A method for calculating corporate carbon emission intensity

Abstract

A method for calculating the carbon emission intensity of an enterprise belongs to the technical field of carbon emission intensity calculation. In order to solve the problem of accurately calculating the carbon emission intensity of an enterprise, the present invention includes constructing an enterprise correlation influence factor based on the geographical distance between enterprises and the relative correlation influence degree between enterprises; using the carbon capture and storage technology parameters and production process complexity parameters adopted by the enterprise, inputting them into a multi-layer perceptron model to calculate the industry carbon emission adjustment coefficient; adjusting the basic carbon emission unit output value of the enterprise, and performing a composite transformation on the adjusted carbon emission unit output value of the enterprise, including high-dimensional mapping and standardization transformation of the data, frequency domain analysis transformation, quantile transformation and adaptive activation transformation, to obtain the transformed carbon emission unit output value of the enterprise, and further adjusting the weight using the attention mechanism to obtain the final comprehensive carbon emission intensity of the enterprise. The present invention can more accurately calculate the carbon emission intensity of an enterprise.

Description

Technical Field

[0001] The present invention belongs to the technical field of carbon emission intensity calculation, and specifically relates to a method for calculating the carbon emission intensity of an enterprise. Background Art

[0002] Current methods for calculating corporate carbon emission intensity suffer from numerous shortcomings. They tend to be partial in their consideration of influencing factors, often focusing solely on a single emission value. For example, they consider only a company's production scale, ignoring the complex inter-company relationships and the combined impact of industry characteristics on carbon emissions. This leads to biased calculations and a lack of comparison within the same industry. Data processing technology is relatively backward, lacking in-depth exploration and effective utilization of the data's underlying characteristics. Traditional methods lack in-depth exploration of the underlying characteristics of corporate carbon emission data, relying solely on simple data processing techniques. This makes it difficult to fully capture the complex information contained in the data, resulting in calculations that fail to accurately reflect a company's true carbon emissions. While some methods consider spatial distances between companies and industry relationships, these approaches remain superficial, failing to deeply analyze how these factors interact and influence carbon emission intensity, hindering their full value in the calculation process. This makes it difficult to accurately reflect the dynamic changes in a company's carbon emission intensity, thus hindering the scientific and effective nature of carbon emission management decisions. Summary of the Invention

[0003] The problem to be solved by the present invention is to accurately calculate the carbon emission intensity of an enterprise, and a method for calculating the carbon emission intensity of an enterprise is proposed.

[0004] To achieve the above object, the present invention is implemented through the following technical solutions:

[0005] A method for calculating the carbon emission intensity of an enterprise comprises the following steps:

[0006] S1. Construct enterprise correlation influence factors based on the geographical distance between enterprises and the relative correlation influence between enterprises;

[0007] S2. Utilize the carbon capture and storage technology parameters and production process complexity parameters adopted by the enterprise and input them into a multi-layer perceptron model to calculate the industry carbon emission adjustment coefficient;

[0008] S3. Using the enterprise-related impact factor obtained in step S1 and the industry carbon emission adjustment coefficient obtained in step S2, adjust the enterprise's basic carbon emissions per unit of output value to obtain the adjusted enterprise carbon emissions per unit of output value;

[0009] S4. Performing a composite transformation on the adjusted enterprise carbon emissions per unit of output value obtained in step S3, including high-dimensional data mapping and normalization transformation, frequency domain analysis transformation, quantile transformation, and adaptive activation transformation, to obtain the transformed enterprise carbon emissions per unit of output value;

[0010] S5. The transformed carbon emissions per unit output value of the enterprise obtained in step S4 are further weighted using the attention mechanism to obtain the final comprehensive carbon emission intensity of the enterprise.

[0011] Furthermore, the specific implementation method of step S1 includes the following steps:

[0012] S1.1. Calculate the geographic distance between enterprises: Obtain the address information of each enterprise from its registration information. Using the coordinate picking function on the online map, enter the address to obtain the longitude and latitude coordinates of the enterprise and calculate the straight-line distance between the enterprises.

[0013] S1.2. Calculate the relative degree of influence between companies. Collect information on the companies' core products or services, lists of upstream and downstream partners, and the scale of their collaborations from their annual reports, publicly available bidding documents, and information released by industry associations. The sum of the proportions of procurement or sales between companies will be used as the relative degree of influence.

[0014] S1.3. Construct an enterprise correlation impact factor based on the geographical distance between enterprises and the relative degree of correlation between them. The expression is:

[0015] ;

[0016] in, For enterprises With other companies The impact factor of enterprise association between For enterprises With other companies geographical distance, For enterprises With other companies The relative degree of influence of association; is the maximum geographical distance, is the total number of enterprises in the enterprise alliance involved in the calculation.

[0017] Furthermore, the specific implementation method of step S2 includes the following steps:

[0018] S2.1. Calculate carbon capture and storage technical parameters. Collect carbon capture equipment capacity data from company equipment archives and operating records. Divide the company's capture capacity by the industry average to obtain the carbon capture and storage technical parameters.

[0019] S2.2. Calculate the production process complexity parameter. Count the number of production processes from the company's production process documentation and the number of equipment types from the equipment list. Add the number of production processes and the number of equipment types, then divide by the industry average to obtain the production process complexity parameter.

[0020] S2.3. Calculate the industry carbon emission adjustment coefficient using the multi-layer perceptron model. The expression is:

[0021] ;

[0022] in, For enterprises The industry carbon emission adjustment coefficient, For enterprises Carbon capture and storage technology parameters, For enterprises The production process complexity parameters, is a multi-layer perceptron model, is the hyperbolic tangent function, Represents the maximum value of carbon capture and storage technology parameters among all companies, Represents the maximum value of the production process complexity parameter among all enterprises.

[0023] Furthermore, the specific implementation method of step S3 includes the following steps:

[0024] S3.1. Calculate the company's basic carbon emissions per unit of output using the following formula:

[0025] ;

[0026] in, It is an enterprise of carbon emissions, It is an enterprise The output value, It is an enterprise Basic carbon emissions per unit of output value;

[0027] S3.2. Adjust the enterprise's basic carbon emissions per unit of output value obtained in step S3.1 to obtain the adjusted enterprise carbon emissions per unit of output value. The calculation formula is:

[0028] ;

[0029] in, It is an adjusted enterprise Carbon emissions per unit of output value, Other companies Basic carbon emissions per unit of output value.

[0030] Furthermore, the specific implementation method of step S4 includes the following steps:

[0031] S4.1. Construct a high-dimensional feature space, map the adjusted carbon emissions per unit output value to a high-dimensional feature vector, and introduce a feature enhancement factor, expressed as:

[0032] ;

[0033] in, It is an enterprise The feature enhancement factor, All companies The mean of is the adjustment parameter;

[0034] Will and Multiply the elements and get the enterprise The enhanced high-dimensional feature vector , the expression is:

[0035] ;

[0036] S4.2. Perform a frequency-quantile hybrid transform on the enhanced high-dimensional feature vector obtained in step S4.1;

[0037] Will Convert to the frequency domain and weight the frequency domain features. The expression is:

[0038] ;

[0039] in It is an enterprise The frequency domain eigenvector of is the weight matrix;

[0040] Then perform quantile mapping, the expression is:

[0041] ;

[0042] in, It is an enterprise The frequency-quantile mixed eigenvector of is the indicator function, when When the indicator function is 1, otherwise the indicator function is 0; is the quantile number.

[0043] S4.3. For enterprises iThe frequency domain-quantile mixed eigenvector of the transformed enterprise carbon emissions per unit output value is adaptively activated and the expression is obtained as follows:

[0044] ;

[0045] in, For the transformed enterprise i Carbon emissions per unit of output value, are the first adaptive parameter and the second adaptive parameter respectively, All companies The mean of .

[0046] Furthermore, the specific implementation method of step S5 includes the following steps:

[0047] S5.1. Calculate the transformed carbon emissions per unit of output value for all enterprises, constructing the input set in a time series manner, namely:

[0048] ;

[0049] in, is the input set constructed from the transformed carbon emissions per unit of output value of all enterprises, The transformed enterprise Carbon emissions per unit of output value;

[0050] S5.2. Use the attention mechanism to adjust the weights of the input set constructed from the transformed corporate carbon emissions per unit output value obtained in step S5.1 to obtain the final weights , the calculation formula is:

[0051] ;

[0052] in, represents the attention calculation function, is an adjustable power parameter;

[0053] S5.3. Calculate the final enterprise-wide carbon emission intensity based on the final weights obtained in step S5.2 using the following formula:

[0054] ;

[0055] in, For enterprises The final comprehensive carbon emission intensity of the enterprise.

[0056] Beneficial effects of the present invention:

[0057] The method for calculating the carbon emission intensity of an enterprise described in the present invention comprehensively considers multiple factors such as the geographical distance between enterprises, the degree of industry correlation, and industry characteristics, and combines advanced data processing technologies, including high-dimensional mapping, frequency domain-quantile hybrid transformation, and adaptive nonlinear activation, to deeply mine the value of data. This can not only more accurately calculate the carbon emission intensity of an enterprise, but also provide a scientific and reliable decision-making basis for environmental regulatory authorities, helping them to formulate more targeted carbon emission policies. At the same time, it provides enterprises with a clear direction for improving carbon emissions, helps enterprises optimize production processes, improve energy utilization efficiency, and thus promote the development of the entire industry in a green and low-carbon direction, playing an important role in responding to climate change and achieving sustainable development.

[0058] The method for calculating the carbon emission intensity of an enterprise described in the present invention accurately provides insights into new problems with corporate carbon emissions: Under the traditional carbon emission intensity calculation model, many companies find it difficult to accurately locate their own carbon emission status in the industry. Because the calculation method fails to fully consider key factors such as the relationship between enterprises and industry characteristics, some companies have long ignored potential carbon emission problems. With the help of the technical solution of the present invention, the carbon emission intensity of 50 manufacturing companies in a certain area was recalculated. In the comparative analysis with peer companies, as many as 20 companies (accounting for 40%) discovered problems that had not been noticed before. Although some companies have similar production scales, their actual carbon emission intensity is much higher than that of their peers due to differences in corporate relationships and industry characteristics. These newly discovered problems provide a clear direction for companies to subsequently formulate targeted energy-saving and emission reduction strategies and optimize production processes, helping companies to make precise efforts on the path of low-carbon development.

[0059] The method for calculating the carbon emission intensity of an enterprise described in the present invention effectively assists enterprises in making energy-saving and emission reduction decisions: the technical solution of the present invention can provide enterprises with more accurate carbon emission intensity data, helping enterprises to clearly understand their own carbon emission status. For example, before a chemical company adopted this technical solution, its energy-saving and emission reduction measures were not effective because it did not have a clear grasp of the key links of its own carbon emissions. With the help of the calculation results of the present invention, the company found that the carbon emissions of specific links in its production process accounted for a high proportion, and then optimized the process in a targeted manner. Within one year after the upgrade, the company's carbon emission intensity was reduced by 15%, energy conservation and emission reduction results were significant, production costs also decreased, and the company's market competitiveness was enhanced.

[0060] The method for calculating corporate carbon emission intensity described in the present invention optimizes environmental regulatory efficiency: For environmental regulatory authorities, traditional calculation methods make it difficult to accurately assess the carbon emissions of enterprises in a region, resulting in a lack of targeted supervision. After applying the technical solution of the present invention, regulatory authorities can clearly grasp the carbon emission intensity of each enterprise and its changing trends. Based on the calculation results, regulatory authorities focus on regulating enterprises with high carbon emission intensity, conduct special inspections of these high-emission enterprises within a year, and encourage relevant enterprises to actively rectify and improve. This has promoted the improvement of regional environmental quality.

[0061] The method for calculating corporate carbon emission intensity, described in this invention, promotes the green transformation and development of industries. By accurately calculating carbon emission intensity, this technical solution encourages companies to be aware of their carbon emission levels within their industry. Within a particular industry, when companies clearly identify high carbon emission intensity, they proactively increase their investment in green technology research and development. As companies actively improve production technologies and processes, the carbon emission intensity of the entire industry gradually decreases, the industrial structure continues to optimize, and the industry's transformation and development towards green, low-carbon, and sustainable development is driven. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 This is a flow chart of a method for calculating corporate carbon emission intensity according to the present invention. DETAILED DESCRIPTION

[0063] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present invention and are not intended to limit the present invention. That is, the specific embodiments described herein are only some embodiments of the present invention, not all embodiments. Generally, the components of the specific embodiments of the present invention described and illustrated in the drawings herein can be arranged and designed in various different configurations, and the present invention can also have other embodiments.

[0064] Therefore, the following detailed description of the specific embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but is merely representative of selected specific embodiments of the present invention. All other specific embodiments obtained by those skilled in the art based on the specific embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0065] In order to further understand the content, features and effects of the present invention, the following specific embodiments are given as examples, and the attached Figure 1 The detailed instructions are as follows:

[0066] Example 1:

[0067] A method for calculating the carbon emission intensity of an enterprise comprises the following steps:

[0068] S1. Construct enterprise correlation influence factors based on the geographical distance between enterprises and the relative correlation influence between enterprises;

[0069] To reflect the complex relationships between enterprises, the geographical distance between enterprises and the degree of industry correlation are introduced. This calculation is to avoid the carbon emission measurement deviation between different industries due to industry characteristics caused by single enterprise calculations. For example, simply comparing the carbon emissions of steel mills and electronics factories is not comparable, so an enterprise management impact calculation is constructed.

[0070] Furthermore, the specific implementation method of step S1 includes the following steps:

[0071] S1.1. Calculate the geographic distance between enterprises: Obtain the address information of each enterprise from its registration information. Using the coordinate picking function on the online map, enter the address to obtain the longitude and latitude coordinates of the enterprise and calculate the straight-line distance between the enterprises.

[0072] S1.2. Calculate the relative degree of influence between companies. Collect information on the companies' core products or services, lists of upstream and downstream partners, and the scale of their collaborations from their annual reports, publicly available bidding documents, and information released by industry associations. The sum of the proportions of procurement or sales between companies will be used as the relative degree of influence.

[0073] For example: Enterprise i To other companies j The purchase amount accounts for 20% and the sales amount accounts for 10%. i For gas companies j The industry relevance score is 0.2+0.1=0.3;

[0074] S1.3. Construct an enterprise correlation impact factor based on the geographical distance between enterprises and the relative degree of correlation between enterprises. The expression is:

[0075] ;

[0076] in, For enterprises With other companies The impact factor of enterprise association between For enterprises With other companies geographical distance, For enterprises With other companies The relative degree of influence of association; is the maximum geographical distance, is the total number of enterprises in the enterprise alliance involved in the calculation;

[0077] Furthermore, the molecules As the geographical distance between enterprises increases Relative to the maximum geographical distance As increases, the degree of correlation decreases exponentially; reflects the degree of industry correlation; the denominator is all other companies of The purpose of summing is to normalize the comprehensive correlation impact factors to accurately reflect the enterprise With other companies The relative correlation influence of parameters Represents the total number of companies in the enterprise alliance that participates in the calculation.

[0078] Furthermore, suppose there are five companies, A, B, C, D, and E, in an industrial park. The impact of inter-company connections on carbon emission intensity calculations needs to be assessed. Without the proposed method, traditional approaches would measure connections solely based on geographic distance, assuming that closer distances have a greater impact. For example, if A and D are only 0.5 km apart (the park cafeteria) and A and B are 3 km apart (their auto parts supply relationship), traditional methods would incorrectly determine that AD has a stronger connection than AB. The subsequent carbon emission intensity calculation would give cafeteria D an unreasonably high weight and auto parts manufacturer B a low weight, significantly deviating from the actual carbon emission transmission relationship. Using the proposed method, both geographic distance and industry connections are considered: The geographical distances between the five companies are collected (maximum 5 km), and the degree of industry connection is scored based on business collaboration (AB has an industry connection of 0.9 due to supply chain ties; AD has an industry connection of 0.1 due to its lack of business connection). This is then substituted into the formula to calculate the connection impact factor. After calculation, AB's strong industry connection results in a much greater connection factor than AD, accurately capturing the logic of strong connections between upstream and downstream of the industrial chain. The subsequent calculation of carbon emission intensity using this factor can reasonably assign a high weight to B and a low weight to D. The result is more in line with reality and assists in the carbon supervision of the park.

[0079] The calculation of the comprehensive correlation impact factor comprehensively considers the geographic distance and industry correlation between enterprises. Leveraging the geographical distance and industry correlation of enterprises, an exponential function is used to reflect the exponentially attenuated effect of geographic distance on the correlation degree. Logarithmic operations are used to further refine the effect of changes in geographic distance, while also taking into account the value of industry correlation. The advantage of this calculation method is that it can more accurately depict the complex correlation characteristics between enterprises, avoiding the need to evaluate correlation relationships from only a single dimension. Its significance lies in providing more realistic correlation weights for subsequent enterprise carbon emission intensity calculations, so that the final carbon emission intensity results can more truly reflect the comprehensive carbon emission impact of enterprises in the industry and geographic space, providing a more valuable reference basis for environmental supervision and the formulation of corporate carbon emission governance strategies.

[0080] S2. Utilize the carbon capture and storage technology parameters and production process complexity parameters adopted by the enterprise and input them into a multi-layer perceptron model to calculate the industry carbon emission adjustment coefficient;

[0081] Furthermore, the specific implementation method of step S2 includes the following steps:

[0082] S2.1. Calculate carbon capture and storage technical parameters. Collect carbon capture equipment capacity data from company equipment archives and operating records. Divide the company's capture capacity by the industry average to obtain the carbon capture and storage technical parameters.

[0083] For example, if a company has a capacity of 500 units and the industry average is 1,000 units, the carbon capture and storage technical parameter is 500 / 1,000=0.5.

[0084] S2.2. Calculate the production process complexity parameter. Count the number of production processes from the company's production process documentation and the number of equipment types from the equipment list. Add the number of production processes and the number of equipment types, then divide by the industry average to obtain the production process complexity parameter.

[0085] For example, if a company has 15 production processes, 5 types of equipment, and an industry average of 10, the production process complexity parameter is (15 + 5) / 10 = 2;

[0086] S2.3. Calculate the industry carbon emission adjustment coefficient using the multi-layer perceptron model. The expression is:

[0087] ;

[0088] in, For enterprises The industry carbon emission adjustment coefficient, For enterprises Carbon capture and storage technology parameters, For enterprises The production process complexity parameters, is a multi-layer perceptron model, is the hyperbolic tangent function, Represents the maximum value of carbon capture and storage technology parameters among all companies, Represents the maximum value of the production process complexity parameter among all enterprises.

[0089] Furthermore, the multi-layer perceptron model A three-layer neural network is designed. The first layer is the input layer, whose number of neurons is set to 2 based on the input data: carbon capture and storage technology parameters and production process complexity parameters. The second layer is the hidden layer, with 5 neurons. Neurons are connected by weights, which are continuously adjusted and optimized during model training. The neuron activation function uses a simple ReLU function (i.e., the output is the larger of the input value and 0). The third layer is the output layer, with 1 neuron. The output value is processed by a sigmoid function to compress it between 0 and 1, ultimately resulting in the industry carbon emission adjustment coefficient.

[0090] The hyperbolic tangent operation further adjusts the coefficient based on the comprehensive characteristics of the company's technology and process parameters, highlighting the impact of differences in technology and process on the carbon emission adjustment coefficient. By learning and processing these parameters, it outputs a value that reflects the impact of industry characteristics on carbon emissions regulation.

[0091] Based on the five enterprises in the park, A, B, C, D, and E, the industry carbon emission adjustment coefficient needs to be calculated for each enterprise. Traditional calculations focus on a single indicator, such as calculating the adjustment coefficient based solely on "carbon capture technology parameters." For example, if enterprise A has excellent carbon capture technology but a simple production process, while enterprise B has average carbon capture technology but a complex process, traditional methods will overestimate the regulatory effect of enterprise A on carbon emissions and underestimate the regulatory effect of enterprise B on carbon emissions. This will lead to significant discrepancies between the subsequent carbon emission intensity calculations and actual results, and fail to accurately reflect the enterprises' true emission reduction potential. Using the method described in the example, calculate each enterprise's carbon capture and storage technology parameter (e.g., enterprise A's value of 8 represents a comprehensive evaluation of carbon capture equipment efficiency and storage capacity) and its production process complexity parameter (e.g., enterprise A's value of 6, derived from a combination of production processes and equipment types). Find the maximum value of these two parameters across all enterprises, assuming the maximum carbon capture value is 10 and the maximum production process complexity is 8. A three-layer neural network is used for this analysis. The input layer contains two data points, the middle hidden layer has five neurons (similar to optimizing data at a "transfer station"), and the output layer has one neuron to compress the results to a range between 0 and 1, simulating the combined impact of enterprise technology and processes. Hyperbolic tangent adjustment involves substituting technology and process parameters into a formula, then combining the maximum values ​​of these parameters across companies to further adjust the results. For example, the formula can comprehensively reflect the characteristics of Company A, which has "good carbon capture technology but average process," resulting in a more reasonable adjustment coefficient. Using this method, the adjustment coefficients for the five companies accurately reflect the true impact of their technology and processes on carbon emissions. Subsequent calculations of carbon emission intensity are more realistic, helping regulators and companies identify the right path to reduce emissions.

[0092] S3. Using the enterprise-related impact factor obtained in step S1 and the industry carbon emission adjustment coefficient obtained in step S2, adjust the enterprise's basic carbon emissions per unit of output value to obtain the adjusted enterprise carbon emissions per unit of output value;

[0093] Furthermore, the specific implementation method of step S3 includes the following steps:

[0094] S3.1. Calculate the company's basic carbon emissions per unit of output using the following formula:

[0095] ;

[0096] in, It is an enterprise of carbon emissions, It is an enterprise The output value, It is an enterprise Basic carbon emissions per unit of output value;

[0097] S3.2. Adjust the enterprise's basic carbon emissions per unit of output value obtained in step S3.1 to obtain the adjusted enterprise carbon emissions per unit of output value. The calculation formula is:

[0098] ;

[0099] in, It is an adjusted enterprise Carbon emissions per unit of output value, Other companies Basic carbon emissions per unit of output value;

[0100] This part indicates that according to the enterprise With other companies The degree of association, and other enterprises The basic carbon emissions per unit output value and the industry carbon emissions adjustment coefficient are In this way, the complex relationships between enterprises based on geographical location and industry attributes are taken into account, while also incorporating the impact of industry characteristics on carbon emissions.

[0101] S4. Performing a composite transformation on the adjusted enterprise carbon emissions per unit of output value obtained in step S3, including high-dimensional data mapping and normalization transformation, frequency domain analysis transformation, quantile transformation, and adaptive activation transformation, to obtain the transformed enterprise carbon emissions per unit of output value;

[0102] Furthermore, the specific implementation method of step S4 includes the following steps:

[0103] S4.1. Construct a high-dimensional feature space, map the adjusted carbon emissions per unit output value to a high-dimensional feature vector, and introduce a feature enhancement factor, expressed as:

[0104] ;

[0105] in, It is an enterprise The feature enhancement factor, All companies The mean of is the adjustment parameter;

[0106] Will and Multiply the elements and get the enterprise The enhanced high-dimensional feature vector , the expression is:

[0107] ;

[0108] S4.2. Perform a frequency-quantile hybrid transform on the enhanced high-dimensional feature vector obtained in step S4.1;

[0109] Will Convert to the frequency domain and weight the frequency domain features. The expression is:

[0110] ;

[0111] in It is an enterprise The frequency domain eigenvector of is the weight matrix;

[0112] Then perform quantile mapping, the expression is:

[0113] ;

[0114] in, It is an enterprise The frequency-quantile mixed eigenvector of is the indicator function, when When the indicator function is 1, otherwise the indicator function is 0; is the quantile number.

[0115] S4.3. For enterprises i The frequency domain-quantile mixed eigenvector of the transformed enterprise carbon emissions per unit output value is adaptively activated and the expression is:

[0116] ;

[0117] in, For the transformed enterprise Carbon emissions per unit of output value, are the first adaptive parameter and the second adaptive parameter respectively, All companies The mean of .

[0118] Further, The data is extracted through historical training using the gradient descent method. The design of the adaptive nonlinear activation function fully considers the characteristics and dynamics of corporate carbon emissions data. It addresses the issue of varying data characteristics, striving to dynamically and flexibly adjust the activation level based on the data's distribution. Through specialized computational methods, the function transforms the data into an appropriate range, making the differences between different data more pronounced. Furthermore, the function utilizes a logarithmic transformation to further explore hidden features within the data. This series of composite transformations not only achieves high-dimensional mapping and standardization of the data, but also, through frequency domain analysis, quantile adjustment, and adaptive activation, fully exploits the underlying characteristics of corporate carbon emissions data, providing a superior data foundation for subsequent carbon emission intensity calculations.

[0119] Based on the five enterprises in the park, A, B, C, D, and E, their carbon emissions per unit of output value need to be optimized through feature transformation. Previously, the original carbon emissions per unit of output value data was used directly without optimization. For example, the data of enterprise A is greatly affected by occasional factors (such as temporary high-energy-consuming orders), while the data of enterprise B has little fluctuation. The original data cannot highlight these differences, and subsequent analysis is prone to "confusing them together", affecting the accuracy of carbon emission assessment. Using the feature enhancement method in the embodiment, the feature enhancement factor of each enterprise is first calculated. The relevant data of carbon emissions per unit of output value of all enterprises are counted, the mean is calculated, and the enhancement factor is calculated for enterprise A in combination with the adjustment parameters (for example, due to large data fluctuations, the enhancement factor makes the effective features more prominent), and then multiplied with the original data to obtain the enhanced high-dimensional features, so that the features of enterprise A affected by occasional factors are clearer. The frequency domain-quantile transformation transfers the enhanced features to the frequency domain, weights them with a weight matrix (highlighting key frequency features), and then performs quantile mapping. For example, for Company A, frequency-domain weighting highlights the frequency characteristics of its carbon emissions fluctuations, while quantile mapping further enhances the distributional differences between different company data. Adaptive activation dynamically adjusts the activation level based on factors such as the mean of all company data. Given the unique data distribution of Company A, the activation function was specifically adjusted to ensure that the transformed carbon emissions per unit of output value more closely reflect its actual characteristics. This series of transformations accurately captures the differences in data characteristics across the five companies, making subsequent carbon emissions analysis more accurate.

[0120] S5. The transformed carbon emissions per unit output value of the enterprise obtained in step S4 are further weighted using the attention mechanism to obtain the final comprehensive carbon emission intensity of the enterprise.

[0121] Furthermore, the specific implementation method of step S5 includes the following steps:

[0122] S5.1. Calculate the transformed carbon emissions per unit of output value for all enterprises, constructing the input set in a time series manner, namely:

[0123] ;

[0124] in, is the input set constructed from the transformed carbon emissions per unit of output value of all enterprises, The transformed enterprise Carbon emissions per unit of output value;

[0125] S5.2. Use the attention mechanism to adjust the weights of the input set constructed from the transformed corporate carbon emissions per unit output value obtained in step S5.1 to obtain the final weights , the calculation formula is:

[0126] ;

[0127] in, represents the attention calculation function, is an adjustable power parameter;

[0128] Further, By inputting data The relative importance of the standardized carbon emission value of each enterprise is calculated based on the analysis of The results of attention calculation are exponentially transformed to highlight the differences; the denominator is for all companies Sum and normalize the weights so that The value ranges from 0 to 1, and the sum of all enterprise weights is 1.

[0129] S5.3. Calculate the final enterprise-wide carbon emission intensity based on the final weights obtained in step S5.2 using the following formula:

[0130] ;

[0131] in, For enterprises The final comprehensive carbon emission intensity of the enterprise.

[0132] Furthermore, the weights obtained through the attention mechanism are multiplied by the standardized carbon emissions per unit output value, so that the weights can be dynamically allocated according to the dynamic changes in the company's carbon emissions, more accurately reflecting the company's comprehensive carbon emission intensity and providing a more scientific reference for environmental supervision and the company's own carbon emission management.

[0133] For the five enterprises in the industrial park, A, B, C, D, and E, the comprehensive carbon emission intensity needs to be calculated. Previously, the method for calculating comprehensive carbon emission intensity simply assigned an average weight to each enterprise (for example, 0.2). However, in reality, enterprise A's carbon emission data fluctuates significantly and has a strong overall impact, while enterprise B's data is stable and has a weaker impact. This average weighting would obscure the key impact of enterprise A, causing the comprehensive intensity result to deviate from the true situation. Using the attention mechanism described in the example to calculate weights, the carbon emission data of the five enterprises are first analyzed. An attention function is then used to determine the relative importance of each enterprise's data. Enterprise A's production fluctuates significantly, and its carbon emission changes have a significant impact on the overall situation. After exponential transformation and normalization, a weight is calculated for A (for example, 0.3). Enterprise B, with its weaker impact, might receive a weight of 0.1. This dynamically assigns appropriate weights to different enterprises, unlike the traditional "one-size-fits-all" approach. To calculate the comprehensive carbon emission intensity, the weights obtained above are multiplied by each enterprise's carbon emissions per unit of output value. Given a high weight for enterprise A and its combined emission data, the comprehensive intensity emphasizes its unique characteristics. A low weight for enterprise B has a minimal impact on the results. After the calculations for the five companies are completed, the comprehensive intensity can accurately reflect each company's true contribution to overall carbon emissions, helping regulators and companies find key areas for emission reduction.

[0134] The comprehensive calculation results are as follows:

[0135]

[0136] The carbon emission intensity calculations for the five companies mentioned above reveal that traditional methods, relying solely on baseline data and a single dimension of correlation (such as geographic distance), lead to significant deviations from the companies' true emission characteristics. For example, Company D, due to its close geographic proximity, was misjudged as highly correlated. The traditional method's calculated result (2.0 tons per 10,000 yuan) was consistent with the baseline value, failing to reflect its actual low emissions and weak industry correlations. Furthermore, due to the geographical distance between companies A and B, the traditional method underestimated the impact of their strong industry correlations on carbon emissions, resulting in results of 1.8 tons per 10,000 yuan and 1.9 tons per 10,000 yuan, respectively, deviating from their true emission intensities (1.5 tons per 10,000 yuan and 1.6 tons per 10,000 yuan, respectively, based on the new method).

[0137] This embodiment introduces enterprise-related influencing factors, industry carbon emission adjustment coefficients, and adaptive weight adjustments to accurately depict the industrial chain relationships between enterprises (for example, the supply chain binding of AB increases the association factor by 30%), technological process differences (the carbon capture technology parameter of 0.8 of enterprise A reduces the adjustment coefficient by 20%), and data dynamic characteristics (the fluctuating data of enterprise A is given a weight of 0.3 through the attention mechanism). The final result is an average reduction of 15%-25% compared with traditional methods, and the differences between enterprises are more significant (for example, enterprise D is reduced to 1.3 tons / 10,000 yuan), providing a more scientific quantitative basis for carbon regulation and corporate emission reduction.

[0138] It should be noted that relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

[0139] Although the present application has been described above with reference to specific embodiments, various modifications may be made thereto and components may be substituted with equivalents without departing from the scope of the present application. In particular, as long as there are no structural conflicts, the various features of the embodiments disclosed herein may be combined with each other in any manner, and the omission of an exhaustive description of these combinations in this specification is solely for the sake of space and resource conservation. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions within the scope of the claims.

Claims

1. A method for calculating corporate carbon emission intensity, characterized by: The steps include: S1. Construct enterprise correlation influence factors based on the geographical distance between enterprises and the relative correlation influence between enterprises; S2. Utilize the carbon capture and storage technology parameters and production process complexity parameters adopted by the enterprise and input them into a multi-layer perceptron model to calculate the industry carbon emission adjustment coefficient; S3. Using the enterprise-related impact factor obtained in step S1 and the industry carbon emission adjustment coefficient obtained in step S2, adjust the enterprise's basic carbon emissions per unit of output value to obtain the adjusted enterprise carbon emissions per unit of output value; The specific implementation method of step S3 includes the following steps: S3.

1. Calculate the company's basic carbon emissions per unit of output using the following formula: ； in, is the carbon emissions of enterprise i, is the output value of firm i, is the basic carbon emission per unit output value of enterprise i; S3.

2. Adjust the enterprise's basic carbon emissions per unit of output value obtained in step S3.1 to obtain the adjusted enterprise carbon emissions per unit of output value. The calculation formula is: ； in, is the adjusted carbon emissions per unit output value of enterprise i, is the basic carbon emission per unit output value of other enterprise j; S4. Performing a composite transformation on the adjusted enterprise carbon emissions per unit of output value obtained in step S3, including high-dimensional data mapping and normalization transformation, frequency domain analysis transformation, quantile transformation, and adaptive activation transformation, to obtain the transformed enterprise carbon emissions per unit of output value; S5. Use the attention mechanism to adjust the weight of the transformed enterprise carbon emissions per unit output value obtained in step S4 to obtain the final enterprise comprehensive carbon emission intensity.

2. The method for calculating the carbon emission intensity of an enterprise according to claim 1, characterized in that: The specific implementation method of step S1 includes the following steps: S1.

1. Calculate the geographic distance between enterprises: Obtain the address information of each enterprise from its registration information. Using the coordinate picking function on the online map, enter the address to obtain the longitude and latitude coordinates of the enterprise and calculate the straight-line distance between the enterprises. S1.

2. Calculate the relative degree of influence between companies. Collect information on the companies' core products or services, lists of upstream and downstream partners, and the scale of their collaborations from their annual reports, publicly available bidding documents, and information released by industry associations. The sum of the proportions of procurement or sales between companies will be used as the relative degree of influence. S1.

3. Construct an enterprise correlation impact factor based on the geographical distance between enterprises and the relative degree of correlation between them. The expression is: ； in, is the enterprise linkage influencing factor between enterprise i and other enterprises j, is the geographical distance between enterprise i and other enterprises j, is the relative correlation influence between enterprise i and other enterprises j; is the maximum geographical distance, and m is the total number of enterprises in the enterprise alliance participating in the calculation.

3. A method for calculating corporate carbon emission intensity according to claim 1 or 2, characterized in that: The specific implementation method of step S2 includes the following steps: S2.

1. Calculate carbon capture and storage technical parameters. Collect carbon capture equipment capacity data from company equipment archives and operating records. Divide the company's capture capacity by the industry average to obtain the carbon capture and storage technical parameters. S2.

2. Calculate the production process complexity parameter. Count the number of production steps from the company's production process documentation and the number of equipment types from the equipment list. Add the number of production steps and the number of equipment types, then divide by the industry average to obtain the production process complexity parameter. S2.

3. Calculate the industry carbon emission adjustment coefficient using the multi-layer perceptron model. The expression is: ； in, is the industry carbon emission adjustment coefficient of enterprise i, is the carbon capture and storage technology parameter of enterprise i, is the production process complexity parameter of enterprise i, is a multi-layer perceptron model, is the hyperbolic tangent function, Represents the maximum value of carbon capture and storage technology parameters among all companies, Represents the maximum value of the production process complexity parameter among all enterprises.

4. The method for calculating the carbon emission intensity of an enterprise according to claim 3, characterized in that: The specific implementation method of step S4 includes the following steps: S4.

1. Construct a high-dimensional feature space, map the adjusted carbon emissions per unit output value to a high-dimensional feature vector, and introduce a feature enhancement factor, expressed as: ； in, is the characteristic enhancement factor of enterprise i, All companies The mean of is the adjustment parameter; Will and Perform element-wise multiplication to obtain the enhanced high-dimensional feature vector of enterprise i , the expression is: ； S4.

2. Perform a frequency-quantile hybrid transform on the enhanced high-dimensional feature vector obtained in step S4.1; Will Convert to the frequency domain and weight the frequency domain features. The expression is: ； in, is the frequency domain eigenvector of enterprise i, is the weight matrix; Then perform quantile mapping, the expression is: ； in, is the frequency domain-quantile mixed eigenvector of enterprise i, is the indicator function, when When the indicator function is 1, otherwise the indicator function is 0; P is the quantile number; S4.

3. Perform adaptive nonlinear activation on the frequency-quantile hybrid eigenvector of enterprise i to obtain the transformed expression for the enterprise's carbon emissions per unit of output value: ； in, is the carbon emission per unit output value of enterprise i after transformation, are the first adaptive parameter and the second adaptive parameter respectively, All companies The mean of .

5. The method for calculating the carbon emission intensity of an enterprise according to claim 4, characterized in that: The specific implementation method of step S5 includes the following steps: S5.

1. Calculate the transformed carbon emissions per unit of output value for all enterprises, constructing the input set in a time series manner, namely: ； in, is the input set constructed from the transformed carbon emissions per unit of output value of all enterprises, is the carbon emission per unit output value of enterprise n after transformation; S5.

2. Use the attention mechanism to adjust the weights of the input set constructed from the transformed corporate carbon emissions per unit output value obtained in step S5.1 to obtain the final weights , the calculation formula is: ； in, represents the attention calculation function, is an adjustable power parameter; S5.

3. Calculate the final enterprise-wide carbon emission intensity based on the final weights obtained in step S5.2 using the following formula: ； in, is the final comprehensive carbon emission intensity of enterprise i.

Images