A method and system for analyzing carbon footprint of industrial products based on energy efficiency labeling database

Abstract

The present invention discloses a method for analyzing the carbon footprint of industrial products based on an energy efficiency label database, including: reading the data from the energy efficiency label database; splitting the total carbon emission data over the years to obtain the total carbon emission of the product, calculating the carbon emission at each stage in the product's entire life cycle, and taking the ratio of the two as the carbon emission benefit; performing a coefficient of variation analysis and screening out key parameters; searching for relevant emission factors based on the key parameters, and using the carbon emission benefit to correct the emission factors to obtain the carbon emissions at each stage of the product's actual life cycle; and then weighting the overall carbon emissions of the enterprise after the coefficient of variation is corrected to obtain the carbon emissions during the product's entire life cycle. On this basis, a grayscale model is used to screen and normalize the data, and then a modified BP neural network prediction model is constructed for prediction. The present invention makes full use of the data from the energy efficiency label database, and the results can be used for carbon footprint calculation and carbon footprint prediction.

Description

Technical Field

[0001] The present invention relates to the field of carbon footprint analysis, and in particular to an industrial product carbon footprint analysis method and system based on an energy efficiency labeling database. Background Art

[0002] The concept of carbon footprint is proposed on the basis of ecological footprint. It is mainly used to calculate carbon emissions. It measures the pressure of human activities on natural resources during the entire life cycle, that is, the degree and intensity of the use of natural resources. It directly measures the response of natural systems to carbon emissions in human activities.

[0003] Carbon footprint analysis includes carbon footprint measurement (calculating carbon emissions) and carbon footprint prediction (predicting carbon emissions).

[0004] Carbon footprint measurement can be roughly divided into system measurement method and non-system measurement method according to the method used. System measurement method can be divided into life cycle method (LCA, Life Circle Assessment), input-output method (IOA, Input-Output Analysis), etc.; non-system measurement method can be divided into actual measurement method, mass balance method (Mass-Balance approach) and emission factor method.

[0005] The life cycle method is created to analyze products and services. It measures carbon dioxide emissions and resource consumption at different stages of the entire life cycle of a product or thing, from production to use to destruction. It is a micro-level measurement method.

[0006] The input-output method uses the input-output table compiled by various countries or the World Bank as a basis to calculate the direct consumption coefficient and complete consumption coefficient of the product to measure the direct and indirect emissions of carbon dioxide. At the same time, it reflects the relationship between indicators such as initial input, intermediate input, intermediate product, and total output through the balance equation. It is a bottom-up calculation method.

[0007] The mass balance method, also known as the material balance method, studies the entire industrial production process or the gas emissions in a local production process. Based on the law of conservation of mass, it is a method for calculating the total gas emissions of an enterprise based on its production process, gas emissions in the production process, and the corresponding environmental governance. It is mainly used in specific enterprises or industries.

[0008] The emission factor method is also known as the emission coefficient method or the IPCC method. This method involves all areas and processes of social production and life, and has become an internationally recognized and widely used carbon footprint calculation method. According to the "2006 IPCC National Greenhouse Gas Inventory Guidelines" compiled and issued by the Intergovernmental Panel on Climate Change (IPCC), carbon emissions are the product of emission factors and energy consumption, that is, the calculation formula: carbon emissions = emission factor × energy consumption.

[0009] Among them, the best way is to calculate through the whole life cycle. Conventional LCA methods need to determine the boundaries of direct and indirect emissions and collect information within the boundaries. However, in actual work, the availability and accuracy of data need to be considered. The acquisition of a large amount of data, especially supply chain and raw material mining data, is not easy to obtain, and it is difficult to ensure its accuracy. Secondly, data will change with the upgrading of industry scale and environmental impact, and timely data updates are also very necessary. Changes in application scenarios will also cause changes in the carbon footprint of products, and the final calculation results will also change. In addition, the calculation of recycled and renewable parts is difficult to calculate clearly due to the high difficulty of tracking, and it is difficult to make accurate calculations for subsequent future emissions, so it is impossible to plan for policy formulation and industrial distribution.

[0010] At present, there are three mainstream methods for predicting carbon emissions: grayscale model, machine learning method, and scenario analysis method. The characteristics of gray prediction model are low training difficulty and high prediction accuracy. If nonlinear related variables are involved, the prediction accuracy of grayscale model is worse than that of machine learning. However, for existing related studies, most of them still have incomplete consideration of carbon emission influencing factors, simple models, and insufficient effects. The advantage of machine learning is to obtain important information and features from the current data set, and then accurately predict future data. Among them, BP neural network, ELM, SVM and other algorithms are used more. Sufficient practical research has proved that machine learning is better in prediction accuracy, which also shows that the idea of ​​predicting carbon emissions through machine learning is reliable and reasonable, but the large amount of data, high screening difficulty, and low sensitivity to policies and technologies are problems. Summary of the invention

[0011] In view of this, the present invention provides an industrial product carbon footprint analysis method based on an energy efficiency labeling database, which at least partially solves the problems existing in the prior art in carbon footprint measurement and carbon footprint prediction.

[0012] In a first aspect, the present invention provides an industrial product carbon footprint analysis method based on an energy efficiency labeling database, which is used to calculate the carbon emissions of products, comprising:

[0013] Read the energy efficiency and energy consumption related data from the energy efficiency label database; analyze and obtain the total carbon emission data over the years; use the index decomposition method to split the total carbon emission data over the years to obtain the total carbon emission of the product, and calculate the carbon emissions of the raw material stage, product production stage, product application stage and product recycling stage in the whole life cycle of the product, and use the ratio of the carbon emissions of each stage to the total carbon emissions of the product as the carbon emission benefits of the product at different stages;

[0014] The coefficient of variation is analyzed by using the statistical standard deviation and mean ratio to obtain the significance of the variation between the carbon emission benefits of the product at different stages and the actual observed values, and to screen out the global and local key parameters;

[0015] According to key parameters, relevant emission factors are searched through the IPCC database, and the emission factors are corrected using the carbon emission benefits of the product at different stages to obtain the actual carbon emissions in each stage of the product's life cycle. The carbon emissions of the company as a whole after correction of the coefficient of variation are then weighted to obtain the carbon emissions during the entire life cycle of the product.

[0016] According to a specific implementation of an embodiment of the present invention, the carbon emissions are calculated for the raw material stage, product production stage, product application stage and product recycling stage in the whole life cycle of the product respectively, including:

[0017] 1) Raw materials stage: Use the energy efficiency data in the energy efficiency label database to analyze the proportion of raw materials and conduct data analysis and summary of carbon emissions from the transportation and production processes of raw materials;

[0018] 2) Product production stage: Based on the energy consumption limit standard requirements, the consumption of water, electricity, gas and coal of industrial products is considered, and the carbon emissions in the production stage are determined according to the production process and energy efficiency indicators, and data analysis and summary are conducted;

[0019] 3) Product application stage: Combine product energy efficiency data and product application scenarios to analyze and summarize data on carbon emissions generated during the application process and carbon emissions during transportation, repair, and maintenance;

[0020] 4) Product recycling stage: Utilize the renewable raw material information in the energy efficiency label database to analyze and summarize the carbon emissions during the product remanufacturing and recycling process.

[0021] According to a specific implementation of an embodiment of the present invention, the method for calculating the carbon emission benefits of the product at different stages includes:

[0022] After reading the energy efficiency and energy consumption data from the energy efficiency label database, the total energy and total energy intensity are calculated using the addition model. The formula is as follows:

[0023]

[0024] In the formula, E is energy consumption, which is a quantity indicator; V is energy intensity, which is an intensity indicator; Q is output; n represents the number of corresponding energy efficiency and energy consumption indicators from the database; S is the whole life cycle stage; energy intensity includes energy consumption, and energy consumption includes the impact of user use; I represents the energy consumption per unit product, I n Indicates the unit product energy consumption corresponding to the nth indicator of energy efficiency and energy consumption in the database;

[0025] The following formula is used to analyze the carbon emissions of products at different life cycle stages;

[0026]

[0027] Based on the above formula, the energy intensity V of a single product i in a single stage j in the life cycle can be analyzed: i,j , at this stage the output Q j The carbon emission in this case is C i,j ; In the formula, C is the total carbon emissions, S is the whole life cycle stage, Q is the output, and V is the energy intensity; C i,j is the carbon emission of a single product i in a single stage j of its life cycle, S i represents the life cycle stage of a single product i, Q j represents the output of a single stage j in the life cycle, V i,j It represents the energy intensity of a single product i in a single stage j during its life cycle;

[0028] Based on the above calculations, the total carbon emissions of the product and the carbon emissions at each stage of the life cycle can be obtained, and the ratio is used as the carbon emission benefit of the product at different stages;

[0029] According to a specific implementation of an embodiment of the present invention, the emission factor is corrected using the carbon emission benefits of the product at different stages to obtain the carbon emissions at each stage of the product's actual life cycle; then the total carbon emissions of the enterprise after the coefficient of variation is corrected are weighted to obtain the calculation formula for the carbon emissions during the entire life cycle of the product as follows:

[0030]

[0031] δ=ρ i δ i

[0032] τ i =C j / C

[0033] Where δ is the emission factor, δ i is the emission factor of a single product i; ρi is the coefficient of variation of a single product i; τ is the carbon emission benefit, τ i The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is: E is energy consumption, C is the total carbon emission, and C j is the carbon emission in a single stage j in the life cycle, E i is the energy consumption of a single product i, Q is the output, Q i represents the output of a single product i, I i Represents the unit product energy consumption of a single product i.

[0034] According to a specific implementation of an embodiment of the present invention, the analysis method is also used to predict the carbon emissions of products, including:

[0035] Data screening and normalization:

[0036] The grayscale model is used to screen and the results are tested for cointegration, and the single integer series energy efficiency and energy consumption parameters of the normalized indicator variables related to carbon emissions are obtained as normalized data;

[0037] Construct a modified BP neural network prediction model:

[0038] The hidden layer of the BP neural network prediction model is simplified, and the output of the raw material stage in the product life cycle is used as the input of the product production stage, the output of the product production stage is used as the input of the product application stage, and the output of the product application stage is used as the input of the product recycling stage;

[0039] A linear model of a multi-layer perceptron is added to the hidden layer of the BP neural network prediction model. Normalized data is used as prediction pooling data. The lowest carbon emissions in the entire life cycle and the highest energy efficiency of the product itself are considered. The carbon emissions of the product with the highest energy efficiency are first found through global optimization. Then, the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method to find the best product with the best carbon emissions.

[0040] Then the optimized data is passed into the modified BP neural network prediction model to be solved in space;

[0041] Carbon emissions forecast:

[0042] The modified BP neural network prediction model is used to predict product carbon emissions.

[0043] According to a specific implementation of an embodiment of the present invention, the grayscale model is used for screening and the results are subjected to cointegration test to obtain single integer series energy efficiency and energy consumption parameters of normalized indicator variables related to carbon emissions, and the normalized data includes:

[0044]

[0045] In the formula, ii represents the product category and j represents a single stage in the life cycle; It represents the normalized value of energy consumption of the product of category ii in a single stage j during its life cycle, E ii，j is the energy consumption of a single stage j in the life cycle of a product of category ii, max E ii,j , min E ii，j Indicates the maximum and minimum values ​​of energy consumption in this data class, V represents the normalized value of energy consumption of the product of category II in a single stage j during its life cycle. ii，j is the energy intensity of a single phase j of the life cycle of a product of category ii, maxV ii,j , minV ii，j Indicates the maximum and minimum values ​​of energy intensity in this data class;

[0046] Based on the above expression, the energy efficiency and energy consumption data are screened, and then the same root unit root test and different root unit root test are used to analyze and obtain the first-order single integer series. After screening by the grayscale model, the data in the unstable transition interval are abandoned for data consistency cointegration check, and finally the normalized single integer series data is obtained for subsequent prediction of pooled data.

[0047] According to a specific implementation of an embodiment of the present invention, the hidden layer of the BP neural network prediction model is simplified, and the corresponding algorithm of using the output of the raw material stage in the whole life cycle of the product as the input of the product production stage, using the output of the product production stage as the input of the product application stage, and using the output of the product application stage as the input of the product recycling stage includes:

[0048] c t =tanh(W c (h t-1 , x t )+b c )

[0049] c t =c t-1 f t +ra t c t

[0050] ra t =σ(W x，ra (h t-1 , x t )+b ra )

[0051] pr t =σ(W x，pr(h t-1 , x pr )+b pr )

[0052] ap t =σ(W x，ap (h t-1 , x ap )+b ap )

[0053] h t =ap t tanh(c t )

[0054] Where W is the weight coefficient matrix, b is the bias vector, h is the activation function, c is the hidden layer sequence variable calculation variable, t is the calculation step length, x is the life cycle sequence variable, ra, pr, ap are the raw materials, production, and application stages in the product life cycle, ra t ,pr t ,ap t represents the calculation output of the raw materials, production, and application stages at time t, h t and h t-1 is the activation function output at two moments before and after, x t is the output of the life cycle sequence variable at time t, x pr , x ap represents the life cycle sequence variable value in the production and application stages, c t and c t-1 Calculate the value of the hidden layer sequence variable for the two moments before and after, W c is the weight coefficient matrix of the hidden layer sequence variables, W x,ra ,W x,pr ,W x,ap is the weight coefficient matrix of the life cycle sequence variables of raw materials, production and application stages, and σ is the cumulative function.

[0055] According to a specific implementation of an embodiment of the present invention, a linear model of a multilayer perceptron is added to the hidden layer of the BP neural network prediction model, normalized data is used as prediction pooling data, and the lowest carbon emissions in the entire life cycle and the highest energy efficiency of the product itself are considered to meet the conditions at the same time. The carbon emissions of the product with the highest energy efficiency are first found through global optimization, and then the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method. Finding the best product carbon emissions includes:

[0056] The tanh function is used to optimize the minimum sum of errors of sample N. The optimization algorithm is as follows:

[0057]

[0058] In the formula, y i is the calculated output value at time m, st is the minimum optimization step length, v m is the correction factor at the mth moment, f is the calculated output value at the m-1th moment, and N is the total calculation time;

[0059] Add a multi-layer perceptron MLP and use a particle swarm algorithm to optimize the error sum, classify and clean the data to obtain the best of the same kind, and optimize the clustering method to minimize the median value of the objective function of each mode to cluster analysis, and aim to achieve the lowest energy consumption and carbon emission values. The energy efficiency evaluation mathematical model is expressed as:

[0060]

[0061] In the formula, E is energy consumption, C is total carbon emissions, ep represents the pth product attribute that affects energy efficiency and carbon emissions in the eth stage, g op is the pth attribute of the oth domain center; ω oe is the domain center and stage attribute judgment value. It takes 1 when the domain center and stage value of the product attribute belong to the same category, and takes 0 when they are not of the same category. The subscript o is the number of domain centers of the attribute, ranging from 1 to n. o , subscript e is the number of stages of the attribute, ranging from 1 to n e ; The subscript p is the number of attributes, ranging from 1 to n; n, n o , n e They are the number of corresponding energy efficiency and energy consumption indicators, the number of attribute domain centers, and the number of attribute stages in the database;

[0062] According to the above expression, local and global searches are performed in the domain, the cluster center and objective function value are calculated for each variable, and the result set is transformed;

[0063] The optimized data is passed into the modified BP neural network prediction model to solve in space, considering the influence of emission benefits and coefficient of variation, and the solution matrix is ​​updated as follows:

[0064] τ i (t+Δ)=(1-ρ)τ i (t)+βτ i

[0065] Where τ is the carbon emission benefit, τ i The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is the output value of the BP network; t is the solution time, Δ is the solution step, ρ is the coefficient of variation, βτ i It is the carbon emission benefit variable in the BP network calculation process.

[0066] In a second aspect, an embodiment of the present invention further provides an industrial product carbon footprint analysis system based on an energy efficiency identification database, comprising at least one processor; and

[0067] a memory communicatively connected to the at least one processor; wherein,

[0068] The memory stores instructions that can be executed by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor executes the aforementioned industrial product carbon footprint analysis method based on the energy efficiency identification database.

[0069] In a third aspect, an embodiment of the present invention further provides a computer program product, which includes a computer program / instructions, and when the computer program / instructions are executed by a processor, the aforementioned industrial product carbon footprint analysis method based on the energy efficiency identification database is implemented.

[0070] Under the premise of fully understanding and evaluating my country's mandatory energy efficiency standards and energy consumption limit standards, the present invention makes full use of the energy efficiency label database data to predict and analyze carbon emission intensity and carbon emission influencing factors. The results can be used for carbon footprint calculation and carbon footprint prediction. In terms of carbon footprint calculation, the calculation of the corresponding data can be completed by comparing the existing influencing factor prediction results with the existing carbon footprint analysis report. In terms of carbon footprint prediction, the existing emission intensity can be combined with big data scenarios for analysis to complete future predictions. The results are very beneficial to carbon emission monitoring and the country's formulation of corresponding policy deployments, as well as the improvement of energy consumption and energy efficiency standards. It can be applied to the measurement process for relevant research institutions and testing institutions conducting carbon research. The calculation process is simple. The analysis using the energy efficiency label registration database can simplify the calculation process and facilitate data capture and updating throughout the life cycle. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The above is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0072] Figure 1 A simplified diagram of the carbon footprint distribution of products throughout their life cycle;

[0073] Figure 2 This is a schematic diagram of carbon footprint analysis based on energy efficiency labeling database;

[0074] Figure 3 This is the uncorrected BP product carbon footprint prediction algorithm model;

[0075] Figure 4 This is the revised BP product carbon footprint prediction algorithm model;

[0076] Figure 5 This is the product carbon footprint BP neural network prediction model diagram;

[0077] Figure 6 It is a BP neural network prediction model modified by using emission benefits and coefficient of variation. DETAILED DESCRIPTION

[0078] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0079] In the following description, specific details are provided to facilitate a thorough understanding of the examples. However, one skilled in the art will appreciate that the aspects described may be practiced without these specific details.

[0080] The industrial product carbon footprint analysis method provided by the embodiment of the present invention is an analysis method based on an energy efficiency label database. For ease of understanding, the current status of my country's energy efficiency, energy consumption standards and energy efficiency labeling is introduced below:

[0081] (1) Current status of energy efficiency standards

[0082] Energy efficiency standards mainly measure the energy utilization efficiency of products and ensure that products are used efficiently.

[0083] (2) Energy consumption limit standards

[0084] The energy consumption limit standard stipulates the energy consumption limit of the system under specific conditions to ensure the overall energy efficiency of the system in the process of producing products. Through the energy consumption limit standard limit value and access value, enterprises are forced to eliminate backward production capacity, backward process and backward energy-consuming products and equipment; through the advanced value of the energy consumption limit standard, enterprises are guided to follow the benchmark, optimize the process technology route, implement energy-saving transformation to improve the energy efficiency of production and operation, and promote the green transformation of enterprises. It further promotes the supply and application of high-efficiency energy-saving products and advanced energy-saving and carbon-reduction technologies, and lays a solid foundation for energy-saving and carbon-reduction renovation.

[0085] (3) Current status of energy efficiency labeling

[0086] Energy efficiency label, also known as energy efficiency label, is an information label attached to energy-consuming products or their smallest packaging, indicating the product's energy efficiency level and other performance indicators. The purpose is to provide users and consumers with necessary information for purchasing decisions, so as to guide and help consumers choose high-efficiency and energy-saving products. All products included in the catalog must be registered on the China Energy Efficiency Label Network and affixed with the corresponding energy efficiency label in accordance with the format specified in the new version of the "Energy Efficiency Label Management Measures". Currently, 41 categories of products are included, including almost all industrial products. The latest version of the label contains a QR code area. Consumers and law enforcement agencies can scan the QR code to enter the energy efficiency label information platform, obtain the energy efficiency information of energy-consuming products, energy efficiency registration number, product energy efficiency quality spot check, and verify product quality. At the same time, the platform integrates information such as raw materials and manufacturers, which is convenient for tracing back; for industrial products, energy efficiency label information needs to be presented when bidding, and user information will be recorded after scanning.

[0087] The above energy efficiency standards are mainly used to measure the energy efficiency of products to ensure that the products are used efficiently; the energy consumption limit standards stipulate the energy consumption limits of the system producing products under specific conditions to ensure the overall energy efficiency of the system in the process of producing products. Although the two types of standards have the same goals, they are to improve energy utilization, reduce resource and energy waste, and reduce environmental impact. However, in reality, the two types of standards are basically unrelated to ensure that they do not have unrealistic impacts in different fields. To achieve the combination of the two and effectively promote energy conservation and emission reduction, it is necessary to introduce a full life cycle analysis, use energy consumption limit standards to restrict the raw material production stage, introduce the application of energy efficiency standards in product design, and use the idea of ​​recycling in the subsequent use process. Coordination is achieved from product design, manufacturing, transportation, use and post-processing.

[0088] The QR code information of the above energy efficiency label includes product specifications, product types, product testing methods, energy efficiency levels, manufacturer information, and raw material information. During the bidding process, once the enterprise scans the QR code, it can record the information of the application enterprise, understand the purpose, subsequent recycling process, and remanufacturing process. After scanning the code, relevant information will be recorded to understand the subsequent destination, which basically covers the entire life cycle of the product. The relevant data are also generated in real time or on-site, with high credibility and real-time update. However, this part of the data has not been directly used in the carbon footprint calculation and has not been well combined.

[0089] In the process of carbon footprint analysis, there are many links and many specific influencing factors. In the raw material production stage, mining and extraction require energy, and transportation requires energy, especially long-distance transportation, which leads to higher greenhouse gas emissions. In the product production stage, electricity, fuel and other energy are used in large quantities, and different chemicals and materials are also affected by the manufacturing process and the environment, producing greenhouse gases. In the product use stage, maintenance and daily use will generate carbon emissions. In the recycling stage, although recycling is a means to reduce carbon emissions, the waste treatment and recycling process in the process are energy-consuming processes. If there is incineration or landfill, greenhouse gases will also be released. Therefore, how to combine the existing energy efficiency labeling data at each stage, conduct big data analysis, find out the unified common influencing factors, and make calculations more convenient, has not been analyzed much at present.

[0090] The present invention's method for analyzing the carbon footprint of industrial products based on the energy efficiency label database still has the following difficulties:

[0091] (1) Carbon footprint calculation model for products with different energy efficiency levels

[0092] The energy efficiency label database has many data parameters with low statistical correlation, including: production information, raw material information, etc., and the origin information must also be analyzed to ensure carbon emissions during transportation. If only the carbon emission coefficient method is used to calculate carbon emission data under certain data, there may also be problems such as imperfect emission factors, incomplete testing, and untimely data updates. If only the mass balance method is used to calculate the total gas emissions of the enterprise based on the enterprise's production process, gas emissions during the production process, and corresponding environmental governance, it is mainly calculated for emissions in the industry and the enterprise, and cannot be refined to the product level. Therefore, when considering the complex weakly correlated energy efficiency label filing data parameters, it is one of the difficulties of the present invention to use the actual updated data and enterprise industry emission data to comprehensively calculate the carbon footprint calculation method and model of industrial products.

[0093] (2) Predicting the carbon footprint of a product over its entire life cycle using product energy efficiency

[0094] Since the energy efficiency identification database has a large amount of data and small correlation, in order to meet the needs of fast and accurate prediction, it is necessary to do data consistency screening in advance. How to use the existing energy efficiency data and the macro and meso data of carbon emission intensity in previous years to complete the screening of product carbon footprint data is the focus and difficulty of the subsequent BP neural network model. If the consistency of the data is low, the BP neural network return data considering the dynamic model cannot verify the results, which will lead to erroneous results in subsequent predictions. Therefore, after considering the dynamic weighting factor of the coefficient of variation, the calculation method of efficient convergence that maintains robustness is one of the difficulties of the present invention.

[0095] The overall design concept of the present invention is:

[0096] The energy efficiency label database information formed by the energy efficiency standards and energy consumption limit standards contains information such as product energy efficiency and energy consumption. Therefore, according to the energy efficiency and energy consumption related to the product, the carbon footprint of the entire product life cycle can be linked together, such as Figure 1 As shown in the figure, the green part is the carbon compensation part, and the light green part is the partial carbon compensation part. The remanufacturing process uses recycled products, so it is considered as the compensation part. The black solid line part is the carbon emission process, the black dotted line part is the possible carbon emission process, the green solid line is the part that can be directly applied to the carbon emission calculation in the energy efficiency and energy consumption database, and the green dotted line is the part that can be traced to carbon emissions or carbon emissions can be checked through data. The energy efficiency and energy consumption databases can be compared with each other through carbon emissions and overall emissions over the years for possible subsequent screening.

[0097] The consideration for making this invention is that, since the carbon emission intensity data of a single product in its entire life cycle is difficult to obtain through one-to-one measurement, the calculation of the product's carbon footprint remains in the analysis field after the scene setting; a large number of carbon footprint verifications and applications are carbon emission assessments for industries, enterprises or macro performance. Although this method is beneficial for the country to use macro data when formulating an overall strategy, it is difficult to obtain detailed data for micro products, especially industrial products with large energy consumption and carbon emissions, and it is impossible to guide the application of certain types of products in industries and enterprises, as well as energy efficiency improvement and carbon emission reduction from the details.

[0098] Based on the above analysis, in view of the large amount of public data retained in my country's energy efficiency labeling management system and the complete energy efficiency standards and energy consumption limit standard system, the present invention uses the energy utilization data in the labeling database, as well as upstream raw material information, downstream user information and the registration ratio of each energy efficiency product, and selects the industrial product field with a short life cycle stage, traceable raw material carbon emission data and clear use as the entry point. Through energy efficiency-related content, combined with the carbon emission statistics of previous years, the product emission benefits are indirectly inferred, and combined with the actual observation values, a dynamic product carbon footprint database is formed, and the model is trained to realize the carbon footprint prediction of the same type of products, thereby realizing a product carbon footprint calculation method based on existing energy efficiency labeling data.

[0099] The present invention can not only predict future carbon emissions based on existing data, but also make accurate predictions and feedback on the future carbon emissions and energy efficiency and consumption indicators of a certain type of product after decomposition according to policy requirements. This can not only better promote product carbon emission reduction, but also effectively guide the updating of standards.

[0100] like Figure 2As shown, the industrial product carbon footprint analysis method in the embodiment of the present invention mainly reads the information of energy efficiency level, energy efficiency index, energy consumption limit index and other information through the information of energy efficiency identification database, divides the total carbon emission data of previous years by quantitative analysis, and uses the logarithmic mean index method to combine the above data to split various factors according to industrial composition, energy use and economic level to obtain the carbon emission benefits of the product; the statistical standard deviation and mean ratio are used to analyze the coefficient of variation, and the significance of the variation between the product carbon emission benefits and the observed values ​​is obtained, and the global (macro) and local (meso) key parameters are screened out. The emission factors used in the carbon emission coefficient method are corrected by the product carbon emission benefits, and the carbon emissions in each stage of the actual life cycle of the product are obtained, and then the overall carbon emissions of the enterprise after the coefficient of variation is corrected are weighted to obtain the carbon footprint results of the product during the entire life cycle.

[0101] By comparing the existing energy efficiency data with the macro and meso data of carbon emission intensity in previous years, combined with the calculated results of product carbon emissions, the grayscale model is used for screening and the results are tested for cointegration, and the single integer sequence energy efficiency and energy consumption parameters of the normalized indicator variables related to carbon emissions are obtained, which are brought into the BP neural network model for calculation. In the weighting process of the middle layer of the network, the fixed weighting factor is optimized, and the influence of the coefficient of variation is incorporated to form a dynamic weighting factor to calculate the changing environmental parameters and energy efficiency indicator data. Since the lowest carbon emissions throughout the life cycle must be considered, and the highest energy efficiency of the product itself must be guaranteed, the convergence criterion is set to the lowest carbon emission value and the highest energy efficiency value. To ensure high-speed convergence and stability of computational convergence, the present invention uses a combination of global optimization and local optimization. The carbon emissions of the product with the highest energy efficiency are first found through global optimization, and then the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method. The best product carbon emissions are found to complete the model prediction.

[0102] The following is a detailed description of the product carbon emission calculation and carbon emission prediction in the industrial product carbon footprint analysis method based on the energy efficiency labeling database of the present invention in conjunction with the accompanying drawings:

[0103] 1. Calculation of product carbon emissions:

[0104] This embodiment provides an industrial product carbon footprint analysis method based on an energy efficiency label database, which is used to calculate the carbon emissions of products, including:

[0105] Read the energy efficiency and energy consumption related data from the energy efficiency label database; analyze and obtain the total carbon emission data over the years; use the index decomposition method to split the total carbon emission data over the years to obtain the total carbon emission of the product, and calculate the carbon emissions of the raw materials stage, product production stage, product application stage and product recycling stage in the entire life cycle of the product, and use the ratio of the carbon emissions of each stage to the total carbon emissions of the product as the carbon emission benefits of the product at different stages.

[0106] The coefficient of variation is analyzed by using the statistical standard deviation and mean ratio to obtain the significance of the variation between the carbon emission benefits of the product at different stages and the actual observed values, and to screen out the global and local key parameters;

[0107] According to key parameters, relevant emission factors are searched through the IPCC database, and the emission factors are corrected using the carbon emission benefits of the product at different stages to obtain the actual carbon emissions in each stage of the product's life cycle. The carbon emissions of the company as a whole after correction of the coefficient of variation are then weighted to obtain the carbon emissions during the entire life cycle of the product.

[0108] More specifically:

[0109] 1.1 Index decomposition to obtain the carbon emission benefits of different stages of the product

[0110] The index decomposition method is to decompose the abstract research object into specific influencing factors, determine the method according to different weights, and then determine the contribution of each factor to the research object. After reading the information in the energy efficiency label database, it is divided into two layers for analysis. The first layer is to divide the energy efficiency and energy consumption data into the total carbon emissions of the product. The second layer is to divide the product into different stages of the life cycle and examine the carbon emissions at different stages, so as to obtain the carbon emission benefits of the product at different stages.

[0111] First-level calculation: After reading the data, the total energy and total energy intensity are calculated using the addition model.

[0112]

[0113] In the formula, E is energy consumption, which is a quantity indicator; V is energy intensity, which is an intensity indicator; Q is output; n represents the number of corresponding energy efficiency and energy consumption indicators from the database; S is the whole life cycle stage; energy intensity includes energy consumption, and energy consumption includes the impact of user use; I represents the energy consumption per unit product, I n Represents the unit product energy consumption corresponding to the nth indicator of energy efficiency and energy consumption in the database.

[0114] Using the above results, we can modify the Kaya formula. The original Kaya formula was only a macro analysis of GDP, and the formula is as follows:

[0115]

[0116] Among them, C is carbon emissions, P is population, G is gross national product, and E is energy consumption.

[0117] The above formula is improved to analyze the carbon emissions of products at different life cycle stages.

[0118]

[0119] Based on the above formula, the energy intensity V of a single product i in a single stage j in the life cycle can be analyzed: i,j , at this stage the output Q j The carbon emission in this case is C i,j ; In the formula, C is the total carbon emissions, S is the whole life cycle stage, Q is the output, and V is the energy intensity; C i,j is the carbon emission of a single product i in a single stage j of its life cycle, S i represents the life cycle stage of a single product i, Q j represents the output of a single stage j in the life cycle, V i,j It represents the energy intensity of a single product i in a single stage j during its life cycle;

[0120] Through the above calculations, the key emissions of the product and the emissions in each stage of its life cycle can be obtained, and the ratio is used as the carbon emission benefit parameter.

[0121] It should be noted here that the carbon emissions calculation for the raw material stage, product production stage, product application stage and product recycling stage in the product life cycle includes:

[0122] 1) Raw materials stage: Use the energy efficiency data in the database to analyze the proportion of raw materials and conduct data analysis and summary of carbon emissions from the transportation and production processes of raw materials;

[0123] 2) Product production stage: Based on the energy consumption limit standard requirements, the consumption of water, electricity, gas and coal of industrial products is considered, and the carbon emissions in the production stage are determined according to the production process and energy efficiency indicators, and data analysis and summary are conducted;

[0124] 3) Product application stage: Combine product energy efficiency data and product application scenarios to analyze and summarize data on carbon emissions generated during the application process and carbon emissions during transportation, repair, and maintenance;

[0125] 4) Product recycling stage: Utilize the renewable raw material information in the energy efficiency label database to analyze and summarize the carbon emissions during the product remanufacturing and recycling process.

[0126] 1.2 Coefficient of variation analysis

[0127] The coefficient of variation analysis is expressed by standard deviation and mean, and the formula is:

[0128] ρ=σμ

[0129] Where ρ is the coefficient of variation, σ is the standard deviation of carbon emissions calculated from existing historical data, and μ is the expected value of carbon emissions calculated from existing historical data. The significance of the variation between the product carbon emission benefits and the observed values ​​is obtained. If the coefficient of variation changes beyond the set interval, it means that there is a significant difference between the standard deviation and the expected value, thereby screening out the key parameters of the industry (macro) and the whole life cycle process stage (local) contained in the historical data.

[0130] 1.3 Revised IPCC and Mass Balance Approach

[0131] According to the key parameters, the relevant emission factors can be found through the IPCC database. Since the emphasis of the parameters is different, the weighting coefficients are different. Taking into account the different carbon emission benefits in each stage, the present invention combines the carbon emission benefits with the emission factors as a correction to the different weighting coefficients to obtain the carbon emissions in each stage of the actual life cycle of the product, and then weights the overall carbon emissions of the enterprise after the coefficient of variation is corrected to obtain the carbon footprint result of the product during its entire life cycle.

[0132]

[0133] δ=ρ i δ i

[0134] τ i =C j / C

[0135] where δ is the emission factor, δ i is the emission factor of a single product i; ρ i is the coefficient of variation of a single product i; τ is the carbon emission benefit, τ i The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is: E is energy consumption, C is the total carbon emission, and C j is the carbon emission in a single stage j in the life cycle, E i is the energy consumption of a single product i, Q is the output, Q i represents the output of a single product i, I i Represents the unit product energy consumption of a single product i.

[0136] 2. Prediction of product carbon emissions

[0137] This embodiment provides an industrial product carbon footprint analysis method based on an energy efficiency label database, which is further used to predict product carbon emissions, including:

[0138] Data screening and normalization:

[0139] The grayscale model is used to screen and the results are tested for cointegration, and the single integer series energy efficiency and energy consumption parameters of the normalized indicator variables related to carbon emissions are obtained as normalized data;

[0140] Construct a modified BP neural network prediction model:

[0141] The hidden layer of the BP neural network prediction model is simplified, and the output of the raw material stage in the product life cycle is used as the input of the product production stage, the output of the product production stage is used as the input of the product application stage, and the output of the product application stage is used as the input of the product recycling stage;

[0142] A linear model of a multi-layer perceptron is added to the hidden layer of the BP neural network prediction model. Normalized data is used as prediction pooling data. The lowest carbon emissions in the entire life cycle and the highest energy efficiency of the product itself are considered. The carbon emissions of the product with the highest energy efficiency are first found through global optimization. Then, the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method to find the best product with the best carbon emissions.

[0143] Then the optimized data is passed into the modified BP neural network prediction model to be solved in space;

[0144] Carbon emissions forecast:

[0145] The modified BP neural network prediction model is used to predict product carbon emissions.

[0146] More specifically:

[0147] 2.1 Data screening and normalization

[0148] Data normalization is currently widely used in the evaluation parameter indicators of analytical models. The purpose is to eliminate the defects of different data units. After standardization, each data becomes a pure value without unit, which is convenient for subsequent comparative analysis of various indicators. In the process of data processing, the relevant data needs to be converted accordingly. The unit of energy consumption is 10,000 tons of standard coal, so the unit needs to be converted during calculation. Standard coal is also called coal equivalent, and the calorific value has the same standard value. The coefficients used in this invention are all from the IPCC guideline document to facilitate everyone's subsequent use.

[0149] The present invention compares the existing energy efficiency data with the macro and meso data of carbon emission intensity over the years, combines the calculated results of product carbon emissions, uses the gray model to screen and perform cointegration test on the results, and obtains the single integer sequence energy efficiency and energy consumption parameters of the normalized indicator variables related to carbon emissions. The details are as follows:

[0150]

[0151] In the formula, ii represents the product category and j represents a single stage in the life cycle; It represents the normalized value of energy consumption of the product of category ii in a single stage j during its life cycle, E ii,j is the energy consumption of a single stage j in the life cycle of a product of category ii, max E ii,j , min E ii,j Indicates the maximum and minimum values ​​of energy consumption in this data class, V represents the normalized value of energy consumption of the product of category II in a single stage j during its life cycle. ii,j is the energy intensity of a single phase j of the life cycle of a product of category ii, maxV ii,j， minV ii,j Indicates the maximum and minimum values ​​of energy intensity in this data class;

[0152] Based on the above expression, the energy efficiency and energy consumption data are screened, and then the same root unit root test and different root unit root test are used to analyze and obtain the first-order single integer series. After screening by the grayscale model, the data in the unstable transition interval are abandoned for data consistency cointegration check, and finally the normalized single integer series data is obtained for subsequent prediction of pooled data.

[0153] 2.2 Constructing the modified BP neural network prediction model (1):

[0154] Early artificial neural networks used neurons as units to integrate and abstract data to establish models. The nodes and weights between nodes were repeatedly revised to obtain the final prediction results. However, it is still difficult to make accurate predictions for complex problems with multiple hidden layers. In order to solve the shortcomings of traditional neural network models, such as inflexible structure, insensitivity to time series data, and easy to fall into local minimum values, the BP model is used as the basic model in the present invention and the model is transformed.

[0155] In the normal calculation process, the function of the neuron is first recorded based on the Cell state, and the update information is input at the same time. The forward and reverse propagation calculations are performed respectively, and the weights and error biases are considered multiple times to obtain the final prediction value. The calculation is complex and slow, and it is not applicable to the decomposition of the entire life cycle and a large number of cyclic calculations. In order to ensure fast calculations at each stage, the hidden layer is simplified in the present invention.

[0156] For a given life cycle sequence x = (x1, x2, ..., x n ) After applying the standard iteration, the hidden layer sequence h = (h1, h, ..., h n ) and the output sequence y = (y1, y2, ..., y n ) are:

[0157] h t =f n (W xh x t +W hh x t-1 +b h )

[0158] y t =W hy h t +b y

[0159] Where: W is the weight coefficient matrix, b is the bias vector, f n is the activation function, and the subscript t is the product life cycle stage.

[0160] Raw material stage status: i t =σ(W xi x t +W hi h t-1 +W ci c i-1 +b i )

[0161] Manufacturing stage status: f t =σ(W xf x t +W hf h t-1 +W cf c t-1 +b f )

[0162] Application phase status: c t =f t c t-1 +i t tanh(W xc x t +W hc h t-1 +b c )

[0163] Activation function status: h t =o t tanh(c t )

[0164] The corresponding status is Figure 3 shown.

[0165] In order to ensure the calculation speed, the state of each gate is adjusted, and the carbon emission output in a single state is not considered, but the emission output in the whole life cycle is considered. That is to say, in the calculation of cel, no positive and negative verification is performed on a single stage, and no parameter adjustment is made, so as to maximize the data transmission. Considering that the retained content may be different according to the number of stages, resulting in different calculation layers, the cell state activation function is optimized for the retained part in the following way, thereby bringing more carbon emission parameters of the upper and lower stages into the calculation.

[0166] c t =tanh(W c (h t-1 , x t )+b c )

[0167] c t =c t-1 f t +ra t c t

[0168] It can be seen from the above formula that although different stages have discrete characteristics, the previous neuron can be used to the maximum extent to update the current cell state and reduce the amount of calculation. When a single product circulates in different life cycle stages, only the current calculation moment is considered, thereby achieving maximum information transmission. That is, the output of the raw material stage in the entire life cycle of the product is used as the input of the product production stage, the output of the product production stage is used as the input of the product application stage, and the output of the product application stage is used as the input of the product recycling stage. This form of cell state corresponds to the state diagram as follows Figure 4 As shown by Figure 4 It can be seen that it is obviously much simpler.

[0169] Along with the calculation of cell state changes, the expressions of other gates evolve as follows:

[0170] ra t =σ(W x，ra (h t-1 , x t )+b ra )

[0171] pr t =σ(W x，pr (h t-1 , x pr )+b pr )

[0172] ap t =σ(W x，ap (h t-1, x ap )+b ap )

[0173] In the above gate expression and activation function h t =ap t tanh(c t ), W is the weight coefficient matrix, b is the bias vector, h is the activation function, c is the hidden layer sequence variable calculation variable, t is the calculation step length, x is the life cycle sequence variable, ra, pr, ap are the raw materials, production, and application stages in the product life cycle, ra t ,pr t ,ap t represents the calculation output of the raw materials, production, and application stages at time t, h t and h t-1 is the activation function output at two moments before and after, x t is the output of the life cycle sequence variable at time t, x pr , x ap represents the life cycle sequence variable value in the production and application stages, c t and c t-1 Calculate the value of the hidden layer sequence variable for the two moments before and after, W c is the weight coefficient matrix of the hidden layer sequence variables, W x,ra ,W x,pr ,W x,ap is the weight coefficient matrix of the life cycle sequence variables of raw materials, production and application stages, and σ is the cumulative function.

[0174] The transfer process still uses the tanh function, so h t It can also be seen from the expression that the number of accumulation and multiplication is reduced after the improvement, and the output part chooses to retain each state and cell state, and accumulates the entire life cycle process through the σ function. When the requirements are met, the relevant information is passed down to improve the efficiency of each calculation.

[0175] The calculation diagram is as follows Figure 5 As shown, the solid line is the current operation i, and the dotted line is the next operation unit i+1, until the loop meets the conditions and jumps out of the loop.

[0176] 2.3 Constructing a modified BP neural network prediction model (2) - Carbon footprint model prediction considering dynamic weighting and energy efficiency index changes

[0177] In the present invention, a linear model of a multi-layer perceptron (MLP) is considered to be added to the hidden layer. In order to minimize the prediction loss, normalized single integer sequence data is used as the prediction pooling data, and the carbon emissions of the entire life cycle are minimized and the energy efficiency of the product itself is maximized. The above-mentioned tanh function is used to optimize the minimum error sum of sample N. The optimization algorithm is as follows:

[0178]

[0179] In the formula, y i is the calculated output value at time m, st is the minimum optimization step length, v m is the correction factor at the mth moment, f is the calculated output value at the m-1th moment, and N is the total calculation time;

[0180] Adding a multi-layer perceptron (MLP) and using a particle swarm algorithm for error sum optimization aims to classify and clean the data to obtain the best of the same kind, and then pass the optimized data into the model for solving in space. The optimal clustering method minimizes the objective function value of each pattern to cluster analysis, and aims to achieve the lowest energy consumption and carbon emission values. The mathematical model of energy efficiency evaluation is expressed as:

[0181]

[0182] In the formula, E is energy consumption, C is total carbon emissions, ep represents the pth product attribute that affects energy efficiency and carbon emissions in the eth stage, g op is the pth attribute of the oth domain center; ω oe is the domain center and stage attribute judgment value. It takes 1 when the domain center and stage value of the product attribute belong to the same category, and takes 0 when they are not of the same category. The subscript o is the number of domain centers of the attribute, ranging from 1 to n. o , subscript e is the number of stages of the attribute, ranging from 1 to n e ; The subscript p is the number of attributes, ranging from 1 to n; n, n o , n e They are the number of corresponding energy efficiency and energy consumption indicators, the number of attribute domain centers, and the number of attribute stages in the database;

[0183] According to the above expression, local and global searches are performed in the domain, the cluster center and objective function value are calculated for each variable, and the result set is transformed. The influence of emission benefits and coefficient of variation is considered, and the matrix is ​​updated as follows:

[0184] τ i (t+Δ)=(1-ρ)τ i (t)+βτ i

[0185] Where τ is the carbon emission benefit, τi The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is the output value of the BP network; t is the solution time, Δ is the solution step, ρ is the coefficient of variation, βτ i It is the carbon emission benefit variable in the BP network calculation process.

[0186] The BP neural network prediction model modified by emission benefits and coefficient of variation is as follows: Figure 6 shown.

[0187] Since BP back propagation also needs to satisfy the gradient change and maintain a certain correlation with the computing unit and data, you should pay attention to the following when selecting data:

[0188] (1) After data cleaning, it needs to remain discrete but continuous in time series, which means that the product life cycle process is complete, not that there is a gap due to lack of data in the middle. That is, the loss function must satisfy the partial derivatives of the error and weight matrix at each time point and satisfy the gradient descent after residual transmission.

[0189]

[0190]

[0191] (2) Before data is input, it must be processed by a particle swarm algorithm. The particle swarm algorithm will distinguish different categories and find the optimal values ​​of different categories. However, if the particle swarm algorithm cannot successfully find the optimal value locally and globally, then the data of this type needs to be re-cleaned. The present invention uses the moving average method to process the singular point at time t in a time series data. Because these data come from real data, but the magnitude or index difference is too large, it will lead to outliers. In order to take them all into account, use x aver To replace the singular values, the singular values ​​not only contain individual information, but also the information of the previous and next time points, as well as local and global data. Their influence will be gradually reflected in the calculation process.

[0192]

[0193] Where b represents the time of the abnormal sampling point.

[0194] If you do not have full life cycle data but want to calculate, you need to skip a reasonable step size and perform a global check on the data, or manually enter the gap value to ensure data smoothness. i There are missing data in the data set, which are smoothed using the following method.

[0195]

[0196] Among them, step is the step value (deprecation time), x t+step represents the predicted value of the missing part, a t b t c t are smoothing parameters respectively. Derivatives of different orders in the formula represent different smoothing values, and α is the weight function.

[0197] In addition, an embodiment of the present invention also provides an industrial product carbon footprint analysis system based on an energy efficiency identification database, comprising at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions that can be executed by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor executes the above-mentioned industrial product carbon footprint analysis method based on the energy efficiency identification database. No further details will be given here.

[0198] Furthermore, an embodiment of the present invention further provides a computer program product, which includes a computer program / instructions, and when the computer program / instructions are executed by a processor, the above-mentioned industrial product carbon footprint analysis method based on the energy efficiency identification database is implemented.

[0199] In summary, the present invention realizes the calculation and prediction of carbon emissions of industrial products based on the energy efficiency representation database.

[0200] In the calculation of carbon emissions, based on the information of the energy efficiency label database, the full life cycle process analysis method is used to analyze the following: 1) Raw material side: Use the energy efficiency situation in the database to analyze the proportion of raw materials and analyze and summarize the carbon emissions of the transportation process and production process of raw materials. 2) Product production side: Based on the energy consumption limit standard limit value requirements, consider the various consumptions of water, electricity, gas, and coal of industrial products, and determine the carbon emissions of its production stage according to the production process and energy efficiency indicators, and analyze and summarize the data. 3) Product application side: Combined with product energy efficiency data and product application scenarios, analyze and summarize the carbon emissions generated during the application process and the carbon emissions during transportation, repair, and maintenance. 4) Product recycling side: Use the renewable raw material information in the energy efficiency label database to summarize the carbon emissions during product remanufacturing and recycling. Based on the above data, quantitative analysis and coefficient of variation difference analysis are used, and the improved carbon emission coefficient method and mass balance method are used in combination. Combined with the correlation of the energy efficiency level proportion of energy efficiency label filing data and the differences in the physical space of the production area, the carbon emission intensity of industrial products is analyzed to form a carbon footprint analysis.

[0201] In the prediction of carbon emissions: according to the energy efficiency and energy consumption data of the energy efficiency label, the grayscale model is used to compare with the carbon emission intensity of previous years. After the cointegration test, the single integer sequence energy efficiency and energy consumption parameters with consistent indicator variables are screened out, and the dynamic weighting factor of the coefficient of variation is considered. The optimized BP neural network model is used to predict the carbon footprint, and the change analysis and prediction statistics of the carbon footprint of industrial products are obtained.

[0202] The industrial product carbon footprint analysis method and system based on the energy efficiency label database of the present invention have strong application and implementation potential, which is reflected in:

[0203] (1) Application scenarios

[0204] Future development is inseparable from the real economy and industry. Industry is still a strong point to boost economic growth. Obviously, the increase in total carbon emissions brought by the industrial sector is inevitable. Although industrial energy efficiency is gradually improving, the indicators of energy efficiency standards and energy consumption limit standards are also increasing year by year, and the management of energy efficiency labels is also progressing steadily, there is almost no connection between the two types of standards, and the large amount of information recorded by energy efficiency labels cannot be widely used. To ensure that the low-carbon goals are achieved in my country's industrial sector, it is necessary to deepen the development of deep emission reduction work on the basis of the foundation.

[0205] Under the premise of fully understanding and evaluating my country's mandatory energy efficiency standards and energy consumption limit standards, the present invention makes full use of the data of the energy efficiency label management system to predict and analyze carbon emission intensity and carbon emission influencing factors. The results can be used for carbon footprint accounting and carbon footprint prediction. In terms of carbon footprint accounting, the corresponding data can be calculated by comparing the existing influencing factor prediction results with the existing carbon footprint analysis report. In terms of carbon footprint prediction, the existing emission intensity can be combined with big data scenarios for analysis to complete future predictions. The results are very beneficial to carbon emission monitoring and the country's formulation of corresponding policy deployments, as well as the improvement of energy consumption and energy efficiency standards. It can be applied to the measurement process for relevant research institutions and testing institutions conducting carbon research. The calculation process is simple. The analysis is performed using the energy efficiency label filing database, which can simplify the calculation process and facilitate data capture and updating throughout the life cycle.

[0206] (2) Potential for energy conservation and emission reduction

[0207] The industrial sector is a major energy consumer and carbon emitter in China. Over the years, the proportion of energy consumption in the industrial sector in the country has always been higher than 65%, and the proportion of carbon emissions in the total has exceeded 70%, making it a key area for addressing climate change. Due to the complexity of industrial processes and the high cost of carbon reduction, industry has always been widely regarded as a difficult area for carbon reduction. With the country's emphasis on energy conservation and emission reduction, advanced technologies that can save energy or reduce carbon dioxide emissions have gradually been put into use by enterprises. Analyzing the energy conservation and emission reduction potential of advanced technologies is of great significance for enterprises to apply technology.

[0208] The present invention studies specific products and industries in the industrial field, measures the energy-saving and emission-reduction potential and cost generated by the use of specific technologies, analyzes the energy-saving and emission-reduction potential and cost of advanced technologies in specific industries, and provides effective data support for enterprises to invest in technology use.

[0209] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A method for analyzing the carbon footprint of industrial products based on an energy efficiency label database, characterized in that: Used for calculating product carbon emissions, including: Read the energy efficiency and energy consumption related data from the energy efficiency label database; analyze and obtain the total carbon emission data over the years; use the index decomposition method to split the total carbon emission data over the years to obtain the total carbon emission of the product, and calculate the carbon emissions of the raw material stage, product production stage, product application stage and product recycling stage in the whole life cycle of the product, and use the ratio of the carbon emissions of each stage to the total carbon emissions of the product as the carbon emission benefits of the product at different stages; The coefficient of variation is analyzed by using the statistical standard deviation and mean ratio to obtain the significance of the variation between the carbon emission benefits of the product at different stages and the actual observed values, and to screen out the global and local key parameters; According to key parameters, relevant emission factors are searched through the IPCC database, and the emission factors are corrected using the carbon emission benefits of different stages of the product to obtain the carbon emissions in each stage of the product's actual life cycle; then the carbon emissions of the enterprise are weighted in combination with the corrected coefficient of variation to obtain the carbon emissions during the entire life cycle of the product; The calculation method of carbon emission benefits of different stages of the product includes: After reading the energy efficiency and energy consumption data from the energy efficiency label database, the total energy and energy intensity are calculated using the addition model. The formula is as follows: In the formula, E is energy consumption, which is a quantity indicator; V is energy intensity, which is an intensity indicator; Q is output; n represents the number of corresponding energy efficiency and energy consumption indicators from the database; S is the whole life cycle stage; energy intensity includes energy consumption, and energy consumption includes the impact of user use; I represents the energy consumption per unit product, I n Indicates the unit product energy consumption corresponding to the nth indicator of energy efficiency and energy consumption in the database; The following formula is used to analyze the carbon emissions of products at different life cycle stages; Based on the above formula, the energy intensity V of a single product i in a single stage j in the life cycle can be analyzed: i,j , at this stage the output Q j The carbon emission in this case is C i,j ; In the formula, C is the total carbon emissions, S is the whole life cycle stage, Q is the output, and V is the energy intensity; C i,j is the carbon emission of a single product i in a single stage j of its life cycle, S i represents the life cycle stage of a single product i, Q j represents the output of a single stage j in the life cycle, V i,j It represents the energy intensity of a single product i in a single stage j during its life cycle; Based on the above calculations, the total carbon emissions of the product and the carbon emissions at each stage of the life cycle can be obtained, and the ratio is used as the carbon emission benefit of the product at different stages; The carbon emission benefits of different stages of the product are used to correct the emission factor to obtain the carbon emissions in each stage of the product's actual life cycle; then combined with the overall carbon emissions of the enterprise after the coefficient of variation is corrected for weighting, the calculation formula for carbon emissions during the entire life cycle of the product is as follows: d=r i d i t i =C j / C Where δ is the emission factor, δ i is the emission factor of a single product i; ρ i is the coefficient of variation of a single product i; τ is the carbon emission benefit, τ i The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is: E is energy consumption, C is the total carbon emission, and C j is the carbon emission in a single phase j in the life cycle, E i is the energy consumption of a single product i, Q is the output, Q i represents the output of a single product i, I i Represents the unit product energy consumption of a single product i.

2. The industrial product carbon footprint analysis method based on energy efficiency labeling database according to claim 1 is characterized in that: Carbon emissions calculations are performed for the raw material stage, product production stage, product application stage, and product recycling stage during the product life cycle, including: 1) Raw materials stage: Use the energy efficiency data in the energy efficiency label database to analyze the proportion of raw materials and conduct data analysis and summary of carbon emissions from the transportation and production processes of raw materials; 2) Product production stage: Based on the energy consumption limit standard requirements, the consumption of water, electricity, gas and coal of industrial products is considered, and the carbon emissions in the production stage are determined according to the production process and energy efficiency indicators, and data analysis and summary are conducted; 3) Product application stage: Combine product energy efficiency data and product application scenarios to analyze and summarize data on carbon emissions generated during the application process and carbon emissions during transportation, repair, and maintenance; 4) Product recycling stage: Utilize the renewable raw material information in the energy efficiency label database to analyze and summarize the carbon emissions during the product remanufacturing and recycling process.

3. The industrial product carbon footprint analysis method based on the energy efficiency labeling database according to claim 1 or 2, characterized in that: It is also used to predict product carbon emissions, including: Data screening and normalization: The grayscale model is used to screen and the results are tested for cointegration, and the single integer series energy efficiency and energy consumption parameters of the normalized indicator variables related to carbon emissions are obtained as normalized data; Construct a modified BP neural network prediction model: The hidden layer of the BP neural network prediction model is simplified, and the output of the raw material stage in the product life cycle is used as the input of the product production stage, the output of the product production stage is used as the input of the product application stage, and the output of the product application stage is used as the input of the product recycling stage; A linear model of a multi-layer perceptron is added to the hidden layer of the BP neural network prediction model. Normalized data is used as prediction pooling data. The lowest carbon emissions in the entire life cycle and the highest energy efficiency of the product itself are considered. The carbon emissions of the product with the highest energy efficiency are first found through global optimization. Then, the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method to find the best product with the best carbon emissions. Then the optimized data is passed into the modified BP neural network prediction model to be solved in space; Carbon emissions forecast: The modified BP neural network prediction model is used to predict product carbon emissions.

4. The industrial product carbon footprint analysis method based on energy efficiency labeling database according to claim 3 is characterized in that: The grayscale model is used to screen and the results are tested for cointegration, to obtain single integer series energy efficiency and energy consumption parameters of normalized indicator variables related to carbon emissions, as normalized data including: In the formula, ii represents the product category and j represents a single stage in the life cycle; It represents the normalized value of energy consumption of the product of category ii in a single stage j during its life cycle, E ii，j is the energy consumption of a single stage j in the life cycle of a product of category ii, max E ii，j , min E ii，j Indicates the maximum and minimum values ​​of energy consumption in this data class, V represents the normalized value of energy intensity of the product of category ii in a single stage j of its life cycle. ii，j is the energy intensity of a single stage j in the life cycle of a product of category ii, max V ii，j , min V ii，j Indicates the maximum and minimum values ​​of energy intensity in this data class; Based on the above expression, the energy efficiency and energy consumption data are screened, and then the same root unit root test and different root unit root test are used to analyze and obtain the first-order single integer series. After screening by the grayscale model, the data in the unstable transition interval are abandoned for data consistency cointegration check, and finally the normalized single integer series data is obtained for subsequent prediction of pooled data.

5. The industrial product carbon footprint analysis method based on energy efficiency labeling database according to claim 3 is characterized in that: The hidden layer simplification of the BP neural network prediction model, using the output of the raw material stage in the product life cycle as the input of the product production stage, using the output of the product production stage as the input of the product application stage, and using the output of the product application stage as the input of the product recycling stage, includes: c t =tanh(W c (h t-1 ,x t )+b c ) c t =c t-1 f t +ra t c t ra t =σ(W x，ra (h t-1 ,x t )+b ra ) pr t =σ(W x，pr (h t-1 ,x pr )+b pr , ap t =σ(W x，ap (h t-1 ,x ap )+b ap , h t =ap t fishy t ) Where W is the weight coefficient matrix, b is the bias vector, h is the activation function, c is the hidden layer sequence variable calculation variable, t represents the time, x represents the life cycle sequence variable, ra, pr, ap represent the raw material, production, and application stages in the product life cycle, ra t ,pr t ,ap t represents the calculation output of the raw materials, production, and application stages at time t, h t and h t-1 is the activation function output at two moments before and after, x t is the output of the life cycle sequence variable at time t, x pr , x ap represents the life cycle sequence variable value in the production and application stages, c t and c t-1 Calculate the value of the hidden layer sequence variable for the two moments before and after, W c is the weight coefficient matrix of the hidden layer sequence variables, W x,ra ,W x,pr ,W x,ap is the weight coefficient matrix of the life cycle sequence variables of raw materials, production and application stages, and σ is the cumulative function.

6. The industrial product carbon footprint analysis method based on energy efficiency labeling database according to claim 5 is characterized in that: In the hidden layer of the BP neural network prediction model, a linear model of a multilayer perceptron is added, normalized data is used as prediction pooling data, and the lowest carbon emissions in the entire life cycle and the highest energy efficiency of the product itself are considered. The carbon emissions of the product with the highest energy efficiency are first found through global optimization, and then the carbon emissions of the product with the highest energy efficiency are predicted according to the local optimization method. The optimal carbon emissions of the product are found to include: The tanh function is used to optimize the minimum sum of errors of N samples. The optimization algorithm is as follows: In the formula, y m is the calculated output value at time m, st is the minimum optimization step length, v m is the correction factor at the mth moment, f is the calculated output value at the m-1th moment, and N is the number of samples; Add a multi-layer perceptron MLP and use a particle swarm algorithm to optimize the error sum, classify and clean the data to obtain the best of the same kind, and optimize the clustering method to minimize the objective function value of each pattern to cluster analysis, and to achieve the lowest energy consumption and carbon emission values. The energy efficiency evaluation mathematical model is expressed as: In the formula, E is energy consumption, C is total carbon emissions, ep represents the pth product attribute that affects energy efficiency and carbon emissions in the eth stage, g op is the pth attribute of the oth domain center; ω oe is the domain center and stage attribute judgment value. It takes 1 when the domain center and stage value of the product attribute belong to the same category, and takes 0 when they are not. The subscript o is the number of domain centers of the attribute, ranging from 1 to n. o , subscript e is the number of stages of the attribute, ranging from 1 to n e ; The subscript p is the number of attributes, ranging from 1 to n; n, n o , n e They are the number of corresponding energy efficiency and energy consumption indicators, the number of attribute domain centers, and the number of attribute stages in the database; According to the above expression, local and global searches are performed in the domain, the cluster center and objective function value are calculated for each variable, and the result set is transformed; The optimized data is passed into the modified BP neural network prediction model to solve in space, considering the influence of emission benefits and coefficient of variation, and the solution matrix is ​​updated as follows: t i (t+Δ)=(1-ρ)τ i (t)+βτ i Where τ is the carbon emission benefit, τ i The carbon emissions of a single product i in a single stage j of its life cycle are C j The carbon emission benefit is the output value of the BP network; t is the time, Δ is the solution step, ρ is the coefficient of variation, βτ i It is the carbon emission benefit variable in the BP network calculation process.

7. An industrial product carbon footprint analysis system based on energy efficiency labeling database, characterized in that: comprising at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and when the instructions are executed by the at least one processor, the at least one processor executes the industrial product carbon footprint analysis method based on the energy efficiency identification database as described in any one of claims 1 to 6.

8. A computer program product, characterized in that The computer program product comprises a computer program / instruction, which, when executed by a processor, implements the industrial product carbon footprint analysis method based on an energy efficiency identification database as claimed in any one of claims 1 to 6.

Images