Intelligent optimization control method for desulfurization slurry circulating pump with dynamic load response

Abstract

The present application relates to the field of coal-fired power plant desulfurization system control technology, and discloses a dynamic load response desulfurization slurry circulating pump intelligent optimization control method, comprising the following steps: S1, data acquisition; S2, data preprocessing; S3, model construction and model training; S4, model iteration optimization; S5, circulating pump optimization control. The dynamic load response desulfurization slurry circulating pump intelligent optimization control method comprehensively collects the big data of desulfurization system equipment operation transmission, removes the noise and abnormal values in the data by using the data cleaning method, thereby ensuring the accuracy of model construction, the reliability of model training and prediction, and dividing the data reasonably, constructing a sample pool, thereby improving the generalization ability of the model, using support vector machine to build a prediction model, making the model have high fitting precision, and enhancing the optimization operation mode of the slurry circulating pump to adapt to the complexity of boiler load and working condition change.

Description

Technical Field

[0001] This invention relates to the field of desulfurization system control technology in coal-fired power plants, specifically to an intelligent optimization and control method for desulfurization slurry circulation pumps with dynamic load response. Background Technology

[0002] The desulfurization slurry circulation pump is one of the most important pieces of equipment in the wet flue gas desulfurization (FGD) subsystem, and its energy consumption accounts for a large proportion of the FGD system's energy consumption. The start-up and shutdown of the circulation pump significantly affect the FGD system's desulfurization efficiency and energy consumption.

[0003] To minimize the energy consumption of wet desulfurization units while ensuring stable and compliant outlet concentrations, it is necessary to pre-construct a prediction model for the outlet concentration of wet desulfurization units, study the variation law of sulfur dioxide concentration at the outlet of the desulfurization tower under different slurry circulation pump configurations, and obtain the optimal circulation pump operation configuration under different operating conditions. Therefore, it is necessary to invent a dynamic load-response intelligent optimization and control method for desulfurization slurry circulation pumps to achieve the above-mentioned solution. Summary of the Invention

[0004] This invention provides the following technical solution: a method for intelligent optimization and control of desulfurization slurry circulation pumps with dynamic load response, comprising the following steps:

[0005] S1, Data Acquisition

[0006] From the equipment operation of the desulfurization system, based on the DCS data source, big data is collected on boiler load, sulfur dioxide concentration at the inlet of the absorption tower, flue gas flow rate, flue gas temperature, raw flue gas pressure, dust concentration at the tower inlet, absorption tower liquid level, oxidation air volume, gypsum slurry density, limestone slurry flow rate, slurry pH value, and the number of circulating pumps in operation.

[0007] S2, Data Preprocessing

[0008] The data collected in step S1 is cleaned to remove outliers and noise. Then, the data is divided into training and testing data, and a sample pool is constructed.

[0009] S3. Build and train the model.

[0010] Based on support vector machine, a model for predicting sulfur dioxide concentration at the desulfurization tower outlet is constructed using the data in the sample pool built in step S2. The model is trained and its parameters are adjusted using field test data. The model performance is then evaluated using model evaluation indicators, and the model is continuously optimized.

[0011] S4, Model Iterative Optimization

[0012] By selecting historical operating data of the desulfurization system in stages to simulate real-time operating data of the power plant, the constructed sulfur dioxide concentration prediction model at the outlet of the desulfurization tower is iteratively applied to obtain the operation optimization algorithm of the slurry circulation pump.

[0013] S5, Circulation Pump Optimization and Control

[0014] The real-time operating data of the desulfurization system is input into a trained nonlinear dynamic model based on support vector machine. The configuration of the slurry circulation pump is optimized and adjusted in real time, the optimized operating mode of the slurry circulation pump is output, and the command is transmitted to the DCS system to control the start and stop of each slurry circulation pump and complete the corresponding control flow program for the start and stop of the slurry circulation pump.

[0015] Preferably, in step S2, the data cleaning uses the following formula to detect and process outliers in the data:

[0016] ;

[0017] ;

[0018] ;

[0019] Where Q1 is the lower quartile of the data, Q3 is the upper quartile of the data, IQR is the interquartile range, k is a set coefficient with a value of 1.5, Lower Bound is the lower bound, Upper Bound is the upper bound, and data smaller than Lower Bound and larger than Upper Bound are considered outliers and are treated as outliers.

[0020] Preferably, in step S3, the objective function of the support vector machine model is:

[0021] ;

[0022] The constraints are:

[0023] ;

[0024] Where ω is the weight vector and b is the bias. C is a slack variable, and C is a penalty parameter. For input data, For the corresponding tags, This is a mapping function that maps input data to a high-dimensional space.

[0025] Preferably, in step S3, during the model evaluation process, the root mean square error, mean absolute error, and absolute coefficient are used as model evaluation indicators, and the calculation formula is as follows:

[0026] ;

[0027] ;

[0028] ;

[0029] Where n is the number of samples. This is the actual value. For predicted values, This is the average of the actual values.

[0030] Preferably, in step S4, the historical operating data of the desulfurization system is selected in stages using the sliding window technique. The data window length of each stage is L, and the sliding step size is S. The model is iteratively trained and optimized by continuously updating the data in the data window.

[0031] Preferably, in step S5, based on the sulfur dioxide concentration prediction model at the desulfurization tower outlet and real-time operating data, a greedy algorithm is used to optimize and adjust the configuration of the slurry circulation pump in real time.

[0032] Preferably, in steps S1-S5, data is collected using sensors, and a real-time monitoring step of the desulfurization system's operating status is also included. When abnormal fluctuations or exceeding set thresholds are detected in the operating parameters of the desulfurization system, the model retraining and parameter adjustment process is triggered to ensure the model's accuracy and adaptability.

[0033] Preferably, in step S5, in the step of transmitting instructions to the DCS system to control the start and stop of each slurry circulation pump, the industrial Ethernet communication protocol is used to achieve reliable data transmission.

[0034] Preferably, the data collected in step S1 also includes the sulfur dioxide concentration at the absorber outlet, the slurry circulation flow rate, and the system power supply voltage, in order to expand the model input dimensions.

[0035] Preferably, during model iterative optimization in step S4, the penalty parameter C and kernel function parameters of the support vector machine are adaptively adjusted in conjunction with the particle swarm optimization algorithm to improve the dynamic response accuracy of the model.

[0036] Compared with the prior art, the present invention provides an intelligent optimization and control method for desulfurization slurry circulation pumps with dynamic load response, which has the following beneficial effects:

[0037] 1. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response comprehensively collects big data transmitted from the operation of desulfurization system equipment, and uses data cleaning methods to remove noise and outliers from the data, thereby ensuring the accuracy of model construction and the reliability of model training and prediction. The data is reasonably divided into circulation and testing to build a sample pool, thereby improving the generalization ability of the model.

[0038] 2. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response adopts support vector machine to build a prediction model, which makes the model have high fitting accuracy and enhances the optimization operation mode of slurry circulation pump to adapt to the complexity of boiler load and operating condition changes.

[0039] 3. The intelligent optimization and control method for desulfurization slurry circulation pumps with dynamic load response uses a greedy algorithm to optimize and adjust the configuration of slurry circulation pumps in real time, aiming to minimize the operating cost of the desulfurization system. It transmits instructions to the DCS system to control the start and stop of each slurry circulation pump and completes the corresponding control process program for the start and stop of the slurry circulation pumps. Under the premise of ensuring that the sulfur dioxide concentration at the outlet of the desulfurization tower does not exceed the standard, the operating cost of the desulfurization system is optimized. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the process structure of the present invention. Detailed Implementation

[0041] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] Please see Figure 1 This invention provides a technical solution: a method for intelligent optimization and control of desulfurization slurry circulation pumps with dynamic load response, comprising the following steps:

[0043] S1, Data Acquisition

[0044] From the equipment operation of the desulfurization system, based on the DCS data source, big data is collected on boiler load, sulfur dioxide concentration at the inlet of the absorption tower, flue gas flow rate, flue gas temperature, raw flue gas pressure, dust concentration at the tower inlet, absorption tower liquid level, oxidation air volume, gypsum slurry density, limestone slurry flow rate, slurry pH value, and the number of circulating pumps in operation.

[0045] S2, Data Preprocessing

[0046] The data collected in step S1 is cleaned to remove outliers and noise. Then, the data is divided into training and testing data, and a sample pool is constructed.

[0047] Data cleaning uses the following formula to detect and handle outliers:

[0048] ;

[0049] ;

[0050] ;

[0051] Where Q1 is the lower quartile of the data, Q3 is the upper quartile of the data, IQR is the interquartile range, k is a set coefficient with a value of 1.5, Lower Bound is the lower bound, Upper Bound is the upper bound, and data smaller than Lower Bound and larger than Upper Bound are considered outliers and are treated as outliers.

[0052] S3. Build and train the model.

[0053] Based on support vector machine, a model for predicting sulfur dioxide concentration at the desulfurization tower outlet is constructed using the data in the sample pool built in step S2. The model is trained and its parameters are adjusted using field test data. The model performance is then evaluated using model evaluation indicators, and the model is continuously optimized.

[0054] The objective function of the support vector machine model is:

[0055] ;

[0056] The constraints are:

[0057] ;

[0058] Where ω is the weight vector and b is the bias. C is a slack variable, and C is a penalty parameter. For input data, For the corresponding tags, This is a mapping function that maps input data to a high-dimensional space.

[0059] In the model evaluation process, root mean square error, mean absolute error, and absolute coefficient are used as model evaluation indicators, and the calculation formulas are as follows:

[0060] ;

[0061] ;

[0062] ;

[0063] Where n is the number of samples. This is the actual value. For predicted values, This is the average of the actual values.

[0064] S4, Model Iterative Optimization

[0065] By selecting historical operating data of the desulfurization system in stages to simulate real-time operating data of the power plant, the constructed sulfur dioxide concentration prediction model at the outlet of the desulfurization tower is iteratively applied to obtain the operation optimization algorithm of the slurry circulation pump.

[0066] The sliding window technique is used to select historical operating data of the desulfurization system in stages. The data window length of each stage is L, and the sliding step size is S. The model is iteratively trained and optimized by continuously updating the data in the data window.

[0067] S5, Circulation Pump Optimization and Control

[0068] The real-time operating data of the desulfurization system is input into a trained nonlinear dynamic model based on support vector machine. The configuration of the slurry circulation pump is optimized and adjusted in real time, the optimized operating mode of the slurry circulation pump is output, and the command is transmitted to the DCS system to control the start and stop of each slurry circulation pump and complete the corresponding control flow program for the start and stop of the slurry circulation pump.

[0069] Based on the sulfur dioxide concentration prediction model at the desulfurization tower outlet and real-time operating data, a greedy algorithm is used to optimize and adjust the configuration of the slurry circulation pump in real time.

[0070] The specific steps in constructing the greedy algorithm are as follows:

[0071] First, there are m slurry circulation pumps, each with a power of Pi (i=1, 2, ..., m), an electricity price of c, and an operating time of t. The total electricity cost is:

[0072] ;

[0073] in, Let i be the operating state variable of the i-th pump, when A value of 1 indicates that the pump is running. A value of 0 indicates that the pump has stopped. ∈{0,1},i=1,2,…,m.

[0074] Let the predicted sulfur dioxide concentration at the desulfurization tower outlet be C, and the maximum permissible sulfur dioxide emission concentration be C. max Then C≤C max .

[0075] The greedy algorithm calculates the cost changes and impact on sulfur dioxide concentration under different pump start / stop combinations based on the current state at each selection. Assuming we are currently considering starting or stopping the j-th pump, it calculates the predicted sulfur dioxide concentration Cnew and cost Cost after starting or stopping this pump. new ;

[0076] If the j-th pump is turned on:

[0077] ;

[0078] And calculated using a sulfur dioxide prediction model ;

[0079] If the j-th pump is shut down:

[0080] ;

[0081] And calculated using a sulfur dioxide prediction model .

[0082] Based on the above calculation formula, under the condition of satisfying the sulfur dioxide concentration constraint, the operation that minimizes the cost (turning on or off the j-th pump) is selected. The above process is repeated until the optimal slurry circulation pump configuration is obtained.

[0083] In the process of transmitting commands to the DCS system to control the start and stop of each slurry circulation pump, the industrial Ethernet communication protocol is used to achieve reliable data transmission.

[0084] Furthermore, in steps S1-S5, data is collected using sensors, and real-time monitoring of the desulfurization system's operating status is also included. When abnormal fluctuations or exceeding set thresholds are detected in the desulfurization system's operating parameters, the model's retraining and parameter adjustment process is triggered to ensure the model's accuracy and adaptability.

[0085] Furthermore, the data collected in step S1 also includes the sulfur dioxide concentration at the absorber outlet, the slurry circulation flow rate, and the system power supply voltage, in order to expand the model input dimensions.

[0086] Furthermore, during the model iterative optimization in step S4, the penalty parameter C and kernel function parameters of the support vector machine are adaptively adjusted using the particle swarm optimization algorithm to improve the dynamic response accuracy of the model.

[0087] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent optimization and control of desulfurization slurry circulation pump with dynamic load response, characterized in that, Includes the following steps: S1, Data Acquisition Based on DCS data sources, big data is collected from the equipment operation of the desulfurization system, including boiler load, sulfur dioxide concentration at the inlet of the absorption tower, flue gas flow rate, flue gas temperature, raw flue gas pressure, dust concentration at the tower inlet, absorption tower liquid level, oxidation air volume, gypsum slurry density, limestone slurry flow rate, slurry pH value, and the number of circulating pumps in operation. S2, Data Preprocessing The data collected in step S1 is cleaned to remove outliers and noise. Then, the data is divided into training and testing data, and a sample pool is constructed. In step S2, the data cleaning process uses the following formula to detect and process outliers in the data: ， ， ， Where Q1 is the lower quartile of the data, Q3 is the upper quartile of the data, IQR is the interquartile range, k is a set coefficient with a value of 1.5, Lower Bound is the lower bound, Upper Bound is the upper bound, and data smaller than Lower Bound and larger than Upper Bound are considered outliers and are treated as outliers. S3. Build and train the model. Based on support vector machine, a sulfur dioxide concentration prediction model for desulfurization tower outlet is constructed using data from the sample pool built in step S2. The model is trained and parameters are adjusted using field test data. The model performance is then evaluated using model evaluation indicators, and the model is continuously optimized. In step S3, the objective function of the support vector machine model is: ； The constraints are: ； Where ω is the weight vector and b is the bias. C is a slack variable, and C is a penalty parameter. For input data, For the corresponding tags, A mapping function that maps input data to a high-dimensional space; S4, Model Iterative Optimization Historical operating data of the desulfurization system was selected in stages to simulate real-time operating data of the power plant. The constructed sulfur dioxide concentration prediction model at the outlet of the desulfurization tower was iteratively applied to obtain the operation optimization algorithm of the slurry circulation pump. In step S3, during the model evaluation process, the root mean square error, mean absolute error, and absolute coefficient are used as model evaluation indicators, and the calculation formula is as follows: ， ， ； Where n is the number of samples. This is the actual value. For predicted values, The average of the actual values; S5, Circulation Pump Optimization and Control The real-time operating data of the desulfurization system is input into a trained nonlinear dynamic model based on support vector machine. The configuration of the slurry circulation pump is optimized and adjusted in real time, the optimized operating mode of the slurry circulation pump is output, and the command is transmitted to the DCS system to control the start and stop of each slurry circulation pump and complete the corresponding control flow program for the start and stop of the slurry circulation pump.

2. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, In step S4, the historical operating data of the desulfurization system is selected in stages using the sliding window technique. The data window length of each stage is L, and the sliding step size is S. The model is iteratively trained and optimized by continuously updating the data in the data window.

3. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, In step S5, based on the sulfur dioxide concentration prediction model at the desulfurization tower outlet and real-time operating data, a greedy algorithm is used to optimize and adjust the configuration of the slurry circulation pump in real time.

4. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, In steps S1-S5, sensors are used to collect data, and a real-time monitoring step of the desulfurization system's operating status is also included. When abnormal fluctuations or exceeding set thresholds are detected in the operating parameters of the desulfurization system, the model retraining and parameter adjustment process is triggered to ensure the model's accuracy and adaptability.

5. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, In step S5, during the step of transmitting commands to the DCS system to control the start and stop of each slurry circulation pump, the industrial Ethernet communication protocol is used to achieve reliable data transmission.

6. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, The data collected in step S1 also includes the sulfur dioxide concentration at the absorber outlet, the slurry circulation flow rate, and the system power supply voltage, in order to expand the input dimensions of the model.

7. The intelligent optimization and control method for desulfurization slurry circulation pump with dynamic load response according to claim 1, characterized in that, In step S4, during model iterative optimization, the penalty parameter C and kernel function parameters of the support vector machine are adaptively adjusted using the particle swarm optimization algorithm to improve the dynamic response accuracy of the model.

Images