A steel production facility-level carbon emission real-time accounting method and system based on dynamic factor iteration

Abstract

The application provides a kind of steel production facility level carbon emission real-time accounting method and system based on dynamic factor iteration, relating to carbon emission monitoring technical field, by building facility level dynamic emission factor calculation model containing trainable parameters, replace traditional static factor, make accounting result real-time adapt to production condition change, through facility level carbon flow topology network modeling and carbon flow balance equation, form global mass conservation constraint;Through the first loss function and the second loss function, the dynamic emission factor calculation model parameters are optimized, and the trained facility level dynamic emission factor calculation model is obtained, based on real-time preprocessing sensor data and the final dynamic emission factor calculation model, the real-time dynamic emission factor is calculated, and based on the real-time dynamic emission factor, the real-time and accuracy of carbon emission accounting are improved, carbon flow tracking and closed-loop optimization in steel production are realized, to drive energy efficiency improvement and precise carbon reduction.

Description

Technical Field

[0001] This invention relates to the field of carbon emission monitoring technology, and more specifically, to a method and system for real-time calculation of carbon emissions at the steel production facility level based on dynamic factor iteration. Background Technology

[0002] The steel industry accounts for a large proportion of global carbon emissions. Its complex processes and fluctuating operating conditions pose a significant challenge to accurately, in real-time, comprehensively, and comparablely quantifying carbon emissions.

[0003] On the one hand, traditional carbon emission accounting relies on static emission factors and manual sampling verification. Static emission factors refer to emission coefficients that remain constant over a certain period. Their core function is to provide a standardized and simplified benchmark tool for carbon emission calculation. However, static factors ignore the real-time impact of equipment load, fuel composition, and temperature changes, leading to high calculation errors. Although existing carbon emission accounting platforms have introduced IoT data collection, the data from multiple sensor sources are asynchronous and heterogeneous. For example, gas concentration is measured in seconds, while temperature is measured in minutes, lacking a spatiotemporal alignment mechanism and unable to support accurate real-time calculations. On the other hand, current carbon emission accounting calculates carbon emissions as isolated equipment, without constructing the carbon flow topology between processes. It only relies on the Continuous Emission Monitoring System (CEMS) to provide feedback on optimization factor parameters. However, CEMS uses a single equipment as a monitoring unit and does not link the carbon flow relationship between upstream and downstream processes, resulting in a large deviation between upstream output carbon and downstream input carbon. It is impossible to locate leak points, such as valve leakage or unmetered energy consumption, and ignores process carbon balance constraints. Local optimization and global true emissions result in calculation errors. Summary of the Invention

[0004] To address the issues of poor real-time performance and low accuracy in existing carbon emission accounting methods, this invention proposes a real-time carbon emission accounting method and system based on dynamic factor iteration for steel production facilities. This method improves the real-time performance and accuracy of carbon emission accounting, enables carbon flow tracking and closed-loop optimization in steel production, and drives energy efficiency improvement and precise carbon reduction.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] Firstly, this application proposes a real-time carbon emission accounting method for steel production facilities based on dynamic factor iteration, comprising the following steps: A facility-level dynamic emission factor calculation model is constructed, which includes several parameters to be optimized and trained. Obtain initial facility-level dynamic emission factor values ​​and initial calculated carbon emission values ​​for each process; A carbon flow topology network model at the steel production facility level was constructed, and a carbon flow balance equation was generated. Based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, a first loss function is constructed, and a second loss function is constructed based on the carbon flow balance equation. Based on the first loss function and the second loss function, the parameters of the dynamic emission factor calculation model are iteratively optimized and trained. During the optimization process, the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production are updated. When the optimization training is terminated, the parameters of the trained dynamic emission factor calculation model are obtained. Based on the parameters, the final dynamic emission factor calculation model is obtained. Deploy multi-source heterogeneous sensing and data acquisition equipment at key facilities to collect and preprocess key dynamic parameters of steel production in real time. Based on the preprocessed data and the final dynamic emission factor calculation model, the real-time dynamic emission factor is calculated, and based on the real-time dynamic emission factor, facility-level carbon emissions are calculated in real time.

[0007] Preferably, the expression for the facility-level dynamic emission factor calculation model is:

[0008] in, Indicates facility-level dynamic emission factors. , , and All of these represent the parameters to be optimized during training, namely the baseline emission factor weight parameter, the CO concentration change rate weight parameter, the temperature correction weight parameter, and the combustion efficiency weight parameter. This represents the facility's baseline emission factor, which is either the industry default value or a historical statistical value. T_real / T_std represents the rate of change of CO concentration per unit time in the facility, reflecting the degree of incomplete combustion; T_real / T_std represents the ratio of the facility's real-time temperature to the design temperature, correcting for thermal efficiency deviations. This represents the combustion efficiency coefficient based on vibration spectrum analysis.

[0009] Preferably, the process of obtaining the initial facility-level dynamic emission factor value and the initial calculated carbon emission values ​​for each process is as follows: Obtain initial values, including: initial values ​​for several parameters to be optimized and trained in the facility-level dynamic emission factor calculation model, and initial CO concentration per unit time. Initial facility real-time temperature T_real0, initial value of combustion efficiency coefficient based on vibration spectrum analysis. 0; Based on the dynamic emission factor calculation model, the initial facility-level dynamic emission factor value is calculated. ; Based on the initial facility-level dynamic emission factor value Combined with the facility's real-time gas flow rate The initial calculated values ​​of the dynamic carbon emission factors for each process in steel production are given by the following expression: .

[0010] Preferably, the process of modeling the carbon flow topology network at the steel production facility level is as follows: Using specific facilities in high-speed rail production as nodes and carbon carriers as edges, a carbon flow topology relationship is constructed between nodes; the nodes include: blast furnace, converter, and rolling mill, and the carbon carriers include: molten iron, steel billet, and coal gas. Based on the law of conservation of mass, the carbon balance equations for each node are constructed, and the expressions are as follows:

[0011] in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

[0012] Preferably, the process of constructing the first loss function is as follows: Based on the initial calculated values ​​of dynamic carbon emission factors for each process in steel production Compared with CEMS measured emission values To determine the deviation, penalize local accounting errors and construct the first loss function. The expression is:

[0013] The process of constructing the second loss function based on the carbon flow balance equation is as follows: Based on the carbon balance equations of each node, a second loss function is constructed to penalize carbon flow imbalances between processes while satisfying global carbon flow topological constraints. The expression is:

[0014] in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

[0015] Preferably, the process of iteratively optimizing and training the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function is as follows: Constrain and initialize the parameters of the dynamic emission factor calculation model. , , and All satisfy: , , and ∈[0,1]; Based on the first and second loss functions, the parameters of the dynamic emission factor calculation model are obtained using the gradient descent method. , , and Perform iterative optimization training to satisfy:

[0016] in, Including parameters of the dynamic emission factor calculation model , , and Iterative optimization is achieved through gradient descent; This represents the parameter value for the current iteration step, which is updated during the iteration process. This represents the updated parameter value after the next iteration step, based on... Obtained by gradient calculation; This represents the gradient operator. This represents the learning rate, which is a preset value. This represents the carbon flow constraint weight, which is a hyperparameter. Update facility-level dynamic emission factor values, changing the facility-level dynamic emission factor values ​​from their initial values. Iterative updates to , ,..., The calculated values ​​of dynamic carbon emission factors for each process in steel production were changed from the initial values. Iterative updates to , , ,..., n represents the number of training iterations. During iterative training, the iteration terminates when the value of the first loss function is less than a first preset value and the value of the second loss function is less than a second preset value, and the final parameters are output. , , and .

[0017] Preferably, the key facilities for steel production include blast furnace gas outlet, converter flue, and rolling mill heating furnace. The multi-source heterogeneous sensing and acquisition equipment includes: laser gas analyzer, thermocouple array, mass flow meter, and vibration sensor. The key dynamic parameter data for steel production includes: CO concentration per unit time. Real-time temperature of the facility, combustion efficiency coefficient based on vibration spectrum analysis, and real-time gas flow rate of the facility. .

[0018] Preferably, the process of performing real-time facility-level carbon emission accounting is as follows: The pre-processed key dynamic parameter data of steel production are input into the final dynamic emission factor calculation model to calculate the facility-level dynamic emission factor value in real time. ; Based on the facility-level dynamic emission factor value Based on the real-time gas flow rate of the facility, the carbon emission values ​​of each process in steel production are calculated in real time, as expressed by:

[0019] in, This indicates the carbon emission values ​​of each process in steel production.

[0020] Preferably, the method further includes: constructing a digital twin based on the carbon flow topology network of steel production facilities to visualize and display real-time carbon emission data of each facility.

[0021] Secondly, this application also proposes a real-time carbon emission accounting system for steel production facilities based on dynamic factor iteration, the system comprising: The first construction module is used to construct a facility-level dynamic emission factor calculation model, which includes several parameters to be optimized and trained. The first acquisition module is used to acquire the initial facility-level dynamic emission factor value and the initial calculated value of carbon emissions for each process. The second building module is used to model the carbon flow topology network at the steel production facility level and construct the carbon flow balance equation. The third construction module constructs a first loss function based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, and constructs a second loss function based on the carbon flow balance equation. The training module iteratively optimizes the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function. During the optimization process, it updates the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production. When the optimization training is terminated, it obtains the parameters of the trained dynamic emission factor calculation model and obtains the final dynamic emission factor calculation model based on the parameters. The second acquisition module deploys multi-source heterogeneous sensing and acquisition equipment at key facilities to collect key dynamic parameter data of steel production in real time and perform preprocessing. The accounting module calculates the real-time dynamic emission factor based on the preprocessed data and the final dynamic emission factor calculation model, and performs real-time facility-level carbon emission accounting based on the real-time dynamic emission factor.

[0022] Compared with the prior art, the beneficial effects of the technical solution of the present invention are: This invention provides a method and system for real-time carbon emission accounting at the steel production facility level based on dynamic factor iteration. By constructing a facility-level dynamic emission factor calculation model with trainable parameters to replace traditional static factors, the calculation results are adapted to changes in production conditions in real time. A global material conservation constraint is formed through facility-level carbon flow topology network modeling and carbon flow balance equations. The parameters of the dynamic emission factor calculation model are collaboratively optimized using a first loss function and a second loss function to obtain a trained facility-level dynamic emission factor calculation model. Based on real-time preprocessed sensor data and the final dynamic emission factor calculation model, real-time dynamic emission factors are calculated. Based on these real-time dynamic emission factors, the real-time performance and accuracy of carbon emission accounting are improved, enabling carbon flow tracking and closed-loop optimization in steel production, driving energy efficiency improvement and precise carbon reduction. Attached Figure Description

[0023] Figure 1 A flowchart illustrating the real-time carbon emission accounting method for steel production facilities based on dynamic factor iteration proposed in Embodiment 1 of the present invention; Figure 2 A schematic diagram illustrating the carbon flow topology network modeling at the steel production facility level proposed in Embodiment 2 of the present invention; Figure 3 This diagram illustrates the real-time carbon emission accounting system for steel production facilities based on dynamic factor iteration proposed in Embodiment 3 of the present invention. Detailed Implementation

[0024] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this application. To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions. It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.

[0025] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0026] Example 1 like Figure 1 As shown in the figure, this embodiment presents a real-time carbon emission accounting method for steel production facilities based on dynamic factor iteration, which includes the following steps: S1: Construct a facility-level dynamic emission factor calculation model, which includes several parameters to be optimized and trained; S2: Obtain initial facility-level dynamic emission factor values ​​and initial calculated carbon emission values ​​for each process; S3: Perform carbon flow topology network modeling at the steel production facility level and construct carbon flow balance equations; S4: Based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, a first loss function is constructed, and a second loss function is constructed based on the carbon flow balance equation. S5: Based on the first loss function and the second loss function, iteratively optimize and train the parameters of the dynamic emission factor calculation model. During the optimization process, update the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production. When the optimization training is terminated, obtain the parameters of the trained dynamic emission factor calculation model. Based on the parameters, obtain the final dynamic emission factor calculation model. S6: Deploy multi-source heterogeneous sensing and acquisition equipment at key facilities to collect key dynamic parameter data of steel production in real time and perform preprocessing. S7: Based on the preprocessed data and the final dynamic emission factor calculation model, calculate the real-time dynamic emission factor, and perform real-time facility-level carbon emission accounting based on the real-time dynamic emission factor.

[0027] In this embodiment, considering the shortcomings of static emission factors in real-time carbon emission accounting, a facility-level dynamic emission factor calculation model is constructed for steel production. Several parameters to be optimized and trained are introduced to quantify the impact of different operating conditions on the emission factor. Then, based on empirical settings or historical data fitting, initial values ​​for the parameters to be optimized and trained are obtained. Based on these initial values, initial facility-level dynamic emission factor values ​​are obtained, and further, initial calculated carbon emission values ​​for each process are obtained. Considering that current carbon emission accounting calculates carbon emissions as isolated equipment, without constructing inter-process carbon flow topology relationships, and only relying on feedback from the continuous emission monitoring system to optimize factor parameters, while CEMS uses a single equipment as a monitoring unit and does not connect the carbon flow relationships of upstream and downstream processes, a facility-level carbon flow topology network model is performed for steel production, and a carbon flow balance equation is constructed to form a global mass conservation constraint. In this embodiment, CEMS is deployed at the chimneys or flues of facilities such as blast furnaces, converters, and heating furnaces. Each device serves as a monitoring unit, directly providing real-time emission data and a true benchmark value for the first loss function. Based on this, the first loss function is constructed to achieve local accuracy constraints. Based on the facility-level carbon balance equation, an imbalance in carbon flow between processes is penalized, and a second loss function is constructed to achieve global carbon balance constraints. The parameters of the dynamic emission factor calculation model are iteratively optimized and trained based on the first and second loss functions to obtain well-trained parameters that satisfy both local accuracy and global carbon balance constraints. Finally, based on the collected key dynamic parameter data of steel production and the final dynamic emission factor calculation model, real-time facility-level carbon emission accounting is performed. The real-time carbon emission accounting method for steel production facilities based on dynamic factor iteration provided in this embodiment reduces the poor real-time performance caused by operating condition fluctuations and avoids low accuracy caused by isolated equipment accounting through online optimization of dynamic factors and carbon flow topology constraints, providing highly reliable data for carbon reduction decisions in the steel industry.

[0028] Example 2 In this embodiment, the expression for the facility-level dynamic emission factor calculation model, constructed for specific facilities or processes in steel production, such as blast furnaces and converters, is as follows:

[0029] in, Indicates facility-level dynamic emission factors. , , and All of these represent the parameters to be optimized during training, namely the baseline emission factor weight parameter, the CO concentration change rate weight parameter, the temperature correction weight parameter, and the combustion efficiency weight parameter. This represents the facility's baseline emission factor, which is either the industry default value or a historical statistical value. T_real / T_std represents the rate of change of CO concentration per unit time in the facility, reflecting the degree of incomplete combustion; T_real / T_std represents the ratio of the facility's real-time temperature to the design temperature, correcting for thermal efficiency deviations. This represents the combustion efficiency coefficient based on vibration spectrum analysis. In this embodiment, these parameters are adjustment parameters of the dynamic emission factor model, which can quantify the influence of different operating condition variables on the emission factor. The idle time of the industry default benchmark emission factor is controlled. The larger the size, the more the model relies on historical experience; The contribution of the rate of change in Co concentration reflects the degree of incomplete combustion. The larger the value, the more sensitive the model is to fluctuations in combustion state. The contribution of the temperature correction term reflects the deviation in thermal efficiency. The larger the value, the more sensitive the model is to temperature fluctuations. The contribution of combustion efficiency control reflects the equipment's operating status. The larger the value, the more sensitive the model is to equipment failure and aging.

[0030] In this embodiment, the process of obtaining the initial facility-level dynamic emission factor value and the initial calculated carbon emission values ​​for each process is as follows: Obtain initial values, including: initial values ​​for several parameters to be optimized and trained in the facility-level dynamic emission factor calculation model, and initial CO concentration per unit time. Initial facility real-time temperature T_real0, initial value of combustion efficiency coefficient based on vibration spectrum analysis. 0; Based on the dynamic emission factor calculation model, the initial facility-level dynamic emission factor value is calculated. ; Based on the initial facility-level dynamic emission factor value Combined with the facility's real-time gas flow rate The initial calculated values ​​of the dynamic carbon emission factors for each process in steel production are given by the following expression: .

[0031] like Figure 2 As shown, the process of modeling the carbon flow topology network at the steel production facility level is as follows: Using specific high-speed rail production facilities as nodes and carbon carriers as edges, a carbon flow topology is constructed between the nodes. The nodes include blast furnaces, converters, and rolling mills, and the carbon carriers include molten iron, steel billets, and coal gas. Specifically, in this embodiment, the topology is defined as follows: [See...] Figure 2 With blast furnace, converter, and rolling mill as core equipment, each node also includes auxiliary facilities such as gas holders and power generation boilers. In the carbon flow topology network, with blast furnace as the source node and converter as the target node, molten iron serves as the carbon carrier; with converter as the source node and rolling mill as the target node, steel billet serves as the carbon carrier; with blast furnace as the source node and gas holder as the target node, blast furnace gas serves as the carbon carrier; and with converter as the source node and gas holder as the target node, converter gas serves as the carbon carrier.

[0032] Then, based on the law of conservation of mass, the carbon balance equations for each node are constructed, with the following expressions:

[0033] in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

[0034] The process of constructing the first loss function is as follows: Based on the initial calculated values ​​of dynamic carbon emission factors for each process in steel production Compared with CEMS measured emission values To determine the deviation, penalize local accounting errors and construct the first loss function. ( (This is continuously updated during subsequent iterations and optimization training), the expression is:

[0035] The process of constructing the second loss function based on the carbon flow balance equation is as follows: Based on the carbon balance equations of each node, a second loss function is constructed to penalize carbon flow imbalances between processes while satisfying global carbon flow topological constraints. The expression is:

[0036] in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

[0037] In this embodiment, during the iterative optimization training of the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function, the initial calculated value of the dynamic carbon emission factor of each process in steel production is the starting point value of the model optimization, which is calculated from the initial parameters, rather than the final value. The initial value will be continuously adjusted as the model parameters are optimized.

[0038] The process of iteratively optimizing and training the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function is as follows: Constrain and initialize the parameters of the dynamic emission factor calculation model. , , and All satisfy: , , and ∈[0,1]. In this embodiment, the parameter Take 0.7, initial value Take 0.2, initial value Take 0.5, initial value Take 0.05.

[0039] Based on the first and second loss functions, the parameters of the dynamic emission factor calculation model are obtained using the gradient descent method. , , and Perform iterative optimization training to satisfy:

[0040] in, Including parameters of the dynamic emission factor calculation model , , and The unknown value is optimized iteratively using gradient descent. This represents the parameter value at the current iteration step, which is updated during the iteration process, such as in the k-th iteration. k , , k and k , This represents the updated parameter value after the next iteration step, based on... Obtained by gradient calculation; This represents the gradient operator. This represents the learning rate, which is a preset value. λ represents the carbon flow constraint weight, which is a hyperparameter. In this embodiment, it needs to be set according to business requirements. For example, if the carbon balance requirement is high, then λ should be increased.

[0041] In this embodiment, parameters are used. Taking gradient calculation as an example, the process satisfies:

[0042] Update via gradient descent ,optimization The accuracy, to meet and .

[0043] Update facility-level dynamic emission factor values, changing the facility-level dynamic emission factor values ​​from their initial values. Iterative updates to , ,..., The calculated values ​​of dynamic carbon emission factors for each process in steel production were changed from the initial values. Iterative updates to , , ,..., n represents the number of training iterations. During iterative training, the iteration terminates when the value of the first loss function is less than a first preset value and the value of the second loss function is less than a second preset value, and the final parameters are output. , , and .

[0044] Key facilities in steel production include blast furnace gas outlet, converter flue, and rolling mill heating furnace. The multi-source heterogeneous sensing and acquisition equipment includes: laser gas analyzer, thermocouple array, mass flow meter, and vibration sensor. Key dynamic parameter data for steel production includes: CO concentration per unit time. Real-time temperature of the facility, combustion efficiency coefficient based on vibration spectrum analysis, and real-time gas flow rate of the facility. Specifically, the preprocessing includes: interpolating and aligning the data collected by the multi-source heterogeneous sensing devices based on the equipment process timing, and cleaning abnormal data.

[0045] The process of performing real-time facility-level carbon emission accounting is as follows: The pre-processed key dynamic parameter data of steel production are input into the final dynamic emission factor calculation model to calculate the facility-level dynamic emission factor value in real time. ; Based on the facility-level dynamic emission factor value Based on the real-time gas flow rate of the facility, the carbon emission values ​​of each process in steel production are calculated in real time, as expressed by:

[0046] in, This indicates the carbon emission values ​​of each process in steel production.

[0047] In this embodiment, the method further includes: constructing a digital twin based on the carbon flow topology network at the steel production facility level, and visually displaying real-time carbon emission data for each facility. Specifically, firstly, dynamic rendering of carbon flow is performed based on Unity3D, according to the following correspondence:

[0048] It displays real-time changes in thickness and can issue a red alert when carbon flow is abnormal.

[0049] Example 3 like Figure 3 As shown, this embodiment also proposes a real-time carbon emission accounting system for steel production facilities based on dynamic factor iteration. The system includes: The first construction module is used to construct a facility-level dynamic emission factor calculation model, which includes several parameters to be optimized and trained. The first acquisition module is used to acquire the initial facility-level dynamic emission factor value and the initial calculated value of carbon emissions for each process. The second building module is used to model the carbon flow topology network at the steel production facility level and construct the carbon flow balance equation. The third construction module constructs a first loss function based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, and constructs a second loss function based on the carbon flow balance equation. The training module iteratively optimizes the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function. During the optimization process, it updates the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production. When the optimization training is terminated, it obtains the parameters of the trained dynamic emission factor calculation model and obtains the final dynamic emission factor calculation model based on the parameters. The second acquisition module deploys multi-source heterogeneous sensing and acquisition equipment at key facilities to collect key dynamic parameter data of steel production in real time and perform preprocessing. The accounting module calculates the real-time dynamic emission factor based on the preprocessed data and the final dynamic emission factor calculation model, and performs real-time facility-level carbon emission accounting based on the real-time dynamic emission factor.

[0050] The same or similar labels correspond to the same or similar parts; The terms used to describe positional relationships in the accompanying drawings are for illustrative purposes only and should not be construed as limiting this application. Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A real-time carbon emission accounting method for steel production facilities based on dynamic factor iteration, characterized in that, Includes the following steps: A facility-level dynamic emission factor calculation model is constructed, which includes several parameters to be optimized and trained. Obtain initial facility-level dynamic emission factor values ​​and initial calculated carbon emission values ​​for each process; A carbon flow topology network model at the steel production facility level was constructed, and a carbon flow balance equation was generated. Based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, a first loss function is constructed, and a second loss function is constructed based on the carbon flow balance equation. Based on the first loss function and the second loss function, the parameters of the dynamic emission factor calculation model are iteratively optimized and trained. During the optimization process, the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production are updated. When the optimization training is terminated, the parameters of the trained dynamic emission factor calculation model are obtained. Based on the parameters, the final dynamic emission factor calculation model is obtained. Deploy multi-source heterogeneous sensing and data acquisition equipment at key facilities to collect and preprocess key dynamic parameters of steel production in real time. Based on the preprocessed data and the final dynamic emission factor calculation model, the real-time dynamic emission factor is calculated, and based on the real-time dynamic emission factor, facility-level carbon emissions are calculated in real time.

2. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 1, characterized in that, The expression for the facility-level dynamic emission factor calculation model is as follows: in, Indicates facility-level dynamic emission factors. , , and All of these represent the parameters to be optimized during training, namely the baseline emission factor weight parameter, the CO concentration change rate weight parameter, the temperature correction weight parameter, and the combustion efficiency weight parameter. This represents the facility's baseline emission factor, which is either the industry default value or a historical statistical value. T_real / T_std represents the rate of change of CO concentration per unit time in the facility, reflecting the degree of incomplete combustion; T_real / T_std represents the ratio of the facility's real-time temperature to the design temperature, correcting for thermal efficiency deviations. This represents the combustion efficiency coefficient based on vibration spectrum analysis.

3. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 2, characterized in that, The process of obtaining the initial facility-level dynamic emission factor value and the initial calculated carbon emission values ​​for each process is as follows: Obtain initial values, including: initial values ​​for several parameters to be optimized and trained in the facility-level dynamic emission factor calculation model, and initial CO concentration per unit time. Initial facility real-time temperature T_real0, initial value of combustion efficiency coefficient based on vibration spectrum analysis. 0; Based on the dynamic emission factor calculation model, the initial facility-level dynamic emission factor value is calculated. ; Based on the initial facility-level dynamic emission factor value Combined with the facility's real-time gas flow rate The initial calculated values ​​of the dynamic carbon emission factors for each process in steel production are given by the following expression: 。 4. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 1, characterized in that, The process of modeling the carbon flow topology network at the steel production facility level is as follows: Using specific high-speed rail production facilities as nodes and carbon carriers as edges, we construct the carbon flow topology between nodes; The nodes include: blast furnace, converter, and rolling mill; the carbon carrier includes: molten iron, steel billet, and coal gas. Based on the law of conservation of mass, the carbon balance equations for each node are constructed, and the expressions are as follows: in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

5. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 1, characterized in that, The process of constructing the first loss function is as follows: Based on the initial calculated values ​​of dynamic carbon emission factors for each process in steel production Compared with CEMS measured emission values To determine the deviation, penalize local accounting errors and construct the first loss function. The expression is: The process of constructing the second loss function based on the carbon flow balance equation is as follows: Based on the carbon balance equations of each node, a second loss function is constructed to penalize carbon flow imbalances between processes while satisfying global carbon flow topological constraints. The expression is: in, Indicates the amount of carbon input upstream of the facility. Indicates the amount of carbon output downstream of the facility. Indicates the carbon sequestration content of the facility's products. This indicates the calculated carbon emissions of the facility.

6. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 5, characterized in that, The process of iteratively optimizing and training the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function is as follows: Constrain and initialize the parameters of the dynamic emission factor calculation model. , , and All satisfy: , , and ∈[0,1]; Based on the first and second loss functions, the parameters of the dynamic emission factor calculation model are obtained using the gradient descent method. , , and Perform iterative optimization training to satisfy: in, Including parameters of the dynamic emission factor calculation model , , and Iterative optimization is achieved through gradient descent; This represents the parameter value for the current iteration step, which is updated during the iteration process. This represents the updated parameter value after the next iteration step, based on... Obtained by gradient calculation; This represents the gradient operator. This represents the learning rate, which is a preset value. This represents the carbon flow constraint weight, which is a hyperparameter. Update facility-level dynamic emission factor values, changing the facility-level dynamic emission factor values ​​from their initial values. Iterative updates to , ,..., The calculated values ​​of dynamic carbon emission factors for each process in steel production were changed from the initial values. Iterative updates to , , ,..., n represents the number of training iterations. During iterative training, the iteration terminates when the value of the first loss function is less than a first preset value and the value of the second loss function is less than a second preset value, and the final parameters are output. , , and .

7. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 6, characterized in that, Key facilities in steel production include blast furnace gas outlet, converter flue, and rolling mill heating furnace. The multi-source heterogeneous sensing and acquisition equipment includes: laser gas analyzer, thermocouple array, mass flow meter, and vibration sensor. Key dynamic parameter data for steel production includes: CO concentration per unit time. Real-time temperature of the facility, combustion efficiency coefficient based on vibration spectrum analysis, and real-time gas flow rate of the facility. .

8. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 7, characterized in that, The process of performing real-time facility-level carbon emission accounting is as follows: The pre-processed key dynamic parameter data of steel production are input into the final dynamic emission factor calculation model to calculate the facility-level dynamic emission factor value in real time. ; Based on the facility-level dynamic emission factor value Based on the real-time gas flow rate of the facility, the carbon emission values ​​of each process in steel production are calculated in real time, as expressed by: in, This indicates the carbon emission values ​​of each process in steel production.

9. The method for real-time carbon emission calculation at the steel production facility level based on dynamic factor iteration according to claim 1, characterized in that, Also includes: A digital twin is constructed based on the carbon flow topology network at the steel production facility level to visualize real-time carbon emission data of each facility.

10. A real-time carbon emission accounting system for steel production facilities based on dynamic factor iteration, characterized in that, The system includes: The first construction module is used to construct a facility-level dynamic emission factor calculation model, which includes several parameters to be optimized and trained. The first acquisition module is used to acquire the initial facility-level dynamic emission factor value and the initial calculated value of carbon emissions for each process. The second building module is used to model the carbon flow topology network at the steel production facility level and construct the carbon flow balance equation. The third construction module constructs a first loss function based on the deviation between the calculated values ​​of the dynamic carbon emission factors of each process in steel production and the measured values ​​of CEMS, and constructs a second loss function based on the carbon flow balance equation. The training module iteratively optimizes the parameters of the dynamic emission factor calculation model based on the first loss function and the second loss function. During the optimization process, it updates the facility-level dynamic emission factor value and the calculated value of the carbon emission dynamic factor of each process in steel production. When the optimization training is terminated, it obtains the parameters of the trained dynamic emission factor calculation model and obtains the final dynamic emission factor calculation model based on the parameters. The second acquisition module deploys multi-source heterogeneous sensing and acquisition equipment at key facilities to collect key dynamic parameter data of steel production in real time and perform preprocessing. The accounting module calculates the real-time dynamic emission factor based on the preprocessed data and the final dynamic emission factor calculation model, and performs real-time facility-level carbon emission accounting based on the real-time dynamic emission factor.

Images