor calculation module 403 is specifically used for: Calculate the provincial average carbon emission factor based on the preprocessed dataset; Based on the provincial average carbon emission factor, and combined with the electricity consumption and green electricity data of each region in the preprocessed dataset, a multi-level dynamic optimization model of carbon emission factor at the city-district-county-transformer level is constructed. The dynamic carbon emission factors of various cities, districts, counties, and sub-districts at different times were calculated based on a multi-level dynamic optimization model of carbon emission factors.

[0131] For example, the provincial average carbon emission factor is calculated based on the preprocessed dataset and the first formula. The first formula includes:

[0132]

[0133] in, for p Average carbon dioxide emission factor of provincial power generation; for p Average carbon dioxide emission factor of provincial thermal power plants; for p Direct carbon dioxide emissions from power generation in the province; To p Saves net power output n Average carbon dioxide emission factor of provincial power generation; for n Province p Saves net power output; To p The average carbon dioxide emission factor of power generation in country K based on the province's net export electricity. for k Guo Xiang p Electricity exported from the province; For regional power grid r The average carbon dioxide emission factor; For regional power grid r Towards p Saves net power output; for p Total annual electricity consumption of the province; for p Total annual power generation of the province; for pTotal annual renewable energy power generation in the province; p Target province; n To p Other provinces that have reduced net electricity transmission; k To p Countries that reduce net electricity exports; r for p The regional power grid where the province is located.

[0134] For example, the total carbon emissions calculation module 404 is specifically used for: Based on the enterprise's power supply attributes, the enterprise will be matched to the corresponding dynamic carbon emission factor level; Based on the enterprise's electricity consumption data and the corresponding dynamic carbon emission factor level, the indirect carbon emissions generated by the enterprise's electricity consumption are calculated. The dynamic carbon emission factor level includes districts / counties and transformer substations. For enterprises connected to public transformer substations, the dynamic carbon emission factor of the substation to which the enterprise belongs is used for calculation. For enterprises connected to dedicated transformer substations or belonging to dedicated line users, the dynamic carbon emission factor of the district / county to which the enterprise belongs is used for calculation. Obtain enterprise energy consumption data; Calculate the company's direct carbon emissions based on its energy consumption data; The total carbon emissions of an enterprise are determined by combining indirect and direct carbon emissions.

[0135] For example, the product carbon emission factor calculation module 405 is specifically used for: A product carbon emission regression model is constructed using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables. The product carbon emission regression model introduces a physical range constraint on the product carbon emission coefficient, and takes minimizing the error between the predicted total carbon emissions of enterprises and the actual total carbon emissions of enterprises as the optimization objective. The product carbon emission regression model is solved by a gradient-based optimization algorithm, and the product carbon emission factor is output.

[0136] For example, gradient-based optimization algorithms include the Lagrange multiplier method or sequential quadratic programming.

[0137] For example, using the output data of various products of an enterprise as independent variables and the total carbon emissions of the enterprise as the dependent variable, a product carbon emission regression model is constructed, including: Using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables, a product carbon emission regression model is constructed in conjunction with the second formula. The second formula includes:

[0138] in, Y Total carbon emissions of enterprises; For the company's product output; It is the first p The unit carbon emission factor of a product type, that is, the carbon emission amount corresponding to each unit of production of this type of product; It is a collection of other factors that affect carbon emissions; It is the first O The coefficients of other influencing factors characterize the degree of their impact on carbon emissions; It is a random error term.

[0139] For example, the data preprocessing module 402 is specifically used for: The OneClassSVM algorithm is used to identify outliers in the carbon emission data of enterprises, remove the identified outliers, and mark them as missing values. The missing values ​​are filled using the K-nearest neighbor classification algorithm to form the preprocessed dataset.

[0140] For example, the data preprocessing module 402 is further configured to: Construct a numerical data matrix of electricity consumption based on the company's carbon emission data; Based on the numerical data matrix of electricity, a hypersphere model is established using the OneClassSVM algorithm; Based on the hypersphere model, a nonlinear objective function is constructed; The optimized hypersphere model is obtained by optimizing the hypersphere model based on the nonlinear objective function; Calculate the projection distance from each data point in the numerical data matrix of electrical energy to the center of the hypersphere in the optimized hypersphere model; Based on distance and a set anomaly detection threshold, outliers are identified.

[0141] The beneficial effects of this embodiment of a multi-level dynamic enterprise product carbon emission calculation device are described in the section on the beneficial effects of a multi-level dynamic enterprise product carbon emission calculation method.

[0142] Figure 7 This is a schematic diagram of an electronic device provided in an embodiment of the present invention. For example... Figure 7 As shown, the electronic device 5 of this embodiment includes a processor 50 and a memory 51. The memory 51 stores a computer program 52. When the processor 50 executes the computer program 52, it implements the steps in the various method embodiments described above. Alternatively, when the processor 50 executes the computer program 52, it implements the functions of each module / unit in the various device embodiments described above.

[0143] For example, computer program 52 may be divided into one or more modules / units, which are stored in memory 51 and executed by processor 50 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 52 in electronic device 5.

[0144] Electronic device 5 may include, but is not limited to, processor 50 and memory 51. Those skilled in the art will understand that... Figure 7 This is merely an example of electronic device 5 and does not constitute a limitation on electronic device 5. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device 5 may also include input / output devices, network access devices, buses, etc.

[0145] For the sake of simplicity and clarity, only the above-described functional modules / units are used as examples. In practical applications, the functions described above can be assigned to different functional modules / units as needed. These modules / units can be implemented in hardware, software, or a combination of both.

[0146] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not detailed or described in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Unless otherwise specified or in conflict with logic, the terminology and / or descriptions between different embodiments are consistent and can be referenced interchangeably. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships.

[0147] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A multi-level dynamic method for calculating enterprise product carbon emissions, characterized in that, include: Obtain carbon emission data from enterprises; The carbon emission-related data of the aforementioned enterprises are preprocessed to obtain a preprocessed dataset. Based on the preprocessed dataset, a dynamic carbon emission factor is obtained using a multi-level carbon emission factor dynamic optimization model. Calculate the total carbon emissions of the enterprise based on the aforementioned dynamic carbon emission factors; Based on the total carbon emissions of the enterprise, the carbon emission factor of the product is determined using a product carbon emission regression model. The product carbon emission factor represents the carbon emission per unit of the company's products.

2. The multi-level dynamic enterprise product carbon emission calculation method according to claim 1, characterized in that, The dynamic carbon emission factor is obtained based on the preprocessed dataset using a multi-level dynamic optimization model, including: Based on the preprocessed dataset, calculate the provincial average carbon emission factor for thermal power. Based on the provincial average carbon emission factor, and combined with the electricity consumption and green electricity data of each region in the preprocessed dataset, a multi-level dynamic optimization model of carbon emission factors at the prefecture-city, district / county, and transformer substation levels is constructed. The dynamic carbon emission factors of various cities, districts, counties, and sub-districts at different times are calculated based on the multi-level carbon emission factor dynamic optimization model.

3. The multi-level dynamic enterprise product carbon emission calculation method according to claim 2, characterized in that, The step of calculating the provincial average carbon emission factor based on the preprocessed dataset includes: Based on the preprocessed dataset, the provincial average carbon emission factor of thermal power is calculated using the first formula. The first formula includes: in, for p Average carbon dioxide emission factor of provincial power generation; for p Average carbon dioxide emission factor of provincial thermal power plants; for p Direct carbon dioxide emissions from power generation in the province; To p Saves net power output n Average carbon dioxide emission factor of provincial power generation; for n Province p Saves net power output; To p Save net export electricity k National average carbon dioxide emission factor for power generation; for k Guo Xiang p Electricity exported from the province; For regional power grid r The average carbon dioxide emission factor; For regional power grid r Towards p Saves net power output; for p Total annual electricity consumption of the province; for p Total annual renewable energy power generation in the province; p Target province; n To p Other provinces that have reduced net electricity transmission; k To p Countries that reduce net electricity exports; r for p The regional power grid where the province is located.

4. The multi-level dynamic enterprise product carbon emission calculation method according to claim 1, characterized in that, The calculation of total corporate carbon emissions based on the dynamic carbon emission factor includes: Based on the power supply attributes of the enterprise, the enterprise will be matched to the corresponding dynamic carbon emission factor level; Based on the enterprise's electricity consumption data and the corresponding dynamic carbon emission factor level, the indirect carbon emissions generated by the enterprise's electricity consumption are calculated; the dynamic carbon emission factor level includes districts / counties and transformer substations; for enterprises connected to public transformer substations, the dynamic carbon emission factor of the substation to which the enterprise belongs is used for calculation; for enterprises connected to dedicated transformer substations or belonging to dedicated line users, the dynamic carbon emission factor of the district / county to which the enterprise belongs is used for calculation. Obtain enterprise energy consumption data; Calculate the company's direct carbon emissions based on its energy consumption data; The total carbon emissions of the enterprise are determined by combining the indirect carbon emissions and the direct carbon emissions.

5. The multi-level dynamic enterprise product carbon emission calculation method according to claim 1, characterized in that, The determination of product carbon emission factors based on the total carbon emissions of the enterprise using a product carbon emission regression model includes: A product carbon emission regression model is constructed using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables. The product carbon emission regression model introduces a physical range constraint on the product carbon emission coefficient, and takes minimizing the error between the predicted total carbon emissions of enterprises and the actual total carbon emissions of enterprises as the optimization objective. The product carbon emission regression model is solved by a gradient-based optimization algorithm, and the product carbon emission factor is output.

6. The multi-level dynamic enterprise product carbon emission calculation method according to claim 5, characterized in that, The gradient-based optimization algorithm is either the Lagrange multiplier method or the sequential quadratic programming algorithm.

7. The multi-level dynamic enterprise product carbon emission calculation method according to claim 5, characterized in that, The aforementioned method uses the production data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as the dependent variable to construct a product carbon emission regression model, including: Using the output data of various products of the enterprise as independent variables and the total carbon emissions of the enterprise as dependent variables, a product carbon emission regression model is constructed in conjunction with the second formula. The second formula includes: in, Y This refers to the total carbon emissions of the aforementioned enterprise; For the company's product output; It is the first p The unit carbon emission factor of a product type, that is, the carbon emission amount corresponding to each unit of production of this type of product; It is a collection of other factors that affect carbon emissions; It is the first O The coefficients of other influencing factors characterize the degree of their impact on carbon emissions; It is a random error term.

8. The multi-level dynamic enterprise product carbon emission calculation method according to claim 1, characterized in that, The preprocessing of the carbon emission-related data of the enterprise to obtain the preprocessed dataset includes: The OneClassSVM algorithm is used to identify outliers in the carbon emission-related data of the enterprise, remove the identified outliers, and mark them as missing values. The missing values ​​are filled using the K-nearest neighbor classification algorithm to form the preprocessed dataset.

9. The multi-level dynamic enterprise product carbon emission calculation method according to claim 8, characterized in that, The identification of outliers in the enterprise's carbon emission-related data based on the OneClassSVM algorithm includes: A numerical data matrix of electricity consumption is constructed based on the carbon emission data of the aforementioned enterprises; Based on the aforementioned power numerical data matrix, a hypersphere model is established using the OneClassSVM algorithm; Based on the aforementioned hypersphere model, a nonlinear objective function is constructed; The hypersphere model is optimized based on the nonlinear objective function to obtain the optimized hypersphere model; Calculate the projection distance from each data point in the numerical data matrix of the electrical quantity to the center of the hypersphere of the optimized hypersphere model; Based on the distance and the set anomaly detection threshold, abnormal values ​​are identified.

10. A multi-level dynamic enterprise product carbon emission calculation device, characterized in that, include: The data acquisition module is used to acquire carbon emission-related data from enterprises. The data preprocessing module is used to preprocess the carbon emission-related data of the enterprise to obtain a preprocessed dataset; The dynamic carbon emission factor calculation module is used to obtain the dynamic carbon emission factor based on the preprocessed dataset using a multi-level dynamic optimization model for carbon emission factors. The total carbon emissions calculation module is used to calculate the total carbon emissions of an enterprise based on the dynamic carbon emission factor. The product carbon emission factor calculation module is used to determine the product carbon emission factor based on the total carbon emissions of the enterprise using a product carbon emission regression model. The product carbon emission factor represents the carbon emission per unit of the company's products.

Images