Optimization control methods for SO2 concentration at the outlet of the desulfurization system of thermal power units and pH of the absorption tower

Abstract

The present application relates to a kind of outlet SO2 concentration of thermal power unit desulfurization system and pH optimization control method of absorption tower, the method comprises: parameter prediction model is constructed based on historical data, and pH set value calculation model and slurry circulating pump combination calculation model are constructed;Based on parameter prediction model, slurry pump frequency optimization and slurry circulating pump combination optimization are carried out;Based on the optimization result obtained, slurry pump and slurry circulating pump are controlled respectively, and the control target of outlet SO2 concentration standard and absorption tower pH stability is realized.Compared with prior art, the present application has improved the automation level of desulfurization system, reduced the operation of slurry pump and slurry circulating pump by manual intervention of operating personnel, and has advantages such as reducing the burden of operating personnel.

Description

Technical Field

[0001] This invention relates to the field of process control of coal-fired power units and intelligent power plants, and in particular to a method for optimizing the control of SO2 concentration at the outlet of the desulfurization system and pH of the absorption tower in a coal-fired power unit. Background Technology

[0002] Thermal power units commonly employ wet flue gas desulfurization (FGD) systems to regulate the SO2 concentration in the outlet flue gas, ensuring that the SO2 concentration in the emitted flue gas meets environmental standards. This involves introducing the combusted flue gas into an absorption tower, where limestone slurry is added to react with the SO2 in the flue gas, reducing the SO2 emissions. During the reaction, an increase in SO2 concentration lowers the pH of the absorption tower, while an increase in the amount of limestone slurry raises the pH. The change in the pH value of the absorption tower is strongly correlated with the sufficiency of the reaction. Due to the large scale and complex disturbances of large thermal power units, most units currently use the pH value of the absorption tower slurry as the controlled variable, adjusting the opening of the limestone slurry supply pump for single-objective control. By maintaining the pH value of the absorption tower within a reasonable range, the efficiency of SO2 absorption is ensured to be within normal limits. Because desulfurization systems have significant delays and are difficult to automate, most units still use manual control. In manual mode, to ensure that the outlet SO2 concentration does not exceed the standard, excessive amounts of limestone slurry are often added, leading to material waste and low economic efficiency of the desulfurization system. Therefore, a scientifically sound and reasonable automatic control method is urgently needed.

[0003] To achieve automatic control of desulfurization systems, most units currently employ basic single-loop or cascade feedback control schemes, automatically controlling the slurry pump based on the current outlet SO2 concentration and the current pH value of the absorber. Some schemes incorporate feedforward strategies, calculating the theoretical slurry supply based on the reaction mechanism and current parameters. However, in actual operation, the SO2 concentration at the desulfurization unit outlet fluctuates significantly. When these fluctuations are large, the theoretical slurry supply calculated by the feedforward often fails to match the actual supply. When the outlet SO2 concentration is about to exceed the limit, operators often manually activate additional slurry circulation pumps, resulting in a low level of automation.

[0004] A search of Chinese Patent Publication No. CN112426869A reveals a control system and method for the slurry concentration of a wet desulfurization absorption tower in a thermal power unit. Specifically, it discloses an electromagnetic flowmeter installed on a limestone slurry inlet pipe, a slurry circulation pump connected to the bottom of the absorption tower, a concentration meter connected to the outlet of the slurry circulation pump, a slurry discharge pump connected to the bottom of the absorption tower, and a slurry regulating valve connected to the outlet of the slurry discharge pump. The signal terminals of the electromagnetic flowmeter and the concentration meter are connected to the input terminal of the control system, and the output terminal of the control system is connected to the regulating valve. When the slurry concentration in the absorption tower increases, the control system can fully consider the change in the limestone slurry inlet flow rate, accurately controlling the gypsum slurry discharge rate while avoiding the wear and tear caused by frequent opening and closing of the gypsum slurry regulating valve, thus ensuring the control quality of the absorption tower slurry concentration. However, this existing patent does not solve the above-mentioned problems.

[0005] Therefore, the existing technology has the following problems: 1) From the perspective of the overall structure of the control system, due to the large lag and complex interference factors of the desulfurization absorption tower, the parameters change significantly in real time, making it difficult to match the theoretical slurry supply calculated at present with the actual situation. The accuracy of the feedforward is not high, the ability to overcome disturbances is poor, and the lag of the feedback value leads to the occurrence of abnormal parameters from time to time, which ultimately leads to a decrease in desulfurization efficiency and a deterioration in slurry quality.

[0006] 2) From a data-driven perspective, current technical solutions all adopt traditional control ideas and control based on current feedback data. They do not make full use of the massive historical operating data accumulated by the desulfurization system. If the historical data is deeply mined and a model is built, the robustness of the control algorithm under different operating conditions can be greatly improved.

[0007] Therefore, how to solve the problem of large delays and many interference factors in desulfurization systems, which leads to poor performance of traditional control schemes, has become a technical problem that needs to be solved. Summary of the Invention

[0008] The purpose of this invention is to overcome the defects of the prior art by providing a method for optimizing the control of SO2 concentration at the outlet of a thermal power unit desulfurization system and pH of the absorption tower.

[0009] The objective of this invention can be achieved through the following technical solutions: According to one aspect of the present invention, a method for optimizing and controlling the SO2 concentration at the outlet of a desulfurization system in a thermal power unit and the pH of the absorption tower is provided, the method comprising: A parameter prediction model was constructed based on historical data, and a calculation model for pH setpoint and a combined calculation model for slurry circulation pump were also constructed. Optimization of slurry pump frequency and slurry circulation pump combination based on parameter prediction model; Based on the optimization results, the slurry supply pump and slurry circulation pump were controlled to achieve the control objectives of meeting the SO2 concentration standard at the outlet and stabilizing the pH of the absorption tower.

[0010] As a preferred technical solution, the method specifically includes the following steps: Step S1: Obtain historical data and preprocess it to construct model input and output data; Step S2: Construct an outlet SO2 concentration prediction model and an absorption tower pH prediction model; Step S3: Construct a pH setpoint calculation model; Step S4: Construct a combined calculation model for the slurry circulation pump; Step S5: Input the real-time read boiler main control command and unit load command into the slurry circulation pump combination calculation model to obtain the recommended slurry circulation pump combination to be started; at the same time, predict whether the future outlet SO2 concentration will exceed the set threshold according to the SO2 concentration prediction model. If yes, start the recommended slurry circulation pump combination and the additional slurry circulation pump; if no, start the recommended slurry circulation pump combination to be started. Step S6: Gradually reduce the upper limit of the slurry pump frequency. H and lower limit O L The pH prediction value of the absorption tower at the corresponding slurry pump frequency is calculated based on the pH prediction model of the absorption tower until the absolute value of the difference between the slurry pH prediction value and the pH setting value is less than the setting value. The slurry pump frequency at this time is then output as the actual slurry pump frequency command.

[0011] As a preferred technical solution, step S1 specifically includes: Step S101: Collect historical data of input and output variables and perform data preprocessing. The input and output variables include FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual pH value of the absorber tower, FGD outlet flue gas SO2 concentration, slurry circulation pump current and slurry supply pump frequency. Step S102: The current of the slurry circulation pump is binarized and combined into a combined value of the slurry circulation pump. Step S103: Eliminate outliers in pH measurements by setting a rate of change threshold; Step S104: If the historical data collection interval is less than one minute, take the average of multiple data within one minute to obtain the historical data used for modeling.

[0012] As a preferred technical solution, the binarization process in step S102 specifically includes: This is the current value of the slurry circulation pump. This represents the maximum current of the slurry circulation pump. This represents the minimum current for the slurry circulation pump. This is the binarized start signal for the slurry circulation pump; Assuming the heads of the four slurry circulation pumps are A, B, C, and D from largest to smallest, then after binarization... Re-convert from 1 to 8, 4, 2, 1: This is the start signal for slurry circulation pump A. Code for slurry circulation pump A. This is the code for the pre-treated slurry circulation pump combination.

[0013] As a preferred technical solution, step S103 specifically includes: The time interval between two adjacent data collections. The rate limit for determining abnormal data changes is determined manually and is set to 0.1 based on experience. The total number of data collected. For the i-th data, This is the (i+1)th data point.

[0014] As a preferred technical solution, step S2 specifically includes: Step S201: The FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control data, limestone slurry flow rate, actual absorber pH value, and slurry pump frequency feedback data for adjacent N minutes are grouped into one group. The first n minutes are used as known values ​​input into the prediction model, and the actual absorber pH value for the next Nn minutes is used as the unknown predicted value, i.e., the expected output of the prediction model. After the data division is completed, the process moves to the next minute and continues to build input and output data. After all the data is divided, 70% of the data is sent to the neural network algorithm for time series prediction for modeling, 15% of the data is used to adjust the model parameters during the modeling process, and 15% of the data is used to verify the prediction effect after the model is built, thereby constructing the absorber pH prediction model. Step S202: The FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control, limestone slurry flow rate, FGD outlet flue gas SO2 concentration, and slurry circulation pump combination encoding data for adjacent N minutes are grouped into one group. The first n minutes are used as known values ​​input into the prediction model, and the FGD outlet flue gas SO2 concentration for the next Nn minutes is used as unknown predicted values, i.e., the expected output of the prediction model. After the data division is completed, the process moves to the next minute and continues to build input and output data. After all the data is divided, 70% of the data is sent to the neural network algorithm for time series prediction for modeling, 15% of the data is used to adjust the model parameters during the modeling process, and 15% of the data is used to verify the prediction effect after the model is built, thereby constructing the outlet flue gas SO2 concentration prediction model.

[0015] As a preferred technical solution, step S3 specifically includes: Historical data of boiler main control commands and unit load commands are used as input, and the actual pH value at the corresponding time is used as output. The pH setpoint under different load commands and boiler main control conditions is obtained by fitting the data through a linear regression algorithm.

[0016] As a preferred technical solution, step S4 specifically includes: Historical data of boiler main control commands and unit load commands are used as input, and the corresponding slurry circulation pump combination codes are used as output. K-means clustering is used to obtain the slurry circulation pump codes under different loads. The recommended slurry circulation pump combinations for different loads are obtained by restoring the slurry circulation pump codes.

[0017] As a preferred technical solution, in step S5, the inlet flue gas flow rate of FGD is input into the SO2 concentration prediction model to obtain the outlet SO2 concentration at a future set time.

[0018] As a preferred technical solution, step S6 specifically includes: Step S601: Specify the upper limit of the slurry pump frequency. O H and lower limit O L， Will O H and O L These frequencies are used as the future set slurry pump frequencies, and the predicted pH value of the absorption tower is calculated by combining the real-time read boiler main control commands. Step S62, if O L If the predicted pH value of the slurry is closer to the pH set value of the absorption tower, then ( O L + O H) / 2 as the new O H And calculate the new O H The corresponding predicted pH value of the slurry is compared with the latest O H and O L Otherwise ( O L + O H ) / 2 as the new O L and calculate ( O L + O H ) / 2 corresponds to the predicted pH value of the slurry, compared to the latest O L and O H， Repeat the above process to gradually reduce the size. O L and O H The range is adjusted until the absolute value of the difference between the predicted pH value and the set pH value of the slurry is less than the set value. At this point, the slurry pump frequency is output as the actual slurry pump frequency command. When entering the next control cycle, the above prediction and optimization process is repeated, continuously outputting slurry circulation pump start / stop commands and slurry pump frequency commands, and performing continuous optimization control in the actual desulfurization system.

[0019] Compared with the prior art, the present invention has the following advantages: 1) This invention utilizes the massive historical operating data accumulated by the desulfurization system and, based on learning from the historical data of the absorption tower, constructs a predictive model to predict the future pH value of the absorption tower slurry and the SO2 concentration at the outlet. On the basis of achieving accurate prediction, it forms a closed-loop control, realizes automatic slurry supply and automatic adjustment of the slurry circulation pump, decouples the automatic control of slurry pH value and outlet SO2 concentration, realizes the optimized control of the desulfurization system, and can effectively solve the problems of complex disturbances and large lag in the desulfurization system.

[0020] 2) This invention addresses the problem that most thermal power units still use the pH value of the absorber slurry as the controlled variable for single-objective control, resulting in a low level of automation in the desulfurization control system. It adopts neural network predictive control to adjust the frequency of the slurry supply pump, thereby achieving stable adjustment of the slurry pH value. At the same time, it adjusts the opening degree of the slurry circulation pump to control the outlet SO2 concentration, forming an optimized control method. This improves the automation level of the desulfurization system, reduces the need for manual intervention by operators in the operation of the slurry supply pump and slurry circulation pump, and alleviates the burden on operators. Attached Figure Description

[0021] Figure 1This is a flowchart illustrating the control method of the present invention. Figure 2 This is a schematic diagram of the actual current value and binary conversion criterion value of the slurry circulation pump of the present invention, wherein (a) is a schematic diagram of slurry circulation pump A, (b) is a schematic diagram of slurry circulation pump B, (c) is a schematic diagram of slurry circulation pump C, and (d) is a schematic diagram of slurry circulation pump D. Figure 3 This is a schematic diagram showing the original pH measurement value and the corrected pH measurement value of the present invention; Figure 4 This is a schematic diagram illustrating the test results of the pH prediction model for the absorption tower of the present invention; Figure 5 This is a schematic diagram illustrating the test results of the SO2 concentration prediction model for flue gas outlet of the present invention. Figure 6 This is a schematic diagram of the calculation results of the pH setpoint calculation model of the present invention; Figure 7 This is a schematic diagram of the calculation results of the slurry circulation pump combination calculation model of the present invention, where (a) is the clustering result and (b) is the calculated slurry circulation pump group code; Figure 8 This is a schematic diagram comparing the control effects of the present invention and the prior art, where (a) shows the control effect of the existing manual method and (b) shows the control effect of the present invention. Detailed Implementation

[0022] 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.

[0023] The pH value of the absorber slurry and the SO2 concentration at the outlet of the desulfurization system of a 660MW generator unit in a power plant exhibit significant lag and inertia, making automatic control difficult using traditional PID algorithms. This necessitates dedicated operators manually adjusting the slurry supply pump and slurry circulation pump to maintain pH stability and ensure the outlet SO2 concentration meets standards. Because manual operation makes it difficult to predict parameter changes in advance, adjustments are made based on current deviations, ultimately leading to large fluctuations in the slurry pH value and outlet SO2 concentration. To improve the unit's automation level and reduce the workload of operators, like Figure 1 As shown, the desulfurization optimization control method implemented using the technical solution of this invention is as follows: 1. Identify relevant input and output variables and perform data preprocessing. Collect historical data for the relevant variables in Table 1, and perform linear interpolation on outliers and missing values ​​determined by human experience.

[0024] Table 1 The current of the slurry circulation pump is binarized and then combined into a combined value of the slurry circulation pump based on the head code. The binarization process is as follows: This is the current value of the slurry circulation pump. This represents the maximum current of the slurry circulation pump. This represents the minimum current for the slurry circulation pump. This is the binarized start signal of the slurry circulation pump. The actual current values ​​of slurry circulation pumps A, B, C, and D and the binarization criterion values ​​are as follows: Figure 2 As shown: The four slurry circulation pumps of this unit have heads D, C, A, and B from largest to smallest. After binarization... Re-convert from 1 to 8, 4, 2, 1: This is the start signal for slurry circulation pump A. Code for slurry circulation pump A. This is the code for the pre-treated slurry circulation pump combination.

[0025] Because periodic flushing of pH measuring devices can cause abrupt changes in pH readings at fixed intervals, outliers can be eliminated by setting a rate of change threshold. The time interval between two adjacent data collections. The rate limit for determining abnormal data changes is determined manually and can be set to 0.1 based on experience. The total number of data collected. For the i-th data, This is the (i+1)th data point.

[0026] Original pH measurement and corrected pH measurement as follows Figure 3 As shown: Historical data was collected at 10-second intervals, and the average of six data points within one minute was used as the historical data for the final modeling.

[0027] 2. Establish a prediction model for outlet SO2 concentration and absorber pH. Historical data was reorganized, grouping adjacent 15-minute intervals of FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control data, limestone slurry flow rate, actual absorber pH value, and slurry pump frequency feedback data into a single group. The first 10 minutes were used as known values ​​input to the prediction model, while the actual absorber pH value of the last 5 minutes was used as the unknown predicted value, i.e., the expected output of the prediction model. After data division, the process moved to the next minute to continue building input and output data. After all data was divided, 70% of the data was fed into a neural network algorithm such as a Long Short-Term Memory (LSTM) network for time-series prediction modeling; 15% of the data was used to adjust model parameters during modeling; and 15% of the data was used to verify the prediction effect after model building. Finally, an absorber pH prediction model was constructed. The test results of the absorber pH prediction model are shown below. Figure 4 As shown: Historical data was reorganized, and data from consecutive 15-minute intervals were grouped into sets based on the following parameters: FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control parameters, limestone slurry flow rate, FGD outlet flue gas SO2 concentration, and slurry circulation pump combination encoding. The first 10 minutes were used as known values ​​input to the prediction model, while the last 5 minutes' FGD outlet flue gas SO2 concentration was used as the unknown predicted value, representing the expected output of the prediction model. After data partitioning, the process moved to the next minute to continue building input and output data. After all data was partitioned, 70% of the data was fed into a neural network algorithm (such as a Long Short-Term Memory network) for time-series prediction modeling, 15% was used to adjust model parameters during modeling, and 15% was used to verify the prediction effect after model building. Finally, an outlet flue gas SO2 concentration prediction model was constructed. The test results of the outlet flue gas SO2 concentration prediction model are shown below. Figure 5 As shown: 3. Establish a pH setpoint calculation model. Using historical data from boiler main control commands and unit load commands as input, and the actual pH value at the corresponding time as output, the pH setpoint under different load commands and boiler main control conditions is obtained through linear regression and other regression algorithms. Figure 6 As shown.

[0028] 4. Establish a calculation model for the slurry circulation pump combination. Using historical data of boiler main control commands and unit load commands as input, and the corresponding slurry circulation pump combination codes as output, clustering algorithms such as K-means clustering are used to obtain the slurry circulation pump codes under different loads. The recommended slurry circulation pump combinations for different loads are then derived from these codes. The final slurry circulation pump combination calculation model is shown below. Figure 7 As shown, (b) is the calculated code for the slurry circulation pump group, and (a) is the clustering result: 5. Read and store real-time data, and calculate the slurry circulation pump start-up command. Input the real-time boiler main control command and unit load command into the slurry circulation pump combination model to obtain the recommended slurry circulation pump combination to start. At the same time, input the FGD inlet flue gas flow rate and other data into the outlet SO2 concentration prediction model. If the predicted outlet SO2 concentration in the next 10 minutes does not exceed 20 mg / m3, start the corresponding slurry circulation pump group according to the recommended slurry circulation pump combination. If the predicted outlet SO2 concentration in the next 10 minutes tends to exceed 20 mg / m3, then add a slurry circulation pump with a smaller head.

[0029] 6. Read and store real-time data, and calculate the slurry pump frequency command. Specify the upper limit of the slurry pump frequency. O H and lower limit O L First of all O H and O L These values ​​are used as the future set slurry pump frequencies, combined with real-time read boiler main control commands, to calculate the predicted pH value of the absorption tower. If... O L If the predicted pH value of the slurry is closer to the pH set value of the absorption tower, then ( O L + O H ) / 2 as the new O H And calculate the new O H The corresponding predicted pH value of the slurry is compared with the latest O H and O L Otherwise ( O L + O H ) / 2 as the new O L and calculate ( O L + O H ) / 2 corresponds to the predicted pH value of the slurry, compared to the latest O L and O H Repeat the above steps to gradually reduce the size. O L and O HThe predicted pH value of the slurry is adjusted until it approaches the set pH value, at which point the slurry pump frequency is output as the actual slurry pump frequency command. This prediction and optimization process is repeated before entering the next control cycle. The control effect is compared to... Figure 8 As shown, compared to the manual method, this solution achieves automatic control, with a smooth output frequency of the slurry pump, meeting the requirements for pH stability and SO2 concentration at the outlet, thus improving the operational economy of the desulfurization system.

[0030] 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 method for optimizing and controlling the SO2 concentration at the outlet of a desulfurization system in a thermal power unit and the pH of the absorption tower, characterized in that, The method includes: A parameter prediction model was constructed based on historical data, and a calculation model for pH setpoint and a combined calculation model for slurry circulation pump were also constructed. Optimization of slurry pump frequency and slurry circulation pump combination based on parameter prediction model; Based on the optimization results, the slurry supply pump and slurry circulation pump were controlled respectively to achieve the control objectives of meeting the outlet SO2 concentration standard and stabilizing the pH of the absorption tower. The method specifically includes the following steps: Step S1: Obtain historical data and preprocess it to construct model input and output data; Step S2: Construct an outlet SO2 concentration prediction model and an absorption tower pH prediction model; Step S3: Construct a pH setpoint calculation model; Step S4: Construct a combined calculation model for the slurry circulation pump; Step S5: Input the real-time read boiler main control command and unit load command into the slurry circulation pump combination calculation model to obtain the recommended slurry circulation pump combination to be started; at the same time, predict whether the future outlet SO2 concentration will exceed the set threshold according to the SO2 concentration prediction model. If yes, start the recommended slurry circulation pump combination and the additional slurry circulation pump; if no, start the recommended slurry circulation pump combination to be started. Step S6: Gradually reduce the upper limit of the slurry pump frequency. H and lower limit O L The pH prediction value of the absorption tower at the corresponding slurry pump frequency is calculated based on the pH prediction model of the absorption tower until the absolute value of the difference between the slurry pH prediction value and the pH setting value is less than the setting value. The slurry pump frequency at this time is then output as the actual slurry pump frequency command.

2. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, Step S1 specifically involves: Step S101: Collect historical data of input and output variables and perform data preprocessing. The input and output variables include FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual pH value of the absorber tower, FGD outlet flue gas SO2 concentration, slurry circulation pump current and slurry supply pump frequency. Step S102: The current of the slurry circulation pump is binarized and combined into a combined value of the slurry circulation pump. Step S103: Eliminate outliers in pH measurements by setting a rate of change threshold; Step S104: If the historical data collection interval is less than one minute, take the average of multiple data within one minute to obtain the historical data used for modeling.

3. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 2, is characterized in that, The binarization process in step S102 specifically involves: This is the current value of the slurry circulation pump. This represents the maximum current of the slurry circulation pump. This represents the minimum current for the slurry circulation pump. This is the binarized start signal for the slurry circulation pump; Assuming the heads of the four slurry circulation pumps are A, B, C, and D from largest to smallest, then after binarization... Re-convert from 1 to 8, 4, 2, 1: This is the start signal for slurry circulation pump A. Code for slurry circulation pump A. This is the code for the pre-treated slurry circulation pump combination.

4. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 2, characterized in that, Step S103 specifically involves: The time interval between two adjacent data collections. The rate limit for determining abnormal data changes is determined manually and is set to 0.1 based on experience. The total number of data collected. For the i-th data, This is the (i+1)th data point.

5. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, Step S2 specifically involves: Step S201: The FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control data, limestone slurry flow rate, actual absorber pH value, and slurry pump frequency feedback data for adjacent N minutes are grouped into one group. The first n minutes are used as known values ​​input into the prediction model, and the actual absorber pH value for the next Nn minutes is used as the unknown predicted value, i.e., the expected output of the prediction model. After the data division is completed, the process moves to the next minute and continues to build input and output data. After all the data is divided, 70% of the data is sent to the neural network algorithm for time series prediction for modeling, 15% of the data is used to adjust the model parameters during the modeling process, and 15% of the data is used to verify the prediction effect after the model is built, thereby constructing the absorber pH prediction model. Step S202: The FGD inlet flue gas flow rate, FGD inlet flue gas temperature, FGD inlet flue gas SO2 concentration, actual unit power, boiler main control, limestone slurry flow rate, FGD outlet flue gas SO2 concentration, and slurry circulation pump combination encoding data for adjacent N minutes are grouped into one group. The first n minutes are used as known values ​​input into the prediction model, and the FGD outlet flue gas SO2 concentration for the next Nn minutes is used as unknown predicted values, i.e., the expected output of the prediction model. After the data division is completed, the process moves to the next minute and continues to build input and output data. After all the data is divided, 70% of the data is sent to the neural network algorithm for time series prediction for modeling, 15% of the data is used to adjust the model parameters during the modeling process, and 15% of the data is used to verify the prediction effect after the model is built, thereby constructing the outlet flue gas SO2 concentration prediction model.

6. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, Step S3 specifically involves: Historical data of boiler main control commands and unit load commands are used as input, and the actual pH value at the corresponding time is used as output. The pH setpoint under different load commands and boiler main control conditions is obtained by fitting the data through a linear regression algorithm.

7. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, Step S4 specifically involves: Historical data of boiler main control commands and unit load commands are used as input, and the corresponding slurry circulation pump combination codes are used as output. K-means clustering is used to obtain the slurry circulation pump codes under different loads. The recommended slurry circulation pump combinations for different loads are obtained by restoring the slurry circulation pump codes.

8. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, In step S5, the inlet flue gas flow rate of FGD is input into the SO2 concentration prediction model to obtain the outlet SO2 concentration at a future set time.

9. The method for optimizing and controlling the SO2 concentration at the outlet of the desulfurization system of a thermal power unit and the pH of the absorption tower according to claim 1, characterized in that, Step S6 specifically involves: Step S601: Specify the upper limit of the slurry pump frequency. O H and lower limit O L， Will O H and O L These frequencies are used as the future set slurry pump frequencies, and the predicted pH value of the absorption tower is calculated by combining the real-time read boiler main control commands. Step S62, if O L If the predicted pH value of the slurry is closer to the pH set value of the absorption tower, then ( O L + O H ) / 2 as the new O H And calculate the new O H The corresponding predicted pH value of the slurry is compared with the latest O H and O L Otherwise ( O L + O H ) / 2 as the new O L and calculate ( O L + O H ) / 2 corresponds to the predicted pH value of the slurry, compared to the latest O L and O H， Repeat the above process to gradually reduce the size. O L and O H The range is set until the absolute value of the difference between the predicted pH value and the set pH value of the slurry is less than the set value, at which point the slurry pump frequency is output as the actual slurry pump frequency command.

Images