Carbon cost-fuel characteristic coupled coal blending combustion dynamic optimization method

Abstract

The invention discloses a carbon cost-fuel characteristic coupled coal blending combustion dynamic optimization method, and relates to the related technical field of carbon combustion optimization, and the method comprises the steps: building a fuel combustion time sequence analysis model; acquiring information of fuel to be blended, wherein the information comprises a fuel name, fuel components and cost; establishing an optimization model which takes the ratio of the fuel to be blended as a decision variable and takes the total cost minimization as a target; setting constraint conditions including total heat value constraint, sulfur emission constraint, volatile component safety constraint and ash fusion point safety constraint, wherein the sum of the ratio of the to-be-blended fuel is 100%; and obtaining a proportioning optimization strategy of the to-be-blended fuel. The technical problems that in the prior art, fuel combustion and carbon emission are regarded as static processes, consequently, a coal blending combustion optimization model and the actual dynamic combustion process are disjointed, and the carbon cost and the instantaneous combustion characteristic cannot be accurately coupled are solved, and the purposes of achieving fine optimization of coal blending combustion and improving the coal blending combustion efficiency are achieved. The operation stability of the boiler is improved, and the actual carbon emission is reduced.

Description

Technical Field

[0001] This invention relates to the field of carbon combustion optimization technology, and in particular to a dynamic optimization method for coal blending and combustion that couples carbon cost and fuel characteristics. Background Technology

[0002] Optimizing boiler operation through the blending and combustion of multiple fuels is a key technological means for coal-fired power plants to reduce costs, increase efficiency, and lower pollutant and carbon emissions. However, traditional coal blending optimization focuses primarily on static, average-characteristic-based cost and environmental performance optimization, which has significant limitations. It typically treats fuel combustion and carbon emission characteristics as constants, using linear or nonlinear programming methods to find the lowest-cost blending ratio under average constraints such as calorific value, sulfur content, and volatile matter. This approach fails to fully consider the dynamic, time-varying characteristics of fuel combustion within the boiler. For example, different fuels have different ignition points, burnout times, and carbon release patterns during combustion, directly affecting boiler stability, combustion efficiency, and instantaneous carbon emission intensity. This leads to problems such as unstable combustion, efficiency deviations from expectations, and excessive instantaneous pollutant and carbon emissions in actual operation, causing a discrepancy between theoretical optimization results and actual operational performance.

[0003] At present, the relevant technologies treat fuel combustion and carbon emissions as static processes, which leads to a disconnect between the coal blending optimization model and the actual dynamic combustion process, and makes it impossible to accurately couple carbon costs and instantaneous combustion characteristics. Summary of the Invention

[0004] This application provides a dynamic optimization method for coal blending and combustion that couples carbon cost and fuel characteristics. This method solves the technical problem in the prior art where fuel combustion and carbon emissions are treated as static processes, leading to a disconnect between the coal blending and combustion optimization model and the actual dynamic combustion process, and making it impossible to accurately couple carbon cost and instantaneous combustion characteristics. This method achieves the technical effect of fine optimization of coal blending and combustion, improving boiler operation stability, and reducing actual carbon emissions.

[0005] This application provides a dynamic optimization method for coal blending that couples carbon cost and fuel characteristics. The method includes: establishing a time-series analytical model of fuel combustion to describe the time-series quantitative relationship between fuel combustion characteristics and carbon emission characteristics during the combustion process; obtaining information on the fuels to be blended, including fuel name, fuel composition, and cost; establishing an optimization model based on the time-series analytical model and the information on the fuels to be blended, with the proportion of the fuels to be blended as the decision variable and the goal of minimizing total cost; setting constraints on the optimization model, including total calorific value constraints, sulfur emission constraints, volatile matter safety constraints, ash melting point safety constraints, and the total proportion of the fuels to be blended being 100%; and solving the optimization model to obtain the optimization strategy for the proportion of the fuels to be blended.

[0006] In a possible implementation, the dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further includes the following processing: collecting combustion description parameters and carbon emission monitoring parameters of fuel during combustion; fitting a first quantitative relationship of combustion description parameters and a second quantitative relationship of carbon emission monitoring parameters according to the changes of the combustion description parameters and carbon emission monitoring parameters over time during combustion; and performing coupled analysis on the first quantitative relationship and the second quantitative relationship to establish the time-series analytical model.

[0007] In a possible implementation, the dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further performs the following processing: aligning the first quantification relationship with the second quantification relationship according to the time sequence; fitting the quantification correlation between the first quantification relationship and the second quantification relationship based on the time alignment relationship, including co-directional coupling and reverse coupling; and coupling the first quantification relationship with the second quantification relationship according to the identified coupling relationship to construct a dynamic mapping function from combustion state to carbon emissions, forming the core algorithm of the time-series analytical model.

[0008] In a possible implementation, the dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further performs the following processing: using a data-driven method based on machine learning, a nonlinear mapping relationship between combustion parameters and carbon emission parameters is learned through a neural network to establish the dynamic mapping function.

[0009] In a possible implementation, the aforementioned dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further performs the following processing: extracting time-domain statistical features and frequency-domain energy features from time-series data of combustion description parameters and carbon emission monitoring parameters; screening combustion description parameters that are strongly correlated with carbon emissions based on Pearson correlation analysis, training a neural network model using the screened features, and optimizing the network weights through a backpropagation algorithm; using the trained neural network model as the core prediction engine of the time-series analytical model to establish a nonlinear mapping relationship between combustion description parameters and carbon emission parameters.

[0010] In a possible implementation, the carbon cost-fuel characteristic coupled coal blending dynamic optimization method further performs the following processing: the total cost includes fuel cost and carbon cost, wherein the fuel cost is the sum of the product of the unit price of each fuel and the proportion of each fuel, and the carbon cost is the sum of the product of the carbon content of each fuel and the carbon price of each fuel.

[0011] In a possible implementation, the aforementioned dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further performs the following processing: based on the fuel composition, the carbon content of the fuel is determined, and the carbon conversion efficiency curve of the fuel blending scheme during the combustion process is predicted according to the carbon content using a time-series analytical model; the carbon conversion efficiency curve is integrated over the entire combustion cycle to obtain the total carbon emissions; and the product of the total carbon emissions and the real-time carbon trading price is used as the carbon cost corresponding to the fuel blending scheme.

[0012] In a possible implementation, the aforementioned dynamic optimization method for carbon cost-fuel characteristic coupling coal blending further includes the following steps: simulating the combustion process of the blending optimization strategy using a digital twin simulation space; evaluating the safety, environmental protection, and economic indicators of the blending optimization strategy based on the simulation results; and adjusting the constraints of the optimization model and resolving the problem when the evaluation indicators do not meet the preset requirements, iteratively optimizing until a fuel blending strategy that meets the preset requirements is obtained.

[0013] In a possible implementation, the aforementioned dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics further performs the following processing: collecting key parameters of actual combustion, including boiler efficiency, pollutant emission concentration, and carbon emission intensity, and calculating the deviation between the actual collected operating data and the predicted values ​​of the time-series analytical model; when the deviation exceeds a threshold, triggering a model update, wherein incremental learning and parameter correction are performed on the time-series analytical model based on the latest operating data; and resolving the optimization model using the updated time-series analytical model to dynamically adjust the fuel blending strategy.

[0014] This application proposes a dynamic optimization method for coal blending that couples carbon cost and fuel characteristics. The method establishes a time-series analytical model of fuel combustion; obtains information on the fuels to be blended, including fuel name, fuel composition, and cost; establishes an optimization model with the proportion of the fuels to be blended as the decision variable and minimizing total cost as the objective; sets constraints, including total calorific value constraints, sulfur emission constraints, volatile matter safety constraints, ash fusion point safety constraints, and a total proportion of the fuels to be blended of 100%; and obtains an optimization strategy for the proportion of the fuels to be blended. This method solves the technical problem in existing technologies that treat fuel combustion and carbon emissions as static processes, leading to a disconnect between the coal blending optimization model and the actual dynamic combustion process, and an inability to accurately couple carbon cost and instantaneous combustion characteristics. It achieves precise optimization of coal blending, improves boiler operational stability, and reduces actual carbon emissions. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments of the present invention will be briefly described below. Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed precisely in sequence. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from these processes.

[0016] Figure 1 This is a flowchart illustrating a dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics, provided in an embodiment of this application. Detailed Implementation

[0017] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structure, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0018] This application provides a dynamic optimization method for coal blending that couples carbon cost and fuel characteristics, such as... Figure 1 As shown, the method includes: Step S100: Establish a time-series analytical model of fuel combustion to describe the time-series quantitative relationship between fuel combustion characteristics and carbon emission characteristics during the combustion process.

[0019] Step S100 further includes step S110, collecting combustion description parameters and carbon emission monitoring parameters of fuel during combustion; step S120, fitting a first quantitative relationship of combustion description parameters and a second quantitative relationship of carbon emission monitoring parameters according to the changes of the combustion description parameters and carbon emission monitoring parameters over time during combustion; step S130, performing coupled analysis on the first quantitative relationship and the second quantitative relationship to establish the time series analytical model.

[0020] Preferably, during boiler combustion experiments or actual operation, combustion description parameters and carbon emission monitoring parameters are continuously measured and recorded. Combustion description parameters are physicochemical quantities describing the combustion state and characteristics, and may include temperature curves at different locations within the furnace, volatile matter emission rates, flue gas component concentrations, changes in the proportion of fixed carbon burned from the fuel over time, flame images / signals, and other combustion stability indicators. Carbon emission monitoring parameters refer to data directly related to carbon dioxide generation, including the instantaneous concentration of carbon dioxide in the flue gas or the carbon emission rate, which is calculated using flue gas flow rate and carbon dioxide concentration. By analyzing the changes in combustion description parameters and carbon emission monitoring parameters over time during combustion, a first quantitative relationship for the combustion description parameters is fitted, representing the inherent time-dependent variation of a single combustion description parameter. For example, data may show that the volatile matter emission rate peaks at 5 seconds after combustion begins and then decays exponentially. A second quantitative relationship for the carbon emission monitoring parameters is then fitted, representing the inherent time-dependent variation of the carbon emission parameters themselves. For example, data may show that the carbon emission rate is highest in the middle of combustion, forming a bell-shaped curve.

[0021] Preferably, the first and second quantitative relationships are coupled and analyzed to determine the mutual influence between changes in combustion description parameters and changes in carbon emission parameters. The first and second quantitative relationships are integrated through a dynamic mapping function of machine learning to generate a time-series analytical model that describes the time-series quantitative relationship between fuel combustion characteristics and carbon emission characteristics during the combustion process. That is, the combustion state at a certain moment is taken as input, and the carbon emission situation at that moment and the next moment is predicted as output. This allows for dynamic evaluation of the carbon emission situation of the entire combustion process under different ratio schemes, thereby calculating a more accurate carbon cost and ensuring that the combustion process is both safe and efficient.

[0022] Furthermore, step S130 also includes step S131, aligning the first quantization relationship and the second quantization relationship according to the time sequence; step S132, fitting the quantization correlation between the first quantization relationship and the second quantization relationship based on the time alignment relationship, including unidirectional coupling and anti-directional coupling; step S133, coupling the first quantization relationship and the second quantization relationship according to the identified coupling relationship, constructing a dynamic mapping function from combustion state to carbon emissions, forming the core algorithm of the time sequence analysis model.

[0023] Preferably, time alignment refers to precisely aligning the time-series data of combustion description parameters with the time-series data of carbon emission monitoring parameters on a unified time axis. This eliminates potential system timestamp errors during data acquisition and ensures that, when analyzing correlations, data points at the same moment in both datasets truly correspond to the same physical instant in the combustion process. Based on the time-aligned data, cross-correlation analysis and regression analysis are used to quantify the statistical correlation strength and direction between the two relationships, including co-directional coupling and inverse coupling. Specifically, co-directional coupling means that an increase in a combustion description parameter is statistically significantly correlated with an increase in a carbon emission parameter, exhibiting a positive correlation. Inverse coupling means that an increase in a combustion description parameter is statistically significantly correlated with a decrease in a carbon emission parameter, exhibiting a negative correlation. The coupling relationship is used to couple the first and second quantification relationships to construct a dynamic mapping function from combustion state to carbon emissions. The input is the combustion description parameter, and the output is the predicted value of the carbon emission parameter. This transforms the identified statistical correlations into a predictive model that can dynamically and continuously predict the corresponding carbon emission characteristics based on the current and past combustion states. This ultimately forms the core algorithm of the time-series analytical model, enabling it to simulate and predict the dynamic behavior of the entire combustion process.

[0024] Furthermore, step S133 also includes using a data-driven method of machine learning to learn the nonlinear mapping relationship between combustion parameters and carbon emission parameters through a neural network, and establishing the dynamic mapping function.

[0025] Preferably, without relying on preset physicochemical equations, a data-driven approach using machine learning is adopted. A neural network is selected as the specific model architecture and used as a function approximator to characterize the complex nonlinear mapping relationship between combustion parameters and carbon emission parameters. The input variables are multiple combustion parameters measured at the same time, defining the combustion state at that moment. The output variable is the carbon emission parameter measured at the same time point as the input variables. Using a prepared dataset, the connection weights and bias terms between neurons within the neural network are automatically and iteratively adjusted through backpropagation algorithms to optimize the parameters. The parameters are adjusted to minimize the mean square error (MSE) between the neural network output and the actual measurement value. The neural network with fixed optimal parameters obtained after learning and training is the dynamic mapping function, which can dynamically predict the corresponding carbon emission situation based on the real-time combustion state.

[0026] Furthermore, step S133 also includes extracting time-domain statistical features and frequency-domain energy features from the time-series data of combustion description parameters and carbon emission monitoring parameters; screening combustion description parameters that are strongly correlated with carbon emissions based on Pearson correlation analysis; training a neural network model using the screened features; optimizing the network weights through a backpropagation algorithm; and using the trained neural network model as the core prediction engine of the time-series analytical model to establish a nonlinear mapping relationship between combustion description parameters and carbon emission parameters.

[0027] Preferably, time-domain statistical feature extraction and frequency-domain energy feature extraction are performed from the time-series data of combustion description parameters and carbon emission monitoring parameters, respectively. The time-domain statistical feature extraction involves calculating multiple statistical features within a sliding time window from the original, time-series recorded combustion parameter and carbon emission parameter data streams, including but not limited to mean, variance, standard deviation, peak value, valley value, slope, and curvature. The frequency-domain energy feature extraction involves applying Fast Fourier Transform or Wavelet Transform to the data within the same time window to convert the signal from the time domain to the frequency domain, extracting the amplitude or energy of the main frequency components and the spectral entropy used to measure the degree of disorder in the signal spectrum.

[0028] Preferably, the Pearson correlation coefficient between all extracted combustion description parameter features and carbon emission parameter features is calculated to quantify the strength and direction of the linear correlation between the combustion description parameter features and carbon emission parameter features. A correlation threshold is set, and then combustion description parameter features that show a strong linear correlation with carbon emissions are selected as candidate inputs for the neural network model. The neural network model is then trained. Specifically, the selected strongly correlated feature set is used as the input layer node of the neural network, and the carbon emission parameters are used as the output layer node. The backpropagation algorithm is used for supervised learning. The training samples are input into the network to obtain the predicted output. The error between the predicted output and the actual carbon emission value is calculated. The error is backpropagated from the output layer to the input layer, and the gradient descent method is used to optimize and adjust the connection weights and bias terms of all nodes in the neural network model according to the magnitude of the error.

[0029] Preferably, after training is completed and the prediction accuracy is verified to meet the standard, the neural network model is used as the core prediction engine of the time series analysis model. Specifically, in practical applications, the combustion parameters collected in real time are input into the engine after the same feature extraction and screening process, and high-precision carbon emission parameter prediction values ​​are immediately output. Finally, a nonlinear mapping relationship between combustion description parameters and carbon emission parameters is established, which can capture the complex and high-order interactions represented by the intermediate hidden layer.

[0030] Step S200: Obtain information on the fuel to be blended, including fuel name, fuel composition, and cost.

[0031] Preferably, all attribute data of the fuels to be blended are collected for coal blending calculations, including fuel name, fuel composition, and cost. Specifically, the fuel name refers to the unique identifier of each fuel to be blended to clearly distinguish different fuels, such as 40% Shenhua coal, 30% Indonesian coal, and 30% biomass pellets. The fuel composition refers to the detailed data of the industrial analysis composition and elemental analysis composition of each fuel, usually expressed as a mass percentage or unit mass content, which may include the content of moisture, ash, volatile matter, fixed carbon, or carbon, hydrogen, oxygen, nitrogen, and sulfur. This is used to set constraints for the optimization model, such as volatile matter for volatile matter safety constraints, sulfur for sulfur emission constraints, ash fusion point for ash fusion point safety constraints, and calorific value for total calorific value constraints. The cost refers to obtaining economic data for each fuel, including the fuel unit price and transportation, storage, or pretreatment costs.

[0032] Step S300: Based on the time series analytical model and the information of the fuel to be blended, establish an optimization model with the ratio of the fuel to be blended as the decision variable and the goal of minimizing the total cost.

[0033] Step S300 further includes that the total cost includes fuel cost and carbon cost, wherein the fuel cost is the sum of the product of the unit price of each fuel and the proportion thereof, and the carbon cost is the sum of the product of the carbon content of each fuel and the carbon price thereof.

[0034] Preferably, the fuel blending ratio refers to the mass percentage of each fuel to be blended in the final blended fuel. The total cost includes fuel cost and carbon cost. The fuel cost is the sum of the product of the unit price of each fuel and the blending ratio, and the carbon cost is the sum of the product of the carbon content of each fuel and the carbon price. That is, based on the final fuel blending ratio, the dynamic carbon emission behavior of the blended fuel during combustion is simulated, and the total carbon emission is calculated by integration and then multiplied by the carbon trading price to obtain the carbon cost. Then, with the fuel blending ratio as the decision variable and the goal of minimizing the total cost, an optimization model is obtained by mathematical integration with the time series analytical model and the information of the fuel to be blended. When all constraints are met, the optimization model dynamically and accurately predicts the carbon emission characteristics of the blended fuel, and minimizes the sum of the fuel purchase cost and the carbon cost dynamically calculated by the time series analytical model. Finally, the optimal fuel blending ratio scheme is output.

[0035] Furthermore, step S300 also includes step S310, determining the carbon content of the fuel based on the fuel composition, and predicting the carbon conversion efficiency curve of the fuel blending scheme under combustion conditions using a time-series analytical model based on the carbon content; step S320, integrating the carbon conversion efficiency curve over the entire combustion cycle to obtain the total carbon emissions; and step S330, using the product of the total carbon emissions and the real-time carbon trading price as the carbon cost corresponding to the fuel blending scheme.

[0036] Preferably, the carbon content of the fuel is determined from the fuel composition and input into a time-series analytical model for prediction. Based on combustion dynamics such as volatile analysis and burnout kinetics, the model simulates the rate at which carbon in the fuel is converted into carbon dioxide under real, time-varying combustion conditions, outputting a carbon conversion efficiency curve. The Y-axis represents carbon conversion efficiency, and the X-axis represents time, used to describe the rhythm and intensity of carbon emissions throughout the entire process from ignition to burnout of the mixed fuel. The carbon conversion efficiency curve is integrated over the entire combustion cycle, i.e., the area under the curve is summed, to obtain the total carbon emissions actually generated by the fuel blending scheme. The product of the total carbon emissions and the real-time carbon trading price is used as the carbon cost corresponding to the fuel blending scheme, thereby directly converting physical environmental indicators into economic indicators and accurately reflecting the carbon trading costs generated by a specific fuel blending scheme due to its unique combustion behavior.

[0037] Step S400: Set the constraints of the optimization model. The constraints include total calorific value constraints, sulfur emission constraints, volatile matter safety constraints, ash melting point safety constraints, and the total proportion of the fuels to be blended is 100%.

[0038] Preferably, physical, environmental, safety, and logical constraints are set for the optimization model that minimizes total cost to ensure that the optimal fuel blending scheme is feasible, safe, and compliant. These constraints include total calorific value constraints, sulfur emission constraints, volatile matter safety constraints, ash fusion point safety constraints, and a total blending ratio of 100% for the fuels to be blended. Specifically, the total calorific value constraint is an energy supply guarantee constraint, meaning that the weighted average lower heating value of the blended fuels must be equal to or higher than the minimum calorific value required for boiler design and operation. This ensures that the blended fuels can provide sufficient energy input to meet the load requirements of the generator set and prevents insufficient boiler output or unstable combustion due to excessively low calorific value. The sulfur emission constraint is an environmental compliance constraint, meaning that the weighted average sulfur content of the blended fuels must be lower than the upper limit of fuel sulfur content corresponding to the emission standards allowed by environmental regulations. This ensures that the sulfur dioxide concentration in the flue gas after combustion, after passing through the desulfurization system, still meets environmental compliance requirements. The emissions must meet atmospheric pollutant emission standards. Volatile matter safety constraints are for combustion stability and equipment safety. This means the weighted average volatile matter content of the mixed fuels must be within a safe range. The lower limit prevents excessively low volatile matter content, which can lead to ignition difficulties and unstable combustion. The upper limit prevents excessively high volatile matter content, which can cause the ignition point to be too close and combustion too intense, potentially leading to nozzle slagging or burner damage. Ash melting point safety constraints are for boiler heating surface safety. This means the ash melting point of the mixed fuels must be higher than the furnace temperature during boiler operation, or the melting point of the mixed ash must be brought to a safe range through coal blending. This prevents ash from melting at high temperatures during combustion and adhering to the heating surface, forming severe slagging, reducing heat transfer efficiency, affecting normal boiler operation, and even causing safety accidents. The total percentage of the fuels to be blended must be 100%, a logical constraint. This means the sum of the percentages of all fuels involved in the blending must equal 1, ensuring the optimization result is a complete fuel formulation scheme.

[0039] Step S500: Solve the optimization model to obtain the blending optimization strategy for the fuel to be blended.

[0040] Preferably, a feasible solution space is constructed using multiple different fuel blending schemes to meet constraints such as calorific value, sulfur content, volatile matter, and ash fusion point. Then, within the feasible solution space, intelligent search and iteration are performed to determine the fuel blending combination that minimizes the total cost of fuel cost and dynamic carbon cost. This generates a blending optimization strategy for the fuels to be blended, i.e., specific quantitative instructions. This may include the optimal blending scheme, i.e., the optimal mass percentage of each fuel to be blended in the mixed fuel; and the calculation of expected key indicators based on the optimal blending scheme, such as expected total cost and cost composition, expected characteristics of the blended fuel, and expected total carbon emissions. This provides direct, data-driven decision-making basis for fuel procurement, inventory management, and boiler operation, guiding production personnel to blend fuels according to the calculated optimal ratio, so as to achieve safe and environmentally friendly fine optimization of coal blending and combustion, improve boiler operation stability, and reduce actual carbon emissions.

[0041] Furthermore, step S500 also includes step S510, simulating the combustion process of the ratio optimization strategy through a digital twin simulation space; step S520, evaluating the safety, environmental protection and economic indicators of the ratio optimization strategy based on the simulation results; and step S530, when the evaluation indicators do not meet the preset requirements, adjusting the constraints of the optimization model and resolving it, and iterating until a fuel ratio strategy that meets the preset requirements is obtained.

[0042] Preferably, the digital twin simulation space is a virtual model of the boiler combustion process driven by physicochemical principles and actual data. It can simulate the comprehensive behavior of the entire boiler system, including fluid dynamics, heat transfer, and pollutant generation. The optimized fuel ratio strategy is input into the digital twin simulation space for combustion process simulation, fully rehearsing the entire dynamic process of the ratio scheme in the boiler from ignition to burnout. The output data is used as the simulation result, and a multi-dimensional evaluation is conducted to determine safety, environmental, and economic indicators. Safety indicators refer to assessing the risks of slagging, corrosion, fire extinguishing, and overheating. Environmental indicators refer to assessing whether instantaneous and cumulative pollutant emissions and carbon emission intensity consistently meet environmental standards. Economic indicators refer to a comprehensive evaluation of fuel and carbon composition. The analysis considers the operating costs and their impact on boiler efficiency, equipment lifespan, and auxiliary power consumption. Then, it compares the safety, environmental, and economic indicators of the fuel blending optimization strategy with the preset requirements for safety, environmental protection, and economy. If any indicator fails to meet the standard, the strategy is deemed unqualified. Based on the problems revealed by the simulation results, the constraints in the optimization model are adjusted. For example, if the ash melting point issue leads to safety concerns, the constraint range for the ash melting point is tightened; if sulfur dioxide emissions exceed the standard, the upper limit of the sulfur emission constraint in the model is lowered. The optimization model is then re-run using the adjusted constraints to solve the problem. This iterative optimization continues until a fuel blending strategy that meets the preset requirements is obtained, ensuring that the fuel blending strategy is absolutely safe, reliable, and efficient.

[0043] Furthermore, step S500 also includes step S540, collecting key parameters of actual combustion, including boiler efficiency, pollutant emission concentration, and carbon emission intensity, and calculating the deviation between the actual collected operating data and the predicted values ​​of the time series analytical model; step S550, when the deviation exceeds a threshold, triggering a model update, wherein incremental learning and parameter correction are performed on the time series analytical model based on the latest operating data; step S560, using the updated time series analytical model to resolve the optimization model, and dynamically adjusting the fuel ratio strategy.

[0044] Preferably, key parameters reflecting the actual combustion state, including boiler efficiency, pollutant emission concentration, and carbon emission intensity, are collected in real time from the boiler's distributed control system, online flue gas monitoring system, and online carbon emission monitoring system. Boiler efficiency is a comprehensive economic indicator for measuring energy conversion, pollutant emission concentration is used to measure environmental performance, and carbon emission intensity is used to measure low-carbon performance. The actual collected operating data is compared with the predicted values ​​of the time-series analytical model, and the deviation is calculated. An acceptable deviation range is preset. When the actual deviation continues or significantly exceeds the preset deviation range, it indicates that the prediction of the time-series analytical model has deviated from the actual situation, and the model update program is automatically triggered, including incremental learning and parameter correction of the time-series analytical model based on the latest operating data. Specifically, the latest, real-world operational data is used as new training samples to fine-tune the parameters of the time-series analytical model. Simultaneously, using the new operational data, the weights and biases of the network are slightly adjusted through backpropagation. This allows the time-series analytical model to learn online new patterns caused by unknown factors such as slow changes in fuel characteristics, boiler equipment scaling, seasonal environmental changes, and instrument drift, maintaining the accuracy and timeliness of its predictive capabilities. Finally, the updated time-series analytical model is used to resolve and optimize the model, dynamically adjusting the fuel blending strategy to determine the optimal fuel blending strategy based on the latest actual conditions. This strategy guides the next stage of fuel blending and co-firing operations, thereby continuously maximizing the cost, environmental, and safety benefits throughout the entire lifecycle.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics, characterized in that, include: Establish a time-series analytical model of fuel combustion to describe the time-series quantitative relationship between fuel combustion characteristics and carbon emission characteristics during the combustion process; Obtain information on the fuel to be blended, including fuel name, fuel composition, and cost; Based on the time series analytical model and the information of the fuel to be blended, an optimization model is established with the ratio of the fuel to be blended as the decision variable and the goal of minimizing the total cost. The optimization model is set with constraints, including total calorific value constraints, sulfur emission constraints, volatile matter safety constraints, ash melting point safety constraints, and the total proportion of the fuels to be blended being 100%. Solve the optimization model to obtain the optimal ratio strategy for the fuel to be blended.

2. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 1, characterized in that, Establish a time-series analytical model of fuel combustion, including: Collect combustion description parameters and carbon emission monitoring parameters of fuel during the combustion process; Based on the changes in combustion description parameters and carbon emission monitoring parameters over time during the combustion process, a first quantitative relationship for the combustion description parameters and a second quantitative relationship for the carbon emission monitoring parameters are fitted. The first quantization relationship and the second quantization relationship are coupled and analyzed to establish the time series analytical model.

3. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 2, characterized in that, The first quantization relationship and the second quantization relationship are coupled for analysis to establish the time series analytical model, including: The first quantization relationship and the second quantization relationship are time-aligned according to their temporal sequence. Based on the time alignment relationship, the quantization correlation between the first quantization relationship and the second quantization relationship is fitted, including co-coupling and inverse coupling; Based on the identified coupling relationship, the first quantification relationship is coupled with the second quantification relationship to construct a dynamic mapping function from combustion state to carbon emissions, forming the core algorithm of the time series analytical model.

4. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 3, characterized in that, Construct a dynamic mapping function from combustion state to carbon emissions, including: A data-driven approach using machine learning is employed to learn the nonlinear mapping relationship between combustion parameters and carbon emission parameters through a neural network, thereby establishing the dynamic mapping function.

5. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 4, characterized in that, Using a data-driven approach employing machine learning, a nonlinear mapping relationship between combustion parameters and carbon emission parameters is learned through a neural network, establishing the dynamic mapping function, including: Extract time-domain statistical features and frequency-domain energy features from time-series data of combustion description parameters and carbon emission monitoring parameters; Based on Pearson correlation analysis, combustion description parameters that are strongly correlated with carbon emissions were screened, and the screened features were used to train a neural network model. The network weights were then optimized using the backpropagation algorithm. The trained neural network model is used as the core prediction engine of the time-series analytical model to establish a nonlinear mapping relationship between combustion description parameters and carbon emission parameters.

6. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 1, characterized in that, The total cost includes fuel cost and carbon cost, wherein fuel cost is the sum of the product of the unit price of each fuel and its proportion, and carbon cost is the sum of the product of the carbon content of each fuel and its carbon price.

7. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 6, characterized in that, Methods for calculating carbon costs include: Based on the fuel composition, the carbon content of the fuel is determined, and the carbon conversion efficiency curve of the fuel ratio scheme under combustion conditions is predicted by a time-series analytical model according to the carbon content. The total carbon emissions are obtained by integrating the carbon conversion efficiency curve over the entire combustion cycle. The product of total carbon emissions and real-time carbon trading price is taken as the carbon cost corresponding to the fuel blending scheme.

8. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 1, characterized in that, After obtaining the optimal blending strategy for the fuel to be blended, the following steps are taken: The combustion process of the ratio optimization strategy is simulated using a digital twin simulation space; The safety, environmental friendliness, and economic efficiency of the proposed ratio optimization strategy are evaluated based on simulation results. When the evaluation indicators do not meet the preset requirements, the constraints of the optimization model are adjusted and the solution is recalculated. The optimization is iterated until a fuel ratio strategy that meets the preset requirements is obtained.

9. The dynamic optimization method for coal blending coupled with carbon cost and fuel characteristics according to claim 1, characterized in that, Also includes: Key parameters of actual combustion are collected, including boiler efficiency, pollutant emission concentration and carbon emission intensity, and the deviation between the actual collected operating data and the predicted values ​​of the time series analytical model is calculated. When the deviation exceeds the threshold, the model is updated, which involves incremental learning and parameter correction of the time series analytical model based on the latest running data. The optimization model is re-solved using the updated time-series analytical model, and the fuel ratio strategy is dynamically adjusted.

Images