Desulfurization system carbon emission data statistics and carbon management and control system

Abstract

The invention relates to the field of industrial process control and data management, and particularly discloses a desulfurization system carbon emission data statistics and carbon management and control system which comprises a carbon flow panoramic sensing module, a multi-scale carbon accounting engine, a carbon flow collaborative optimization module and a carbon management and control execution and feedback module. Panoramic carbon flow metering is realized by constructing a full-flow carbon atom tracking model, a carbon flow coupling relationship with upstream and downstream processes is established, a dynamic carbon management and control instruction is generated by using a multi-target collaborative optimization algorithm, and a closed-loop management and control system integrating sensing, accounting, optimization and execution is formed, so that the carbon statistical integrity and the cross-process collaborative emission reduction capability are improved.

Description

Technical Field

[0001] This invention belongs to the field of industrial process control and data management, specifically relating to a carbon emission data statistics and carbon control system for desulfurization systems. Background Technology

[0002] In the field of industrial environmental protection and energy conservation and emission reduction, accurate statistics and intelligent management of carbon emissions during the production process are the core links in achieving the "dual carbon" goal. Among them, flue gas desulfurization systems in industries such as thermal power, steel, and chemicals are important environmental protection facilities, and their operation is accompanied by significant indirect carbon emissions. Effective management of these emission data is a key technological direction for carbon emission reduction in these industries.

[0003] The carbon emission data statistics and carbon management system for desulfurization systems aims to collect, calculate, and analyze carbon data generated throughout the entire desulfurization process, and implement optimized control accordingly. Existing technologies typically treat desulfurization systems as independent pollution control units, resulting in significant limitations in carbon emission management. Current carbon emission accounting for desulfurization processes primarily focuses on indirect emissions such as electricity consumption, neglecting the carbon flows contained in desulfurization reaction byproducts like gypsum or wastewater treatment, leading to incomplete carbon footprint statistics. Furthermore, existing control strategies lack coordinated analysis of carbon flows between the desulfurization system and upstream combustion and downstream product resource utilization, rendering the system merely a passive "end-of-pipe treatment" unit, unable to integrate into the overall carbon cycle and optimization system of the plant. This fragmented data and siloed management makes it difficult for enterprises to tap into effective carbon reduction potential from the desulfurization process, and also hinders precise carbon quota management and trading decisions. Summary of the Invention

[0004] The purpose of this invention is to provide a carbon emission data statistics and carbon control system for desulfurization systems, so as to solve the technical contradictions in the prior art, such as incomplete carbon emission statistics, isolated control strategies, and inability to be integrated into the overall carbon cycle optimization system of the plant.

[0005] This invention provides a carbon emission data statistics and carbon control system for desulfurization systems. The system includes a panoramic carbon flow perception module, a multi-scale carbon accounting engine, a carbon flow collaborative optimization module, and a carbon control execution and feedback module. By constructing a carbon atom tracking model covering the entire material and energy flow of the desulfurization system, the system achieves panoramic carbon flow measurement from carbon input, carbon conversion to carbon output. Based on this, it establishes the carbon flow coupling relationship between the desulfurization system and upstream and downstream processes, and generates dynamic carbon control commands through a multi-objective collaborative optimization algorithm, ultimately forming a closed-loop carbon control system integrating perception, accounting, optimization, and execution.

[0006] The carbon flow panoramic sensing module is used to collect all direct and indirect carbon-related data in the entire desulfurization process in real time. This module is deployed with a high-precision sensor network and data interface, and the collected data is divided into three categories. Category 1 is process material flow carbon data, including the instantaneous flow rate and carbon content of desulfurizing agents such as limestone or ammonia water, the concentration and flow rate of carbon dioxide and carbon monoxide in the raw flue gas, the yield and carbon content of desulfurization byproducts such as gypsum or ammonium sulfate, and the discharge volume of desulfurization wastewater and its dissolved organic and inorganic carbon concentrations. Category 2 is process energy flow carbon data, including the real-time power of all power-consuming equipment in the desulfurization system, such as pumps, fans, agitators, and oxidation fans, as well as the system's steam consumption. Category 3 is related process carbon data, acquired through a plant-level data bus, including carbon element analysis data of boiler coal combustion, power generation load, and energy and material data of downstream gypsum calcination or ammonium sulfate refining processes. The carbon flow panoramic perception module preprocesses all the above raw data, including outlier removal, noise filtering and timestamp alignment, and converts them into a standardized carbon flow rate signal in kilograms of carbon per hour.

[0007] The multi-scale carbon accounting engine connects to the carbon flow panoramic perception module to receive standardized carbon flow rate signals and perform multi-level carbon accounting from the equipment level to the system level and then to the plant level. The engine incorporates a carbon flow tracking model and a dynamic accounting rule base. The carbon flow tracking model is a mathematical model based on the law of conservation of mass. This model abstracts the desulfurization system as a network containing multiple carbon nodes, each node representing a process unit, and the connections between nodes representing carbon flow paths. Based on the input carbon flow rate signal, the model calculates and tracks the changes in the quantity and flow of carbon atoms in various nodes such as the desulfurization absorption tower, oxidation tank, dehydrator, and wastewater treatment unit in real time. The dynamic accounting rule base stores carbon emission factors that comply with the latest national and industry standards, such as carbon emission factors for electricity from different sources, carbon emission factors for steam of different qualities, and carbon emission coefficients for chemical reactions such as calcium carbonate decomposition and sulfite oxidation.

[0008] The multi-scale carbon accounting engine operates as follows: First, at the equipment level, it calculates the indirect carbon emissions of a single device, such as a circulating pump, as a real-time power multiplied by the corresponding electricity carbon emission factor. Second, at the system level, it integrates all equipment indirect carbon emissions and superimposes the process direct carbon emissions. The process direct carbon emissions are calculated using a carbon flow tracking model, specifically the total carbon flow entering the system minus the total carbon flow solidified in byproducts and the total carbon flow discharged with wastewater. Finally, at the plant level, it couples the net carbon emissions of the desulfurization system with the associated process carbon data obtained from the carbon flow panoramic perception module to calculate the marginal impact of the desulfurization system's operation on the overall carbon balance of the plant. For example, it assesses the trade-off between the increased power consumption due to improved desulfurization efficiency and the downstream environmental governance carbon costs avoided by reducing sulfur dioxide in flue gas. The accounting results are output in structured data format, including real-time carbon emission intensity, cumulative carbon emissions, carbon flow distribution maps, and a carbon footprint contribution analysis report.

[0009] The carbon flow co-optimization module connects to the multi-scale carbon accounting engine to generate a globally optimal carbon management strategy based on the accounting results and preset optimization objectives. The core of this module is a multi-objective dynamic optimization controller, whose optimization process is built upon a carbon flow coupling model of the desulfurization system and upstream and downstream processes. This carbon flow coupling model quantitatively describes the carbon correlations between desulfurizing agent consumption, desulfurization efficiency, by-product quality, system energy consumption, front-end combustion conditions, and downstream product resource utilization pathways. The multi-objective dynamic optimization controller simultaneously handles three interrelated and potentially conflicting optimization objectives: the first objective is to minimize the total carbon emission cost of the desulfurization system operation, which includes the carbon cost of energy consumption and the carbon cost of chemical agent consumption. The second objective is to maximize the carbon gain from the resource utilization of desulfurization by-products, i.e., improving the purity and stability of by-products such as gypsum to obtain higher carbon offset value in downstream applications such as building materials. The third objective is to meet dynamically changing environmental constraints, i.e., ensuring that sulfur dioxide emission concentrations meet standards in real time. The controller uses a hybrid algorithm combining constrained particle swarm optimization and linear programming for solution.

[0010] In each optimization cycle, the controller receives real-time carbon data from the multi-scale carbon accounting engine, production plan information from the plant scheduling system, and sulfur dioxide emission limits from the environmental monitoring system. Under the constraints of the carbon-fluid coupling model, the controller iteratively adjusts a set of decision variables, including limestone slurry supply flow rate, circulating slurry pH setpoint, oxidation air flow rate, and dewatering machine operating frequency. The algorithm simulates the impact of different combinations of these decision variables on the three optimization objectives, finds the Pareto optimal solution set, and selects the optimal operating point from this solution set according to preset priority weights. This optimal operating point is then transformed into a specific set of carbon control instructions.

[0011] The carbon management execution and feedback module connects to the carbon flow collaborative optimization module and the distributed control system of the desulfurization system. This module decomposes the carbon management instruction set into executable control signals and distributes them to the corresponding actuators in the desulfurization system, while simultaneously monitoring the execution effect and forming a closed-loop feedback. This module includes an instruction translator, an execution agent, and a feedback evaluation unit. The instruction translator translates high-level optimization objectives in the carbon management instruction set, such as "reduce system carbon intensity by 5%", into specific setpoint adjustments in the lower-level control loop, such as adjusting the absorber slurry pH setpoint from 5.5 to 5.3 or reducing the circulating pump frequency by 2 Hz. The execution agent is responsible for safely and smoothly writing these setpoint adjustments into the corresponding control modules of the distributed control system.

[0012] The feedback evaluation unit continuously monitors the actual operating status of the system after command execution. Key monitoring parameters include actual carbon emission intensity, sulfur dioxide outlet concentration, and the moisture content and purity of the byproduct gypsum. The feedback evaluation unit compares the monitored actual effects with the expected effects of the carbon flow co-optimization module and calculates the deviation value. If the deviation value continues to exceed a preset threshold, for example, if the actual carbon intensity reduction does not reach the expected target of 80%, the feedback evaluation unit sends a re-optimization trigger signal to the carbon flow co-optimization module, along with deviation analysis data, prompting the optimization module to start a new round of optimization calculations based on the new actual data, thereby achieving continuous adaptive correction of the carbon management strategy.

[0013] In one embodiment of the present invention, the device for measuring the carbon content of desulfurization byproducts in the carbon flow panoramic sensing module employs an online near-infrared spectrometer. This analyzer scans the material on the dehydrated gypsum conveyor belt at a frequency of once per second. The spectral data is analyzed in real time using a partial least squares regression model to determine the content of calcium carbonate impurities and organic carbon in the gypsum, and the analysis results are converted into a carbon content percentage signal, which is then transmitted to the data aggregation unit of the carbon flow panoramic sensing module.

[0014] As one embodiment of the present invention, the carbon flow tracking model in the multi-scale carbon accounting engine employs an extended Kalman filter for state estimation. This filter takes the measurements from the carbon flow panoramic sensing module as observation input and uses the state-space equations constructed based on chemical reaction kinetics and material balance as the prediction model. Through recursive calculation, it optimally estimates the carbon flow state variables that cannot be directly measured in the model and simultaneously updates the covariance matrix of the estimation error, thereby significantly improving the overall accuracy and anti-interference capability of carbon flow calculation.

[0015] In one embodiment of the present invention, the multi-objective dynamic optimization controller of the carbon flow collaborative optimization module introduces the carbon emission trading price as a dynamic weight coefficient when solving for the Pareto optimal solution set. The controller obtains the carbon quota price from the national or local carbon emission trading market in real time through a data interface and uses this price as a key coefficient in the first optimization objective, namely the carbon emission cost objective function. When the carbon price rises, the controller automatically assigns a higher weight to the emission reduction objective, and the generated strategy is more inclined towards energy conservation and emission reduction; when the carbon price falls, the weight of the by-product resource utilization benefit objective is appropriately increased, and the generated strategy is more inclined towards improving the quality of by-products to gain a market advantage.

[0016] In one embodiment of the present invention, the feedback evaluation unit of the carbon control execution and feedback module incorporates a prediction model based on a long short-term memory network. This model, trained using historical data, can predict the trajectory of key parameters such as carbon intensity and sulfur dioxide concentration within the next 30 minutes based on currently issued control commands and system operating conditions. The feedback evaluation unit compares the predicted trajectory with the expected target trajectory in advance. If the prediction indicates a significant deviation, it can send an early warning signal to the carbon flow collaborative optimization module without waiting for the actual deviation to occur, initiating preventative re-optimization and thus achieving proactive carbon control.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention, by constructing a panoramic carbon flow perception module and a multi-scale carbon accounting engine, achieves for the first time the full-process carbon atom tracking and quantification of a desulfurization system, from input chemicals and energy consumption to output byproducts and wastewater. This system not only accounts for indirect energy emissions but also accurately measures the conversion, fixation, and loss of carbon elements during the process. It overcomes the fundamental shortcomings of traditional methods, such as incomplete carbon footprint statistics and omissions of process carbon flows, providing enterprises with a true and comprehensive carbon ledger for the desulfurization process and laying a solid data foundation for precise carbon management.

[0018] 2. This invention innovatively establishes a carbon flow coupling model between the desulfurization system and the front-end combustion and back-end resource recovery processes through a carbon flow collaborative optimization module, placing the operation optimization of the desulfurization system within the framework of the overall plant carbon cycle. Through a multi-objective optimization algorithm, the system comprehensively considers emission reduction costs, by-product value, and environmental constraints. The resulting control strategy breaks the limitation of the desulfurization system as an isolated "end-of-pipe treatment" system, transforming it into a collaborative unit that actively participates in the optimal allocation of carbon resources across the entire plant. This enables the discovery of hidden carbon emission reduction potential and economic benefits across processes.

[0019] 3. This invention establishes a complete "perception-decision-execution-evaluation" closed loop through a carbon control execution and feedback module. The system not only generates optimized strategies but also automatically translates them into underlying control actions, monitoring execution effectiveness in real time for feedback correction. This closed-loop adaptive mechanism ensures that the carbon control strategy can dynamically respond to internal and external disturbances such as changes in raw materials, fluctuations in equipment operating conditions, and changes in market carbon prices, continuously maintaining optimal or near-optimal carbon control performance. This significantly improves the system's practicality, robustness, and long-term emission reduction benefits in complex industrial environments. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall technical solution architecture of the carbon emission data statistics and carbon control system for desulfurization systems proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the multi-scale carbon accounting engine in this invention; Figure 3 This is a logical flow diagram of the carbon flow collaborative optimization module in this invention; Figure 4 This is a schematic diagram of the closed-loop control framework of the carbon control execution and feedback module in this invention. Detailed Implementation

[0021] The carbon emission data statistics and carbon control system for desulfurization systems provided by this invention is described in the appendix. Please refer to the appendix for the overall technical architecture. Figures 1 to 4 This system is an integrated hardware and software platform designed to monitor, calculate, optimize, and control carbon flow throughout the entire lifecycle of flue gas desulfurization systems in coal-fired power plants or similar industrial facilities. By constructing a complete tracking and control closed loop from carbon atom input to output, the system transforms the desulfurization system from a passive pollutant treatment unit into an intelligent, collaborative unit that actively participates in the optimal allocation of carbon resources across the entire plant. The following detailed explanation, based on the system's core modules and accompanied by accompanying diagrams, will provide a comprehensive overview.

[0022] First, the cornerstone of the system is the carbon flow panoramic sensing module. This module is responsible for providing the entire system with real-time, accurate, and comprehensive carbon-related data. Its physical deployment includes a high-precision sensor network covering the entire desulfurization process, as well as a series of standard interfaces for data exchange with other information systems in the plant. The data acquisition scope of this module is strictly divided into three logical layers to ensure the integrity of carbon flow information.

[0023] The first level involves the acquisition of carbon data for process material flow. This level directly tracks the flow and transformation of carbon elements in the desulfurization process in material form. Specifically, electromagnetic flowmeters and densitometers are installed on the desulfurizing agent preparation and supply pipelines to continuously measure the instantaneous volumetric flow rate and density of limestone slurry or ammonia solution. Combined with reports from periodic laboratory analyses of slurry solid content and calcium carbonate purity, carbon content parameters are acquired in real time via a data interface, thereby calculating the input rate of carbon elements in the desulfurizing agent. At the inlet of the original flue gas duct, a non-dispersive infrared gas analyzer and a Pitot tube flowmeter are deployed to simultaneously measure the volumetric concentrations of carbon dioxide and carbon monoxide in the flue gas, as well as the flue gas velocity, temperature, and pressure, once per second. Using the ideal gas law and carbon atomic mass, the mass flow rate of carbon elements is calculated in real time. At the desulfurization byproduct output point, for gypsum, an online near-infrared spectrometer is installed above the discharge belt of the vacuum belt dewatering machine.

[0024] The analyzer performs a line scan of the gypsum filter cake moving on the conveyor belt once per second, collecting the near-infrared spectrum of the material in a specific wavelength band. The collected raw spectral data is immediately transmitted to the built-in embedded processor, which has a pre-set spectral quantitative analysis model trained based on the partial least squares regression algorithm. This model establishes a mathematical mapping relationship between spectral characteristics and the content of key impurities in gypsum, such as residual calcium carbonate and organic carbon. The processor executes the model calculations and outputs two key indicators in real time: the percentage of calcium carbonate content and the percentage of total organic carbon content in the gypsum. Multiplying these two percentage signals by the instantaneous output signal from the belt scale yields the carbon output rate solidified in the gypsum. For desulfurization wastewater, online total organic carbon analyzers and inorganic carbon analyzers are installed on the wastewater discharge pipeline, along with an ultrasonic flow meter, to measure the concentration of dissolved organic and inorganic carbon in the wastewater and the wastewater discharge flow rate in real time, thereby calculating the rate of carbon loss with the wastewater.

[0025] The second level involves the acquisition of process energy flow carbon data. This level calculates the indirect carbon emissions corresponding to the energy consumed by the desulfurization system. During implementation, intelligent power monitoring terminals are installed in the distribution cabinets of all major power-consuming equipment in the desulfurization system, including the absorption tower circulating pump, oxidation fan, slurry discharge pump, process water pump, agitator, and booster fan. These terminals collect instantaneous values ​​of three-phase current and voltage through current transformers and voltage transformers. Internal calculation modules calculate the active power, reactive power, power factor, and other electrical parameters of each device in real time, and upload the real-time active power values ​​of each device to the data aggregation server of the carbon flow panoramic sensing module via industrial Ethernet at a frequency of one data packet per second. For steam consumption, vortex flow meters and temperature and pressure transmitters are installed on the steam pipelines entering the desulfurization system to measure the mass flow rate, temperature, and pressure of the steam in real time. The enthalpy of the steam is calculated using a lookup table method or formula, providing basic data for subsequent carbon emission factor conversion.

[0026] The third level involves acquiring carbon data related to the processes. This level is achieved through the factory-level manufacturing execution system data bus or a unified real-time database interface. The carbon flow panoramic perception module is equipped with a dedicated data acquisition engine, which actively reads upstream and downstream process data closely related to the carbon flow of the desulfurization system from the factory data bus at a polling frequency of once every 5 seconds. Upstream data mainly includes real-time data from the boiler combustion side, such as hourly carbon element analysis data of the coal fed into the furnace, real-time power generation load of the boiler, and total fuel consumption. Downstream data depends on the by-product processing path. If the by-product is gypsum, it reads the natural gas consumption, electricity consumption, and calcination output of the gypsum calciner; if the by-product is ammonium sulfate, it reads the energy consumption data of the crystallizer, centrifuge, and dryer, as well as the ammonium sulfate output and quality data.

[0027] Upon receiving all the raw data streams, the core data processing unit of the carbon flow panoramic sensing module immediately initiates a standardized preprocessing pipeline. This pipeline first performs timestamp alignment, uniformly aligning all data points from different sampling frequencies and communication delays to a global time series with 1-second intervals using an interpolation algorithm. Next, outlier removal is performed using rules based on statistical process control. For example, if a data point deviates from its moving average by more than three standard deviations, it is marked as invalid and replaced by the previous valid value or a linear interpolation result. Then, noise filtering is performed, using a first-order low-pass digital filter to smooth the volatile flow and concentration signals. The filtering time constant is preset to 10 to 30 seconds based on the signal characteristics. Finally, unit conversion is performed. The preprocessing pipeline incorporates a complex unit conversion and calculation subroutine that converts all cleaned raw data—whether it's limestone flow rate in kilograms per hour, carbon dioxide concentration in milligrams per cubic meter, or equipment power in kilowatts—into a unified intermediate variable: a standardized carbon flow rate signal in kilograms of carbon per hour. For example, the conversion formula for flue gas carbon dioxide flow rate is: carbon dioxide mass flow rate multiplied by the ratio of carbon atomic mass to carbon dioxide molecular weight. For indirect carbon flow corresponding to equipment power consumption, the conversion logic is to first accumulate real-time power into time-period energy consumption, and then multiply it by the corresponding carbon emission factor. However, at this stage, only the power signal to be multiplied by the factor is output. After all preprocessing is completed, these standardized carbon flow rate signals are encapsulated into data frames with a fixed structure. The data frame contains four fields: data source identifier, timestamp, carbon flow value, and data quality identifier, and is pushed to the downstream multi-scale carbon accounting engine in real time via a high-speed internal communication bus.

[0028] The multi-scale carbon accounting engine is the core computational brain of the system; please refer to the appendix for its core principle framework. Figure 2 The engine is deployed on a high-performance industrial server and continuously receives standardized carbon flow rate signal streams from the carbon flow panoramic perception module. Two core components run within the engine: a carbon flow tracking model and a dynamic calculation rule base.

[0029] The carbon flow tracking model is a highly abstract mathematical model built upon the law of conservation of mass. It maps the entire desulfurization system and its key process units as a directed graph network. In this network, nodes represent specific process equipment or virtual carbon accumulation points, such as the absorption tower reaction zone, slurry tank, oxidation tank, hydrocyclone, vacuum belt dewatering machine, and wastewater treatment pond. The directed edges between nodes represent possible carbon flow paths, such as carbon flow from the absorption tower to the slurry tank, from the slurry tank to the dewatering machine, from the dewatering machine to the gypsum storage tank, and from each stage to the wastewater treatment pond. Each node maintains a dynamic carbon stock state variable, and each edge is associated with a carbon flow state variable. The model input is the carbon flow rate signal provided by the carbon flow panoramic perception module, acting on the network boundary nodes, such as the carbon flow input signal of flue gas entering the absorption tower and the carbon flow input signal of limestone slurry. The core algorithm of the model is a set of material balance equations based on difference equations. These equations describe the rate of change of carbon stock at each node as equal to the sum of all carbon flows into that node minus the sum of all carbon flows out of that node. In practical implementation, to address the issues of measurement noise and the fact that some carbon flows cannot be directly measured, this embodiment employs an extended Kalman filter as the online state estimator for the carbon flow tracking model.

[0030] The implementation of this extended Kalman filter involves the following steps. First, the system's state vector is defined, containing all carbon stock and key internal carbon flow variables that need to be estimated. Second, state-space equations are established, including state transition equations and observation equations. The state transition equations are constructed based on discretized material balance equations and simplified chemical reaction kinetic equations. For example, at the absorber node, the state transition equations describe the impact of the reaction of dissolved carbon dioxide and sulfite to form sulfate on the carbon stock. The observation equations correlate the direct measurements from the carbon flow panoramic sensing module, such as flue gas outlet carbon dioxide concentration, gypsum production, and carbon content, with the corresponding variables in the state vector. The filter operates as a recursive prediction and update process. In each calculation cycle, for example, every 5 seconds, the filter performs a prediction step: based on the state estimate from the previous moment and the state transition equation, it predicts the current state value and its estimation error covariance matrix. Then, an update is performed: the actual observations from the carbon flow panoramic sensing module are compared with the predicted observations, and the Kalman gain is calculated.

[0031] The Kalman gain is a matrix that dynamically adjusts based on the prediction error covariance and the observation noise covariance, determining the extent to which model predictions are trusted versus actual measurements. Finally, the predicted state is corrected using the Kalman gain to obtain the optimal state estimate for the current moment, and the covariance matrix of the state estimate error is updated synchronously to prepare for the next cycle. In this way, the carbon flow tracking model can not only calculate observable carbon flows but also optimally estimate internal carbon flows that cannot be directly measured by sensors, such as the carbon content in intermediate products reacting within the absorber tower. This significantly improves the overall accuracy, consistency, and anti-interference capability of the entire system's carbon flow measurement.

[0032] The dynamic carbon emission accounting rule base is an online, updatable database that stores all the conversion factors and rules required for carbon emission accounting. These rules strictly adhere to the latest "Guidelines for Enterprise Greenhouse Gas Emission Accounting and Reporting" issued by the relevant national authorities, as well as related supplementary regulations for the power industry. The main contents of the rule base include: the national power grid average carbon emission factor, the specific emission factor of the local power grid, the emission factor of the enterprise's self-owned power plant; the carbon emission factor for steam at different pressure levels; and emission factors for key chemical processes, such as the theoretical carbon dioxide generation factor from the decomposition of calcium carbonate in an absorption tower, and the carbon monoxide emission factor that may be generated during the oxidation of sulfites. The rule base is updated through an administrator interface or a standard data interface. When the state releases new emission factors, the administrator can manually import the update files, or the system can be configured to automatically retrieve updates from authoritative data sources periodically.

[0033] The multi-scale carbon accounting engine operates in a hierarchical manner. The first scale is device-level carbon accounting. The engine obtains real-time power signals for each power-consuming device from the carbon flow panoramic perception module, for example, a circulating pump with a power of 150 kW. Simultaneously, the engine queries the dynamic accounting rule base for the currently applicable electricity carbon emission factor, assumed to be 0.5703 kg CO2 per kilowatt-hour. The calculation subroutine within the accounting engine performs the following calculations: First, it integrates the real-time power value over time to calculate the energy consumption of the device in the current accounting cycle, typically set to 1 hour. Then, it multiplies this energy consumption by the electricity carbon emission factor to obtain the device's indirect carbon emissions in that cycle, expressed in kilograms of CO2 equivalent. The calculation logic for steam consumption is similar: the steam mass flow rate is converted into heat consumption, and then multiplied by the steam carbon emission factor.

[0034] The second scale is system-level carbon accounting. This is the core of the accounting, integrating indirect carbon emissions from all equipment and overlaying direct carbon emissions from the process. The calculation of direct process carbon emissions relies entirely on the results output by the carbon flow tracing model. The specific calculation formula is as follows:

[0035] in, This indicates the direct carbon emissions from the desulfurization system during the accounting period, expressed in kilograms of carbon. This represents the total amount of carbon entering the system through the raw flue gas during the cycle, derived from the carbon flow tracking model's integration of the flue gas carbon inflow. This indicates the total amount of carbon entering the system through the desulfurizing agent during the cycle. This indicates the total amount of carbon that is solidified in desulfurization byproducts and carried out of the system during the cycle. This represents the total amount of carbon carried out of the system during the cycle by wastewater discharge. The essence of this formula is the system's carbon mass balance; carbon that is not fixed by byproducts or discharged with wastewater is considered as carbon directly emitted into the atmosphere in the form of carbon dioxide, etc. The calculation engine multiplies the calculated direct carbon emissions by the carbon dioxide to carbon molecular weight ratio of 44 / 12, converting it to carbon dioxide equivalents. Finally, the total system-level carbon emissions equal the sum of all indirect carbon emissions from equipment plus the direct carbon emissions from the process.

[0036] The third scale is plant-level marginal carbon impact analysis. This scale evaluates the desulfurization system within the plant-wide carbon flow network. The calculation engine utilizes relevant process carbon data acquired by the carbon flow panoramic perception module. For example, it analyzes the impact of improving desulfurization efficiency on the plant-wide carbon balance: improving desulfurization efficiency typically requires increasing the speed of the circulating pump or increasing the limestone slurry supply, which leads to increased carbon emissions from the desulfurization system's own energy and material consumption. However, higher desulfurization efficiency means lower sulfur dioxide concentrations in the flue gas. The marginal benefit is a reduced burden on downstream flue gas treatment or a reduction in the amount of subsequent treatment chemicals needed to meet stricter environmental standards, thus avoiding this portion of the "environmental governance carbon cost." The calculation engine quantitatively estimates this trade-off relationship through a carbon flow coupling model, outputting analytical conclusions such as "For every additional 1 ton of sulfur dioxide reduction, the desulfurization system increases carbon dioxide emissions by X tons, but avoids environmental governance costs equivalent to Y tons of carbon dioxide for the entire plant." All accounting results, including real-time carbon emission intensity, cumulative carbon emissions, carbon flow distribution maps, and contribution analysis reports at various scales, are output in structured JSON or XML data formats and generated into dynamic dashboards through visualization components. At the same time, core data is pushed to the carbon flow collaborative optimization module.

[0037] The carbon flow collaborative optimization module is the system's decision center; please refer to the appendix for its logical flow framework. Figure 3The core task of this module is to solve for a set of process operation parameters that achieve comprehensive optimization of multiple objectives, based on real-time carbon data provided by the multi-scale carbon accounting engine, factory production plans, and environmental constraints. Its optimization foundation is the carbon flow coupling model, which is more macroscopic than the carbon flow tracking model. It quantitatively describes the carbon correlation between internal variables of the desulfurization system and upstream and downstream process variables. For example, the model describes the relationship between limestone slurry flow rate and desulfurization efficiency, the relationship between desulfurization efficiency and outlet sulfur dioxide concentration, the relationship between slurry pH value and oxidation efficiency and gypsum quality, the relationship between oxidation air flow rate and system power consumption, and the relationship between dewatering machine frequency and gypsum moisture content and power consumption. More importantly, it incorporates the impact of upstream boiler load changes and coal quality fluctuations on flue gas volume and sulfur dioxide concentration, as well as the sensitivity requirements of downstream gypsum calcining furnaces to gypsum purity and moisture content, as boundary conditions or correlation functions into the model.

[0038] The core algorithm component of this module is a multi-objective dynamic optimization controller. The controller needs to handle three optimization objectives simultaneously.

[0039] Objective 1: Minimize the total carbon emission cost of operating the desulfurization system. This cost includes the carbon cost of energy consumption and the carbon cost of chemical agent consumption.

[0040] Objective 2: Maximize the carbon gains from the resource utilization of desulfurization byproducts. This means optimizing operations to produce gypsum with higher purity, lower moisture content, and better stability, thereby enabling it to obtain higher market value or more clearly defined carbon offset certification when used as a building material raw material. This portion of the gains is converted into negative carbon costs.

[0041] Objective 3: Meet dynamically changing environmental constraints. This means ensuring that the sulfur dioxide concentration in the flue gas at the desulfurization tower outlet is always below the limit stipulated by environmental regulations; this is a mandatory constraint.

[0042] In its implementation, the controller employs a hybrid optimization algorithm that combines the global search capability of particle swarm optimization with the fast local convergence capability of linear programming. At the start of each optimization cycle, the controller receives the latest carbon data from the multi-scale carbon accounting engine, the production plan for the next few hours from the plant scheduling system, and the real-time sulfur dioxide emission limits from the environmental monitoring system. The controller encodes the current operating status of the desulfurization system, including all adjustable decision variables such as limestone slurry supply flow rate, circulating slurry pH setpoint, oxidation air flow rate, operating frequency of each circulating pump, and speed of the vacuum belt dewatering machine, into a multi-dimensional vector called "particle position".

[0043] The optimization process takes place in a simulated decision space. The particle swarm optimization algorithm initializes a swarm of "particles," each representing a set of possible combinations of decision variables. Each particle flies and searches within the decision space based on its own historical best position and the historical best position of the entire particle swarm. In each iteration, the algorithm evaluates the impact of each particle's corresponding plan on the three objectives. This evaluation is performed by substituting the particle's decision variable values ​​into a carbon flux coupling model. The model simulates the future operating state of the system under this set of operating parameters and calculates the corresponding carbon emission costs, byproduct resource recovery benefits, and export sulfur dioxide concentration.

[0044] For objective one, the calculation of carbon emission costs incorporates external market signals as dynamic weights. The controller obtains the latest transaction price of carbon allowances from the national or local carbon emission trading market in real time through a secure data interface. Assuming the current carbon price is 60 yuan per ton of CO2 equivalent, then the cost coefficient per ton of carbon emissions in the objective function is 60 yuan. When the carbon price rises to 80 yuan, the economic value of emission reduction increases, and the controller automatically assigns a higher implicit weight to the emission reduction target during optimization.

[0045] For objective two, the revenue from resource recovery is calculated based on a model that correlates the quality of by-products with market benchmark prices.

[0046] For objective three, the sulfur dioxide concentration at the outlet must be below the limit; otherwise, the corresponding scheme for that particle will be marked as infeasible.

[0047] Through multiple iterations, the algorithm eventually converges to a Pareto optimal solution set. Each solution in this set represents an objective that cannot be further optimized without compromising other objectives. The controller has a pre-defined priority and weighting strategy, such as prioritizing cost targets when carbon prices are high and prioritizing emission compliance targets during environmental inspections. Based on the current priority strategy, the controller selects the final optimal operating point from the Pareto optimal solution set. This operating point is then transformed into a specific, quantified set of carbon control instructions, such as "adjust the pH setpoint of the slurry in absorber tower 1 from 5.5 to 5.4", "increase the outlet pressure setpoint of the oxidation blower by 2 kPa", and "decrease the frequency of circulation pump 2 by 1.5 Hz".

[0048] The carbon control execution and feedback module is the final link in the system's interaction with the physical world. Please refer to the appendix for its closed-loop control framework. Figure 4 This module is responsible for safely, reliably, and smoothly translating the strategic instructions generated by the carbon flow collaborative optimization module into the underlying control actions of the desulfurization system's distributed control system, and forming a closed-loop feedback to ensure effective control. This module consists of three sub-units: an instruction translator, an execution agent, and a feedback evaluation unit.

[0049] The instruction translator receives a set of carbon control instructions. These instructions are relatively high-level and geared towards optimization goals. The translator's task is to "translate" them into low-level control loop setpoint adjustments that the distributed control system can understand and execute. The translator internally stores detailed mappings and safe operating ranges for all control loops in the desulfurization system. For example, for the instruction "reduce system carbon intensity by 5%", the translator will break it down into a series of specific actions based on current operating conditions and model knowledge: First, fine-tune the opening of the limestone slurry supply valve to reduce the supply flow by 3%; second, lower the setpoint of the absorber slurry pH control loop by 0.1; simultaneously, assess the redundancy of the oxidation fans and, while ensuring oxidation efficiency, reduce the speed of one fan by 5%. Each decomposed action is accompanied by safety parameters such as maximum adjustment range and adjustment rate limits to prevent impact on the production process.

[0050] The execution agent is a software service responsible for direct communication with the distributed control system. It uses standard industrial communication protocols, such as OPC UA or Modbus TCP, to establish a secure connection with the controller of the distributed control system. Upon receiving a setpoint adjustment command from the translator, the execution agent does not immediately and abruptly overwrite the existing setpoint. Instead, it uses a ramp function generator to smoothly and linearly transition the difference between the target setpoint and the current actual setpoint within a configurable time window, for example, gradually adjusting it to the correct position over 120 seconds. This smooth write mechanism avoids drastic fluctuations in the control system, ensuring production stability. After each write operation, the execution agent reads the controller's feedback value to confirm whether the setpoint has been successfully updated and records the operation in the log.

[0051] The feedback evaluation unit compares the predicted trajectory with the expected target trajectory in advance. If the prediction shows that a serious deviation will occur, it can send an early warning signal to the carbon flow collaborative optimization module in advance without waiting for the actual deviation to occur, and start preventive re-optimization, thereby achieving forward-looking carbon management.

Claims

1. A carbon emission data statistics and carbon control system for desulfurization systems, characterized in that, include: The carbon flow panoramic sensing module is used to collect all direct and indirect carbon-related data in the entire desulfurization process in real time, and to preprocess the collected raw data to generate a standardized carbon flow rate signal. The multi-scale carbon accounting engine is connected to the carbon flow panoramic perception module. It is used to receive the standardized carbon flow rate signal and perform multi-level carbon accounting from the device level to the system level and then to the factory level based on the built-in carbon flow tracking model and dynamic accounting rule library, so as to output structured carbon accounting results. The carbon flow collaborative optimization module is connected to the multi-scale carbon accounting engine. It is used to solve the carbon flow coupling model under the constraints of the built-in multi-objective dynamic optimization controller based on the structured carbon accounting results, preset optimization objectives and external input production plans and environmental constraints, so as to generate a globally optimal carbon management instruction set. The carbon control execution and feedback module is connected to the carbon flow collaborative optimization module and the distributed control system of the desulfurization system. It is used to decompose the carbon control instruction set and convert it into executable control signals and send them to the distributed control system. At the same time, it monitors the actual effect after the instruction is executed and compares it with the expected effect to form a closed-loop feedback and trigger strategy re-optimization.

2. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 1, characterized in that, The data collected by the carbon flow panoramic perception module includes process material flow carbon data, process energy flow carbon data, and associated process carbon data. The process material flow carbon data includes the instantaneous flow rate and carbon content of the desulfurizing agent, the concentration and flow rate of carbon dioxide and carbon monoxide in the raw flue gas, the yield and carbon content of desulfurization by-products, and the discharge volume of desulfurization wastewater and its dissolved organic carbon and inorganic carbon concentration. The process energy flow carbon data includes the real-time power of all power-consuming equipment in the desulfurization system and the system steam consumption. The associated process carbon data is acquired through a factory-level data bus, including carbon element analysis data of boiler coal combustion, power generation load, and energy consumption and material data of downstream by-product processing.

3. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 2, characterized in that, The carbon flow panoramic perception module performs preprocessing on the collected raw data, including outlier removal, noise filtering, and timestamp alignment, and converts all data into the standardized carbon flow rate signal in kilograms of carbon per hour.

4. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 1, characterized in that, The device-level carbon accounting process executed by the multi-scale carbon accounting engine is as follows: based on the real-time power signal of the device in the standardized carbon flow rate signal, combined with the power carbon emission factor stored in the dynamic accounting rule base, the indirect carbon emissions of a single device are calculated. The system-level carbon accounting process is as follows: integrate the indirect carbon emissions of all equipment and superimpose the direct carbon emissions of the process calculated based on the carbon flow tracking model. The formula for calculating the direct carbon emissions of the process is the total carbon flow entering the system minus the total carbon flow solidified in by-products and the total carbon flow discharged with wastewater. The plant-level carbon accounting process involves coupling and analyzing the net carbon emissions of the desulfurization system with the associated process carbon data to calculate the marginal impact of the desulfurization system operation on the overall carbon balance of the plant.

5. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 4, characterized in that, The carbon flow tracing model is a mathematical model based on the law of conservation of mass. This model abstracts the desulfurization system into a network containing multiple carbon nodes, each node representing a process unit, and the connection between nodes representing the carbon flow path. The model calculates and tracks the changes in the quantity and flow of carbon atoms in each process unit node in real time based on the input standardized carbon flow rate signal.

6. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 5, characterized in that, The carbon flow tracking model uses an extended Kalman filter for state estimation. The extended Kalman filter takes the measurement values ​​of the carbon flow panoramic sensing module as the observation input and uses the state-space equations constructed based on chemical reaction kinetics and material balance as the prediction model. Through recursive calculation, it makes the optimal estimation of carbon flow state variables that cannot be directly measured in the model and updates the covariance matrix of the estimation error simultaneously.

7. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 1, characterized in that, The multi-objective dynamic optimization controller in the carbon flow collaborative optimization module simultaneously handles three optimization objectives: minimizing the total carbon emission cost of the desulfurization system operation, maximizing the resource utilization carbon benefits of desulfurization by-products, and meeting the environmental constraints of dynamically changing sulfur dioxide emission concentrations. The controller employs a hybrid algorithm combining constrained particle swarm optimization and linear programming for solution. In each optimization cycle, the controller iteratively adjusts a set of decision variables, including limestone slurry supply flow rate, circulating slurry pH setpoint, oxidation air flow rate, and dewatering machine operating frequency. The algorithm finds a Pareto optimal solution set by simulating the impact of different combinations of decision variables on the optimization objective, and selects an optimal operating point from this solution set according to preset priority weights to transform it into the carbon control instruction set.

8. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 7, characterized in that, When solving for the Pareto optimal solution set, the multi-objective dynamic optimization controller introduces the carbon emission trading price as a dynamic weighting coefficient; the controller obtains the carbon quota price of the carbon emission trading market in real time through a data interface, and uses this price as a key coefficient in the objective function of minimizing the total carbon emission cost.

9. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 1, characterized in that, The carbon control execution and feedback module includes an instruction translator, an execution agent, and a feedback evaluation unit; The instruction translator is used to translate the high-level optimization objectives in the carbon management instruction set into specific setpoint adjustment amounts in the underlying control loop; The execution agent is used to safely and smoothly write the set value adjustment amount into the corresponding control module of the distributed control system; The feedback evaluation unit is used to continuously monitor the actual operating status of the system after the command is executed, compare the actual effect monitored with the expected effect of the carbon flow collaborative optimization module and calculate the deviation value. When the deviation value continues to exceed the preset threshold, a re-optimization trigger signal is sent to the carbon flow collaborative optimization module.

10. The carbon emission data statistics and carbon control system for desulfurization systems according to claim 9, characterized in that, The feedback evaluation unit incorporates a prediction model based on a long short-term memory network. The prediction model predicts the trajectory of key parameters within a set timeframe based on the currently issued control commands and system operating conditions. The feedback evaluation unit compares the predicted trajectory with the expected target trajectory in advance. If the prediction indicates a significant deviation, it sends an early warning signal to the carbon flow collaborative optimization module to initiate preventative re-optimization.

Images