Carbon capture pretreatment control system and method based on multi-parameter predictive model

Abstract

The present application relates to a kind of carbon capture pretreatment control system and method based on multi-parameter prediction model, the system includes: perception layer, real-time acquisition multiple monitoring signals;Control layer, embedded multi-parameter prediction model, according to the inlet and outlet flue gas parameters, scrubber liquid level and circulating liquid pH of real-time acquisition of perception layer, the optimal amount of alkali liquid, flue gas adjustment amount, scrubbing liquid circulation and pollution control strategy are calculated using multi-parameter prediction model, and collaborative control instruction is generated;Execution layer, receive the collaborative control instruction of control layer, drive corresponding execution mechanism to carry out accurate regulation.Compared with prior art, the present application has the advantages of significantly reducing operating cost, improving automation and intelligent level etc..

Description

Technical Field

[0001] This invention relates to carbon capture pretreatment control technology, and in particular to a carbon capture pretreatment control system and method based on a multi-parameter prediction model. Background Technology

[0002] In the carbon capture, utilization, and storage (CCUS) technology chain, the pretreatment unit is used to reduce flue gas temperature and remove acidic impurities such as SO2 from the flue gas. It is a key link to ensure the long-term stable and efficient operation of the subsequent carbon capture absorbent. Currently, desulfurization pretreatment generally uses alkaline solutions (such as Na2CO3 and NaOH solutions) for spray washing, and the desulfurization effect is indirectly controlled by monitoring the pH value of the circulating liquid.

[0003] However, this traditional operating control mode maintains the pH value within a fixed range (e.g., 7-9). This extensive control method has two significant drawbacks: (1) Poor operational economy: In order to prevent the desulfurization efficiency from collapsing due to the pH value falling below the lower limit, a conservative strategy of "better to add more than less" is often adopted in operation. Moreover, the start and stop of the alkali pump and the scrubbing tower drain valve are not related, and they are opened at the same time, which further leads to excessive addition of alkali and serious waste of reagents. At the same time, the frequent start and stop of the alkali pump also increases equipment wear and energy consumption, resulting in high operating costs.

[0004] (2) Low level of automation and slow dynamic response: pH changes lag significantly behind alkali addition and upstream operating condition fluctuations. When key parameters such as inlet flue gas flow rate, SO2 concentration, and temperature are disturbed, control systems based on fixed thresholds or simple PID feedback struggle to achieve rapid and accurate adjustments. This often leads to prolonged pH fluctuations outside the target range, affecting not only desulfurization stability and compliant emissions, but also further exacerbating the ineffective consumption of resources.

[0005] A search of Chinese Patent Publication No. CN206996249U reveals a precise alkali addition device for flue gas acidification tail gas absorption, aiming to solve the problem of inaccurate alkali addition and excessive alkali waste during tail gas absorption. The device includes a tail gas absorption tower with a spray device at the top and a liquid chamber at the bottom. The liquid chamber has an air inlet at the top and a circulating drain device at the bottom, connected to a circulating liquid pH meter. A circulating pump is also connected to the bottom of the liquid chamber, with its outlet connected to the spray device. A discharge liquid pH meter is also installed at the outlet of the circulating pump. An alkali addition pipe is connected to the liquid chamber, with a main alkali addition valve and a secondary alkali addition valve connected in parallel, both interlocked with an intelligent fuzzy controller. This existing patent, through precise control of the pH values ​​of the circulating liquid and discharge liquid, prevents the liquid alkali from neutralizing with the acidic water in the tower, effectively reducing liquid alkali waste. However, this existing patent uses a traditional controller without incorporating intelligent algorithms, resulting in limited control accuracy.

[0006] Therefore, existing technologies lack an intelligent control system that can proactively predict demand and coordinate the regulation of multiple variables, so as to fundamentally solve the technical problems of control lag, resource waste and high operating costs. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the existing technology and provide a carbon capture pretreatment control system and method based on a multi-parameter prediction model.

[0008] The objective of this invention can be achieved through the following technical solutions: According to one aspect of the present invention, a carbon capture pretreatment control system based on a multi-parameter prediction model is provided, the system comprising: The sensing layer collects multiple monitoring signals in real time; The control layer, with its embedded multi-parameter prediction model, calculates the optimal alkali dosage, flue gas regulation, scrubbing liquid circulation volume, and sewage discharge strategy based on the real-time inlet and outlet flue gas parameters, scrubbing tower liquid level, and circulating liquid pH collected by the sensing layer, and generates coordinated control commands. The execution layer receives coordinated control commands from the control layer and drives the corresponding actuators to make precise adjustments.

[0009] As a preferred technical solution, the control layer further includes: Based on the inlet working condition multi-step feedforward prediction loop, it is used to adjust the alkali dosage in real time and proactively according to changes in working conditions. A real-time feedback correction loop based on export status is used to correct the cumulative bias of a multi-parameter prediction model.

[0010] As a preferred technical solution, the multi-parameter prediction model is an optimization decision-making algorithm with the core objective of minimizing the total operating cost of the system. This optimization decision-making algorithm comprehensively considers multiple constraints such as desulfurization efficiency, equipment safety, reagent consumption, and energy consumption, and solves the optimal control instruction set for multiple actuators for inlet flue gas flow, alkali injection acceleration rate, and circulating liquid discharge in an online coordinated manner.

[0011] As a preferred technical solution, the sensing layer includes: The first automatic flue gas monitor, installed on the inlet pipe of the pretreatment tower, is used to measure flue gas flow rate, inlet SO2 concentration, temperature and humidity in real time; A pH meter installed on the circulating fluid pipeline; A second automatic flue gas monitor and a scrubbing tower level gauge are installed on the flue gas pipeline at the outlet of the scrubbing tower.

[0012] As a preferred technical solution, the control layer includes a data acquisition module and a controller connected in sequence. The data acquisition module is connected to each sensor in the perception layer to collect each monitoring signal in real time and send it to the controller. The controller filters and standardizes the received signals before calling a multi-parameter prediction model for processing.

[0013] As a preferred technical solution, the sensing layer includes: an inlet flue gas flow regulating valve, a variable frequency alkali dosing pump and its matching alkali pump regulating valve, a washing liquid circulation pump and its matching circulation pump regulating valve, and a washing liquid drain valve, which respectively receive instructions from the controller to regulate the material flow.

[0014] According to another aspect of the present invention, a method for employing the carbon capture pretreatment control system based on the multi-parameter prediction model is provided, the method comprising the following steps: Step S1: The data acquisition module acquires multiple monitoring signals from the perception layer in real time and sends them to the controller. The controller performs filtering and standardization preprocessing on the acquired signals. Step S2: The controller calculates the current desulfurization efficiency of the system based on real-time data, compares it with the preset efficiency target value, and comprehensively judges whether the system is currently in normal load, high load or low load condition, and monitors whether the pH value and liquid level are within the preset safe operating range. Step S3: The controller calls a multi-parameter prediction model, takes the current real-time data and historical operating data as input, and performs multi-step prediction. The prediction output includes the changing trends of desulfurization efficiency and washing liquid pH under different control strategies in the next few control cycles, the theoretical consumption of alkali solution required to maintain the target pH range, and the changing trend of liquid level in the tower. Step S4: The controller takes minimizing operating costs as the core economic objective and ensures desulfurization efficiency, stable pH value and liquid level as constraints. It constructs a multi-objective optimization function and solves it online through an optimization decision algorithm to obtain a set of cooperative optimal control instructions. In step S5, the controller compares and verifies the collaborative control command generated in step S4 with the safe operating range of each actuator and the permissible mutation rate of the process. If all commands are within the safety constraints, the command is issued; if they exceed the constraints, the controller returns to step S4 to adjust the optimization weights and constraint boundaries and then solves the problem again. In step S6, the controller sends corresponding control signals to the inlet flue gas flow regulating valve, the variable frequency alkali dosing pump, the alkali pump regulating valve, the washing liquid and the washing liquid drain valve, thereby regulating the inlet flue gas flow and balancing the system load, accurately adding the required amount of alkali, draining as needed, and maintaining the liquid level and effective concentration of the circulating liquid.

[0015] As a preferred technical solution, the judgment in step S2 is specifically as follows: a1) Normal state: Desulfurization efficiency meets standards, pH is within the set range. b1) High load condition: Inlet SO2 concentration exceeds the set high threshold. c1) Low load state: The inlet SO2 concentration is lower than the set low threshold; d) Abnormal conditions: abnormal pH or abnormal liquid level.

[0016] As a preferred technical solution, the optimal control instruction set in step S4 specifically includes: a2) Adjustment amount of the opening of the inlet flue gas flow regulating valve; b2) Adjustment of the opening of the washing liquid circulation pump and its associated regulating valve; c2) The instantaneous dosing command of the variable frequency alkali solution dosing pump and the coordinated setting value of the opening of the associated alkali solution pump regulating valve; d2) Instructions on the timing and duration of opening the detergent drain valve.

[0017] As a preferred technical solution, the method further includes: Step S7: After one control cycle, the controller collects the actual response data of the system again, compares the actual response value with the model prediction value in step S3, obtains the deviation, and uses an adaptive algorithm to fine-tune the internal parameters of the multi-parameter prediction model based on the deviation, so as to continuously improve the prediction accuracy of the model and the robustness of the control system; after the correction is completed, the process returns to step S1 and enters the next control loop.

[0018] Compared with the prior art, the present invention has the following advantages: 1) This invention applies a multi-parameter dynamic prediction model to the carbon capture pretreatment stage, overcoming the limitations of traditional single-parameter control and realizing a fundamental shift from "post-event remediation" to "pre-event prediction," which is the theoretical basis for achieving energy saving and precise control.

[0019] 2) This invention integrates a dual control loop of "multi-step feedforward prediction based on inlet operating conditions" and "real-time feedback correction based on outlet status"; the feedforward loop adjusts the amount of alkali added in real time and proactively according to changes in operating conditions, overcoming the problem of large lag; the feedback loop is used to correct the cumulative deviation of model prediction, ensuring long-term control accuracy. 3) The optimization decision algorithm of this invention takes minimizing the total operating cost of the system as its core objective. This algorithm comprehensively considers multiple constraints such as desulfurization efficiency, equipment safety, reagent consumption, and energy consumption. It coordinates and solves the optimal control instruction set for multiple actuators such as inlet flue gas flow, alkali injection acceleration rate, and circulating liquid discharge in online, realizing the leap from single parameter adjustment to multi-variable system collaborative optimization. 4) This invention significantly reduces operating costs: through precise prediction and coordinated control, excessive addition of alkali solution can be avoided; at the same time, smooth equipment operation reduces the energy consumption and mechanical wear of frequent pump start-stop, and the overall operating cost is effectively reduced. 5) This invention improves the level of automation and intelligence: it realizes the transformation from manual experience intervention to fully automatic, adaptive intelligent control, reduces the risk of human error, and improves the stability and reliability of system operation.

[0020] 6) This invention enhances the system's anti-interference capability and stability: multi-parameter feedforward control can quickly respond to drastic fluctuations in inlet operating conditions, stabilizing the pH value within a narrower optimal range, thereby ensuring the stability of desulfurization efficiency and providing more stable intake conditions for subsequent carbon capture units.

[0021] 7) Enhanced equipment reliability and lifespan of the present invention: The stable control strategy avoids large and drastic fluctuations in process parameters, effectively reduces equipment corrosion, scaling and mechanical stress fatigue, thereby extending the service life of key equipment (such as pumps, valves and towers) and reducing maintenance costs. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the control system of the present invention; Figure 2 This is a flowchart illustrating the control method of the present invention. 1 is the first automatic flue gas monitor, 2 is the pH meter, 3 is the second automatic flue gas monitor, 4 is the inlet flue gas flow regulating valve, 5 is the scrubbing tower level gauge, 6 is the scrubbing tower, 7 is the scrubbing liquid circulation pump, 8 is the scrubbing liquid drain valve, 9 is the circulating cooler, 10 is the variable frequency alkali solution dosing pump, 11 is the alkali solution tank, 12 is the alkali solution pump regulating valve, 13 is the data acquisition module, 14 is the controller, and 15 is the circulation pump regulating valve. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] Example 1 This invention provides a carbon capture pretreatment control system based on a multi-parameter prediction model. This system achieves accurate prediction of the circulating liquid pH value by constructing a prediction model that integrates mechanism and data-driven approaches, and establishes a quantitative relationship between alkali demand and multiple operating parameters. The core controller, based on real-time collected inlet and outlet flue gas parameters, scrubbing tower level, and circulating liquid pH, uses this model to predictively calculate the optimal alkali dosage, flue gas regulation rate, scrubbing liquid circulation rate, and wastewater discharge strategy. It then generates coordinated control commands and drives corresponding actuators (such as variable frequency dosing pumps, flue gas regulating valves, circulating pump regulating valves, and wastewater discharge valves) for precise adjustment. Simultaneously, the system incorporates feedback from the outlet pollutant concentration for closed-loop correction. Thus, it upgrades traditional, lagging, single-variable empirical control to proactive, multi-variable collaborative intelligent predictive control, thereby solving the problems of excessive alkali addition, lagging regulation, and low operational efficiency.

[0025] like Figure 1 As shown, the control system of the present invention specifically includes: The sensing layer includes a first automatic flue gas monitor 1 installed on the inlet pipe of the pretreatment tower (which measures flue gas flow, inlet SO2 concentration, temperature, humidity, etc. in real time), a pH meter 2 installed on the circulating liquid pipe, a second automatic flue gas monitor 3 installed on the outlet flue gas pipe of the scrubbing tower, and a scrubbing tower level gauge 5.

[0026] The control layer, whose core is the controller 14, incorporates the prediction model and optimization control algorithm of this invention. It receives all real-time data from the perception layer through the data acquisition module 13 and performs preprocessing.

[0027] The execution layer includes an inlet flue gas flow regulating valve 4, a variable frequency alkali dosing pump 10 and its matching alkali pump regulating valve 12, a washing liquid circulation pump 7 and its matching circulation pump regulating valve 15, and a washing liquid drain valve 8. It receives instructions from the controller and precisely regulates the material flow.

[0028] The main process and auxiliary equipment include the washing tower 6 and the circulating cooler 9, which are the core physical equipment for completing the desulfurization and cooling process.

[0029] This invention addresses the issue of "high operating costs" by shifting from "experience-based over-dosing" to "precise, on-demand coordination." Traditional methods often involve excessive addition of alkali solution to ensure safety. This system, however, uses a multi-parameter prediction model to directly and accurately calculate the theoretical alkali solution requirement based on real-time flue gas parameters, achieving "on-demand dosing" and reducing waste. Simultaneously, its multivariate collaborative optimization algorithm treats flue gas load regulation, alkali solution dosing, and pollution control as a whole for economic optimization, rather than adjusting a single aspect in isolation, thereby minimizing operating costs at the system level.

[0030] This invention addresses the issues of "lagging regulation and low automation" by shifting from "lagging feedback" to "leading feedforward-feedback composite control". Traditional PID control is ineffective for pH adjustment with large time lags. This system innovatively constructs a feedforward-driven control architecture. Utilizing changes in inlet conditions (such as a surge in SO2 concentration) as a feedforward signal, the alkali dosage is adjusted in advance before the pH value fluctuates, actively counteracting disturbances and fundamentally overcoming the time lag effect. Real-time pH feedback is only used for minor model corrections, ensuring long-term accuracy.

[0031] Other implementations of this embodiment are as follows: Alternative implementation schemes for the prediction model: The data-driven part of the fusion model can employ machine learning or deep learning algorithms such as neural networks (e.g., LSTM), support vector machine regression (SVR), and Gaussian process regression (GPR). In scenarios with sufficient data accumulation, adaptive optimization models based on reinforcement learning can also be explored.

[0032] An alternative implementation of the control architecture: The multivariable collaborative control logic described above can be encapsulated and deployed as a standard model predictive controller (MPC). Within each control cycle, the MPC directly performs rolling optimization based on the dynamic model to solve for the optimal control sequence that satisfies multiple constraints, achieving equivalent or even better control performance.

[0033] System deployment and integration variations: The functions of the intelligent controller 14 can be physically integrated into a single industrial PLC / DCS, or a distributed architecture of "edge computing unit + cloud model training" can be adopted. The data acquisition module 13 can be independent hardware or a built-in function of the controller.

[0034] Alternatives for actuators: The actuators for alkali dosing are not limited to the combination of "variable frequency pump + regulating valve". Metering pumps or electric regulating valves can also be used for precise flow control. Flue gas flow regulation can also be achieved by using a variable frequency fan instead of an inlet regulating valve.

[0035] Example 2 like Figure 2 As shown, the present invention also provides a carbon capture pretreatment control method based on a multi-parameter prediction model, which specifically includes the following steps: Step S1: Data Acquisition and Preprocessing The data acquisition module 13 acquires multiple monitoring signals from the sensing layer in real time, including: flue gas flow rate, SO2 concentration (C_in), temperature, and humidity signals from the first automatic flue gas monitor 1; pH value signal from the pH meter 2; outlet SO2 concentration (C_out) and humidity signal from the second automatic flue gas monitor 3; and tower level signal from the scrubbing tower level gauge 5. The controller performs filtering and standardization preprocessing on the acquired signals.

[0036] Step S2: Operational Status Assessment Based on the real-time data, the controller calculates the current system desulfurization efficiency η = (C_in - C_out) / C_in × 100% and compares it with the preset efficiency target value. Simultaneously, it comprehensively determines whether the system is currently operating under normal load, high load, or low load conditions, and monitors whether the pH value and liquid level are within the preset safe operating range.

[0037] a1) Normal state: Desulfurization efficiency meets the standard, and pH is within a reasonable range; b1) High load condition: Inlet SO2 concentration increases significantly; c1) Low load condition: Inlet SO2 concentration is significantly reduced; d1) Abnormal conditions: abnormal pH or abnormal liquid level; Step S3: Multivariate model prediction; The controller invokes the embedded mechanism and data-driven fusion model, using current real-time data and historical operational data as input, to perform multi-step predictions. The prediction output includes: a2) The changing trends of desulfurization efficiency and washing liquid pH under different control strategies over several control cycles in the future; b2) The theoretical amount of alkali solution required to maintain the target pH range; c2) The trend of liquid level change in the tower.

[0038] Step S4: Multi-objective optimization decision The controller prioritizes minimizing operating costs as its core economic objective, while ensuring desulfurization efficiency, stable pH value, and liquid level as constraints. A multi-objective optimization function is constructed. Through online optimization algorithms, a set of cooperative optimal control commands is obtained, including: a3) Adjustment amount of the opening of the inlet flue gas flow regulating valve 4; b3) Adjust the washing liquid circulation pump 7 and its matching regulating valve 15; c3) The instantaneous dosing command of the variable frequency alkali dosing pump 10 and the opening degree of its associated alkali pump regulating valve 12 are set in coordination. d3) Instructions on the timing and duration of opening the washing liquid drain valve 8.

[0039] Step S5: Safety Constraint Check The controller compares and verifies the collaborative control commands generated in step S4 with the safe operating range of each actuator and the permissible rate of change in the process. If all commands are within the safety constraints, the commands are issued. If the constraints are exceeded, the process returns to step S4 to adjust the optimization weights and constraint boundaries before resolving the problem.

[0040] Step S6: Issuance of multivariable coordinated control commands The control bus sends corresponding control signals to the inlet flue gas flow regulating valve 4, the variable frequency alkali dosing pump 10, the alkali pump regulating valve 12, the washing liquid and the washing liquid drain valve 8, thereby regulating the inlet flue gas flow, balancing the system load, accurately adding the required amount of alkali, draining as needed, and maintaining the liquid level and the effective concentration of the circulating liquid.

[0041] Step S7: Closed-loop feedback and adaptive adjustment After one control cycle, the controller again collects the actual system response data (such as the new pH value, liquid level, and outlet concentration), and compares the actual response value with the model prediction value in step S3. Based on this discrepancy, an adaptive algorithm is used to fine-tune the internal parameters of the fusion model to continuously improve the model's prediction accuracy and the robustness of the control system. After the correction is completed, the process returns to step S1 and enters the next control loop.

[0042] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A carbon capture pretreatment control system based on a multi-parameter prediction model, characterized in that, The system includes: The sensing layer collects multiple monitoring signals in real time; The control layer, with its embedded multi-parameter prediction model, calculates the optimal alkali dosage, flue gas regulation, scrubbing liquid circulation volume, and sewage discharge strategy based on the real-time inlet and outlet flue gas parameters, scrubbing tower liquid level, and circulating liquid pH collected by the sensing layer, and generates coordinated control commands. The execution layer receives coordinated control commands from the control layer and drives the corresponding actuators to make precise adjustments.

2. The carbon capture pretreatment control system based on a multi-parameter prediction model according to claim 1, characterized in that, The control layer also includes: Based on the inlet working condition multi-step feedforward prediction loop, it is used to adjust the alkali dosage in real time and proactively according to changes in working conditions. A real-time feedback correction loop based on export status is used to correct the cumulative bias of a multi-parameter prediction model.

3. The carbon capture pretreatment control system based on a multi-parameter prediction model according to claim 1, characterized in that, The multi-parameter prediction model is an optimization decision-making algorithm with the core objective of minimizing the total operating cost of the system. This optimization decision-making algorithm comprehensively considers multiple constraints such as desulfurization efficiency, equipment safety, reagent consumption, and energy consumption, and solves the optimal control instruction set for multiple actuators for inlet flue gas flow, alkali injection acceleration rate, and circulating liquid discharge in an online coordinated manner.

4. The carbon capture pretreatment control system based on a multi-parameter prediction model according to claim 1, characterized in that, The sensing layer includes: The first automatic flue gas monitor (1) installed on the inlet pipe of the pretreatment tower is used to measure the flue gas flow rate, inlet SO2 concentration, temperature and humidity in real time; pH meter (2) installed on the circulating liquid pipeline; The second automatic flue gas monitor (3) and the scrubbing tower level gauge (5) are installed on the flue gas pipeline at the outlet of the scrubbing tower.

5. The carbon capture pretreatment control system based on a multi-parameter prediction model according to claim 4, characterized in that, The control layer includes a data acquisition module (13) and a controller (14) connected in sequence. The data acquisition module (13) is connected to each sensor of the perception layer and is used to collect each monitoring signal in real time and send it to the controller (14). The controller (14) filters and standardizes the received signal and then calls the multi-parameter prediction model for processing.

6. The carbon capture pretreatment control system based on a multi-parameter prediction model according to claim 1, characterized in that, The sensing layer includes: an inlet flue gas flow regulating valve (4), a variable frequency alkali dosing pump (10) and its matching alkali pump regulating valve (12), a washing liquid circulation pump (7) and its matching circulation pump regulating valve (15) and a washing liquid drain valve (8), which respectively receive instructions from the controller (14) to regulate the material flow.

7. A method for a carbon capture pretreatment control system based on a multi-parameter prediction model as described in any one of claims 1-6, characterized in that, The method includes the following steps: Step S1, the data acquisition module (13) acquires multiple monitoring signals from the perception layer in real time and sends them to the controller (14), which performs filtering and standardization preprocessing on the acquired signals; Step S2, the controller (14) calculates the current desulfurization efficiency of the system based on real-time data, compares it with the preset efficiency target value, comprehensively judges whether the system is currently under normal load, high load or low load conditions, and monitors whether the pH value and liquid level are within the preset safe operating range; Step S3, the controller (14) calls the multi-parameter prediction model, takes the current real-time data and historical operating data as input, and performs multi-step prediction. The prediction output includes the trend of desulfurization efficiency and washing liquid pH under different control strategies in the next few control cycles, the theoretical consumption of alkali solution required to maintain the target pH range, and the trend of liquid level change in the tower. Step S4, the controller (14) takes minimizing operating costs as the core economic objective and ensures desulfurization efficiency, stable pH value and liquid level as constraints, constructs a multi-objective optimization function, solves it online through an optimization decision algorithm, and obtains a set of cooperative optimal control instructions; In step S5, the controller (14) compares and verifies the collaborative control command generated in step S4 with the safe working range of each actuator and the allowable mutation rate of the process. If the command is within the safety constraints, the command is issued; if it exceeds the constraints, the controller returns to step S4 to adjust the optimization weights and constraint boundaries and then solves the problem again. In step S6, the controller (14) sends corresponding control signals to the inlet flue gas flow regulating valve (4), the frequency conversion alkali dosing pump (10), the alkali pump regulating valve (12), the washing liquid and the washing liquid drain valve (8), thereby regulating the inlet flue gas flow and balancing the system load, accurately adding the required amount of alkali, draining as needed, and maintaining the liquid level and effective concentration of the circulating liquid.

8. The method according to claim 7, characterized in that, The judgment in step S2 is specifically as follows: a1) Normal state: Desulfurization efficiency meets standards, pH is within the set range. b1) High load condition: Inlet SO2 concentration exceeds the set high threshold. c1) Low load state: The inlet SO2 concentration is lower than the set low threshold; d) Abnormal conditions: abnormal pH or abnormal liquid level.

9. The method according to claim 7, characterized in that, The optimal control instruction set in step S4 specifically includes: a2) Adjustment amount of the opening of the inlet flue gas flow regulating valve (4); b2) Adjustment of the opening of the washing liquid circulation pump (7) and its matching regulating valve (15); c2) The instantaneous dosing command of the variable frequency alkali dosing pump (10) and the opening setting value of its associated alkali pump regulating valve (12); d2) Instructions on the timing and duration of opening the washing liquid drain valve (8).

10. The method according to claim 7, characterized in that, The method further includes: Step S7: After one control cycle, the controller (14) collects the actual response data of the system again, compares the actual response value with the model prediction value in step S3, obtains the deviation, and uses an adaptive algorithm to fine-tune the internal parameters of the multi-parameter prediction model based on the deviation, so as to continuously improve the prediction accuracy of the model and the robustness of the control system; after the correction is completed, the process returns to step S1 and enters the next control cycle.

Images