Industrial enterprise supply chain carbon footprint economic accounting and management method oriented to double carbon targets

Abstract

The invention relates to the technical field of industrial enterprise carbon data management, and discloses an industrial enterprise supply chain carbon footprint economic accounting and management method for a dual-carbon target, and the method comprises the steps: selecting and obtaining a public high-frequency economic agent signal based on a technical premise that core factors for driving industrial product price and carbon emission intensity have homology, dynamically correcting the static reference emission factor through a relevance conduction model; according to the method, a public economic signal is utilized to avoid a traditional mode of relying on static lagging data, the technical problem that the real dynamic state of the carbon intensity of the supply chain cannot be sensed is solved, and the real dynamic state of the carbon intensity of the supply chain can be sensed. The almost real-time carbon risk situation sensing ability of an enterprise is improved, and passive bookkeeping after carbon management is engaged is converted into active intervention fused into daily business.

Description

Technical Field

[0001] This invention relates to an economic accounting and management method for the carbon footprint of industrial enterprise supply chains oriented towards dual carbon objectives, belonging to the field of industrial enterprise carbon data management technology. Background Technology

[0002] Currently, the calculation method is generally based on static emission factors published by the government or industry. This method provides a basis for annual carbon information disclosure and overall corporate carbon asset management. However, industrial production processes are dynamic. The actual carbon emission intensity of suppliers changes continuously due to fluctuations in the power structure, changes in raw material batches, or process adjustments. The annual average factor used in the existing accounting method cannot reflect such high-frequency real fluctuations due to its inherent static attributes. This results in a time scale mismatch between the data and the actual situation when managers make supply chain decisions such as daily procurement.

[0003] To address this issue, requiring suppliers to report carbon emission data in real time is generally not feasible for engineering implementation due to the involvement of commercially sensitive information and increased compliance costs. On the other hand, building complex production process simulations or machine learning models is difficult to widely apply due to its reliance on large amounts of non-public data and high deployment costs. Therefore, there is a lack of effective technical means in this field to meet the data support required for dynamic decision-making without acquiring intrusive data.

[0004] Specifically, existing technologies have the following shortcomings: 1. There is a time disconnect between the accounting results and the dynamic changes in actual production, affecting the effectiveness of supply chain decisions; 2. There is a lack of a technical path to dynamically assess carbon emission intensity using publicly available, non-intrusive information; 3. The use of publicly available economic data remains at a conventional level, failing to explore its intrinsic connection with the dynamic changes in carbon emissions during the production process; Meanwhile, although some technical solutions attempt to optimize the accounting process, their underlying logic still relies on static process data, failing to fundamentally solve the problem of dynamic perception. For example, Chinese invention patent CN115689311B discloses… This invention presents an intelligent carbon data accounting method and system for the industrial product procurement supply chain. This solution aims to systematically process process data throughout the entire lifecycle of industrial products by defining accounting boundaries, developing data selection algorithms, and constructing accounting models. However, this method essentially optimizes the traditional inventory-based accounting logic, still relying on the identification and analysis of actual process data (such as raw material input and energy consumption). This means its effectiveness is highly limited by the availability and timeliness of internal enterprise data, failing to address the key pain point of missing dynamic information due to commercial barriers and data delays. Consequently, it cannot provide enterprises with near-real-time risk insight and decision support capabilities. Therefore, the technical problem this invention aims to solve is how to utilize publicly available high-frequency economic data as proxy signals and establish a correlation transmission model between this data and static emission factors to generate an accounting factor that dynamically reflects changes in carbon emissions at supply chain nodes. Summary of the Invention

[0005] This invention provides an economic accounting and management method for the carbon footprint of industrial enterprise supply chains oriented towards dual carbon objectives. Its main purpose is to solve the problem that existing technologies lack a method for dynamically assessing carbon emission intensity using publicly available, non-intrusive information, which leads to a serious disconnect between the data on which carbon management decisions are based and the actual production situation.

[0006] To achieve the above objectives, this invention provides an economic accounting and management method for the carbon footprint of industrial enterprise supply chains oriented towards dual carbon objectives. This method performs the following steps within a preset calculation cycle: Step a, obtain a static baseline emission factor that characterizes the carbon emission level of a specific node in the supply chain within the baseline period; Step b: Based on the technical premise that the economic factors driving industrial product price fluctuations and the physical factors driving carbon emission intensity have the same driving force, select and obtain a high-frequency economic proxy signal that has a higher time resolution than the calculation period and is related to the same driving force. Step c: Apply the correlation transmission model to generate a dynamic correction value for the static baseline emission factor based on the deviation of the real-time value of the high-frequency economic agent signal from its average value within the baseline period, so as to generate a preliminary dynamic emission factor. Step d, to verify the continued effectiveness of the technological premise of the common driving force in a specific business scenario, the following steps are performed in parallel: Step d1, acquire logistics transportation data orthogonal to the source of the high-frequency economic agent signal, and calculate the orthogonal signal characterizing carbon emissions in the logistics process; Step d2, establish and continuously monitor a verification model to characterize the correlation structure between carbon emissions in the production process estimated by the preliminary dynamic emission factor and the orthogonal signal; Step d3, when the fitting residual of the verification model deviates from a confidence interval preset based on historical statistical data, a warning signal is generated, and the application of the preliminary dynamic emission factor is stopped or the preliminary dynamic emission factor is adjusted.

[0007] Preferably, the correlation transmission model in step c is limited to performing the following operations: substituting the difference between the real-time value of the high-frequency economic agent signal and the baseline average value of the high-frequency economic agent signal into a preset nonlinear transmission function to deterministically calculate the dynamic correction value, and the parameters of the nonlinear transmission function are calibrated by using historical economic data and verified low-frequency carbon emission data for regression analysis.

[0008] Preferably, the verification model in step d2 is a regression model that characterizes the statistical proportional relationship between carbon emissions in the production process and orthogonal signals, and the model's fitting residual is the difference between the real-time estimated value of carbon emissions in the production process and the theoretical value calculated based on the orthogonal signals and the regression model.

[0009] Preferably, the warning signal generated in step d3 further triggers the following steps: capturing and recording the time series of the fitting residual of the verification model and the time series of the high-frequency economic proxy signal within a time window before and after the generation of the warning signal; extracting temporal morphological features, including the steepness of the jump and the frequency of fluctuation, from the two time series to form a real-time anomaly fingerprint; matching the real-time anomaly fingerprint with a preset fingerprint knowledge base that stores a variety of typical anomaly fingerprints and their corresponding root causes; and generating and outputting a warning message containing a diagnostic inference of the root cause of the failure of the same driving force technology premise based on the matching result.

[0010] Preferably, the fingerprint knowledge base includes the following preset fingerprints: a green transition fingerprint used to characterize a supplier's one-time decarbonization transformation, characterized by a one-time sharp drop in the fitting residual time series of the verification model, while the time series of the high-frequency economic proxy signal maintains its original random fluctuation characteristics; and a data contamination fingerprint used to characterize data entry errors, characterized by irregular pulse-like sharp fluctuations in the fitting residual time series of the verification model.

[0011] Preferably, the method further includes the following steps: defining multiple specific nodes with similar attributes in the supply chain as a set of nodes of the same type; periodically statistically analyzing the collective distribution characteristics of all dynamic correction values ​​generated by all nodes in the set of nodes of the same type in the past period; calculating a systematic correction signal characterizing the overall offset of the static baseline emission factor based on the collective distribution characteristics; and applying the systematic correction signal to update the static baseline emission factor itself for subsequent carbon footprint accounting.

[0012] Preferably, the formula for calculating the systematic correction signal in step c is: ,in, It is a systemic correction signal. It is a set of nodes of the same type. It's a past cycle. It is a node In time The generated dynamic correction value, median() is a mathematical operation that takes the median of the set.

[0013] Preferably, the method further includes the following steps: continuously monitoring the time series of multiple economic agent signals corresponding to the same set of nodes, and calculating in real time a coherence metric that characterizes the synchronicity of the collective behavior of multiple time series; when the increase in the value of the coherence metric exceeds a preset impact threshold within a unit time, generating a systemic impact warning signal; in response to the systemic impact warning signal, shortening the time window length used in step b for statistical analysis of collective distribution characteristics, so as to increase the frequency of statistical analysis of collective distribution characteristics.

[0014] Preferably, the high-frequency economic proxy signal in step b is selected from at least one of the following: the industrial electricity price index of the region where the target node is located, the price difference between thermal power and hydropower on-grid electricity in the regional power grid, the coking coal futures price index, the scrap steel price index, and the transportation cost index.

[0015] Preferably, the method further includes the following steps: applying the preliminary dynamic emission factor as a core input parameter to the supply chain procurement decision model when no warning signal is generated, in order to optimize supplier selection and procurement timing.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. This method first uses a publicly released, relatively stable static baseline emission factor as the starting anchor for calculation, ensuring the traceability of subsequent calculations. At the same time, the system continuously acquires a publicly available high-frequency economic proxy signal that is of the same origin as the core driving factor of carbon emissions at a specific node. Through a pre-set correlation transmission model, the economic proxy signal is converted into a dynamic correction value for the aforementioned static baseline emission factor in real time based on the fluctuations over time. The direct result of this process is to generate a dynamic emission factor for enterprises that can reflect the real short-term changes in carbon intensity of upstream suppliers due to factors such as energy structure and process adjustments. This makes subsequent supply chain management and procurement decisions no longer based on the industry average situation lagging by one year, but on the actual carbon emission situation of each supplier's current production and operation.

[0017] 2. Based on the generation of dynamic emission factors, this method further introduces an independent orthogonal signal derived from the enterprise's own logistics and transportation data. This signal is obtained by calculating the transportation carbon footprint of each batch of goods. The system then establishes and continuously monitors a verification model to characterize the correlation structure between the production process carbon emissions estimated by the dynamic emission factors and this orthogonal signal. When the output value of the verification model continuously deviates from its historical statistical distribution, a warning signal is generated. The generation of this signal does not directly determine the level of carbon emissions, but objectively indicates that the correlation between the initially set economic proxy signal and the actual carbon emission intensity may have failed. This provides an inherent logical validity verification mechanism based on multi-dimensional information cross-validation for the entire accounting system.

[0018] 3. To address the issue of the overall drift of the static baseline emission factor over time due to technological advancements or energy structure transformation across the industry, this method defines multiple specific nodes with similar attributes in the supply chain as a set of similar nodes. It then periodically statistically analyzes the collective distribution characteristics of all dynamic correction values ​​generated by all nodes within this set over past periods. Based on these collective distribution characteristics, the system can identify and extract a systematic correction signal characterizing the overall shift of the static baseline emission factor, and apply this signal to update the static baseline emission factor itself. This mechanism enables the benchmark anchor of this method to self-calibrate by continuously learning from the observed group behavior, eliminating reliance on infrequent updates from external authoritative databases and ensuring the long-term effectiveness of the accounting system. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method combining dynamic correction and cross-validation of the present invention; Figure 2 The figure shows the results of a numerical simulation comparison experiment on the effectiveness of the method of the present invention; Figure 3This diagram illustrates the application scenarios and user interaction of the method of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] This invention provides an economic accounting and management method for the carbon footprint of industrial enterprise supply chains, oriented towards dual carbon objectives. The method's execution flow involves using a publicly available economic proxy signal with high time resolution to dynamically correct an officially released static baseline emission factor with low time resolution. Independent logistics data is introduced to continuously cross-validate the effectiveness of this correction process. Within a preset calculation period, such as a working day or a natural week, a preliminary dynamic emission factor reflecting the recent carbon emission intensity of a specific node in the supply chain is output. The method includes a dynamic correction factor generation step, a correlation validity verification step, and a baseline factor adaptive update step. The dynamic correction factor generation step is the core execution part of the method, and the correlation validity verification step is crucial. The validity verification step ensures the operational reliability of the aforementioned core execution part, while the adaptive update step of the benchmark factor ensures the accuracy of the method in long-term operation and prevents systematic deviations due to overall technological advancements in the industry. The procedure for selecting and determining high-frequency economic proxy signals for a specific supply chain node begins by identifying the main carbon emission sources in the node's production process and determining candidate economic indicators directly related to its physical processes based on these emission sources. Subsequently, verified low-frequency unit product carbon emission data for that node over a complete period, such as 24 consecutive months, and high-frequency time series data of all candidate economic indicators within the same period are acquired. By resampling the high-frequency economic indicator data to a resolution consistent with the low-frequency carbon emission data over time, the Pearson correlation coefficient between the two is calculated. and the correlation coefficient The absolute value of the signal exceeds a preset acceptance threshold, such as 0.7, as the criterion for confirming the economic indicator as a valid proxy signal. In the process of generating warning signals, the temporal morphological features extracted from the fitted residual time series of the verification model and the time series of the high-frequency economic proxy signal are specifically calculated as follows: the steepness of the jump is quantified by calculating the first difference value of the residual series, while the fluctuation frequency is characterized by identifying the main frequency component in the energy spectrum after performing a fast Fourier transform on the residual series within a sliding time window. It should be noted that for a newly deployed node, its verification model is in the initial stage of operation, i.e., before... Within a calculation cycle, for example When set to 30, its warning signal generation function is inactive; during this period, the system only records the fitted residual sequence to calculate a stable statistical standard deviation. The standard deviation After the initial phase, it is used to set the confidence intervals required for subsequent operations, for example... .

[0022] In the specific deployment of the method, to address the issue of how to perceive short-term fluctuations in the carbon emission intensity of suppliers without obtaining their internal production data, the system employing this invention is configured to execute a dynamic correction factor generation step within a preset calculation cycle. This step begins with step a, obtaining a static baseline emission factor characterizing the carbon emission level of a specific node in the supply chain within a baseline cycle. This factor can originate from an industry database published by a national authoritative institution. For example, in an application scenario targeting an electrolytic aluminum purchasing company, this static baseline emission factor... The average carbon emission intensity per unit product of the national electrolytic aluminum industry in the previous year can be set as 13.5 tons of CO2 equivalent per ton of aluminum. Then, step b is performed. Based on the technical premise that the economic factors driving industrial product price fluctuations and the physical factors driving carbon emission intensity have the same driving force, a high-frequency economic proxy signal with a time resolution higher than the calculation period and related to the same driving force is selected and acquired. For the aforementioned electrolytic aluminum supplier, considering that its carbon emissions during production mainly originate from electricity consumption, a publicly released economic data reflecting changes in the energy structure of the power grid in its region can be accessed using the existing data interface in the system. For example, the industrial electricity price index of the supplier's region can be selected as the high-frequency economic proxy signal. The signal is updated daily. Next, in step c, a correlation transmission model is applied to generate a dynamic correction value for the static baseline emission factor based on the deviation of the real-time value of the high-frequency economic agent signal from its average value over the baseline period. This generates a preliminary dynamic emission factor. The correlation transmission model is a preset nonlinear transmission function, specifically implemented as a deterministic calculation operation. For example, the preliminary dynamic emission factor... The calculation formula is set as follows: ,in, This is the static baseline emission factor. This represents the real-time value of the high-frequency economic agent signal. The average value of the high-frequency economic proxy signal over the reference period is given, while the parameter is... This is done during the system initialization phase by using historical economic data and verified low-frequency carbon emission data for regression analysis to calibrate the baseline factor. A numerical example illustrates this: for , The sensitivity coefficient was calibrated at 600 yuan / megawatt-hour. It is 0.0005, when the first Daily real-time electricity price If it rises to 650 yuan / MWh, then the preliminary dynamic emission factor for that day... pass The calculation yields a result of approximately This step allows companies to obtain a decision-making basis that is closer to the actual carbon emission situation of their suppliers' current production and operation.

[0023] To address the risk that the aforementioned technological premises might become invalid in specific business scenarios, such as when suppliers sign long-term fixed-price green electricity purchase agreements, the system is configured to execute step d, a procedure for verifying the continued effectiveness of the technological premises driven by the same source. This procedure first, in step d1, acquires logistics transportation data orthogonal to the source of the high-frequency economic agent signal and calculates an orthogonal signal characterizing carbon emissions from the logistics process. Specifically, the system extracts the weight, transportation distance, and transportation mode of each batch of purchased goods from the enterprise transportation management system (TMS), and calculates the transportation carbon footprint of each batch based on a publicly available database of transportation vehicle emission factors. This transportation carbon footprint constitutes a signal independent of electricity market prices. Subsequently, in step d2, a verification model is established and continuously monitored to characterize the correlation structure between production process carbon emissions estimated from preliminary dynamic emission factors and the orthogonal signal. This verification model is a regression model characterizing the statistical proportional relationship between production process carbon emissions and the orthogonal signal. For example, by analyzing historical data, the system establishes a model resembling the carbon emissions per unit of product production. The baseline relationship of carbon emissions per unit of product transportation is established, and the fitting residual of the model under real-time data is continuously calculated. The fitting residual is the difference between the real-time estimated value of carbon emissions in the production process and the theoretical value calculated based on orthogonal signals and regression models. In step d3, when the fitting residual of the verification model deviates from a confidence interval based on historical statistical data, for example, when the absolute value of the fitting residual for five consecutive calculation periods exceeds the 95th percentile of the historical data distribution, a warning signal is generated, and the application of the preliminary dynamic emission factor is stopped or the preliminary dynamic emission factor is adjusted.

[0024] To further enhance the diagnostic value of warning signals and differentiate the root causes of correlated failures, warning signals can trigger an anomaly fingerprint identification process. This process first captures and records the time series of the fitting residuals of the verification model and the time series of the high-frequency economic proxy signal within a time window before and after the generation of the warning signal, for example, within 30 calculation cycles. Subsequently, the system extracts temporal morphological features, including the steepness of the jump and the frequency of fluctuations, from the two time series to form a real-time anomaly fingerprint. This real-time anomaly fingerprint is then matched with a fingerprint knowledge base that stores various typical anomaly fingerprints and their corresponding root causes. The fingerprint knowledge base includes preset fingerprints such as a green transition fingerprint to characterize a supplier's one-time decarbonization transformation, characterized by a one-time sharp drop in the fitting residual time series of the verification model, while the time series of the high-frequency economic proxy signal maintains its original random fluctuation characteristics; and a data contamination fingerprint to characterize data entry errors, characterized by irregular, pulsed, and violent fluctuations in the fitting residual time series of the verification model. Based on the matching results, the system generates and outputs a warning message containing a diagnostic inference about the root cause of the failure of the underlying technology of the same driving force.

[0025] To address the issue of static baseline emission factors drifting over time due to industry technological advancements or energy structure transformations, this method also includes a baseline adaptive update mechanism. This mechanism first defines a set of nodes with similar attributes within the supply chain, such as all cement suppliers, as a cluster of nodes of the same type. Then, periodically, for example, every quarter, it statistically analyzes the collective distribution characteristics of all dynamic correction values ​​generated by all nodes within this cluster over past periods. Based on the technical judgment that the correction value of an individual node should fluctuate around zero in the long run, while a systematic deviation of the group indicates that the baseline itself is inaccurate, the system calculates a systematic correction signal characterizing the overall shift of the static baseline emission factor based on the collective distribution characteristics. The calculation formula is as follows: ,in, For systematic correction signals, A set of nodes of the same type For the past cycle, For nodes In time The generated dynamic correction value, This involves a mathematical operation to obtain the median of the set; finally, the system applies a systematic correction signal to update the static baseline emission factor itself for subsequent carbon footprint calculation. For example, if the calculated... If the value is -0.02, the system will adjust the original static baseline emission factor. Adjusted to This allows the baseline anchor of this method to self-calibrate by learning from the observed group behavior.

[0026] Example 1: In the procurement process of a large equipment manufacturing enterprise, it needs to purchase a batch of high-strength steel for an urgent order. There are two qualified suppliers, Supplier A and Supplier B, to choose from. According to publicly available data released by the industry association in the previous year, the two suppliers have been assigned the same static baseline emission factor. However, it is known that the power grid in the region where Supplier A is located is mainly based on hydropower, and its output fluctuates greatly due to seasonal factors, while Supplier B mainly relies on thermal power, and its energy structure is relatively stable. Given that the procurement decision must be made on the same day and the real-time production data of the two suppliers cannot be obtained, the purchaser cannot determine the carbon emission intensity of each supplier at the current point in time, and therefore faces the risk of incurring unexpected carbon compliance costs.

[0027] To address this situation, the company deployed the accounting and management method of this invention. The system first obtained the same static baseline emission factor for both suppliers and configured different high-frequency economic proxy signals for them. For supplier A, the on-grid electricity price difference between hydropower and thermal power in its regional power grid was selected as the proxy signal, while for supplier B, the domestic coking coal futures price index was selected. On the decision-making day, the system's dynamic correction factor generation step was executed. The calculation results showed that, due to the recent dry season, the proxy signal corresponding to supplier A continuously deviated from its baseline period average, causing its initial dynamic emission factor to be corrected to a value higher than the static baseline emission factor. Meanwhile, supplier B's dynamic emission factor was corrected due to a slight decline in coking coal prices. The initial dynamic emission factor was revised to a value slightly lower than the static baseline emission factor. Faced with this calculation result that contradicted the static data, the system's correlation validity verification step provided the confidence required for decision-making. This step calculated independent orthogonal signals by extracting recent logistics and transportation data from the two suppliers, and then substituted them into their respective verification models for verification. The results showed that the fitting residuals of the verification models of the two suppliers were within the confidence intervals preset based on historical statistical data, and no warning signals were generated. This result confirmed that within this time window, the correlation between the selected economic proxy signal and the carbon emission intensity of the two suppliers remained valid, thus resolving doubts about the reliability of the dynamically revised data.

[0028] Based on the information output from the combined dynamic correction and cross-validation steps, the purchaser ultimately awarded the steel order to supplier B, which had a lower initial dynamic emission factor. In the subsequent supply chain carbon footprint audit, based on the actual energy consumption data disclosed by the suppliers during the corresponding production cycle, it was confirmed that supplier A increased its proportion of purchased thermal power during that period due to insufficient hydropower output, resulting in a higher carbon emission intensity than supplier B. This procurement decision by the company avoided a potentially high-carbon purchase due to information asymmetry. The application of this method transforms the company's procurement management from a supplier selection process relying on static average values ​​to a resource allocation process that optimizes procurement timing based on dynamic carbon intensity signals. The output initial dynamic emission factor is used as the core input parameter of the supply chain procurement decision model, enabling dynamic order allocation among multiple qualified suppliers.

[0029] Example 2: To objectively verify the effectiveness of the method claimed in this invention in dynamically tracking changes in carbon emission intensity at supply chain nodes, this example designed and executed a numerical simulation experiment. The purpose of this experiment was to quantify the difference in accuracy of the method of this invention in estimating carbon emission intensity by comparing it with the method using a static baseline emission factor. The experimental platform was based on a standard computing environment, and a simulation model simulating the operation of a cement production enterprise over 365 consecutive calculation cycles was constructed through software programming. This model could generate three sets of time series data: the first set was a simulated, daily-varying true value sequence of actual emission factors, whose fluctuations were set to be correlated with the source structure of electricity used in production, serving as the evaluation benchmark for all subsequent calculations; the second set was a simulated regional industrial electricity price index sequence coupled with changes in the electricity source structure, serving as a high-frequency economic proxy signal; the third set was a simulated logistics and transportation data sequence related to production and operation activities. The data source is used as the basis for the correlation validity verification step. This experiment sets up two experimental groups: a control group and an experimental group using the method of this invention. The control group uses existing technology, that is, throughout all 365 calculation cycles, it uses the full-cycle arithmetic mean of the actual emission factor true value sequence as a fixed static baseline emission factor for carbon emission accounting. The experimental group uses the method disclosed in this invention. Its initial static baseline emission factor is the same as the control group's set value, but within each calculation cycle, it uses a simulated regional industrial electricity price index as a high-frequency economic proxy signal. Through the correlation transmission model in the specific implementation, the static baseline emission factor is corrected to generate a preliminary dynamic emission factor. The evaluation index of the experiment is to calculate the average absolute error between the emission factor sequences generated by the control group and the actual emission factor true value sequence after all 365 calculation cycles are completed.

[0030] After the experiment was started, the simulation model began to generate data daily, and the experimental group's method processed the input data simultaneously. When the experiment reached the 180th calculation cycle, in order to simulate the scenario where the economic agent signal and carbon emission intensity were decoupled due to the supplier signing a long-term fixed-price green electricity purchase agreement, the simulation model was artificially intervened, interrupting the correlation between the actual emission factor true value sequence and the regional industrial electricity price index sequence, and setting the actual emission factor true value sequence to a lower constant value in subsequent cycles. In the several calculation cycles after this intervention, the fitting residual of the experimental group's correlation validity verification step continuously deviated from the preset confidence interval, thus generating a warning signal. Table 1 shows the key data comparison between the experimental group and the control group on some test dates.

[0031] Table 1: Comparison of key data between the experimental group and the control group on some test dates:

[0032] Referring to Table 1, from day 15 to day 179, there was a persistent deviation between the calculated factors and the true values ​​of the actual emission factors in the control group, while the preliminary dynamic emission factors generated by the experimental group were able to effectively follow the fluctuations of the true values ​​of the actual emission factors. After human intervention on day 180, the preliminary dynamic emission factors of the experimental group continued to output distorted results due to the influence of the electricity price index. By day 184, the correlation validity verification step identified this distortion risk and triggered an alert. In the final error statistics stage, data from the first 179 calculation periods were used for calculation. The mean absolute error of the control group was... The mean absolute error of the experimental group was Experimental data show that the experimental group using the method of this invention can reflect the fluctuation of carbon emission intensity with higher accuracy than the control group using static baseline emission factors, thus reducing the deviation between the calculation results and the actual situation. At the same time, the experimental data also confirms that the correlation validity verification step can identify and output warnings when the correlation between the economic proxy signal and carbon emission intensity fails.

[0033] Example 3: This example combines Figures 1 to 3 This section explains the economic accounting and management methods for the carbon footprint of industrial enterprises' supply chains in response to dual carbon objectives, such as... Figure 1As shown, the process begins with the data input stage, which involves acquiring the static baseline emission factor and high-frequency economic agent signal. It then proceeds to the core dynamic correction factor generation step, where a preliminary dynamic emission factor is generated by applying a correlation transmission model. Simultaneously, orthogonal logistics and transportation data is used as a verification signal, input into the correlation validity verification step to cross-validate the model's effectiveness. If the verification result shows a valid correlation, the generated factor is used in the dynamic emission factor application stage as input to the supply chain procurement decision model. If the verification result shows a invalid correlation, the system generates a warning signal, thereby suspending or adjusting the factor's application and triggering anomaly diagnosis. Furthermore, the process includes a long-term calibration loop for adaptive updating of the baseline factor, which, based on group behavior statistics, self-calibrates the initial static baseline factor.

[0034] like Figure 2 As shown, a series of numerical simulation experiments were conducted to compare the estimation accuracy between the experimental group using the method of this invention and the control group using static factors. The horizontal axis of the graph represents the test date in days, and the vertical axis represents the cumulative mean absolute error, with units of [unit missing]. The experimental results are clearly presented by two curves. The dotted line representing the cumulative average absolute error of the experimental group is lower than the dashed line representing the cumulative average absolute error of the control group. This indicates that the method of the present invention can dynamically track the real fluctuations of carbon emission intensity with higher accuracy. Moreover, after the human intervention point near the 180th test day, the error of the experimental group can quickly identify risks and remain stable, while the error of the control group continues to expand.

[0035] like Figure 3 As shown, the architecture involves two types of core users: procurement / supply chain managers and system administrators / data analysts. Procurement / supply chain managers mainly optimize procurement decisions by obtaining dynamic carbon emission factors and can view the cause diagnosis report after the system issues a risk warning. System administrators / data analysts are responsible for the configuration and maintenance of the backend, including calibrating model parameters, configuring agent signals, and managing the anomaly fingerprint database. The data foundation of the entire system architecture comes from the integration of multiple databases, including industry / official databases, public economic databases, and the company's own enterprise transportation management system (TMS).

[0036] Example 4: In the initial configuration of applying the method of the present invention to a specific cement supplier node, an offline parameter calibration process based on historical data needs to be executed to put the system into operation. This process first calibrates the correlation transmission model. The initial conditions are to obtain two sets of archived historical datasets: one set is the regional industrial electricity price index recorded daily for the past 24 consecutive months by the cement supplier, as a time series of high-frequency economic proxy signals; the other set is the unit product carbon emission intensity data publicly released by the supplier for the past 8 consecutive quarters, as verified carbon emission data. To determine the sensitivity coefficient in the correlation transmission model... The system first performs linear interpolation on the low-frequency quarterly carbon emission intensity data to generate a daily estimated carbon emission intensity sequence aligned with the high-frequency economic proxy signal on the time axis. Subsequently, the average value of this estimated carbon emission intensity sequence is calculated as the static baseline emission factor. The initial value, and the average value of the industrial electricity price index series, are used as the base period average. ;Based on the daily industrial electricity price index and The difference, i.e. As an independent variable, the estimated daily carbon emission intensity relative to The rate of change, i.e. Using the variable as the dependent variable, a least squares linear regression analysis is performed. The slope of the fitted line obtained from this regression analysis is then determined as the sensitivity coefficient. The value; taking a set of values ​​as an example, if the calculated value is... for , If the price is 500 yuan / MWh and the slope of the regression analysis output is 0.0008, then the sensitivity coefficient is... It was set to 0.0008.

[0037] In determining the sensitivity coefficient The process then continues to build the validation model and determine its preset confidence interval, using the calibrated sensitivity coefficients. By comparing historical industrial electricity price indices with the daily preliminary dynamic emission factors over the past 24 months, the system extrapolates daily carbon emissions from the production process. Simultaneously, the system extracts the daily shipment weight and transportation distance of the supplier within the same time period from the company's historical logistics database, calculating daily logistics process carbon emissions as an orthogonal signal. Using daily logistics process carbon emissions as the independent variable and daily production process carbon emissions as the dependent variable, a linear regression analysis is performed again to establish the statistical proportional relationship between the two, and the fitted residual sequence covering the entire historical period is calculated. The system then calculates the standard deviation of this historical fitted residual sequence. The upper and lower limits of the confidence interval are set at three standard deviations, i.e. This interval is used as a threshold to determine whether the correlation has failed during online operation. For the construction of the fingerprint knowledge base, historically known supply chain anomalies with clear causes are retrieved, captured and recorded within the event window. The specific form of the model fitting residual time series and the combined features of the high-frequency economic agent signal time series are verified, and the combined features are stored in the fingerprint knowledge base as the fingerprint of the corresponding anomaly event. After completing the above process, the internal model parameters of the system are all set with values ​​based on historical data calibration, making the entire method ready for online operation.

[0038] Example 5: When the method of the present invention is applied to a set of similar nodes consisting of multiple cement suppliers with similar geographical locations and process routes, in addition to performing carbon emission accounting for each independent node, the system also executes a collective behavior monitoring procedure for the set in parallel. Under this procedure, the system continuously acquires and analyzes multiple high-frequency economic proxy signals corresponding to all nodes in the set, namely the time series of industrial electricity price index and thermal coal futures price index of each supplier's region, and calculates a coherence metric in real time to characterize the synchronicity of the collective behavior of multiple time series. In this example, the metric is determined to be the average Pearson correlation coefficient of multiple time series within a seven-day rolling time window. When the system is running smoothly for a certain calculation cycle, a regional environmental protection production restriction policy is issued, which leads to the unified restriction of the production activities of all cement suppliers in the region in the short term, and is reflected in the highly synchronized fluctuations of their respective thermal coal futures price indices within a few days.

[0039] The system's internal collective behavior monitoring procedure detected that the increase in the value of its calculated coherence metric exceeded a pre-set impact threshold based on historical data statistics within a unit of time. The system then generated a systemic impact early warning signal. In response to this early warning signal, a pre-set adaptive intervention protocol was triggered. This protocol, without interrupting the preliminary dynamic emission factor calculation of each independent node, adjusted the core parameters of the baseline factor adaptive update step. Specifically, the time window length used to statistically analyze the collective distribution characteristics of the dynamic correction values ​​of the same type of node set was shortened from the usual quarter to one week. Through this adjustment, the baseline factor adaptive update step can perform aggregate analysis on recent data that already contains information about this systemic impact at a higher frequency, thereby calculating a systemic correction signal that reflects the impact of this impact more quickly and applying it to the update of the static baseline emission factor itself. This mechanism enables the entire accounting method to switch from quarterly-level adaptation to periodic-level response.

[0040] Example 6: Further, for the warning signal triggered on the 184th day, the system automatically executed the abnormal fingerprint cause inference procedure. This procedure first captured and recorded the time series of the fitting residual of the verification model and the time series of the regional industrial electricity price index within the time window from the 155th to the 184th day. After extracting the temporal morphological features from these two series, the system found that the fitting residual series was characterized by a one-time, steep, and persistently low-level sharp decline, while the time series of the regional industrial electricity price index maintained its original random fluctuation characteristics without synchronous structural changes. The system matched this real-time abnormal fingerprint, which was composed of a unidirectional sharp shift in the residual and no correlation with the proxy signal, with the internally preset fingerprint knowledge base and matched the features of the green transition fingerprint used to characterize the supplier's one-time decarbonization transformation. Finally, the system output a warning message containing a diagnostic inference that the root cause of the anomaly is that the supplier may have undergone a green transition, and prompted the recalibration of the correlation transmission model of this node.

[0041] To further verify from the reverse perspective the crucial role of the correlation validity verification step in the method of this invention in ensuring the reliability of the entire accounting system, the following comparative examples are provided.

[0042] Comparative Example 1: To verify the necessity and crucial role of the correlation validity verification step in the method of this invention in ensuring the reliable operation of the entire accounting system, this comparative example, based on the same numerical simulation test platform and data source as Example 2, designed and executed a comparative verification. The method used in this comparative example is referred to as a dynamic accounting method based solely on economic proxy signals. This method, under all initial conditions and parameter calibrations (including sensitivity coefficients), The values ​​of the emission factors and the calculation model for the dynamic correction factor generation steps are completely consistent with the experimental group in Example 2 of this invention. The essential difference is that the method in this comparative example is set to not execute step d of this invention, that is, the introduction of orthogonal logistics and transportation data and the procedures for constructing and monitoring the correlation validity verification model are completely omitted. The experimental process uses three sets of time series data generated by the simulation model, which are exactly the same as in Example 2. These include: the actual emission factor true value sequence as the evaluation benchmark, the regional industrial electricity price index sequence as the high-frequency input, and a set of logistics and transportation data sequences that are deliberately ignored in this comparative example. The purpose of the experiment is to examine the performance of this method in the absence of a verification mechanism when the correlation between the economic proxy signal and carbon emission intensity, which is the core basis, changes. The experimental results are shown in Table 2. When the experiment reached the 180th calculation cycle, the simulation model was artificially intervened in sync with Example 2 to interrupt the intrinsic correlation between the actual emission factor true value sequence and the regional industrial electricity price index sequence, so as to simulate the real industrial scenario in which the supplier implemented a major decarbonization transformation, but its external economic signals did not change synchronously. See Table 2.

[0043] Table 2: Data Comparison Table between Comparative Example 1 and Embodiment 2 of the Present Invention during Key Test Periods:

[0044] Analysis of the experimental results shows that after the intervention point of the 180th calculation cycle, the true value of the actual emission factor has stabilized at... However, due to the lack of a verification mechanism in the Comparative Example 1 method, its calculated factors still blindly follow the fluctuations of the regional industrial electricity price index, which has no intrinsic correlation. This results in a large and persistent deviation from the actual true value. In periods 180 to 186, the average absolute error of the Comparative Example 1 method was calculated to be as high as... More importantly, this method, by its design principle, cannot self-diagnose this systematic failure, thus continuously outputting severely distorted accounting results to the decision-making system, potentially leading to erroneous high-carbon procurement decisions. In contrast, although the method of Embodiment 2 of this invention also produced distortion in its initial dynamic emission factor from day 180 to day 183, its parallel correlation validity verification step successfully generated a warning signal on day 184 by monitoring the continuous deviation of the fitting residual between the theoretical value calculated from logistics data and the carbon emission calculation value from the production process from the confidence interval, and stopped the application of the erroneous factor. The experimental results of Comparative Example 1 show that, in the absence of a source orthogonal verification mechanism for continuous cross-validation, the method of relying solely on economic proxy signals for dynamic correction has unavoidable technical defects. Although this method can improve accounting accuracy under the premise of effective correlation, it cannot cope with the risk of correlation failure caused by technological transformation or changes in business models, which is common in industrial production practice, and the reliability of its output results cannot be guaranteed.

[0045] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for economic accounting and management of carbon footprint in the supply chain of industrial enterprises with dual carbon objectives, characterized in that, This method performs the following steps within a preset calculation cycle: Step a: Obtain a static baseline emission factor that characterizes the carbon emission level of a preset node in the supply chain within a baseline period; Step b: Based on the technical premise that the economic factors driving industrial product price fluctuations and the physical factors driving carbon emission intensity have the same driving force, select and obtain a high-frequency economic proxy signal that has a higher time resolution than the calculation period and is related to the same driving force. Step c: Apply the correlation transmission model to generate a dynamic correction value for the static baseline emission factor based on the deviation of the real-time value of the high-frequency economic agent signal from its average value within the baseline period, so as to generate a preliminary dynamic emission factor. Step d, to verify the continued effectiveness of the technical premise of the common driving force in the preset business scenario, the following steps are performed in parallel: Step d1, acquire logistics transportation data orthogonal to the source of the high-frequency economic agent signal, and calculate the orthogonal signal characterizing carbon emissions in the logistics process; Step d2, establish and continuously monitor a verification model to characterize the correlation structure between carbon emissions in the production process estimated by the preliminary dynamic emission factor and the orthogonal signal; Step d3, when the fitting residual of the verification model deviates from a preset confidence interval based on historical statistical data, a warning signal is generated, and the application of the preliminary dynamic emission factor is stopped or the preliminary dynamic emission factor is adjusted.

2. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, In step c, the correlation transmission model is limited to performing the following operations: substituting the difference between the real-time value of the high-frequency economic agent signal and the baseline average value of the high-frequency economic agent signal into a preset nonlinear transmission function to deterministically calculate the dynamic correction value, and calibrating the parameters of the nonlinear transmission function by using historical economic data and verified low-frequency carbon emission data for regression analysis.

3. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, The model validation in step d2 is a regression model that characterizes the statistical proportional relationship between carbon emissions in the production process and orthogonal signals. The model's fitting residual is the difference between the real-time estimated value of carbon emissions in the production process and the theoretical value calculated based on the orthogonal signals and the regression model.

4. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, The warning signal generated in step d3 further triggers the following steps: capturing and recording the time series of the fitting residual of the verification model and the time series of the high-frequency economic proxy signal within a time window before and after the generation of the warning signal; extracting temporal morphological features, including the steepness of the jump and the frequency of fluctuation, from the two time series to form a real-time anomaly fingerprint; matching the real-time anomaly fingerprint with a preset fingerprint knowledge base that stores a variety of typical anomaly fingerprints and their corresponding root causes; and generating and outputting a warning message containing a diagnostic inference of the root cause of the failure of the same driving force technology premise based on the matching result.

5. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 4, characterized in that, The fingerprint knowledge base includes the following preset fingerprints: a green transition fingerprint used to characterize a supplier’s one-time decarbonization transformation, characterized by a sharp one-time drop in the time series of the fitting residual of the verification model, while the time series of the high-frequency economic proxy signal maintains its original random fluctuation characteristics; and a data contamination fingerprint used to characterize data entry errors, characterized by irregular, pulse-like sharp fluctuations in the time series of the fitting residual of the verification model.

6. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, The method also includes the following steps: defining multiple specific nodes with similar attributes in the supply chain as a set of nodes of the same type; periodically statistically analyzing the collective distribution characteristics of all dynamic correction values ​​generated by all nodes in the set of nodes of the same type over the past period; calculating a systematic correction signal characterizing the overall offset of the static baseline emission factor based on the collective distribution characteristics; and applying the systematic correction signal to update the static baseline emission factor itself.

7. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 6, characterized in that, The formula for calculating the systematic correction signal in step c is: ,in, It is a systemic correction signal. It is a set of nodes of the same type. It's a past cycle. It is a node In time The generated dynamic correction value, median() is a mathematical operation that takes the median of the set.

8. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 6, characterized in that, The method further includes the following steps: continuously monitoring the time series of multiple economic agent signals corresponding to the same set of nodes, and calculating a coherence metric in real time to characterize the synchronicity of the collective behavior of multiple time series; when the increase in the value of the coherence metric in a unit time exceeds a preset impact threshold, generating a systemic impact warning signal; in response to the systemic impact warning signal, shortening the time window length used in step b for statistical analysis of collective distribution characteristics.

9. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, In step b, the high-frequency economic proxy signal is selected from at least one of the following: the industrial electricity price index of the region where the target node is located, the price difference between thermal power and hydropower on-grid electricity in the regional power grid, the coking coal futures price index, the scrap steel price index, and the transportation cost index.

10. The method for economic accounting and management of carbon footprint in industrial enterprise supply chains oriented towards dual carbon objectives as described in claim 1, characterized in that, The method also includes the following steps: applying the preliminary dynamic emission factor as a core input parameter to the supply chain procurement decision model when no warning signal is generated.

Images