Carbon emission verification method and device, electronic equipment and storage medium

Abstract

The application provides a carbon emission verification method and device, electronic equipment and a storage medium, and relates to the field of carbon emission monitoring. The method comprises the following steps: obtaining enterprise data; wherein the enterprise data at least comprises coal quality, heat supply and power generation; calculating a carbon emission characteristic value according to a carbon emission characteristic correlation and the enterprise data by using a target mechanism model; and fitting a carbon emission parameter interval based on the carbon emission characteristic value by using a target prediction model. The method provided by the application can construct a correlation between carbon emission filling data, couple various carbon emission parameters, and verify the carbon emission data of an enterprise by combining a mechanism model and a prediction model, so that the accuracy of the carbon emission data is improved.

Description

Technical Field

[0001] This application relates to the field of carbon emission monitoring, and more specifically, to a carbon emission verification method, apparatus, electronic device, and storage medium. Background Technology

[0002] For carbon-emitting enterprises, in order to achieve carbon peaking and carbon neutrality targets and thus protect the meteorological environment, decisions need to be made regarding carbon emission planning to reduce corporate carbon emissions. When formulating carbon emission plans, carbon allowances and carbon taxes need to be calculated based on corporate production data. Furthermore, the numerous data fields for reported carbon emissions by enterprises, coupled with the risk of inaccurate reporting, inevitably affect the calculation of carbon allowances and carbon taxes. Therefore, there are challenges in determining the correlation between carbon emission parameters and in accurately predicting the precise true carbon emissions of enterprises. Summary of the Invention

[0003] Based on this, the purpose of this application is to provide a carbon emission verification method, apparatus, electronic device and storage medium, which, by mining the correlation between carbon emission reporting data, couples various carbon emission parameters on the basis of big data, thereby calculating the range of reasonable carbon emission values, and providing data support for the formulation of carbon emission plans.

[0004] In a first aspect, embodiments of this application provide a carbon emission verification method, including:

[0005] Acquire enterprise data; wherein, the enterprise data includes at least coal quality, heat supply, and power generation.

[0006] The carbon emission characteristic values ​​are calculated by the target mechanism model based on the correlation of carbon emission characteristics and the enterprise data.

[0007] The carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission characteristic values.

[0008] In the above implementation process, the correlation between carbon emission reporting data can be established, various carbon emission parameters can be coupled, and the carbon emission data of enterprises can be verified by combining mechanistic models and prediction models, thereby improving the accuracy of carbon emission data.

[0009] Optionally, the carbon emission characteristic value may include carbon emission derived characteristics and resultant characteristics;

[0010] Before the carbon emission derived characteristics are calculated by the target mechanism model based on the correlation of carbon emission characteristics and the enterprise data, the method may further include:

[0011] The carbon emission derived features are constructed based on multiple causal features;

[0012] In the initial mechanism model, feature association relationships are constructed based on the selected causal features and effect features. Multiple historical enterprise data are used as inputs to the initial mechanism model. The causal features are extracted from the historical enterprise data, and the values ​​of the effect features are calculated based on the feature association relationships, as well as the deviation between the values ​​of the effect features and the true values ​​are determined.

[0013] Each of the historical enterprise data, along with the corresponding carbon emission derived characteristics and the deviation of the result characteristics from the actual values, is stored separately to form the target mechanism model.

[0014] Optionally, the carbon emission-derived features may include lower heating value, carbon content on the received basis, coupled coal quality, coal quantity, power generation, carbon oxidation rate and heat supply ratio, and the carbon emission-derived features may include coal quality-derived features and conversion efficiency-derived features;

[0015] The construction of the carbon emission derived features based on multiple genetic features may include:

[0016] Coal quality derived features are constructed based on the lower heating value and the carbon content as received during the coal combustion process. The coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the carbon content as received.

[0017] Based on the coupling of coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio in the coal combustion process, a conversion efficiency derivative feature is constructed.

[0018] In the above implementation process, the characteristic parameters of the carbon emission process can be modeled through mechanism, the basic characteristics can be calculated and the corresponding characteristics can be derived, and these characteristics can be compared with the historical data of enterprises, so that the characteristic data can comprehensively characterize the carbon emission process, which can improve the accuracy and reliability of the data.

[0019] Optionally, before the carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission derived features, the method may include:

[0020] Samples are selected from historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and the selected samples are used as training samples for the initial prediction model.

[0021] The initial prediction model is trained based on the training samples to obtain the target prediction model upon completion of training.

[0022] Optionally, training the initial prediction model based on the training samples may include:

[0023] For the dependent features in the training samples, the corresponding probability density curve is fitted based on kernel density estimation, and the value range of the dependent features is determined based on a preset confidence value.

[0024] For the coal-derived features in the carbon emission-derived features, the relationship line between the received basic carbon content and the lower heating value is determined based on linear regression, and the distance of the coal-derived feature points corresponding to each training sample is determined;

[0025] For the conversion efficiency derivative feature in the carbon emission derivative features, a function of installed capacity, carbon content per unit calorific value, and pressure unit type is obtained by fitting the conversion efficiency corresponding to each training sample.

[0026] In the above implementation process, both supervised and unsupervised learning modes can be combined to train the prediction model. This allows the trained target model to make corresponding predictions and calculations for different features, thereby improving the robustness of the prediction model and the accuracy of carbon emission verification.

[0027] Optionally, the carbon emission parameter range obtained by fitting the target prediction model based on the carbon emission characteristic values ​​may include:

[0028] The system receives carbon emission characteristic values ​​calculated by the target mechanism model, determines whether the carbon emission characteristic values ​​are within a reasonable range based on the data of the training samples, and if so, fits the carbon emission parameter range based on the carbon emission characteristic values; otherwise, it returns an unreasonable characteristic value.

[0029] In the above implementation process, the reasonableness of the data can be judged before the feature value is calculated. If the data is unreasonable, the calculation process can be terminated in advance. The verification process can be carried out after the data is determined to be reasonable. This can improve the calculation efficiency and save computing resources.

[0030] Optionally, before the carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission characteristic values, the method may further include:

[0031] Perform data cleaning on the enterprise data;

[0032] If it is determined that data is missing in the enterprise data, the missing data is filled in based on the median value of the data.

[0033] In the above implementation process, data cleaning and imputation can be performed, and missing values ​​of certain features can be intelligently filled in. Samples that meet the mechanistic model can be selected, thereby improving data reliability.

[0034] Secondly, embodiments of this application provide a carbon emission verification device, which may include:

[0035] The acquisition module is used to acquire enterprise data; wherein the enterprise data includes at least coal quality, heating capacity, and power generation.

[0036] The first prediction module is used to calculate carbon emission characteristic values ​​by the target mechanism model based on the correlation of carbon emission characteristics and the enterprise data.

[0037] The second prediction module is used to obtain the carbon emission parameter range by fitting the target prediction model based on the carbon emission characteristic values.

[0038] Optionally, carbon emission characteristics may include carbon emission derived characteristics and resultant characteristics.

[0039] The first prediction module can also be used to construct the carbon emission derived features based on multiple causal features before the target mechanism model calculates the carbon emission derived features based on the carbon emission feature correlation and the enterprise data; in the initial mechanism model, a feature correlation is constructed based on the selected causal features and the effect features, using multiple historical enterprise data as input to the initial mechanism model, extracting the causal features from the historical enterprise data, and calculating the value of the effect feature and determining the deviation between the value of the effect feature and the true value based on the feature correlation; and saving each historical enterprise data and the corresponding carbon emission derived features and the deviation between the value of the effect feature and the true value to form the target mechanism model.

[0040] Optionally, since the features may include lower heating value, received carbon content, coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio, the carbon emission derived features may include coal quality derived features and conversion efficiency derived features. The first prediction module may be specifically used for:

[0041] Coal quality derived features are constructed based on the lower heating value and the carbon content received during the coal combustion process. The coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the carbon content received. Conversion efficiency derived features are constructed based on the coupling of coal quality, coal quantity, power generation, carbon oxidation rate and heat supply ratio during the coal combustion process.

[0042] Optionally, the second prediction module can also be used for:

[0043] Samples are selected from historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and the selected samples are used as training samples for the initial prediction model; the initial prediction model is trained based on the training samples, so as to obtain the target prediction model when the training is completed.

[0044] Optionally, the second prediction module may be specifically used for:

[0045] For the causal features in the training samples, a corresponding probability density curve is fitted based on kernel density estimation, and the value range of the causal features is determined based on a preset confidence value; for the coal quality-derived features in the carbon emission-derived features, a linear regression is used to determine the relationship line between the received basis carbon content and the lower heating value, and the distance between the coal quality feature points corresponding to each training sample is determined; and for the conversion efficiency-derived features in the carbon emission-derived features, a function of installed capacity, unit calorific value carbon content, and pressure unit type is fitted based on the conversion efficiency corresponding to each training sample.

[0046] Optionally, the second prediction module can also be used for:

[0047] The system receives carbon emission characteristic values ​​calculated by the target mechanism model, determines whether the carbon emission characteristic values ​​are within a reasonable range based on the data of the training samples, and if so, fits the carbon emission parameter range based on the carbon emission characteristic values; otherwise, it returns an unreasonable characteristic value.

[0048] Optionally, the carbon emission verification device may also include a preprocessing module for cleaning the enterprise data; if it is determined that the enterprise data is missing, the missing data is filled in based on the median value of the data.

[0049] Thirdly, embodiments of this application provide an electronic device, which includes a memory and a processor. The memory stores program instructions, and when the processor reads and runs the program instructions, it executes the steps in any of the above implementation methods.

[0050] Fourthly, embodiments of this application also provide a computer-readable storage medium storing computer program instructions, which, when read and executed by a processor, perform the steps in any of the above implementations. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a schematic diagram illustrating the steps of the carbon emission verification method provided in the embodiments of this application;

[0053] Figure 2 This is a schematic diagram of the carbon emission verification process provided in the embodiments of this application;

[0054] Figure 3A schematic diagram illustrating the steps of mechanism modeling provided in the embodiments of this application;

[0055] Figure 4 This is a schematic diagram illustrating the steps for constructing carbon emission derived features as provided in an embodiment of this application;

[0056] Figure 5 A schematic diagram illustrating the linear relationship between lower heating value and carbon content of received substrate, and the distribution of lower heating value and carbon content of received substrate, provided for embodiments of this application.

[0057] Figure 6 A schematic diagram illustrating the steps of training the prediction model provided in the embodiments of this application;

[0058] Figure 7 A schematic diagram illustrating an optional model training step provided in an embodiment of this application;

[0059] Figure 8 This is a schematic diagram illustrating the steps of data preprocessing provided in an embodiment of this application;

[0060] Figure 9 This is a schematic diagram of a carbon emission verification device provided in an embodiment of this application. Detailed Implementation

[0061] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. For example, the flowcharts and block diagrams in the drawings illustrate the architecture, functions, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram may represent a module, program segment, or part of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions. In addition, the functional modules in the various embodiments of the present invention may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0062] During the research process, the applicant discovered that there are currently two main methods for monitoring corporate carbon emissions: one is direct detection of carbon emissions based on sensor data, and the other is based on the law of conservation of mass, i.e., judging corporate carbon emissions through coal quality and quantity. In the first method, the sensor-based measurement scheme is highly dependent on the reliability of the sensors and requires simultaneous monitoring of both flow rate and concentration before calculating the carbon emissions. If any sensor malfunctions, the measured data will have a large error. Actual emissions exceeding the monitored values ​​are detrimental to achieving dual carbon targets, while actual emissions falling short of the monitored carbon emissions impose an economic burden on the company.

[0063] In the second approach, carbon emissions based on the conservation of materials and energy require consideration of more parameters, such as coal consumption, basic carbon content, and the lower heating value of coal. Furthermore, companies may use natural gas or oil as auxiliary energy sources, making the calculation of carbon emissions even more complex.

[0064] Therefore, this application provides a carbon emission verification method that uses a mechanistic model to determine the correlation between carbon emission parameters and combines a deep learning model to verify the carbon emission data of enterprises. Please refer to [link / reference]. Figure 1 , Figure 1 This is a schematic diagram illustrating the steps of the carbon emission verification method provided in this application embodiment. The steps of the carbon emission verification method may include:

[0065] In step S11, enterprise data is obtained.

[0066] Enterprise data can include coal quality, heating capacity, and power generation. In practical applications, it can also include other relevant parameter fields, such as enterprise ID, installed capacity, equipment operating hours, heating ratio, power supply, heating capacity, coal consumption, carbon content per unit of calorific value, lower heating value, purchased electricity, and grid emission factor.

[0067] Carbon-emitting enterprises can include petrochemical enterprises, chemical enterprises, building materials enterprises, steel enterprises, non-ferrous metal enterprises, paper enterprises, power enterprises, and aviation enterprises. In this embodiment of the application, power enterprises, i.e., power plants, are used as carbon-emitting enterprises to illustrate the scheme in this application.

[0068] In step S12, the target mechanism model calculates carbon emission characteristic values ​​based on the correlation of carbon emission characteristics and the enterprise data.

[0069] In step S13, the carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission characteristic values.

[0070] Please Figure 1 Based on the above, refer to Figure 2 , Figure 2This is a schematic diagram of the carbon emission verification process provided in this application embodiment. The overall approach of this application is mainly divided into three parts. The first part is to model the data through a mechanism. Through this step, other features can be calculated from basic features and compared with the data reported by enterprises. The second part is to combine the big data of various enterprises to calculate the distribution of various features and derived features based on the mechanism model and fit them into the corresponding machine learning model. The third part is to verify that when new data is loaded, the model will automatically load the mechanism model to calculate the relevant derived features, and then input them together with the relevant features into the machine learning model. Based on these models, the reasonableness of each field reported by the enterprise will be judged. If the data is unreasonable, the corresponding reasons will be given. At the same time, the model matches the reasonable distribution of these features according to a predetermined algorithm, and further gives the reasonable distribution range of the above parameters. By coupling these distributions, the reasonable range of carbon emission values ​​is calculated.

[0071] In this embodiment of the application, regarding step S12, the technical concept of this application is to model the data through a mechanism. Specifically, other features can be calculated based on basic features by the mechanism model, and then compared with the data reported by the enterprise. Therefore, this embodiment of the application provides an implementation method for mechanism modeling. Figure 1 and Figure 2 Based on the above, refer to Figure 3 , Figure 3 This application provides a schematic diagram of the mechanism modeling steps, which may include:

[0072] In step S31, the carbon emission derived features are constructed based on multiple factor features.

[0073] In this embodiment, carbon emission derivative characteristics, including coal quality derivative characteristics and conversion efficiency derivative characteristics, are used as examples for illustration. Please refer to... Figure 3 Based on the above, refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the steps for constructing carbon emission derived features according to an embodiment of this application. The steps for constructing carbon emission derived features may include:

[0074] Step S311: Construct coal quality derived features based on the lower heating value and the received carbon content during the coal combustion process. The coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the received carbon content.

[0075] Step S312: Construct conversion efficiency derived features based on the coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio during the coal combustion process.

[0076] In this embodiment, the analysis of coal quality selects two basic characteristics: lower heating value and as-received carbon content. Please refer to [link / reference]. Figure 5, Figure 5 This application provides a schematic diagram illustrating the linear relationship between lower heating value and carbon content in the received substrate, as well as the distribution of lower heating value and carbon content in the received substrate. In one implementation process, by plotting the above features in two dimensions, the lower heating value and carbon content in the received substrate can be obtained as shown below. Figure 5 The linear relationship shown can be represented by the distance on the graph from the input to each new point in the mechanistic model. The expression is given, where x0 and y0 are the coordinates of the new point on the coordinate axes, and A and B are the parameters of the line corresponding to the linear relationship.

[0077] Regarding the conversion efficiency derived characteristics, the embodiments of this application are based on fundamental characteristics such as coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio. We obtain, among which, ele p Let n be the amount of electricity generated, n be the type of fuel used, and m be the amount of fuel used. i The mass of fuel used, r heats For the heat supply ratio, h_lowheat i For low heating value, x oxidei ε represents the carbon oxidation rate, and eps is a correction parameter.

[0078] It should be understood that the steps for constructing carbon emission derived features provided in the embodiments of this application are not based on... Figure 4 The order shown is a constraint. In the specific implementation process, step S311 can be implemented first and then step S312, or step S312 can be implemented first and then step S311, or either step S311 or step S312 can be selected for execution.

[0079] In this embodiment of the application, by constructing a derived feature, multiple feature values ​​in the carbon emission process can be coupled. By matching whether the distribution of the features is reasonable, the range of reasonable carbon emission values ​​can be calculated. Furthermore, a derived feature is generated from multiple features. When performing a reasonableness analysis on the derived feature, the analysis can be simplified and more multi-dimensional analysis results can be provided, thereby improving the accuracy of carbon emission calculation.

[0080] In step S32, a feature association relationship is constructed in the initial mechanism model based on the selected causal features and effect features. Multiple historical enterprise data are used as inputs to the initial mechanism model. The causal features are extracted from the historical enterprise data, and the value of the effect feature is calculated based on the feature association relationship, as well as the deviation between the value of the effect feature and the true value is determined.

[0081] In this embodiment, the mechanistic model is constructed based on the physical and chemical conservation relationship to establish a correlation equation based on the characteristics of the carbon emission process. Specifically, the enterprise data can first be screened for causal and effect characteristics to identify the characteristics and basic characteristics used for calculation. The basic characteristics and mechanistic model can be arbitrarily selected. In this embodiment, the installed capacity, equipment operating hours, heating ratio, power supply, heat supply, coal consumption, carbon content per unit calorific value, lower heating value, purchased electricity, and grid emission factor are used as causal characteristics. Carbon emissions from purchased electricity, carbon dioxide emissions from generating units, carbon emission intensity from heating, carbon emission intensity from power supply, fuel heat, coal consumption for heating, gas consumption for heating, carbon emissions from fossil fuel combustion, carbon content per unit calorific value, and load factor are used as effect characteristics for illustration.

[0082] A mechanistic model is established based on the physicochemical correlations of the selected features. Using this mechanistic model and the fundamental features, accounting features (effect features) can be calculated. The causal features can be arbitrarily selected fundamental features of the carbon emission process, while the effect features are carbon emission process features that have a causal relationship with the causal features. For example, if installed capacity is chosen as the causal feature, a larger installed capacity corresponds to a higher load factor; therefore, the load factor can be considered an effect feature of the installed capacity. Furthermore, those skilled in the art should understand that the correspondence between the causal and effect features described above is not limited to the above embodiments. After understanding the above correspondence and logic, those skilled in the art can apply the causal and effect features to other specific scenarios without exceeding the scope of this invention.

[0083] For example, the feature correlation equation established in the embodiments of this application may include:

[0084] a. Carbon emissions from burning fossil fuels: Where n represents the type of fuel used, m i The mass of fuel used, C i x represents the carbon content of the fuel. oxidei Carbon oxidation rate;

[0085] b. Carbon emissions from purchased electricity: m eleco2 =c ele γ, where c ele For purchased electricity, γ is the power grid emission factor;

[0086] c. Coal consumption for power generation: The coal consumption for heating is as follows: Where, r heats The heating ratio is given by 'ele', the power supply is given by 'eps', and the correction parameter is given by 'm'. i The coal used for fuel consumption needs to be converted to standard coal.

[0087] d. Carbon emission intensity of electricity supply: Cefossico2 =m co2 *(1-r heats ) / (ele+eps), Heating carbon emission intensity: Ch fossico2 =m co2 *(1-r heats ) / (ele+eps), where m co2 Carbon emissions;

[0088] e. Carbon content per unit calorific value of fuel: mc heat =(c_recivebase i ) / h_lowheat i Among them, c_recivebase i Indicates the received base carbon content, h_lowheat i Indicates the lower heating value;

[0089] f. Fuel calorific value: mc heati =m i *h_lowwheat i *x oxidei ;

[0090] g. Load factor d load =ele p *T r *C load , among which, ele p For power generation, T r C represents the number of operating hours. load This refers to the installed capacity.

[0091] Based on the feature association relationships established above, the result features corresponding to multiple causal features are calculated, and the deviation values ​​between the result features and the corresponding historical enterprise data are obtained.

[0092] In step S33, the historical enterprise data for each enterprise, as well as the corresponding carbon emission derived characteristics and the deviation of the result characteristics from the actual values, are saved to form the target mechanism model.

[0093] Therefore, the embodiments of this application model the characteristic parameters of the carbon emission process through a mechanism, calculate the basic features and derive the corresponding features, and compare these features with the historical data of the enterprise, so that the feature data can comprehensively characterize the carbon emission process, thereby improving the accuracy and reliability of the data.

[0094] Regarding step S13, this application embodiment also provides an implementation method for training the prediction model, please refer to... Figure 6 , Figure 6The diagram illustrates the steps of training a prediction model according to an embodiment of this application. The implementation of prediction model training may include the following steps:

[0095] In step S61, samples are selected from the historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and the selected samples are used as training samples for the initial prediction model.

[0096] In step S62, the initial prediction model is trained based on the training samples to obtain the target prediction model upon completion of training.

[0097] In this embodiment, the primary purpose of the machine learning model is to record historical sample data. Continuing with the example of the features used in the above steps, before training the model, it is necessary to determine the model training samples. In this embodiment, the model training samples are obtained by filtering the company's historical data. Therefore, the method for filtering model training samples in this embodiment can be:

[0098] After receiving multiple feature values ​​calculated by the initial mechanistic model, the initial prediction model can calculate the percentage deviation of each feature value from historical data. Then, based on kernel density estimation, a corresponding deviation probability density curve is fitted. A reasonable boundary range for each feature value is determined based on a preset confidence threshold. The confidence threshold can be set based on prior knowledge or the company's actual situation, or it can be selected using algorithms such as Isolation Forest or Density Clustering. By filtering the company's historical data using the confidence threshold, suitable data is selected as training samples for the initial prediction model.

[0099] After selecting the training samples, the initial prediction model can be trained. Regarding step S62, this embodiment provides an optional model training implementation method, please refer to... Figure 7 , Figure 7 This is a schematic diagram illustrating an optional model training step provided in an embodiment of this application. This step may include:

[0100] In step S621, for the factor features in the training samples, the corresponding probability density curve is fitted based on kernel density estimation, and the value range of the factor features is determined based on a preset confidence value.

[0101] In step S622, for the coal-derived features in the carbon emission-derived features, a linear regression is used to determine the relationship between the received basic carbon content and the lower heating value, and the distance of the coal-derived feature points corresponding to each training sample is determined.

[0102] In step S623, for the conversion efficiency derivative feature in the carbon emission derivative feature, a function of installed capacity, unit calorific value carbon content and pressure unit type is obtained by fitting the conversion efficiency corresponding to each training sample.

[0103] The following is a detailed description of the training of the initial prediction model in the embodiments of this application. The training of the model in the embodiments of this application can be divided into two categories: supervised learning modeling and unsupervised modeling. For the received carbon content, heating carbon emission intensity, power supply carbon emission intensity, heating coal consumption, power supply coal consumption, and low heating value, a suitable probability density curve is fitted by kernel density estimation. Then, a reasonable range of these values ​​is selected by setting a reasonable confidence level. If necessary, models such as isolated forest and density clustering can also be selected for screening to select reasonable model parameters.

[0104] For the derived feature distance, a linear regression is used to obtain a straight line representing the basic carbon content and lower heating value, and the parameters of the corresponding line are obtained. After calculating the distance of each sample to the line using the above line, it is processed in the same way as in the previous steps.

[0105] As for the conversion efficiency of derived features, it can first be fitted into a function of installed capacity, carbon content per unit calorific value and pressure unit type based on the conversion efficiency of the sample. Depending on the actual needs, the fitting process can use models such as XGBoost, Multilayer Perceptron (MLP) or multiple linear regression.

[0106] After training the initial training model using the aforementioned features and determining the corresponding model parameters, the training can be considered complete, and the target training model can be obtained through file training.

[0107] Therefore, in this embodiment of the application, both supervised and unsupervised learning modes can be combined to train the prediction model, which enables the trained target model to make corresponding predictions and calculations for different features, thereby improving the robustness of the prediction model and the accuracy of carbon emission calculation.

[0108] Before training the model using training samples, you can preprocess the data. Please refer to [link / reference needed]. Figure 8 , Figure 8 This is a schematic diagram illustrating the data preprocessing steps provided in an embodiment of this application. The data preprocessing steps may include:

[0109] In step S81, the enterprise data is cleaned.

[0110] In step S82, if it is determined that data in the enterprise data is missing, the missing data is filled in based on the median value of the data.

[0111] Data cleaning can be achieved through statistical analysis to identify potential errors or outliers, by using a simple rule base (common sense rules, business-specific rules, etc.) to check data values, or by using constraints between different attributes and external data to detect and clean the data. When missing values ​​are detected in the data, intelligent imputation can be performed based on the median value of the feature values ​​to select samples that meet the requirements for model training.

[0112] In addition to preprocessing data during model training, the above methods can also be used to clean and fill data before using historical data to build mechanistic models. This can screen out anomalies and ensure that the final carbon emission calculation is reasonable, thereby improving data reliability.

[0113] In an optional embodiment, for the step of using the target prediction model to predict the carbon emission parameter range in step S13, the carbon emission characteristic value calculated by the target mechanism model can be received, and it can be determined whether the carbon emission characteristic value is within a reasonable range based on the data of the training samples. If it is, the carbon emission parameter range is obtained by fitting the carbon emission characteristic value; otherwise, an unreasonable characteristic value is returned.

[0114] When the sample calculated by the target mechanism model is input into the target prediction model, the parameters obtained from previous training are called to examine the features such as basic carbon content, heating carbon emission intensity, power supply carbon emission intensity, heating coal consumption, power supply coal consumption, and lower heating value in the machine learning model, and analyze whether they are reasonable. If they are not reasonable, the unreasonable feature items are returned for cause analysis.

[0115] For coal quality derived characteristics, the same method as in the above steps is used to calculate the distance between the sample and the line and determine whether the distance is reasonable. If it is reasonable, information indicating that the coal quality is reasonable is returned; if it is unreasonable, information indicating that the coal quality may be abnormal is returned.

[0116] Similarly, the conversion efficiency derived from the corresponding items in the input data is compared with the results calculated by the mechanistic model. If the difference is small, it is considered reasonable. If the efficiency calculated by the mechanistic model is much higher than the value calculated by the prediction model, it is necessary to verify whether the lower heating value of coal is abnormal. If it is normal, it can be assumed that other fuels are present; if it is abnormal, the corresponding abnormality information is returned. Considering the actual situation, it is generally not the case that the efficiency calculated by the mechanistic model is much lower than that calculated by the prediction model. Likewise, the conversion efficiency can be adjusted by coupling the lower heating value and the corresponding carbon content to make it as close as possible to the conversion efficiency fitted by the prediction model. Then, the correct carbon emission value can be obtained through regression analysis using the mechanistic model.

[0117] After the above calculation process, the target prediction model can output the corresponding accounting results, which may include the reasonable range of coal quality, the reasonable range of carbon emissions, the results of derivative characteristic calculation and judgment, and the reasonable range of carbon dioxide emissions, etc.

[0118] In summary, the carbon emission verification method provided in this application can establish the correlation between carbon emission reporting data, couple various carbon emission parameters, and combine mechanistic models and prediction models to verify the carbon emission data of enterprises, thereby improving the accuracy of carbon emission data.

[0119] Based on the same inventive concept, this application also provides a carbon emission verification device 90, please refer to... Figure 9 , Figure 9 This is a schematic diagram of a carbon emission verification device provided in an embodiment of this application. The carbon emission verification device 90 may include:

[0120] The acquisition module 91 is used to acquire enterprise data; wherein the enterprise data includes at least coal quality, heating capacity and power generation.

[0121] The first prediction module 92 is used to calculate carbon emission characteristic values ​​by the target mechanism model based on the correlation of carbon emission characteristics and the enterprise data.

[0122] The second prediction module 93 is used to obtain the carbon emission parameter range by fitting the target prediction model based on the carbon emission characteristic values.

[0123] Therefore, the embodiments of this application can improve the accuracy of carbon emission data by constructing the correlation between carbon emission reporting data, coupling various carbon emission parameters, and combining mechanistic models and prediction models to verify the carbon emission data of enterprises.

[0124] Optionally, carbon emission characteristic values ​​may include carbon emission derived characteristic values ​​and resultant characteristic values.

[0125] The first prediction module 92 can also be used to construct the carbon emission derived features based on multiple causal features before the target mechanism model calculates the carbon emission derived features based on the carbon emission feature correlation and the enterprise data; construct feature correlation based on the selected causal features and the effect features in the initial mechanism model, use multiple historical enterprise data as input to the initial mechanism model, extract the causal features from the historical enterprise data, calculate the value of the effect feature based on the feature correlation, and determine the deviation between the value of the effect feature and the true value; and save each historical enterprise data and the corresponding carbon emission derived features and the deviation between the value of the effect feature and the true value to form the target mechanism model.

[0126] Optionally, since the features may include lower heating value, received carbon content, coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio, the carbon emission derived features may include coal quality derived features and conversion efficiency derived features. The first prediction module 92 may be specifically used for:

[0127] Coal quality derived features are constructed based on the lower heating value and the carbon content received during the coal combustion process. The coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the carbon content received. Conversion efficiency derived features are constructed based on the coupling of coal quality, coal quantity, power generation, carbon oxidation rate and heat supply ratio during the coal combustion process.

[0128] Therefore, the embodiments of this application model the characteristic parameters of the carbon emission process through a mechanism, calculate the basic features and derive the corresponding features, and compare these features with the historical data of the enterprise, so that the feature data can comprehensively characterize the carbon emission process, thereby improving the accuracy and reliability of the data.

[0129] Optionally, the second prediction module 93 can also be used for:

[0130] Samples are selected from historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and the selected samples are used as training samples for the initial prediction model; the initial prediction model is trained based on the training samples, so as to obtain the target prediction model when the training is completed.

[0131] Optionally, the second prediction module 93 may be specifically used for:

[0132] For the causal features in the training samples, a corresponding probability density curve is fitted based on kernel density estimation, and the value range of the causal features is determined based on a preset confidence value; for the coal quality-derived features in the carbon emission-derived features, a linear regression is used to determine the relationship line between the received basis carbon content and the lower heating value, and the distance between the coal quality feature points corresponding to each training sample is determined; and for the conversion efficiency-derived features in the carbon emission-derived features, a function of installed capacity, unit calorific value carbon content, and pressure unit type is fitted based on the conversion efficiency corresponding to each training sample.

[0133] Therefore, in this embodiment of the application, both supervised and unsupervised learning modes can be combined to train the prediction model, which enables the trained target model to make corresponding predictions and calculations for different features, thereby improving the robustness of the prediction model and the accuracy of carbon emission verification.

[0134] Optionally, the second prediction module 93 can also be used for:

[0135] The system receives carbon emission characteristic values ​​calculated by the target mechanism model, determines whether the carbon emission characteristic values ​​are within a reasonable range based on the data of the training samples, and if so, fits the carbon emission parameter range based on the carbon emission characteristic values; otherwise, it returns an unreasonable characteristic value.

[0136] Therefore, the embodiments of this application can improve computational efficiency and save computational resources by judging the reasonableness of data before calculating feature values, ending the calculation process in advance when the data is unreasonable, and then verifying the data after it is determined to be reasonable.

[0137] Optionally, the carbon emission verification device 90 may further include a preprocessing module for cleaning the enterprise data; if it is determined that the enterprise data is missing, the missing data is filled in based on the median value of the data.

[0138] Therefore, the embodiments of this application can perform data cleaning and filling, intelligently fill in certain missing feature values, and screen out samples that meet the mechanistic model, thereby improving data reliability.

[0139] Based on the same inventive concept, this application also provides an electronic device, which includes a memory and a processor. The memory stores program instructions, and when the processor reads and runs the program instructions, it executes the steps in any of the above implementation methods.

[0140] Based on the same inventive concept, embodiments of this application also provide a computer-readable storage medium storing computer program instructions, which, when read and executed by a processor, perform the steps in any of the above implementations.

[0141] The computer-readable storage medium can be any medium capable of storing program code, such as Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM). The storage medium stores the program, and the processor executes the program after receiving an execution instruction. The method executed by the electronic terminal as defined in any embodiment of this invention can be applied to the processor or implemented by the processor.

[0142] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0143] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0144] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0145] It can be replaced and can be implemented, wholly or partially, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, wholly or partially, in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated.

[0146] The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0147] In this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, without necessarily requiring or implying any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0148] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A carbon emission verification method, characterized in that, include: Acquire enterprise data; wherein, the enterprise data includes at least coal quality, heat supply, and power generation. The carbon emission characteristic values ​​are calculated by the target mechanism model based on the correlation of carbon emission characteristics and the enterprise data. The carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission characteristic values; The carbon emission characteristics include carbon emission derived characteristics and effect characteristics. Before the carbon emission derived characteristics are calculated by the target mechanism model based on the carbon emission characteristic correlation and the enterprise data, the method further includes: constructing the carbon emission derived characteristics based on multiple causal characteristics; constructing feature correlations in the initial mechanism model based on the selected causal characteristics and effect characteristics, using multiple historical enterprise data as input to the initial mechanism model, extracting the causal characteristics from the historical enterprise data, calculating the value of the effect characteristics based on the feature correlations, and determining the deviation between the value of the effect characteristics and the true value; and saving each historical enterprise data and the corresponding carbon emission derived characteristics and the deviation between the value of the effect characteristics and the true value to form the target mechanism model. The causal features include lower heating value, as-received carbon content, coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio; the carbon emission derived features include coal quality derived features and conversion efficiency derived features; the construction of the carbon emission derived features based on multiple causal features includes: constructing coal quality derived features based on the lower heating value and as-received carbon content during the coal combustion process; wherein, the coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the as-received carbon content; and constructing conversion efficiency derived features based on the coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio during the coal combustion process. Before the carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission derived features, the method includes: selecting samples from historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and using the selected samples as training samples for the initial prediction model; training the initial prediction model based on the training samples to obtain the target prediction model when the training is completed; The training of the initial prediction model based on the training samples includes: for the causal features in the training samples, fitting the corresponding probability density curve based on kernel density estimation, and determining the value range of the causal features based on a preset confidence value; for the coal quality-derived features in the carbon emission-derived features, determining the relationship line between the received basis carbon content and the lower heating value based on linear regression, and determining the distance of the coal quality feature points corresponding to each training sample; for the conversion efficiency-derived features in the carbon emission-derived features, fitting a function of installed capacity, unit calorific value carbon content, and pressure unit type based on the conversion efficiency corresponding to each training sample.

2. The method according to claim 1, characterized in that, The carbon emission parameter range obtained by fitting the target prediction model based on the carbon emission characteristic values ​​includes: The system receives carbon emission characteristic values ​​calculated by the target mechanism model, determines whether the carbon emission characteristic values ​​are within a reasonable range based on the data of the training samples, and if so, fits the carbon emission parameter range based on the carbon emission characteristic values; otherwise, it returns an unreasonable characteristic value.

3. The method according to claim 1, characterized in that, Before the carbon emission parameter range is obtained by fitting the target prediction model based on the carbon emission characteristic values, the method further includes: Perform data cleaning on the enterprise data; If it is determined that data is missing in the enterprise data, the missing data is filled in based on the median value of the data.

4. A carbon emission verification device, characterized in that, include: An acquisition module is used to acquire enterprise data; wherein, the enterprise data includes at least coal quality, heating capacity, and power generation. The first prediction module is used to calculate carbon emission characteristic values ​​by the target mechanism model based on the correlation between carbon emission characteristics and the enterprise data; The second prediction module is used to obtain the carbon emission parameter range by fitting the target prediction model based on the carbon emission characteristic values; The carbon emission characteristics include carbon emission derived characteristics and effect characteristics. The first prediction module is further configured to: construct the carbon emission derived characteristics based on multiple causal characteristics; construct feature association relationships in the initial mechanism model based on the selected causal characteristics and effect characteristics, use multiple historical enterprise data as input to the initial mechanism model, extract the causal characteristics from the historical enterprise data, calculate the value of the effect characteristics based on the feature association relationships, and determine the deviation between the value of the effect characteristics and the true value; save each historical enterprise data and the corresponding carbon emission derived characteristics and the deviation between the value of the effect characteristics and the true value to form the target mechanism model. The aforementioned features include lower heating value, as-received carbon content, coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio; the carbon emission derived features include coal quality derived features and conversion efficiency derived features; the first prediction module is specifically used to: construct coal quality derived features based on the lower heating value and as-received carbon content during the coal combustion process; wherein, the coal quality derived features are the distances of each coal quality feature point in the coordinate system of the lower heating value and the as-received carbon content; and construct conversion efficiency derived features based on the coupled coal quality, coal quantity, power generation, carbon oxidation rate, and heat supply ratio during the coal combustion process; The second prediction module is further configured to: select samples from historical enterprise data and corresponding feature samples calculated by the initial mechanism model, and use the selected samples as training samples for the initial prediction model; train the initial prediction model based on the training samples, so as to obtain the target prediction model when the training is completed; The second prediction module is specifically used for: for the factor features in the training samples, fitting the corresponding probability density curve based on kernel density estimation, and determining the value range of the factor features based on a preset confidence value; for the coal quality derivative features in the carbon emission derivative features, determining the relationship line between the received basis carbon content and the lower heating value based on linear regression, and determining the distance of the coal quality feature points corresponding to each training sample; for the conversion efficiency derivative features in the carbon emission derivative features, fitting a function of installed capacity, unit calorific value carbon content and pressure unit type based on the conversion efficiency corresponding to each training sample.

5. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores program instructions, and when the processor executes the program instructions, it performs the steps of the method according to any one of claims 1-3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, perform the steps of the method according to any one of claims 1-3.

Images