Industrial enterprise carbon emission weak link identification method and system

Abstract

The invention discloses a carbon emission weak link identification method and system for an industrial enterprise, and relates to the technical field of industrial carbon emission management, and the method comprises the steps: collecting production process data of the industrial enterprise, the production process data comprising a process flow diagram, equipment energy consumption data and material consumption data; an actual sample and a theoretical sample are constructed based on production process data, an XGBoost algorithm is adopted to establish a prediction model of characteristic parameters and carbon emission, and a contribution value of each characteristic parameter to the carbon emission is quantified through an SHAP algorithm. By integrating the process flow diagram, the equipment energy consumption and the material consumption data, a complete carbon emission data system is established, and the problems of scattered data acquisition and non-uniform standards in a traditional method are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial carbon emission management, and in particular to a method and system for identifying weak links in carbon emission of industrial enterprises. Background Art

[0002] Carbon emissions in the industrial sector account for about 70% of the total carbon emissions in the whole society, including direct carbon emissions and indirect carbon emissions. Direct carbon emissions mainly come from fossil fuel combustion processes and industrial production processes, while indirect carbon emissions mainly come from the consumption of purchased electricity and heat. At present, the carbon emission management of industrial enterprises mainly adopts a combination of total control and intensity control, and constrains the carbon emission behavior of enterprises by setting carbon emission quotas and establishing carbon trading markets. Traditional carbon emission management methods mainly rely on energy audits and carbon verification, and identify high-energy-consuming links through manual methods. This method requires a lot of human resources and is difficult to achieve refined management. With the development of industrial Internet and big data technology, some companies have begun to try to use digital means to manage carbon emissions, but in the specific implementation process, there are still problems such as incomplete data collection and unsystematic analysis methods.

[0003] The existing carbon emission management technology for industrial enterprises has the following main deficiencies: First, the process of collecting carbon emission data often has problems such as low data quality and inconsistent collection caliber, which makes it difficult to ensure the accuracy and reliability of carbon emission calculation results; second, the existing carbon emission analysis methods are mostly based on static statistical analysis, which is difficult to reflect the dynamic changes in the production process and cannot accurately identify the key factors affecting carbon emissions; third, the traditional carbon emission management methods lack systematicity and pertinence, making it difficult to provide enterprises with accurate emission reduction suggestions, resulting in poor implementation of emission reduction measures. In addition, when dealing with carbon emission issues in complex process flows, existing technologies often need to rely on expert experience and judgment, and lack scientific quantitative evaluation methods, which makes the formulation of emission reduction plans more subjective and uncertain. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method and system for identifying weak links in carbon emissions of industrial enterprises, which can solve the problems mentioned in the background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: A method for identifying weak links in carbon emissions of industrial enterprises, comprising: collecting production process data of the industrial enterprise, wherein the production process data includes a process flow chart, equipment energy consumption data, and material consumption data;

[0007] Based on the production process data, actual samples and theoretical samples are constructed, a prediction model of characteristic parameters and carbon emissions is established using the XGBoost algorithm, and the contribution of each characteristic parameter to carbon emissions is quantified using the SHAP algorithm;

[0008] The contribution values ​​are classified according to the process stages, and the weak links in carbon emissions are determined by comparing the differences in the contribution values ​​of the process stages of actual samples and theoretical samples.

[0009] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the production process data includes energy production data and material consumption data, and the standard coal consumption of the energy production data is calculated according to the following formula:

[0010]

[0011] Among them, E i is the standard coal consumption, A i is the energy consumption, f i is the discount factor.

[0012] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the construction of the theoretical sample includes establishing a material flow and energy flow model for a unit product based on a process flow chart, and calculating the theoretical natural gas consumption according to the following formula:

[0013]

[0014] Among them, E 天然气 is the natural gas consumption, x 1 is the steam flow rate, x 2 is the gas boiler efficiency, r is the steam saturation enthalpy, h 天然气 is the median calorific value of natural gas.

[0015] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the formula for establishing the prediction model using the XGBoost algorithm is:

[0016]

[0017] Among them, f(X) is the output value of the prediction model, f k (X) is the prediction model of the kth decision tree, K is the total number of trees, and X is the feature parameter set.

[0018] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the feature contribution value calculation of the SHAP algorithm includes the calculation of a single feature contribution value and the calculation of the overall feature contribution value:

[0019] The contribution value of a single feature is calculated according to the following formula:

[0020]

[0021] The overall contribution of a feature is calculated using the following formula:

[0022]

[0023] in, is the single SHAP contribution value of feature i, F is the feature set, S is the feature subset that does not contain the feature, and f x (S) is the model output when only the feature subset S is considered, Φ i is the overall SHAP contribution of feature i, N is the number of data points, is the contribution value of feature i at the jth data point;

[0024] The characteristic overall contribution value is used to determine the carbon emission contribution degree of each process stage.

[0025] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the rule for determining the contribution value of the process section is as follows: if the contribution value of the actual sample process section is higher than the first preset threshold of the contribution value of the theoretical sample process section, it is determined to be a first-level carbon emission weak link; if the contribution value of the actual sample process section is lower than the second preset threshold of the contribution value of the theoretical sample process section, it is determined to be a third-level carbon emission weak link; otherwise, it is determined to be a second-level carbon emission weak link.

[0026] As a preferred solution of the method for identifying weak links in carbon emissions of industrial enterprises described in the present invention, the carbon efficiency improvement suggestions are generated through a preset carbon efficiency improvement decision tree, and the decision tree contains improvement suggestions corresponding to weak links in carbon emissions at all levels.

[0027] To further solve the above technical problems, the present invention provides the following technical solutions: A system for identifying weak links in carbon emissions of industrial enterprises, comprising: a data acquisition module, for collecting production process data of industrial enterprises, wherein the production process data includes process flow charts, equipment energy consumption data, and material consumption data;

[0028] A model analysis module is used to construct actual samples and theoretical samples based on the production process data, establish a prediction model of characteristic parameters and carbon emissions using the XGBoost algorithm, and quantify the contribution of each characteristic parameter to carbon emissions using the SHAP algorithm;

[0029] The link identification module is used to classify the contribution values ​​according to the process stages, and determine the weak links of carbon emissions by comparing the differences in the contribution values ​​of the process stages of actual samples and theoretical samples.

[0030] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the method for identifying weak links in carbon emissions of industrial enterprises as described above are implemented.

[0031] A computer-readable storage medium having a computer program stored thereon, characterized in that when the computer program is executed by a processor, the steps of the method for identifying weak links in carbon emissions of industrial enterprises as described above are implemented.

[0032] The beneficial effects of the present invention are as follows: in the data collection link, by integrating the process flow chart, equipment energy consumption and material consumption data, a complete carbon emission data system is established, overcoming the problems of scattered data collection and inconsistent standards in traditional methods; in the model analysis link, the XGBoost algorithm and the SHAP interpretation framework are innovatively combined, which not only realizes the accurate prediction of carbon emissions, but also quantifies the contribution of each characteristic parameter to carbon emissions through the SHAP value, breaking through the limitation of traditional methods relying on expert experience; in the link identification link, by constructing theoretical samples as a benchmark, a quantitative evaluation system for the contribution value of the process section is established, which realizes the accurate identification of weak links in carbon emissions and avoids the deviation caused by subjective judgment in traditional methods. Overall, the present invention combines data-driven methods with industrial production practices, and while ensuring the interpretability of identification results, it significantly improves the automation level and decision-making scientificity of carbon emission management, and provides reliable technical support for industrial enterprises to implement precise emission reduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0034] Figure 1 is a flow chart of the overall method in the present invention;

[0035] Figure 2 It is a schematic diagram of the identification system in the present invention;

[0036] Figure 3 is a flow chart of the identification steps in the present invention;

[0037] Figure 4 A diagram of a computer device in the present invention. DETAILED DESCRIPTION

[0038] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0039] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0040] Example 1, reference Figures 1 to 3 , as an embodiment of the present invention, provides a method for identifying weak links in carbon emissions of industrial enterprises.

[0041] Figure 1 An overall flow chart of a method for identifying weak links in carbon emissions of industrial enterprises is shown, including: S1. collecting production process data of industrial enterprises, the production process data including process flow charts, equipment energy consumption data and material consumption data;

[0042] S2. Build actual samples and theoretical samples based on production process data, use XGBoost algorithm to establish a prediction model of characteristic parameters and carbon emissions, and quantify the contribution of each characteristic parameter to carbon emissions through SHAP algorithm;

[0043] S3. Classify the contribution values ​​by process stage, and determine the weak links in carbon emissions by comparing the differences in contribution values ​​of the process stages between actual samples and theoretical samples.

[0044] Preferably, the present invention includes three main stages: data collection, model analysis and link identification.

[0045] In the data collection stage, the system first obtains the production process flow chart of the industrial enterprise, which clearly shows the production links and process routes of the enterprise. For the main energy-consuming equipment in the flow chart, its basic parameter information is collected, including installed power, number of equipment, energy type and material input and output type. At the same time, the energy consumption data of the equipment is collected through the energy monitoring management system or specially installed sensors, and the material consumption data is obtained from the raw material collection ledger of the enterprise. This systematic data collection method overcomes the problems of scattered data sources and non-uniform formats in traditional methods, and provides a complete and reliable data foundation for subsequent analysis.

[0046] In the model analysis phase, the system first constructs a theoretical sample. Based on the process flow chart, the theoretical material flow and energy flow of each unit of production are calculated, and the market average energy efficiency index is used as a benchmark. At the same time, the system performs granular synchronous processing of actual sample and theoretical sample data to ensure data comparability. Then, the XGBoost algorithm is used to establish a prediction model for characteristic parameters and carbon emissions, and the SHAP algorithm is used to quantify the contribution of each characteristic parameter to carbon emissions. This machine learning-based analysis method breaks through the limitations of traditional energy audits that rely on manual experience, and realizes the automatic identification and quantitative evaluation of factors affecting carbon emissions.

[0047] In the link identification stage, the system classifies and summarizes the SHAP values ​​of each characteristic parameter by process segment, and identifies the weak links in carbon emissions by comparing the difference in process segment contribution values ​​between actual samples and theoretical samples. For the identified weak links, the system automatically generates corresponding carbon efficiency improvement suggestions based on the preset decision tree. This data-driven identification method avoids the deviation caused by subjective judgment in traditional methods and provides enterprises with more targeted and operational improvement suggestions.

[0048] In summary, in the data collection stage, this method solves the problems of unstable data quality and inconsistent collection standards in traditional methods through a unified data collection framework; in the model analysis stage, by combining XGBoost and SHAP algorithms, it realizes the automatic identification and quantitative evaluation of carbon emission influencing factors, overcoming the defect of over-reliance on expert experience in traditional methods; in the link identification stage, through the comparative analysis of actual samples and theoretical samples, an objective evaluation system is established to avoid the deviation caused by subjective judgment in traditional methods. Overall, this solution realizes the digitization and intelligence of the weak link identification process of carbon emissions of industrial enterprises, significantly improves the identification efficiency and accuracy, and provides reliable technical support for enterprises to implement precise emission reduction. Subsequent examples will explain in detail the specific technical details such as data processing, algorithm implementation and judgment rules.

[0049] After obtaining the production process flow chart of an industrial enterprise, the first step is to systematically divide the process segments. According to the material flow and functional characteristics of the production process, the entire production system is divided into n process segments, numbered i = 1, 2, ..., n. The principles for dividing the process segments are as follows:

[0050] According to the integrity of the material conversion process, each process section should have clear input and output boundaries;

[0051] Consider the independence of the energy supply system and ensure that energy consumption data can be independently measured;

[0052] Combined with the company's existing management unit divisions, it facilitates data collection and management.

[0053] Typical process segment divisions include: raw material pretreatment segment (i=1), reaction conversion segment (i=2), separation and purification segment (i=3), product refining segment (i=4) and utility segment (i=5), etc. This systematic division method lays the foundation for subsequent data processing, model analysis and identification of weak links.

[0054] The production process data includes energy production data and material consumption data. The standard coal consumption of energy production data is calculated according to the following formula:

[0055]

[0056] Among them, E i is the standard coal consumption, A i is the energy consumption, f i is the discount factor.

[0057] The construction of the theoretical sample includes establishing the material flow and energy flow model of the unit product based on the process flow chart, and calculating the theoretical natural gas consumption according to the following formula:

[0058]

[0059] Among them, E 天然气 is the natural gas consumption, x 1 is the steam flow rate, x 2 is the gas boiler efficiency, r is the steam saturation enthalpy, h 天然气 is the median calorific value of natural gas.

[0060] The formula for establishing a prediction model using the XGBoost algorithm is:

[0061]

[0062] Among them, f(X) is the output value of the prediction model, f k (X) is the prediction model of the kth decision tree, K is the total number of trees, and X is the feature parameter set.

[0063] The feature contribution value calculation of the SHAP algorithm includes the calculation of the single feature contribution value and the calculation of the overall feature contribution value:

[0064] The contribution value of a single feature is calculated according to the following formula:

[0065]

[0066] The overall contribution of a feature is calculated using the following formula:

[0067]

[0068] in, is the single SHAP contribution value of feature i, F is the feature set, S is the feature subset that does not contain the feature, and f x (S) is the model output when only the feature subset S is considered, Φ i is the overall SHAP contribution of feature i, N is the number of data points, is the contribution value of feature i at the jth data point;

[0069] The overall characteristic contribution value is used to determine the carbon emission contribution of each process stage.

[0070] The rules for determining the contribution value of the process section are as follows: if the contribution value of the actual sample process section is higher than the first preset threshold of the contribution value of the theoretical sample process section, it is determined to be a first-level carbon emission weak link; if the contribution value of the actual sample process section is lower than the second preset threshold of the contribution value of the theoretical sample process section, it is determined to be a third-level carbon emission weak link; otherwise, it is determined to be a second-level carbon emission weak link.

[0071] Carbon efficiency improvement suggestions are generated through a preset carbon efficiency improvement decision tree, which contains improvement suggestions corresponding to the weak links in carbon emissions at all levels.

[0072] The best solution is to address two core technical issues in the processing of carbon emission data for industrial enterprises: first, there are multiple measurement units for enterprise energy consumption data (such as kilowatt-hours of electricity, cubic meters of natural gas, and tons of steam), and the lack of uniform units makes it difficult to directly use the data for carbon emission calculations; second, in the process of theoretical sample construction, especially when it comes to steam systems, traditional methods often use empirical estimates and lack scientific theoretical calculation basis. These problems seriously affect the accuracy and reliability of carbon emission analysis.

[0073] Therefore, the present invention establishes a standardized processing method for energy consumption data based on standard coal. The specific calculation formula is as follows:

[0074]

[0075] Among them, E i A is the standard coal consumption (tons of standard coal), which is used to keep consistent with the national energy statistics standard; i The original energy consumption, retaining the original measurement unit of each energy type (such as kilowatt-hour of electricity, cubic meter of natural gas); i The conversion coefficient (kg standard coal / original unit) is calculated using the conversion coefficient specified in the national standard. By introducing this standardized calculation method, the problem of unified quantification of data for different energy types is solved, providing a reliable data basis for subsequent carbon emission analysis.

[0076] In addition, in the process of constructing theoretical samples, the present invention focuses on solving the theoretical calculation problem of natural gas consumption in the steam system. Based on the principle of conservation of thermodynamic energy, the following calculation formula is derived:

[0077]

[0078] The parameters in this formula are selected based on the following: 1 (steam flow rate, tons): calculated through process material balance, reflecting actual production demand; r (steam saturation enthalpy, kJ / kg): determined by querying the water vapor property table according to the temperature and pressure required by the process; x 2 (Boiler efficiency, dimensionless): Considering the equipment operating status, determined by measured data; h 天然气 (Natural gas calorific value, kilojoules / standard cubic meter): Use the actual measured value provided by the gas supply company.

[0079] After introducing thermodynamic principles into theoretical calculations, this solution has shown significant technical advantages in practical applications. First, by replacing traditional empirical estimates with rigorous theoretical calculations, a scientific theoretical benchmark has been established, providing a reliable reference standard for energy consumption assessment; second, by incorporating actual operating factors such as equipment efficiency into the calculation model, the consistency between theoretical calculations and actual conditions has been improved; most importantly, this thermodynamics-based calculation method provides a quantifiable comparison basis for energy efficiency assessments of different process stages, making the identification of weak links more objective and accurate.

[0080] In industrial practice, to ensure that the solution can play an effective role, corresponding requirements are put forward for the infrastructure and operation management of enterprises. Enterprises should have a complete energy metering system that can provide accurate process parameters (such as steam temperature, pressure, etc.) and equipment operation data (such as boiler efficiency), and the data collection frequency should be no less than once an hour. Although these requirements put forward higher standards for enterprises, it is precisely the guarantee of these basic conditions that can ensure the reliability of data processing results and the accuracy of analysis conclusions.

[0081] In summary, by organically combining national energy statistical standards with actual engineering needs, a unified data processing system has been established, solving the problem of multi-source data integration; by introducing thermodynamic principles into the theoretical sample construction process, scientific theoretical calculations have been achieved, providing a more accurate data foundation and theoretical support for carbon emission management. These innovations not only improve the scientific nature of data processing, but also lay a solid foundation for the subsequent identification of weak links in carbon emissions.

[0082] In the field of carbon emission analysis of industrial enterprises, how to accurately predict carbon emissions and understand the influencing factors has always been a technical difficulty. Traditional carbon emission prediction methods mainly rely on linear regression or simple statistical analysis, which is difficult to handle the complex nonlinear relationship between multidimensional characteristic parameters, resulting in insufficient prediction accuracy. At the same time, there is a lack of quantitative analysis of the impact mechanism of each characteristic parameter on carbon emissions, which makes it difficult to provide a reliable basis for emission reduction decisions.

[0083] In order to solve the problem of prediction accuracy, the present invention uses the XGBoost algorithm to build a prediction model. The output of the model can be expressed as the sum of multiple decision trees:

[0084]

[0085] Among them, f(X) is the final output value of the prediction model, representing the predicted carbon emissions; f k (X) represents the predicted value of the kth decision tree; K is the total number of decision trees, and the optimal value is determined by cross-validation; X is a set of feature parameters, including input variables of multiple dimensions such as process parameters and equipment parameters.

[0086] During the model training process, the gradient boosting strategy is adopted to optimize the loss function through continuous iteration:

[0087]

[0088] Among them, l is the loss function, which is used to measure the deviation between the predicted value and the actual value; Ω is the regularization term, which is used to control the complexity of the model and prevent overfitting. This method based on ensemble learning can effectively capture the nonlinear relationship between features and significantly improve the prediction accuracy.

[0089] After solving the prediction problem, in order to deeply understand the influence of each characteristic parameter on carbon emissions, the present invention introduces the SHAP algorithm to explain the model. The SHAP value of a single characteristic is calculated by the following formula:

[0090]

[0091] in, represents the SHAP contribution value of feature i; F is the set of all features; S is the feature subset that does not contain feature i; f x (S) represents the model output when only feature subset S is considered.

[0092] For the entire dataset, the overall contribution of feature i is obtained by calculating the average SHAP value:

[0093]

[0094] Where N is the total number of data points; is the SHAP contribution value of feature i at the jth data point. This method can not only quantify the importance of features, but also analyze the direction and degree of feature influence.

[0095] In practical applications, the selection of key parameters of the model is crucial. The learning rate is usually selected to be a small value in the range of 0.01-0.1 to ensure stable model convergence; the depth of the tree is determined by grid search, generally controlled at 3-7 layers to avoid overfitting; the subsampling ratio is set between 0.8-0.9 to enhance the generalization ability of the model; the minimum number of sample splits is determined according to the size of the data set to ensure the statistical reliability of each leaf node.

[0096] This analysis method combining XGBoost and SHAP shows unique advantages in industrial applications. The XGBoost model can automatically handle the interaction effects between features without manually setting complex feature combinations, greatly improving modeling efficiency. At the same time, through SHAP value analysis, companies can clearly understand the impact of various process parameters and equipment parameters on carbon emissions, thereby formulating more targeted emission reduction measures.

[0097] To ensure the reliability of the analysis results, this plan puts forward corresponding requirements for data quality. The data volume should be no less than 1,000 samples, the sampling frequency of the features should be consistent, and the data distribution should be representative. On the basis of meeting these conditions, enterprises can obtain accurate prediction results and explainable analysis conclusions, providing a scientific basis for precise emission reduction.

[0098] In summary, this solution has achieved the organic unity of prediction and explanation by introducing advanced machine learning algorithms into the field of industrial carbon emissions analysis; through scientific parameter configuration and training strategies, the reliability and practicality of the analysis results are guaranteed, providing strong technical support for the refined management of carbon emissions of industrial enterprises.

[0099] Preferably, in the practice of carbon emission management of industrial enterprises, the determination of weak links in carbon emissions often relies on manual experience and lacks systematic and quantitative evaluation standards. Based on the above data processing method and feature analysis results, the present invention establishes a complete set of carbon emission weak link determination rule system.

[0100] The establishment of the judgment rule is first based on the quantitative evaluation of the contribution value of the process stage. The contribution of each characteristic parameter obtained by SHAP value analysis is classified and summarized according to the process flow to calculate the comprehensive contribution value of each process stage. For process stage i, the calculation formula for its comprehensive contribution value is:

[0101]

[0102] Where P i represents the set of all characteristic parameters belonging to process stage i, Φ jis the SHAP contribution value of feature j. This data-driven quantitative method avoids the subjectivity of traditional empirical judgment.

[0103] On a quantitative basis, the present invention adopts a three-level judgment criterion to identify weak links in carbon emissions. The first level of judgment is based on the deviation between the contribution value of the process section and the theoretical benchmark, and the relative deviation is calculated:

[0104]

[0105] Among them A i is the actual contribution value of process section i, B i is the theoretical benchmark value of process section i. The theoretical benchmark value is calculated based on theoretical samples and reflects the carbon emission level of the process section under the optimal operating state. When the relative deviation exceeds the set threshold, the process section is marked as a potential weak link. Taking into account the characteristics of different processes, the threshold is set using a hierarchical strategy, with 15% for the main process section, 20% for the auxiliary process section, and 25% for the common process section.

[0106] The second level of judgment introduces time dimension analysis. For the potential weak links marked, further analyze the time variation characteristics of their contribution values. Set the observation period T (usually one month) and calculate the coefficient of variation:

[0107]

[0108] where σ i and μ i are the standard deviation and mean of the contribution value of process section i in period T. The coefficient of variation reflects the stability of the energy efficiency fluctuation of the process section. i When it is greater than 0.3, it indicates that there are obvious operating fluctuations in this process section and special attention should be paid.

[0109] The third level of determination considers process relevance. For the process sections that pass the first two levels of determination, analyze their degree of relevance to the adjacent process sections. Define the correlation coefficient:

[0110]

[0111] Among them C i and C j is the contribution value of the adjacent process section. ij When | is greater than 0.6, it indicates that there is a significant correlation between the two process stages and the optimization scheme needs to be considered in a coordinated manner.

[0112] Based on the three-level judgment results, the present invention further establishes a mechanism for generating carbon efficiency improvement suggestions. Based on the preset decision tree structure, the system comprehensively considers the contribution value deviation, operation stability and correlation degree of the process segment, and automatically generates targeted improvement suggestions. The construction of the decision tree fully considers the process characteristics and actual constraints to ensure that the generated suggestions are operational.

[0113] In summary, through quantitative evaluation indicators, the identification process of weak links is made more objective and standardized; secondly, the multi-level judgment mechanism can comprehensively consider the influencing factors of different dimensions and improve the reliability of the identification results; finally, the recommendation generation mechanism based on the decision tree provides enterprises with a clear direction for improvement.

[0114] Through multi-level judgment rules and scientific threshold setting, the standardization and intelligence of carbon emission weak link identification is achieved. This method not only improves the accuracy of identification, but also provides reliable decision-making support for the continuous improvement of enterprises. Practice has proved that enterprises adopting this method can more accurately identify energy-saving and emission reduction opportunities, formulate more effective improvement measures, and promote the overall improvement of carbon emission management level.

[0115] Example 2, reference Figure 3 , is an embodiment of the present invention, Figure 3 This is a flow chart of the identification steps of the present invention. The details are as follows:

[0116] Step 0: Identification method of weak links in carbon emissions of any industrial enterprise.

[0117] Step 1: Collect the production process flow chart of industrial enterprises to obtain the basic parameters of the main energy-consuming equipment in the process flow, including static information such as energy type, material input and output type, installed power, number of units, etc.

[0118] Step 2: Obtain energy consumption data of major energy-consuming equipment through energy monitoring and management systems or by installing sensors, readers and other collection devices. Collect raw material requisition records of industrial enterprises and obtain material consumption data of enterprises, including but not limited to metal materials, oil, raw gas and other raw materials required for production.

[0119] Step 3: Use the carbon emission factor method to calculate the total carbon emissions of the enterprise based on the enterprise’s energy consumption data.

[0120] Step 4. Establish a theoretical sample based on the production process flow chart of the industrial enterprise. Use the market average energy efficiency index for energy-consuming equipment to calculate the theoretical material flow and energy flow for every ton of product produced. For example, steam is required in the production process of process section A of a certain product. Define the steam flow rate as x1. Steam is generated by a gas boiler. Define the efficiency of the gas boiler as x2. Then, the natural gas consumption EN natural gas = x1*r / x2 / h natural gas can be calculated, where r is the saturated enthalpy of the steam required for the process, and h natural gas is the median calorific value of the natural gas.

[0121] Step 5: Calculate the energy consumption data and material consumption data of the main energy-consuming equipment in the theoretical sample.

[0122] Step 6: Based on the energy consumption data of the theoretical sample, the carbon emission factor method is used to calculate the total carbon emission data of the theoretical sample.

[0123] Step 7: Preprocess the actual sample and theoretical sample data to synchronize the energy consumption of each device, the data of each material, and the total carbon emissions of the enterprise. The following formula can be used to expand the annual carbon emissions to monthly carbon emissions, thereby synchronizing the granularity with the monthly energy consumption data and monthly material data.

[0124] Energy consumption equivalent to standard coal of each energy type The unit is tons of standard coal

[0125] Monthly energy consumption The unit is tons of standard coal

[0126] Monthly Carbon Emissions The unit is tons of carbon dioxide

[0127] The meanings of each reference parameter are as follows:

[0128] A i The consumption of each energy type, in kilograms, cubic meters, kWh, etc.

[0129] f i The standard reduction factor is expressed in kilograms of standard coal per kWh or kilograms of standard coal per cubic meter, etc.

[0130] C year is the total annual carbon emissions, in tons of carbon dioxide per year

[0131] E total Total annual energy consumption, in tons of standard coal per year

[0132] Step 8: Take the energy consumption of each device and the data of each material as the characteristic parameter X=[X 1 ,X 2 ,…,X n], the total carbon emissions of the enterprise is the target Y, and the XGB model is used to establish the prediction relationship between XY. The XGB model is constructed by the following formula:

[0133]

[0134] Among them, f k (X) refers to the prediction model of the kth tree, and K is the total number of trees.

[0135] Step 9: Use SHAP value to quantify feature parameter X = [X 1 ,X 2 ,…,X n ] to the prediction of total carbon emissions Y, the SHAP value can be calculated using the following formula:

[0136]

[0137] in:

[0138] F is the set of all features

[0139] S is the feature subset that does not contain feature i

[0140] |S| is the number of features in set S

[0141] |F| is the total number of features

[0142] f x (S) is the output of the model when only the features in set S are considered

[0143] f x (S∪{i}) is the output of the model when considering the features in set S plus feature i

[0144] The total SHAP value of each feature can be expressed as the average of the SHAP values ​​of the feature over all data points. i :

[0145]

[0146] in:

[0147] N is the number of data points

[0148] φ j i is the feature X at the jth data point i SHAP value

[0149] Step 10: Use weighted average to count the SHAP values ​​of each characteristic parameter into the corresponding process stage.

[0150] Step 11, compare the process section SHAP values ​​of the actual sample and the theoretical sample. If it is 10% higher than the theoretical sample, it is counted as level 1, if it is within the range of ±10%, it is counted as level 2, and if it is 10% lower than the theoretical sample, it is counted as level 3.

[0151] Step 12: Based on the carbon efficiency rating of each process section, identify the process section with the lowest rating, i.e. the weak link in carbon emissions.

[0152] Step 13: Read the corresponding process section carbon efficiency improvement suggestions from the decision tree for the weak links.

[0153] In summary, the present invention uses a digital approach to achieve carbon emission management, which reduces labor costs and improves analysis efficiency; secondly, through comparative analysis of theoretical samples and actual samples, it achieves accurate identification of weak links in carbon emissions; thirdly, based on the feature contribution analysis of the SHAP algorithm, it provides enterprises with a quantifiable basis for emission reduction decision-making, improving the scientificity and operability of emission reduction plans.

[0154] Example 3, reference Figure 2 , as an embodiment of the present invention, provides a carbon emission weak link identification system for industrial enterprises, comprising:

[0155] Data collection module, used to collect production process data of industrial enterprises, including process flow charts, equipment energy consumption data and material consumption data;

[0156] The model analysis module is used to construct actual samples and theoretical samples based on production process data, use the XGBoost algorithm to establish a prediction model of characteristic parameters and carbon emissions, and quantify the contribution of each characteristic parameter to carbon emissions through the SHAP algorithm;

[0157] The link identification module is used to classify the contribution values ​​according to the process stages and determine the weak links in carbon emissions by comparing the differences in the contribution values ​​of the process stages of actual samples and theoretical samples.

[0158] Example 4, reference Figure 4, is an embodiment of the present invention, which is different from the previous embodiment in that: if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0159] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0160] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0161] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0162] Example 5 is an embodiment of the present invention, which provides a method for identifying weak links in carbon emissions of industrial enterprises. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0163] In order to verify the effectiveness of the present invention, the present invention selected the ethylene production line of a large enterprise as the experimental object, and carried out a 6-month application verification of the carbon emission weak link identification method proposed in the present invention. In the experimental process, the production line was first divided into five main process sections: raw material pretreatment, cracking reaction, quenching separation, compression distillation and public works. In the data collection stage, a multidimensional data collection system including energy consumption, process parameters, equipment operation status, etc. was established, and the sampling frequency was set to 10 minutes / time. In order to ensure the controllability of the experiment, the production line was divided into two parallel lines, A and B, of which Line A applied the method of the present invention to manage carbon emissions, and Line B maintained the original management mode. During the experiment, the process conditions, raw material quality and product specifications of the two production lines remained consistent to ensure the comparability of the test results.

[0164] During the implementation of the experiment, data standardization, XGBoost model construction and SHAP feature analysis were performed on Line A in turn. The model training used historical data from the first five months, and prediction verification was performed in the last month. The model parameters were determined by grid search optimization, where the learning rate was set to 0.05, the maximum tree depth was 5, and the subsampling ratio was 0.85. At the same time, the three-level judgment rules were applied to identify and analyze the weak links in carbon emissions, and the deviation threshold of the process segment contribution value was set to 15%, the coefficient of variation threshold was set to 0.3, and the correlation coefficient threshold was set to 0.6. Throughout the experiment, key indicator data such as energy consumption, carbon emissions, anomaly recognition rate, and optimization suggestion adoption rate were strictly recorded.

[0165] Table 1 Key indicator evaluation table for carbon emission management scheme comparison test

[0166]

[0167] Through in-depth analysis of the test data, it can be seen that the scheme of the present invention shows significant advantages in multiple dimensions. In terms of carbon emission management, the average monthly carbon emissions of Line A are 770 tons of CO2 less than Line B, a decrease of 5.65%, and the carbon emissions per unit product are reduced by 5.61%; in terms of operating efficiency, the scheme of the present invention increases the recognition rate of abnormal working conditions to 92.5%, which is 16.7 percentage points higher than the traditional scheme, and the abnormal response time is significantly shortened from 25.3 minutes to 8.5 minutes, which improves the rapid response capability of the system; in terms of optimization suggestions, the scheme of the present invention generates an average of 28.6 optimization suggestions per month, which is 2.3 times that of the traditional scheme, and the adoption rate of suggestions reaches 85.3%, which is significantly higher than the 60.2% of the traditional scheme; in terms of economic benefits, through the implementation of optimization suggestions, Line A achieves an average monthly energy saving of 285.6 tons of standard coal, and the carbon emission reduction cost is reduced by 29.83% compared with the traditional scheme; in terms of system performance, the data processing time is shortened by 67.06%, and the prediction accuracy is increased to 94.2%, which fully proves the technical advantages and economic value of the scheme of the present invention in practical applications. It is particularly noteworthy that the solution of the present invention performs particularly well in identifying abnormal operating conditions and responding quickly, which is of great significance for improving the carbon emission management level of enterprises.

[0168] It is important to note that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for identifying weak links in carbon emissions of industrial enterprises, characterized in that: include: Collecting production process data of industrial enterprises, including process flow charts, equipment energy consumption data and material consumption data; Based on the production process data, actual samples and theoretical samples are constructed, a prediction model of characteristic parameters and carbon emissions is established using the XGBoost algorithm, and the contribution of each characteristic parameter to carbon emissions is quantified using the SHAP algorithm; The contribution values ​​are classified according to process stages, and the weak links in carbon emissions are determined by comparing the differences in contribution values ​​of the process stages between actual samples and theoretical samples.

2. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 1, characterized in that: The production process data includes energy production data and material consumption data. The standard coal consumption of the energy production data is calculated according to the following formula: Among them, E i is the standard coal consumption, A i is the energy consumption, f i is the discount factor.

3. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 2, characterized in that: The construction of the theoretical sample includes establishing a material flow and energy flow model for a unit product based on the process flow chart, and calculating the theoretical natural gas consumption according to the following formula: Among them, E 天然气 is the natural gas consumption, x1 is the steam flow rate, x2 is the gas boiler efficiency, r is the steam saturation enthalpy, h 天然气 is the median calorific value of natural gas.

4. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 1, characterized in that: The formula for establishing the prediction model by the XGBoost algorithm is: Among them, f(X) is the output value of the prediction model, f k (X) is the prediction model of the kth decision tree, K is the total number of trees, and X is the feature parameter set.

5. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 4, characterized in that: The feature contribution value calculation of the SHAP algorithm includes the calculation of a single feature contribution value and the calculation of the overall feature contribution value: The contribution value of a single feature is calculated according to the following formula: The overall contribution of a feature is calculated using the following formula: in, is the single SHAP contribution value of feature i, F is the feature set, S is the feature subset that does not contain feature i, and f x (S) is the model output when only the feature subset S is considered, Φ i is the overall SHAP contribution of feature i, N is the number of data points, is the contribution value of feature i at the jth data point; The characteristic overall contribution value is used to determine the carbon emission contribution degree of each process stage.

6. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 5, characterized in that: The judgment rule for the contribution value of the process section is: if the contribution value of the actual sample process section is higher than the first preset threshold of the contribution value of the theoretical sample process section, it is judged as a first-level carbon emission weak link; If the actual sample process segment contribution value is lower than the second preset threshold of the theoretical sample process segment contribution value, it is determined as a third-level carbon emission weak link; otherwise, it is determined as a second-level carbon emission weak link.

7. The method for identifying weak links in carbon emissions of industrial enterprises according to claim 6, characterized in that: The carbon efficiency improvement suggestion is generated through a preset carbon efficiency improvement decision tree, and the decision tree contains improvement suggestions corresponding to the weak links of carbon emissions at all levels.

8. A system using the method for identifying weak links in carbon emissions of industrial enterprises as claimed in any one of claims 1 to 7, characterized in that: include: A data collection module is used to collect production process data of industrial enterprises, wherein the production process data includes process flow charts, equipment energy consumption data and material consumption data; A model analysis module is used to construct actual samples and theoretical samples based on the production process data, establish a prediction model of characteristic parameters and carbon emissions using the XGBoost algorithm, and quantify the contribution of each characteristic parameter to carbon emissions using the SHAP algorithm; The link identification module is used to classify the contribution values ​​according to the process stages, and determine the weak links of carbon emissions by comparing the differences in the contribution values ​​of the process stages of actual samples and theoretical samples.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for identifying weak links in carbon emissions of an industrial enterprise according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for identifying weak links in carbon emissions of an industrial enterprise according to any one of claims 1 to 7 are implemented.

Images