Method, device and equipment for process analysis of carbon flow

Abstract

The invention relates to a method, a device and equipment for the process analysis of a carbon flow. The method comprises the following steps: investigating an emission source based on a carbon flow collection range architecture to obtain carbon flow data; constructing a material database of an energy medium, and then constructing a process model library based on a minimum process unit; checking the matching degree of the production process flow and the LCA data, and constructing a basic LCA model library; establishing a two-dimensional contrast diagram of process and carbon flow data, and then screening the basic factor library to obtain a required LCA model library; constructing a carbon flow analysis model, and carrying out calibration through a balance analysis closed-loop calibration mechanism; constructing a carbon flow mathematical model based on the calibrated process model library and the LCA model library; and then based on the carbon flow data, calibrating the carbon flow mathematical model through a carbon flow data closed-loop calibration mechanism to obtain a final carbon flow analysis result. The analysis accuracy of the carbon flow of the product can be improved, and the statistical efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of carbon footprint management, and in particular to a method, device and equipment for process-based analysis of carbon flow. Background Art

[0002] Carbon flow accounting involves a variety of emission sources and activity types, each of which may require a different calculation method. This increases the complexity and workload of accounting, and also places high demands on the professional skills of accounting personnel. Carbon flow accounting requires considerable financial and human resources, including the costs of data collection, processing, analysis, and report preparation, which is a considerable expense for small enterprises with limited resources.

[0003] At the same time, existing standards (such as ISO 14067-2018 "Greenhouse Gas - Product Carbon Footprint - Quantification Requirements and Guidelines" etc.) face many limitations. Although it provides principles, requirements and guidelines for quantifying product carbon footprint, it still relies on the understanding and application of specific implementers. Different organizations or individuals may have differences in understanding and applying the standards, which may lead to inconsistency in carbon footprint assessment results. Although the standard covers all stages of the product life cycle, in actual operation, data in some stages may be difficult to obtain or accurately quantify, such as emission data in the product use stage and waste stage, which may affect the accuracy of carbon footprint assessment. Data quality and availability: Carbon footprint accounting relies on a large amount of data, including raw material consumption, energy consumption, transportation distance, etc. If these data are inaccurate or incomplete, the accounting results will be affected. Especially when the supply chain is complex or involves multiple parties. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device and equipment for process-based analysis of carbon flow that can improve the analysis accuracy of product carbon flow and improve statistical efficiency in response to the above technical problems.

[0005] A method for analyzing carbon flow in a process-based manner, the method comprising: Constructing a carbon flow collection scope framework; investigating emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then constructing a material database of energy media based on the carbon flow data; According to the material database and in combination with the actual production process flow, the minimum process unit is determined, and then a process model library is constructed based on the minimum process unit; Obtaining an LCA master library model, verifying the matching degree between the production process and the LCA data based on the process model library and the LCA master library model, matching the LCA factors, and constructing a basic LCA model library; establishing a two-dimensional comparison chart of process and carbon flow data based on the basic LCA model library, and then screening the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library; Constructing a carbon flow analysis model; based on the carbon flow analysis model, calibrating the process model library and the LCA model library through a balanced analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library; Based on the calibrated process model library and LCA model library, a carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

[0006] A device for process-based analysis of carbon flow, comprising: A data collection module is used to construct a carbon flow collection scope framework; investigate emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then construct a material database of energy media based on the carbon flow data; A process model library construction module is used to determine the minimum process unit according to the material database and in combination with the actual production process flow, and then construct a process model library based on the minimum process unit; An LCA model library construction module is used to obtain an LCA total library model, verify the matching degree between the production process and the LCA data based on the process model library and the LCA total library model, match the LCA factors, and construct a basic LCA model library; according to the basic LCA model library, establish a two-dimensional comparison chart of process and carbon flow data, and then screen the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library; A balance calibration module is used to construct a carbon flow analysis model, and calibrate the process model library and the LCA model library through a balance analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library; The result output module is used to construct a carbon flow mathematical model based on the calibrated process model library and the LCA model library according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

[0007] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method for process-based analysis of carbon flow when executing the computer program.

[0008] The above-mentioned method, device and equipment for process-based analysis of carbon flow are to construct a carbon flow collection range framework; investigate the emission sources based on the carbon flow collection range framework to obtain carbon flow data; then build a material database of energy media based on the carbon flow data; determine the minimum process unit based on the material database and the actual production process flow, and then build a process model library based on the minimum process unit; obtain the LCA total library model, and The total library model verifies the matching degree between the production process and the LCA data, matches the LCA factors, and constructs a basic LCA model library; according to the basic LCA model library, a two-dimensional comparison chart of the process and carbon flow data is established, and then the basic factor library in the basic LCA model library is screened based on the two-dimensional comparison chart to obtain the required LCA model library; a carbon flow analysis model is constructed; based on the carbon flow analysis model, the process model library and the LCA model library are calibrated through a closed-loop calibration mechanism of equilibrium analysis to obtain a calibrated process model library and LCA model library; based on the calibrated process model library and LCA model library, a carbon flow mathematical model is constructed according to a dynamic equation combined with a neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through a closed-loop calibration mechanism of carbon flow data, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

[0009] The present invention achieves full coverage of factors from emission source investigation to energy media through three-level progression from collection scope architecture to material database to process model library, avoiding omissions in traditional single-point sampling, and greatly improving the integrity of carbon data. In the process model library, the process is decomposed into the smallest process unit in combination with the actual production process flow, breaking through the extensiveness of traditional section-level analysis, and can achieve precise emission reduction. In addition, when the enterprise needs to adjust the process, it only needs to make a reasonable combination according to the smallest process unit, or adjust one or more smallest process units. The control range is controllable, and the dynamic update mechanism of the enterprise is effectively guaranteed. Through the two-dimensional comparison chart of the process and carbon flow data, the LCA factor redundancy can be intuitively identified, and the required LCA model library can be obtained to ensure the integrity and accuracy of the subsequent nucleic acid content. At the same time, through the combination of the double closed-loop calibration mechanism of the closed-loop calibration mechanism of the balanced analysis and the closed-loop calibration mechanism of the carbon flow data, the accuracy of the model can be ensured, the deviation can be reduced, and the accuracy of the final analysis result can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0011] Figure 1 A schematic diagram of a process for analyzing a carbon stream in a process-based manner in one embodiment; Figure 2 A schematic diagram of a framework of a method for process-based analysis of carbon flow in one embodiment; Figure 3 A schematic diagram of data flow of a minimum process unit in an embodiment; Figure 4 A structural block diagram of a device for process-based analysis of carbon flow in one embodiment; Figure 5 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment.

[0012] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] It can be understood that the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in the field can implement it. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0015] The implementation modes of the present invention will be described in detail below with reference to the accompanying drawings in the embodiments of the present invention.

[0016] Example 1 This embodiment discloses a method for process-based analysis of carbon flow. Through the three-level progression from the collection scope architecture to the material database to the process model library, full coverage of factors from emission source investigation to energy media is achieved, avoiding the omissions of traditional single-point sampling, and greatly improving the integrity of carbon data. In the process model library, the process is decomposed into the smallest process unit in combination with the actual production process flow, breaking through the extensiveness of the traditional section-level analysis, and accurate emission reduction can be achieved. In addition, when the enterprise needs to adjust the process, it only needs to make a reasonable combination according to the smallest process unit, or adjust one or more smallest process units. The control range is controllable, effectively ensuring the dynamic update mechanism of the enterprise. Through the two-dimensional comparison chart of the process and carbon flow data, the LCA factor redundancy can be intuitively identified, and the required LCA model library can be obtained to ensure the integrity and accuracy of the subsequent nucleic acid content. At the same time, through the combination of the double closed-loop calibration mechanism of the closed-loop calibration mechanism of the balanced analysis and the closed-loop calibration mechanism of the carbon flow data, the accuracy of the model can be ensured, the deviation can be reduced, and the accuracy of the final analysis results can be improved.

[0017] like Figure 1 and Figure 2 As shown, the process-based method for analyzing carbon flow provided in this embodiment includes the following steps: Step 201, construct a carbon flow collection scope framework; investigate emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then construct a material database of energy media based on the carbon flow data.

[0018] Step 202, according to the material database and in combination with the actual production process flow, the minimum process unit is determined, and then a process model library is constructed based on the minimum process unit.

[0019] Step 203, obtain the LCA master library model, verify the matching degree between the production process flow and the LCA data based on the process model library and the LCA master library model, match the LCA factors, and build a basic LCA model library; according to the basic LCA model library, establish a two-dimensional comparison chart of the process and carbon flow data, and then screen the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library.

[0020] Step 204, constructing a carbon flow analysis model; based on the carbon flow analysis model, calibrating the process model library and the LCA model library through a closed-loop calibration mechanism of equilibrium analysis to obtain a calibrated process model library and LCA model library.

[0021] Step 205, based on the calibrated process model library and LCA model library, a carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

[0022] In one embodiment, a carbon stream collection scope architecture is constructed, including: Based on direct emissions, energy consumption, product delivery, raw materials, by-product recycling, transportation and by-product output, a carbon flow collection scope framework is constructed in the form of a table.

[0023] LCA load classification is carried out based on the carbon flow collection range framework. LCA load classification includes direct emission load, energy consumption load, product delivery load, raw material load, by-product recycling load, transportation load and by-product output load.

[0024] Specifically, the established carbon flow collection scope framework covers the entire carbon flow cycle and system plant boundaries, and clarifies the accuracy of the carbon flow collection scope. In terms of the collection scope, the basic characteristics of the carbon flow process are used to refine the division of labor, and the carbon flow collection scope is divided into seven types: direct emissions, energy consumption, product delivery, raw materials, by-product recycling, transportation and by-product output, and a collection scope attribute table is established based on this. The table format is shown in Table 1: Table 1

[0025] It is understandable that using charts instead of simple text descriptions can clarify the survey objects, ensure that the collection scope is clear and accurate, ensure that subsequent content is fully included in the collection scope, reduce the possibility of omissions, and ensure the integrity of the data collection scope. In addition, this method has good versatility and wide adaptability. When the company undergoes dynamic adjustments, the table content can be adjusted accordingly.

[0026] In addition, LCA load classification is carried out based on the carbon flow collection scope framework. LCA load (life cycle assessment load) is a classification of environmental impact loads generated by different activities and processes in the product life cycle, and the main focus is on the environmental impact level. It includes direct emission load, energy consumption load, product transfer load, raw material load, by-product recycling load, transportation load and by-product output load. Among them, direct emission load refers to the load generated by direct emissions during product production; energy consumption load refers to the energy consumption load in each stage of product production; product transfer load refers to the incidental load of intermediate products before entering the organization; raw material load refers to the load of raw materials or auxiliary materials before entering the organization; by-product recycling load refers to the load of inefficient internal circulation during product production; transportation load refers to the load generated by raw materials or products during transportation; by-product output load refers to the load of by-products generated during product production that are output to the outside of the organization.

[0027] In one embodiment, the emission sources are investigated based on the carbon flow collection scope framework to obtain carbon flow data, including: Based on the carbon flow collection scope framework, emissions, product capacity and output, pollution control facility operation, pollutants, greenhouse gas emissions and upstream supply chain data are investigated.

[0028] During the investigation, based on the production process flow, the data relationships and data comparisons in each process flow were divided, the output products, input raw materials, and recycling process were determined, and data were collected in comparison with direct emissions, energy consumption, product delivery, raw materials, by-product recycling, transportation and by-product output to obtain carbon flow data.

[0029] Specifically, according to the carbon flow collection scope framework and carbon flow types, a comprehensive investigation of emission sources is conducted. The survey content includes: emissions, product production capacity and output, pollution control facility operation, pollutants, greenhouse gas emissions and upstream supply chain data. The method of investigation is: based on the actual production process, define the data relationship and data comparison of each production process link, and determine the input raw materials, output products and recycling process. In each production process, as shown in Table 1, the seven types of carbon flow collection scope are compared to fully manage the source of carbon flow, that is, clarify the actual accounting boundaries and the main emission facilities: including the description of the accounting boundary, facility name, facility number, facility specification model, facility installation location, usage status and remarks, etc., specifically fill in the corresponding contents of Table 1 to obtain carbon flow data.

[0030] It can be understood that verifying the collection of carbon flow data based on the production process flow can ensure the accuracy and completeness of the data, ensure the scientific and authoritative requirements of the carbon flow collection workload statistics, clarify the survey objects, provide a reliable basis for carbon flow accounting, and effectively ensure that there is no missing data.

[0031] In one embodiment, a material database is constructed based on energy media, including: Energy media are obtained through production and manufacturing management systems, inspection and testing systems, comprehensive resource utilization systems, procurement management systems and other management systems.

[0032] Based on the energy medium, the carbon flow data is integrated, and then the data structure is unified to build a material database of the energy medium.

[0033] Specifically, the sources of energy media include but are not limited to production and manufacturing management systems, inspection and testing systems, comprehensive resource utilization systems, procurement management systems and other management systems. Energy media use measured values, mainly including carbon content, greenhouse gas emission coefficient, energy calorific value, etc.

[0034] Based on energy media, carbon flow data from different sources are integrated, and data scattered in various systems are brought together for unified processing. During integration, the data is converted into modular library files to unify data rules and achieve unified management of data attributes. There are mainly multiple data core content management such as unit matching, indicator conversion, name relationship library, data type conversion, etc. Among them, unit matching is mainly to ensure that the units used by data from different sources or different representations can match and convert each other; indicator conversion refers to the conversion of different data indicators so that they can be compared and analyzed under the same standard; the name relationship library manages the relationship between data names by establishing a database, because the same thing may have different names in different systems or scenarios. Through this library, the corresponding relationship between them can be clarified to avoid confusion; data type conversion converts different data types according to unified requirements to make the data consistent during storage and processing.

[0035] When establishing a material database for energy media, the core element is to ensure the consistency of data attributes, including data types, names, units, indicators, etc. Since energy media come from the access of multiple data, it may cause various problems such as inconsistent data structure and data duplication. Data governance should be carried out based on a unified data structure and reasonable access to data should be completed based on a data pool.

[0036] It is worth mentioning that after the material database is built, it is necessary to return to the previous step to verify the integrity and accuracy of the carbon flow data to confirm whether the data has been over-entered. The so-called over-entry refers to the same data being entered in different forms or channels, resulting in data duplication. If over-entry occurs, it is necessary to find out the cause of the over-entry, sort out the cause of the over-entry, and remove unnecessary data.

[0037] After the data processing of this embodiment, the attributes of various types of data are normalized, the data duplication is clear at a glance, and the consistency of data attributes and the availability of data are guaranteed.

[0038] It is worth mentioning that in order to ensure the security of the material database, in addition to using the basic network security mechanism and data security mechanism of the data center, a data backup mechanism should also be adopted to avoid potential risks in data management.

[0039] In one embodiment, the minimum process unit is determined based on the material database and the actual production process flow, and then a process model library is constructed based on the minimum process unit, including: According to the material database and combined with the actual production process, the minimum process unit is determined based on the input, output and recycling of the carbon flow, and then a process model library is constructed by combining several minimum process units.

[0040] Specifically, based on the material database and combined with the actual production process, we sort out the data on products, by-products, solid waste, hazardous waste, raw materials, auxiliary materials, energy and energy media, natural resource inputs, atmospheric emissions, water emissions, soil emissions, etc. of each process unit, classify and identify input raw materials, intermediate materials, recycled material media, emission media, carbon fixation products, etc., and establish a complete process model library.

[0041] When establishing a process model library, it is necessary to establish a basic model for each process. In principle, each process should be divided into the smallest process unit. Only by dividing it into the smallest process unit can the corresponding factor library in the LCA model library be better matched and the carbon flow can be finally sorted out.

[0042] like Figure 3 As shown in the figure, the carbon flow information possessed by each minimum process unit includes: input, output, and recycling. The process model library formed by the combination of minimum process units also includes three types of carbon flow information, namely input, output, and recycling. These three types of carbon flow information have completely covered the flow route of carbon flow. In the actual production process, each minimum process unit is intertwined, but its final carbon flow information is included in these three categories. Among them, the difference in input can distinguish different energy medium categories, different energy qualities, different carbon sources, environmental input factors, and also include raw materials for each product. The difference in output can distinguish different energy categories, different emission loads, and different energy uses, and also include product output, by-product output, and waste output. The difference in recycling can distinguish different energy media, the recycling methods and process emissions of each energy medium, the output flow direction of each energy medium, the output of the recycling process, and the new input of the recycling process.

[0043] This embodiment builds a process model library based on the minimum process, which can ensure the precision and accuracy of the model and effectively guarantee the dynamic update mechanism of the enterprise. In addition, when the enterprise only makes small-scale process adjustments, it can only make adjustments to a single or multiple minimum process units, and the adjustment range is controllable; when the enterprise needs to make large-scale process adjustments, it only needs to reconstruct the process, which is simple and applicable.

[0044] In one embodiment, an LCA master library model is obtained, and based on the process model library and the LCA master library model, the matching degree between the production process and the LCA data is verified, the LCA factors are matched, and a basic LCA model library is constructed, including: Get the LCA master library model.

[0045] The consistency check between the process names and system boundaries in the process model library and the LCA master library model was performed to obtain preliminary comparison results.

[0046] Based on the preliminary comparison results, the interference fit data is traced back to verify whether the original data collection scenario in the LCA master library model matches the production process flow in the process model library. If so, the matching LCA factor is obtained; if not, the process model library is returned to be corrected to achieve the matching degree and match the LCA factor; the basic LCA model library is constructed based on the matching LCA factor.

[0047] Specifically, the LCA (Life Cycle Assessment Analysis) library model is an existing model library provided by relevant organizations such as the International Organization for Standardization (ISO), which will not be described in detail here. Based on the process model library and the LCA library model, a preliminary comparison and verification of the matching degree between the production process flow and the LCA data is carried out, and the consistency check of the process name and system boundary is completed, and the appropriate LCA factors are matched to obtain preliminary comparison results.

[0048] Based on the preliminary comparison results, find the mismatched elements, and reverse trace the interference matching data to verify whether the original data collection scenario in the LCA master library model matches the production process in the process model library. If so, obtain the matching LCA factors; if not, mark the mismatched data accordingly, and then return to the process model library to correct the production process to achieve the degree of matching between the production process and the LCA data, and match the LCA factors; then build a basic LCA model library based on all matching LCA factors, and the basic LCA model library contains the basic factor library. It is worth noting that the accounting models of various emission factors and activity elements should be included in the basic LCA model library together with the LCA factors. The accounting models of emission factors and activity elements are constructed based on relevant scientific principles, industry standards and the characteristics of the actual production process. They are conventional technical means and will not be elaborated here.

[0049] Then, according to the basic LCA model library, the minimum process unit in the process model library is allocated based on the type of compliance, and a two-dimensional comparison chart of process and carbon flow data is established as shown in Table 2.

[0050] Table 2

[0051] Then, based on the two-dimensional comparison chart of process and carbon flow data, the carbon flow data are matched in the basic LCA model library, and the most favorable LCA factors are screened out from the basic factor library. The most favorable factors mainly consider the following aspects: matching rationality, the self-interest principle of accounting data, and the compliance of accounting methods. Among them, matching rationality refers to the consistency between the technical parameters, process flow and system boundaries of the LCA factors and the target carbon flow data; the self-interest principle of accounting data refers to the fact that under the compliance framework, the accounting results are beneficial to the enterprise through selective data reference; the compliance of accounting methods means that the accounting process must strictly follow the core requirements of the ISO 14044:2006 "Environmental Management Life Cycle Assessment Requirements and Guidelines" standard. It is worth noting that when screening LCA factors, the most favorable factors are quantified as objective functions, and the NSGA-II algorithm is used to screen the Pareto optimal factor combination.

[0052] After selecting the most favorable LCA factors, the calculation phase begins. The calculation results of the calculation model in the basic LCA model library are accurately matched with the flow chain of the carbon flow data in the two-dimensional comparison diagram. During the matching process, the LCA factors selected by the most favorable factors are strictly calculated to generate complete calculation results and the required LCA model library. It is worth noting that during the calculation, the ISO 14044:2006 "Requirements and Guidelines for Life Cycle Assessment of Environmental Management" standard is used for evaluation, the uncertainty of the calculation is measured, and the final output includes a complete calculation result with a confidence interval of ±15%.

[0053] Through the steps of this embodiment, a basic LCA model library is gradually established, and carbon flow data, process matching and factor library screening are performed to ensure the integrity and accuracy of the accounting content.

[0054] In one embodiment, a carbon flow analysis model is constructed; based on the carbon flow analysis model, a process model library and an LCA model library are calibrated through a closed-loop calibration mechanism of equilibrium analysis to obtain a calibrated process model library and an LCA model library, including: The three-level carbon balance verification framework is composed of process layer, system layer and full-chain layer. The calibration mechanism and verification process are introduced to build a carbon flow analysis model.

[0055] The LCA factors and process parameters in the LCA model library are input into the carbon flow analysis model for calculation, and the carbon flow prediction value is output.

[0056] The predicted carbon flow value is compared with the measured data in the material database. If the comparison result does not reach the threshold, the minimum process unit data in the process model library is adjusted through the closed-loop calibration mechanism of balance analysis, and an updated LCA model library is generated.

[0057] The carbon flow analysis model is calculated based on the updated LCA model library until the termination condition is met, and the calibration of the process model library and the LCA model library is completed to obtain the calibrated process model library and the LCA model library.

[0058] It can be understood that a three-level carbon balance verification framework is built at the process level, system level, and full chain level, and then calibration, optimization, and verification are performed around the framework to build a carbon flow analysis model. It is worth noting that in the constructed carbon flow analysis model, a closed-loop calibration mechanism for balance analysis is introduced, which significantly improves the accuracy and reliability of carbon flow analysis.

[0059] Specifically, in the process layer, dynamic material balance verification is used. In a single production process, a dynamic balance relationship of "input-output-accumulation" is established. All carbon entering the process (such as carbon content of raw materials, energy carbon emissions) must be equal to the sum of the output carbon (such as carbon content of products, exhaust gas emissions) and the temporary carbon stored in the process (such as equipment adsorption, unreacted residues). For example, in the steel smelting process, the carbon content of input materials such as iron ore and coke must be accurately counted and compared with the carbon output of molten steel, slag, and flue gas. The difference is the carbon temporarily stored in the blast furnace. This level requires all data to be labeled with dimensions (such as / process) and control the measurement error within ±5%.

[0060] At the system level, cross-process carbon flow correlation modeling is performed. For production systems consisting of multiple processes, the matrix method is used to quantify the carbon transfer relationship between processes. By constructing a mathematical matrix, it is clear how the carbon emissions of each process affect the downstream links. For example, in the cement production system, the carbon emissions (matrix rows) of the limestone calcining process will be associated with the carbon input (matrix columns) of the clinker grinding process. Matrix elements must distinguish between bio-based carbon and fossil-based carbon according to international standards (such as ASTM D6866-24 "Standard Test Method for Determining the Biobased Content of Solid, Liquid, and Gas Samples Using Radiocarbon Analysis") to ensure data comparability. Finally, matrix operations are used to verify whether the overall carbon flow of the system meets the preset constraints (such as the carbon emission cap for the entire process).

[0061] Uncertainty analysis and optimization are performed at the whole chain level. At the scale of the whole industry chain, data from each layer is integrated to build a visual carbon flow network, and Monte Carlo simulation (more than 10,000 random samplings) is introduced to quantify data uncertainty. For example, when evaluating the carbon emissions of the new energy vehicle battery supply chain, the carbon data fluctuations in the mining, processing, and transportation of battery materials are simulated, and the results are output with a confidence interval of ±15%. Combined with the closed-loop calibration mechanism, the simulation results are fed back to the process and system layers to continuously optimize the LCA factor library and accounting model.

[0062] Next, a dynamic threshold function is set to perform quantitative analysis of the differences, so as to quantitatively evaluate the data differences. The expression of the dynamic threshold function is: ; In the formula, Indicates the maximum data difference tolerance limit allowed during the calibration process; represents the data quality index; Indicates the complexity of the system; It represents the absolute deviation between the model prediction value and the actual measured value.

[0063] Finally, the Levenberg-Marquardt algorithm is used for parameter optimization. The initial value of the regularization coefficient λ is set to 0.01 and decayed by 50% every 5 iterations. The parameters of the carbon flow analysis model are made more realistic through iterative optimization.

[0064] On the basis of the above calibration mechanism, a dynamic calibration protocol is introduced to ensure that the carbon flow analysis model can respond to data changes in a timely and accurate manner. Specifically, for abnormal data marking, the sliding window 3σ criterion (window length ≥ 100 sampling points) is adopted, and in principle, the parameter adjustment response time is ≤ 1 second.

[0065] Furthermore, when constructing the carbon flow analysis model, in order to evaluate and deal with the uncertainty in the model, a large number of samplings were performed by setting the MCMC sampling chain length ≥ 5000 steps to fully capture the uncertainty distribution of the model parameters; and the Gelman-Rubin convergence criterion R<1.1‌25 was used to determine whether the MCMC sampling had converged, so as to improve the reliability of the model.

[0066] Furthermore, when constructing the carbon flow analysis model, an intelligent calibration engine is also set up to improve the calibration capability and intelligence level of the model. First, PCA is used to extract key influencing factors from the LCA model library. Then, the IsolationForest algorithm is used for anomaly detection, which can effectively identify outliers in the data. These outliers may be caused by measurement errors, data entry errors, etc. Timely discovery and processing can avoid interference with the model and ensure the stability and reliability of the model.

[0067] Afterwards, we built an LSTM time series prediction sub-model, set the time step T=24, the number of hidden layer neurons ≥32, and the Dropout rate = 0.2‌16. The LSTM time series prediction sub-model took advantage of the advantages of processing time series data to predict the carbon flow and obtain future carbon footprint data.

[0068] Finally, a knowledge graph is constructed. The process units, material flows, environmental impact categories, etc. are used as nodes, the carbon flow paths, conversion coefficients, uncertainty associations, etc. are used as relationships, and the timestamps, data sources, quality grades, etc. are used as attributes. Various types of information related to carbon footprint accounting are integrated. In this way, the carbon flow analysis model can understand the complex relationships between data, so as to calibrate and predict more accurately.

[0069] Furthermore, when constructing the carbon flow analysis model, a four-stage verification mechanism was established. Through the four-stage verification mechanism, the carbon flow analysis model can be verified from multiple dimensions to ensure the accuracy, reliability and traceability of the model, thereby improving the calibration accuracy and accelerating the model convergence speed to be suitable for complex carbon flow accounting scenarios. Specifically, the four-stage verification mechanism includes a priori verification, cross-validation, empirical verification and traceability verification. Among them, a priori verification is based on the theoretical verification of the law of conservation of matter, which is mainly used for model initialization and theoretical rationality verification process; cross-validation is used for K-fold test of different model output results, which is mainly used for model parameter tuning and overfitting prevention process; empirical verification is the Bland-Altman analysis of model prediction results and measured data, which is mainly used for the credibility confirmation process of model output results; traceability verification is to record the key calibration process, which is mainly used for compliance audit and responsibility tracing process.

[0070] The carbon flow analysis model constructed in the above way can improve the calibration accuracy to more than 98.5% (an increase of about 23% compared to traditional methods), and the model convergence speed is increased by 40%, which is particularly suitable for carbon footprint accounting scenarios of complex industrial systems.

[0071] After the carbon flow analysis model is constructed, a relatively complete and accurate calculation of the current carbon flow state is available. The carbon flow analysis model enters the next step, namely carbon flow prediction.

[0072] The carbon flow analysis model is used to calculate the LCA factors and process parameters in the LCA model library to obtain the carbon flow prediction value, which includes carbon emissions (such as process level emissions, life cycle carbon footprint), carbon conversion pathways (such as the flow ratio of bio-based carbon to fossil-based carbon), and uncertainty quantification results (such as confidence intervals of predicted values).

[0073] Then, through the closed-loop calibration mechanism of equilibrium analysis, the predicted carbon flow value is compared with the measured data in the material database. If the comparison result does not reach the threshold, the minimum process unit data in the process model library is adjusted, and an updated LCA model library is generated. It can be understood that the closed-loop calibration mechanism of equilibrium analysis is a key step to ensure the accuracy of carbon flow analysis. It can effectively identify and correct the deviations in the model by quantitatively describing and balancing the input, output and transformation of the carbon flow analysis model, thereby improving the accuracy and reliability of carbon flow analysis.

[0074] Through the closed-loop calibration mechanism of balance analysis, the data in the process model library is iteratively optimized, and the results are reversely mapped to the LCA model library (such as correcting the LCA factor) until the iteration termination conditions are met (such as dynamic threshold verification passed, convergence criteria met), and finally the calibrated process model library and LCA model library are output. The principles and methods of the closed-loop calibration mechanism of balance analysis are as follows: First, the defect is located. By comparing the residual distribution of the predicted carbon flow value with the measured data in the material database (such as when the reaction conversion rate deviation is >5%), the defect link (such as failure of the thermodynamic parameter assumption) is automatically locked.

[0075] After the defect location is clear, the model is dynamically optimized to achieve iteration. Dynamic optimization is based on the key variables of the defect link (such as temperature and pressure gradient), and tens of thousands of input parameter combinations are randomly generated to cover different working conditions. Multiple data inputs (such as carbon content of raw materials and carbon emissions of energy) are fed into the LSTM time series prediction sub-model to train it to predict indicators such as carbon emission peaks (such as / hour). When the prediction error is less than 5%, the parameter update is triggered and the optimization is coordinated with the main model. If the prediction error is ≥5%, it returns to the dynamic optimization stage and adjusts the multiple data inputs until the prediction error is less than 5%.

[0076] Then, since atypical and uncertain situations are prone to occur during the data calibration process, uncertainty analysis is required. The uncertainty analysis method is to separate the measurement error (such as sensor accuracy ±2%) and the model deviation (such as ±5% error caused by thermodynamic simplification). If the confidence interval exceeds the limit, the high variance source is corrected first (for example, when the measurement error accounts for >70%, the sensor sampling frequency is upgraded).

[0077] Finally, the calibrated process model library and LCA model library are obtained.

[0078] It can be understood that the carbon flow analysis system is divided into process layer, system layer and full chain layer. Among them, the process layer verifies carbon input and output through dynamic balance (error ±5%); the system layer uses matrix method to quantify carbon transfer between processes; the full chain layer evaluates carbon fluctuations in the supply chain based on random simulation (confidence interval ±15%) and reversely calibrates the model. The dynamic threshold function is combined with the optimization algorithm to adjust the parameters, and the abnormal response time is ≤1 second. The intelligent calibration engine extracts key factors and identifies anomalies, the time series model predicts carbon flow trends (step length 24, error <3%), and the knowledge graph integrates process and carbon flow data. Four-stage verification (theoretical, cross, empirical, and traceability) ensures the reliability of the model, and the calibration accuracy reaches 98.5%. By separating measurement errors (such as ±2%) and model deviations, high-impact sources are corrected first to form a closed-loop calibration system suitable for complex industrial carbon accounting.

[0079] In one embodiment, based on the calibrated process model library and LCA model library, a carbon flow mathematical model is constructed according to a dynamic equation combined with a neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through a carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis results are obtained through the calibrated carbon flow mathematical model, including: Based on the calibrated process model library and LCA model library, the carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; the dynamic equation expression is: ; In the formula, Indicates the carbon concentration in the system and reflects the instantaneous carbon concentration change rate; Indicates the input carbon flow, which is collected in real time by sensors; Indicates the output carbon flow, which is collected in real time by sensors; represents the temperature-dependent carbon conversion rate; Represents the carbon storage rate, which is positively correlated with the capacity of the process container. , The unit is kg / h and is collected in real time by sensors. The dynamic equilibrium process of the carbon reserves in the system can be quantified through the carbon flow mathematical model, which constitutes the state variables of the closed-loop control system. Its differential characteristics determine the stability criterion of the system and provide key state parameters for real-time process control.

[0080] The neural network structure includes input layer, hidden layer and output layer. Based on the dynamic equation, the carbon flow data is input into the input layer, and the carbon flow mathematical model is used to calculate the carbon flow data. Perform data integration calculations and then output the predicted values ​​of the carbon flow analysis results through the output layer.

[0081] The predicted values ​​of the carbon flow analysis results are preliminarily calibrated to obtain preliminary calibrated predicted values.

[0082] Through the closed-loop calibration mechanism of carbon flow data, the preliminary calibration prediction value is compared with the carbon flow data, the deviation in the model is calculated, and the carbon flow mathematical model is iteratively optimized based on the deviation until the termination condition is met, and the final carbon flow analysis result is output.

[0083] It can be understood that in order to improve the predictive ability and practicality of the model, a carbon flow mathematical model is constructed by conducting in-depth analysis and quantitative description of the carbon flow process. The established carbon flow mathematical model accurately describes the flow process of carbon in the system, including the source, destination, conversion and storage of carbon. The carbon flow mathematical model has predictive ability and can predict future trends and changes in carbon flow. The accuracy and reliability of the model are then evaluated by comparing and analyzing the actual carbon flow data. If there are deviations or deficiencies in the model, corresponding adjustments and optimizations are made to improve the predictive ability and practicality of the model, thus forming the second mechanism of closed-loop calibration.

[0084] In the carbon flow mathematical model expression, Expressed by the Arrhenius correction term: ; Where k0 is the base rate constant, is the activation energy, is the gas constant; is the temperature in Kelvin (K); is the corrected pre-exponential factor, including, for example, effective reaction area or energy transfer efficiency; is the carbon concentration of the carbon stream during transmission. The combination of temperature and activation energy ( )’s corrective effect.

[0085] The neural network structure adopts the ‌LSTM network structure. The input layer inputs the historical carbon flow data (time window ≥ 72h) and process parameters (temperature, pressure, flow rate) in the carbon flow data; the hidden layer is a 2-layer LSTM unit (64 neurons per layer), and the Dropout rate = 0.2; the output layer outputs the predicted value of the carbon flow in the next 24 hours. When training the model, the training target is MSE < 5%, and the Adam optimizer (learning rate = 1e-3) is used.

[0086] It is worth noting that during the model training process, if the dynamic balance verification carbon input and output error is greater than 5%, the second closed-loop calibration mechanism, namely the carbon flow data closed-loop calibration mechanism, will be triggered.

[0087] The closed-loop calibration mechanism of carbon flow data is mainly based on the predicted value output by the carbon flow mathematical model to update the carbon flow data, and then based on the updated carbon flow data, the carbon flow mathematical model is continuously optimized and improved to achieve the purpose of closed-loop calibration. The calibration process should include steps such as checking the data range and content, identifying and correcting outliers, and filling in missing data.

[0088] Specifically, first obtain carbon flow data, including but not limited to production energy consumption (electricity, fuel), raw material consumption, process flow parameters, etc.; determine the coverage of carbon flow data, including but not limited to direct emissions (combustion, chemical reactions) and indirect emissions (purchased energy, supply chain). Then, check the integrity of the data to confirm that there are no missing data and that all emission links are covered. During verification, compare carbon flow data, industry benchmark values ​​or preliminary calibration predictions to identify outliers. Based on outliers, check the accuracy of calibration instruments, such as sensors and metering equipment, and use regression analysis and interpolation methods to fill in missing carbon flow data and update carbon flow data; then rebuild the material database of energy media based on the updated carbon flow data, and repeat steps 202 to 205 until the iteration termination condition is reached, that is, the dynamic balance verifies that the carbon input and output error is ≤5%, and obtains the final carbon flow analysis results.

[0089] It can be seen that the process-based method for analyzing carbon flow provided by the present invention has a clear definition of the work content and scope of each element of the carbon flow by establishing the scope and content of carbon flow collection. A material database of energy media is established to achieve data validity and authenticity of each energy medium of the carbon flow. An LCA model library is established to clarify the classification of each energy medium and output LCIA results based on the characterization model and the characterization factor library. A carbon flow analysis model is constructed to further calibrate the process model library and the LCA model library by establishing the main element balance and carbon balance of the raw materials. The constructed carbon flow mathematical model accurately describes the flow process of carbon in the system. The model has predictive capabilities and can be pre-adjusted and re-optimized.

[0090] In addition, a dual closed-loop calibration mechanism is introduced, that is, the balance analysis closed-loop calibration mechanism is used in the carbon flow analysis model to realize the first closed-loop calibration mechanism for carbon flow, and the carbon flow data closed-loop calibration mechanism is used in the carbon flow mathematical model to realize the second closed-loop calibration mechanism for carbon flow, which can significantly improve the accuracy and integrity of carbon flow data and effectively ensure the integrity of data. At the same time, the model's predictive ability and practicality are also enhanced, providing more accurate and reliable data support for enterprises' carbon emission management and low-carbon transformation.

[0091] Although this embodiment Figure 1The steps in the process are shown in sequence as indicated by the arrows, but the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of the steps, and the steps can be executed in other orders. Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0092] Example 2 Based on the method for analyzing carbon flow in a process-based manner in Example 1, this embodiment discloses a device for analyzing carbon flow in a process-based manner, such as Figure 4 As shown, the device for process-based analysis of carbon flow includes: a data collection module 401, a process model library construction module 402, an LCA model library construction module 403, a balance calibration module 404 and a result output module 405, wherein: The data collection module 401 is used to construct a carbon flow collection scope framework; investigate emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then construct a material database of energy media based on the carbon flow data.

[0093] The process model library construction module 402 is used to determine the minimum process unit according to the material database and the actual production process flow, and then construct a process model library based on the minimum process unit.

[0094] The LCA model library construction module 403 is used to obtain the LCA total library model, verify the matching degree between the production process flow and the LCA data based on the process model library and the LCA total library model, match the LCA factors, and build a basic LCA model library; according to the basic LCA model library, establish a two-dimensional comparison chart of the process and carbon flow data, and then screen the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library.

[0095] The balance calibration module 404 is used to construct a carbon flow analysis model, and calibrate the process model library and the LCA model library through a balance analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library.

[0096] The result output module 405 is used to construct a carbon flow mathematical model based on the calibrated process model library and the LCA model library according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

[0097] In this embodiment, the specific working process and working principle of the data collection module 401, the process model library construction module 402, the LCA model library construction module 403, the balance calibration module 404 and the result output module 405 are the same as those in the method of Example 1, so they are not described in detail in this embodiment. Each unit module can be implemented in whole or in part by software, hardware and a combination thereof, and each unit module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above unit modules.

[0098] Example 3 like Figure 5 The terminal device disclosed in this embodiment includes a transmitter, a receiver, a memory, and a processor. The transmitter is used to send instructions and data, the receiver is used to receive instructions and data, the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions stored in the memory to implement the method in the above-mentioned embodiment 1.

[0099] It should be noted that the above memory can be independent or integrated with the processor. When the memory is independently provided, the terminal device further includes a bus for connecting the memory and the processor.

[0100] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0101] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] The above-mentioned embodiments only express several implementation modes of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention. It should be pointed out that, for a person of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A method for analyzing carbon flow in a process-based manner, characterized in that: The method comprises: Constructing a carbon flow collection scope framework; investigating emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then constructing a material database based on energy media; According to the material database and in combination with the actual production process flow, the minimum process unit is determined, and then a process model library is constructed based on the minimum process unit; Obtaining an LCA master library model, verifying the matching degree between the production process and the LCA data based on the process model library and the LCA master library model, matching the LCA factors, and constructing a basic LCA model library; establishing a two-dimensional comparison chart of process and carbon flow data based on the basic LCA model library, and then screening the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library; Constructing a carbon flow analysis model; based on the carbon flow analysis model, calibrating the process model library and the LCA model library through a balanced analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library; Based on the calibrated process model library and LCA model library, a carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

2. The method for analyzing carbon flow according to claim 1, characterized in that: Construct carbon flow collection scope architecture, including: According to direct emissions, energy consumption, product delivery, raw materials, by-product recycling, transportation and by-product output, a carbon flow collection scope framework is constructed in the form of a table; And based on the carbon flow collection range framework, LCA load classification is performed, and the LCA load classification includes direct emission load, energy consumption load, product delivery load, raw material load, by-product recycling load, transportation load and by-product output load.

3. The method for analyzing carbon flow according to claim 2, characterized in that: Based on the carbon flow collection scope framework, the emission sources are investigated to obtain carbon flow data, including: Based on the carbon flow collection scope framework, investigate emissions, product capacity and output, pollution control facility operation, pollutants, greenhouse gas emissions and upstream supply chain data; During the investigation, based on the production process flow, the data relationships and data comparisons in each process flow were divided to determine the output products, input raw materials, and recycling process. Data were collected by comparing the direct emissions, energy consumption, product delivery, raw materials, by-product recycling, transportation, and by-product output to obtain carbon flow data.

4. The method for analyzing carbon flow according to any one of claims 1 to 3, characterized in that: Build a material database based on energy media, including: Obtain energy media through production and manufacturing management systems, inspection and testing systems, comprehensive resource utilization systems, procurement management systems and other management systems; Based on the energy medium, the carbon flow data is integrated, and then the data structure is unified to build a material database of the energy medium.

5. The method for analyzing carbon flow according to any one of claims 1 to 3, characterized in that: According to the material database, combined with the actual production process flow, the minimum process unit is determined, and then a process model library is constructed based on the minimum process unit, including: According to the material database, combined with the actual production process, the minimum process unit is determined based on the input, output and recycling of the carbon flow, and then a process model library is constructed by combining several of the minimum process units.

6. The method for analyzing carbon flow according to claim 5, characterized in that: Obtain the LCA master library model, verify the matching degree between the production process and the LCA data based on the process model library and the LCA master library model, match the LCA factors, and build a basic LCA model library, including: Obtain the LCA total library model; Perform consistency check on the process names and system boundaries in the process model library and the LCA master library model to obtain preliminary comparison results; Based on the preliminary comparison results, the interference fit data is reversely traced to check whether the original data collection scenario in the LCA master library model matches the production process in the process model library. If so, the matching LCA factor is obtained; if not, the process model library is returned to be corrected to achieve a degree of matching between the production process and the LCA data, and the LCA factor is matched; a basic LCA model library is constructed based on the matching LCA factors.

7. The method for analyzing carbon flow according to any one of claims 1 to 3, characterized in that: Constructing a carbon flow analysis model; based on the carbon flow analysis model, calibrating the process model library and the LCA model library through a balanced analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library, including: The three-level carbon balance verification framework is based on the process layer, system layer and full chain layer, and the calibration mechanism and verification process are introduced to build a carbon flow analysis model; Inputting LCA factors and process parameters in the LCA model library into the carbon flow analysis model for calculation, and outputting a carbon flow prediction value; Through the closed-loop calibration mechanism of equilibrium analysis, the predicted value of carbon flow is compared with the measured data in the material database. If the comparison result does not reach the threshold, the minimum process unit data in the process model library is adjusted, and an updated LCA model library is generated; The carbon flow analysis model is calculated based on the updated LCA model library until a termination condition is met, and the calibration of the process model library and the LCA model library is completed to obtain the calibrated process model library and LCA model library.

8. The method for analyzing carbon flow according to claim 7, characterized in that: Based on the calibrated process model library and LCA model library, the carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; Then, based on the carbon flow data, calibration is performed through a carbon flow data closed-loop calibration mechanism to obtain the final carbon flow analysis results, including: Based on the calibrated process model library and LCA model library, the carbon flow mathematical model is constructed according to the dynamic equation combined with the neural network structure; the dynamic equation expression is: ； In the formula, Indicates carbon concentration; Indicates the input carbon flow; Indicates the output carbon flow; represents the temperature-dependent carbon conversion rate; Indicates the carbon storage rate; The neural network structure includes an input layer, a hidden layer and an output layer; based on the dynamic equation, the carbon flow data is input into the input layer, and the carbon flow mathematical model is used for calculation, and the output layer outputs the predicted value of the carbon flow analysis result; Preliminarily calibrating the predicted value of the carbon flow analysis result to obtain a preliminary calibrated predicted value; Through the closed-loop calibration mechanism of carbon flow data, the preliminary calibration prediction value is compared with the carbon flow data, the deviation in the model is calculated, and the carbon flow mathematical model is iteratively optimized based on the deviation until the termination condition is met, and the final carbon flow analysis result is output.

9. A device for process-based analysis of carbon flow, characterized in that: The device comprises: A data collection module is used to construct a carbon flow collection scope framework; investigate emission sources based on the carbon flow collection scope framework to obtain carbon flow data; and then construct a material database of energy media based on the carbon flow data; A process model library construction module is used to determine the minimum process unit according to the material database and in combination with the actual production process flow, and then construct a process model library based on the minimum process unit; An LCA model library construction module is used to obtain an LCA total library model, verify the matching degree between the production process and the LCA data based on the process model library and the LCA total library model, match the LCA factors, and construct a basic LCA model library; according to the basic LCA model library, establish a two-dimensional comparison chart of process and carbon flow data, and then screen the basic factor library in the basic LCA model library based on the two-dimensional comparison chart to obtain the required LCA model library; A balance calibration module is used to construct a carbon flow analysis model, and calibrate the process model library and the LCA model library through a balance analysis closed-loop calibration mechanism to obtain a calibrated process model library and LCA model library; The result output module is used to construct a carbon flow mathematical model based on the calibrated process model library and the LCA model library according to the dynamic equation combined with the neural network structure; then based on the carbon flow data, the carbon flow mathematical model is calibrated through the carbon flow data closed-loop calibration mechanism, and the final carbon flow analysis result is obtained through the calibrated carbon flow mathematical model.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method for streamlining the analysis of carbon flow according to any one of claims 1 to 8 are implemented.

Images