An automatic control system for wet flue gas desulfurization based on artificial intelligence

Abstract

The present invention relates to the field of automatic control technology, and more specifically, to an automatic control system for wet flue gas desulfurization based on artificial intelligence, the system comprising: an initialization module for obtaining the control parameter range during desulfurization, and constructing an initial population of a genetic algorithm within the control parameter range, wherein the fitness of each chromosome in the initial population reflects the desulfurization effect under the corresponding control parameter; a desulfurization control module for iteratively updating the genome of each chromosome in the initial population using a genetic algorithm, and at each update, using the weighted mutation rate of each chromosome as the mutation rate at the next update, until a preset update termination condition is met to obtain a target control parameter, so as to achieve desulfurization control. The present invention can improve the accuracy of wet flue gas desulfurization process control.

Description

Technical Field

[0001] The present invention relates to the field of automation control technology, and more specifically, to an artificial intelligence-based wet flue gas desulfurization automation control system. Background Art

[0002] Wet desulfurization technology is used in industrial facilities such as thermal power plants to remove sulfur dioxide ( ). Although the traditional system is effective, it has limitations in desulfurization efficiency, cost and response speed. After the introduction of artificial intelligence technology, through real-time data analysis and intelligent optimization, the system efficiency can be significantly improved, energy consumption can be reduced, and the accuracy and response speed of process control can be improved, thereby improving environmental protection effects and reducing operating costs.

[0003] In the related art, a vulcanization workshop scheduling method based on an improved genetic algorithm is disclosed in a patent application document with publication number CN118348934A, which includes: S1, establishing a vulcanization workshop scheduling model; S2, encoding and decoding; S3, initializing and determining a fitness function; S4, selecting an operation; S5, introducing a mutual learning strategy to replace the crossover operation in the genetic algorithm; S6, mutation operation; S7, calculating the fitness value of the individuals obtained after the mutation operation; S8, the number of iterations t=t+1; S9, encapsulating the improved vulcanization workshop scheduling algorithm into a vulcanization workshop optimization algorithm library.

[0004] However, the above scheme uses the ranking of individual fitness to determine the mutation rate, which usually gives individuals with low fitness more mutation opportunities, which may lead to excessive mutations. Moreover, as evolution proceeds, the mutation rate of individuals with higher fitness continues to decrease, causing the population to converge prematurely. In the later stages of evolution, the entire population tends to be similar, causing individuals to lose diversity and be unable to jump out of the local optimal solution, resulting in a low accuracy of the optimal parameters found by the algorithm, thereby affecting the scheduling of the vulcanization workshop. Summary of the invention

[0005] In order to solve the problem that the optimal control parameters cannot be accurately found, resulting in low control accuracy of the desulfurization process, the present invention provides an artificial intelligence-based wet desulfurization automation control system. The system includes:

[0006] The initialization module is used to obtain the control parameter range during desulfurization and construct the initial population of the genetic algorithm within the control parameter range. The fitness of each chromosome in the initial population reflects the desulfurization effect under the corresponding control parameters.

[0007] The desulfurization control module is used to iteratively update the genome of each chromosome in the initial population using a genetic algorithm. During each update, the weighted mutation rate of each chromosome is used as the mutation rate for the next update until the preset update termination condition is met to obtain the target control parameter to achieve desulfurization control.

[0008] The weighted mutation rate of any chromosome at any update is obtained by calculating the potential factor of the chromosome: , For the Updated at The growth rate of chromosome fitness, For the Updated at the same time as The number of chromosomes whose chromosome distance is less than the preset distance threshold. For the Updated at The fitness of a chromosome, For the Updated at Chromosomes and The distance between chromosomes, is the number of chromosomes, is the normalization function;

[0009] The initial mutation rate is preset, and the initial mutation rate is weighted by weight to obtain the weighted mutation rate. The weight is negatively correlated with the potential factor and the chromosome search efficiency.

[0010] The present invention can adaptively adjust the mutation rate of the corresponding chromosome in combination with the evolution of each chromosome at each update. On the one hand, it can avoid the problem that when the mutation rate is low, the chromosomes with low fitness converge slowly, and the chromosomes with high fitness are difficult to jump out of the local optimal solution; on the other hand, it can avoid the problem that when the mutation rate is high, the chromosomes with low fitness easily miss the optimal solution, and the chromosomes with high fitness are difficult to converge, so that the control parameters with the best desulfurization effect can be found within the control parameter range, and the accuracy of desulfurization process control can be improved.

[0011] Preferably, the weights satisfy the following relationship:

[0012] ;

[0013] In the formula, For the Updated at The weight of the initial mutation rate of each chromosome; For the Updated at Potential factors of chromosomes; For the Updated at The search efficiency of each chromosome.

[0014] The weights set in the present invention can integrate various parameters during chromosome evolution, thereby ensuring the accuracy of the determined weights.

[0015] Preferably, the chromosome search efficiency satisfies the following relationship:

[0016] ;

[0017] In the formula, For the Updated at The search efficiency of each chromosome; For the Updated at The fitness of each chromosome; To preset hyper parameters; For the Updated at The average of the reciprocals of the differences between the genes corresponding to each chromosome and the local optimal chromosome; is the absolute value symbol; is the normalization function.

[0018] The present invention can determine the distance between a chromosome and a local optimal chromosome, and the chromosomes around the local optimal chromosome with lower fitness are basically in a convergence state and have poor search capabilities. This feature is used to calculate the search efficiency and ensure the accuracy of the calculation results.

[0019] Preferably, the method for obtaining the local optimal chromosome includes:

[0020] In any update, the chromosomes whose distance to any chromosome is less than the distance threshold are regarded as similar chromosomes;

[0021] A quantity threshold is preset. If the number of similar chromosomes is greater than the quantity threshold, the chromosome with the highest fitness among the similar chromosomes is taken as the local optimal chromosome.

[0022] Preferably, the method for obtaining the growth degree of chromosome fitness includes:

[0023] Calculate the difference between the fitness of any chromosome at any historical update and the fitness at the last update, multiply all differences greater than zero, and obtain the degree of growth of the fitness of any chromosome.

[0024] Preferably, constructing an initial population of a genetic algorithm within a control parameter range includes:

[0025] A plurality of control parameters are randomly selected within the control parameter range, and each control parameter is used as a genome to construct a plurality of chromosomes, and a set consisting of all chromosomes is used as an initial population.

[0026] Preferably, the method for obtaining fitness includes:

[0027] Calculate the desulfurization efficiency and desulfurization cost when desulfurization is performed under any control parameter, perform weighted summation of the desulfurization efficiency and the desulfurization cost, and obtain the fitness of the chromosome corresponding to any control parameter;

[0028] Among them, the weight of desulfurization efficiency is 1, and the weight of desulfurization cost is -1.

[0029] The present invention can accurately measure the desulfurization effects under different control parameters.

[0030] Preferably, one control parameter includes limestone slurry flow rate, absorber pH, and oxidation zone air flow rate.

[0031] The present invention selects fewer parameters, all of which are key factors affecting the desulfurization effect, which reduces the amount of calculation and can avoid the interference of redundant parameters on the results, thereby ensuring higher control accuracy and a faster optimization process.

[0032] The present invention has the following technical effects:

[0033] The present invention uses a genetic algorithm to find the optimal solution within the range of control parameters. Each time the chromosome genome is updated, the potential factor and search efficiency of the chromosome are calculated based on the evolutionary characteristics of the chromosome during the current update and historical updates, and the mutation rate of the chromosome is adaptively adjusted based on the potential factor and the search efficiency. This can avoid the problem of chromosome convergence difficulties and the chromosome falling into a local optimal solution, ensure that the optimal solution found is a global optimal solution, and thus improve the accuracy of desulfurization process control. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0035] Figure 1 It is a structural schematic diagram of an artificial intelligence-based wet flue gas desulfurization automation control system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0037] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0038] Reference Figure 1 , an artificial intelligence-based wet flue gas desulfurization automation control system 100, including modules 110-120, specifically as follows:

[0039] The initialization module 110 is used to obtain the control parameter range during desulfurization and construct an initial population of the genetic algorithm within the control parameter range. The fitness of each chromosome in the initial population reflects the desulfurization effect under the corresponding control parameters.

[0040] It should be noted that the Genetic Algorithm (GA) is an optimization algorithm that can simulate the principles of natural selection and genetics, and gradually approach the optimal solution to the problem by performing operations such as selection, crossover and mutation on "chromosomes". The initial population of the algorithm is a set of several chromosomes, which are used as the basis for the selection, crossover and mutation operations of chromosomes; the genome of the chromosome represents a possible solution, and the fitness of the chromosome can evaluate the quality of the solution represented by the genome of the chromosome. In the present invention, the genome of the chromosome represents a control parameter, and the fitness of the chromosome represents the desulfurization effect under the corresponding control parameter.

[0041] Among them, one control parameter is composed of limestone slurry flow rate, absorption tower pH value and oxidation zone air flow rate. The control parameter range is the value range of limestone slurry flow rate, absorption tower pH value and oxidation zone air flow rate set according to process standards during wet flue gas desulfurization. The pH value of the absorption tower in the present invention is 5-7.

[0042] It should be noted that limestone slurry can provide sufficient alkaline substances to neutralize the ; The air flow rate in the oxidation zone can provide oxygen to promote the conversion of calcium sulfite into calcium sulfate; The pH value of the absorption tower can ensure the effective progress of the chemical reaction. Therefore, the present invention can accurately evaluate the desulfurization effect based on the limestone slurry flow rate, the pH value of the absorption tower and the air flow rate in the oxidation zone, so as to control the wet desulfurization process based on the optimization results of the genetic algorithm.

[0043] In an exemplary embodiment of the present invention, the construction of the initial population can be achieved by the following steps:

[0044] A plurality of control parameters are randomly selected within the control parameter range, and each control parameter is used as a genome to construct a plurality of chromosomes, and a set consisting of all chromosomes is used as an initial population.

[0045] Optionally, 200 control parameters can be randomly selected from the control parameter range, each control parameter consisting of limestone slurry flow, absorption tower pH value and air flow rate in the oxidation zone, and then the control parameters containing the three kinds of data are used as the genome of the chromosome, thereby obtaining a set consisting of 300 chromosomes, i.e., the initial population. This embodiment does not specifically limit the number of chromosomes in the initial population. It should be noted that in order to avoid premature convergence of the chromosome optimization process, the selected chromosomes are distributed in each interval of the control parameter range.

[0046] In an exemplary embodiment of the present invention, the fitness of a chromosome may be determined by the following steps:

[0047] The desulfurization efficiency and desulfurization cost are calculated under any control parameter, and the desulfurization efficiency and desulfurization cost are weighted and summed to obtain the fitness of the chromosome corresponding to any control parameter; among which, the weight of the desulfurization efficiency is 1, and the weight of the desulfurization cost is -1.

[0048] Optionally, the desulfurization efficiency may be the ratio of the difference in sulfur dioxide concentration before and after desulfurization under any control parameter to the sulfur dioxide concentration before desulfurization; or it may be the normalized value of the difference in sulfur dioxide concentration before and after desulfurization under the control parameter.

[0049] The desulfurization cost can be the operating cost per unit period, such as 2 hours, when desulfurization is performed under any control parameter; or it can be the total operating cost of the desulfurization process. This embodiment does not specifically limit the method for determining the desulfurization efficiency and the desulfurization cost.

[0050] Furthermore, the weight of the desulfurization efficiency and the weight of the desulfurization cost can be used to perform weighted summation on the desulfurization efficiency and the desulfurization cost under the corresponding control parameters of each chromosome to obtain the fitness of the corresponding chromosome.

[0051] The desulfurization control module 120 is used to iteratively update the genome of each chromosome in the initial population using a genetic algorithm. During each update, the weighted mutation rate of each chromosome is used as the mutation rate for the next update until the preset update termination condition is met to obtain the target control parameter to achieve desulfurization control.

[0052] It should be noted that in traditional genetic algorithms, the mutation rate of each chromosome is usually set to a fixed value. However, in the wet flue gas desulfurization process, due to the high nonlinearity of various reactions, the range and complexity of the control parameters increase significantly. In this case, the use of a uniform mutation rate will bring some problems: when the mutation rate is set too low, the chromosomes with lower fitness converge more slowly, and the chromosomes with higher fitness are difficult to jump out of the local optimal solution; when the mutation rate is too high, the chromosomes with lower fitness may miss the optimal solution, and the chromosomes with higher fitness are difficult to converge, which ultimately leads to the genetic algorithm being unable to obtain accurate optimal control parameters. Therefore, the present invention improves the traditional genetic algorithm.

[0053] The specific improvement process is: at each update, the mutation rate of the corresponding chromosome is weighted using the potential factor of each chromosome and the search efficiency, and then the weighted mutation rate of each chromosome at any update is used as the mutation rate of the corresponding chromosome at the next update until the update process is completed.

[0054] Among them, the present invention only improves the method of determining the mutation rate in the genetic algorithm, and does not improve other contents and parameters.

[0055] Specifically, the process of determining the weighted mutation rate of any chromosome during any update includes the following steps:

[0056] Step 1: Calculate the potential factor of any chromosome;

[0057] It should be noted that when the evolutionary direction of any chromosome is accurate and the degree of convergence is low, the potential factor of the chromosome is larger.

[0058] Specifically, the potential factor of any chromosome satisfies the following relationship:

[0059] ;

[0060] In the formula, For the Updated at Potential factors of chromosomes; For the Updated at The degree of growth of the fitness of each chromosome; For the Updated at the same time as The number of chromosomes whose chromosome distance is less than a preset distance threshold, in this embodiment, the distance threshold is 0.2; For the Updated at The fitness of each chromosome; For the Updated at Chromosomes and The distance between chromosomes; is the number of chromosomes; is the normalization function.

[0061] in, Reflects the Updated at The proximity of each chromosome to the chromosome with the highest fitness; Reflects the Updated at the same time as The number of chromosomes that are similar to the chromosomes; when the chromosome is far away from the chromosome with higher fitness and the number of similar chromosomes is small, it means that the possibility of the chromosome having converged is low, and the corresponding potential factor of the chromosome is large.

[0062] Optional, when When the value is larger, the desulfurization effect of the chromosome under the corresponding control parameters gradually becomes better, indicating that the evolutionary direction of the chromosome is more accurate and the corresponding potential factor of the chromosome is relatively large.

[0063] In an exemplary embodiment of the present invention, the determination of the growth degree of the fitness of the chromosome can be achieved by the following steps:

[0064] Calculate the difference between the fitness of any chromosome at any historical update and the fitness at the last update, multiply all differences greater than zero, and obtain the degree of growth of the fitness of any chromosome.

[0065] Specifically, the growth degree of fitness of any chromosome satisfies the following relationship:

[0066] ;

[0067] In the formula, For the Updated at The degree of growth of the fitness of each chromosome; For the Updated at The fitness of each chromosome; For the Updated at The fitness of each chromosome; is an activation function, wherein when the difference is positive, the return value of the activation function is the difference, and when the difference is negative, the return value of the activation function is zero; To find the product symbol.

[0068] Optionally, when the accuracy of the evolutionary direction of any chromosome is high and the possibility that the chromosome has converged is low, that is, the potential factor of the chromosome is large, a smaller mutation rate needs to be set to avoid affecting the evolutionary direction of the chromosome to ensure the accuracy of the chromosome evolutionary direction.

[0069] Step 2: Calculate the chromosome search efficiency;

[0070] It should be noted that when any chromosome is close to the local optimal chromosome and the fitness of the chromosome is low during the historical update process, the search efficiency of the chromosome is low.

[0071] Specifically, the search efficiency of any chromosome satisfies the following relationship:

[0072] ;

[0073] In the formula, For the Updated at The search efficiency of each chromosome; For the Updated at The fitness of each chromosome; To preset hyper parameters, in this embodiment , in order to prevent the denominator from being zero; For the Updated at The average of the reciprocals of the differences between the genes corresponding to each chromosome and the local optimal chromosome; is the absolute value symbol; is the normalization function.

[0074] in, , For the Updated at The first The value of a gene, For the Updated at The local optimal chromosome in the genome The value of a gene, is the number of genes in the chromosome genome, is the number of local optimal chromosomes, is the absolute value symbol, It should be noted that the genome of a chromosome can represent the position of the chromosome in the optimization space. Therefore, calculating the difference in the corresponding gene values ​​in the genomes of any two chromosomes can reflect the distance of the chromosomes, thereby measuring the distance between each chromosome and the local optimal chromosome.

[0075] The distance between the historical update times and the current update times reflects the influence weight of the historical update times. The larger the value, the farther the distance from the current update times. Update to The impact is relatively low.

[0076] In an exemplary embodiment of the present invention, the determination of the local optimal chromosome can be achieved by the following steps:

[0077] During any update, the chromosome whose distance to any chromosome is less than the distance threshold is regarded as a similar chromosome; a preset number threshold is set, and if the number of similar chromosomes is greater than the number threshold, the chromosome with the highest fitness among the similar chromosomes is regarded as the local optimal chromosome.

[0078] Specifically, the quantity threshold may be set to 20. When the number of chromosomes similar to any chromosome is greater than 20, the chromosome with the highest fitness is selected from the chromosome and its similar chromosomes as the local optimal chromosome. This embodiment does not specifically limit the setting of the quantity threshold.

[0079] Step 3: Preset the initial mutation rate, and use the weight to weight the initial mutation rate to obtain the weighted mutation rate. The weight is negatively correlated with the potential factor and the chromosome search efficiency.

[0080] In this embodiment, the initial mutation rate of all chromosomes is 0.3.

[0081] Specifically, the weight of the initial mutation rate of any chromosome satisfies the following relationship:

[0082] ;

[0083] In the formula, For the Updated at The weight of the initial mutation rate of each chromosome; For the Updated at Potential factors of chromosomes; For the Updated at The search efficiency of each chromosome.

[0084] Among them, when the potential factor of any chromosome is large, the evolutionary direction of the chromosome is more accurate and it is in the evolving stage. A smaller weight should be set to reduce the impact of the initial mutation rate on the evolution of the chromosome. At the same time, if the search efficiency of the chromosome is high, it means that the chromosome is far away from the local optimal chromosome and the desulfurization effect of the corresponding control parameters of the chromosome is better. In this case, a smaller weight should be set to reduce the impact of the initial mutation rate on the evolution of the chromosome.

[0085] Furthermore, the weighted mutation rate of each chromosome in any updating process can be used as the mutation rate of the corresponding chromosome in the next updating, until the preset updating termination condition is met, such as the number of chromosome updates is greater than the set value, or the change in the fitness of all chromosomes is less than the set change threshold, etc., and then the chromosome with the highest fitness after stopping the updating is taken as the optimal chromosome, so as to control the wet desulfurization process based on the control parameters corresponding to the optimal chromosome and realize the automatic control of wet desulfurization.

[0086] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.

[0087] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.

Claims

1. An artificial intelligence-based wet flue gas desulfurization automation control system, characterized in that: include: An initialization module is used to obtain a control parameter range during desulfurization, and to construct an initial population of a genetic algorithm within the control parameter range, wherein the fitness of each chromosome in the initial population reflects the desulfurization effect under the corresponding control parameters; A desulfurization control module is used to iteratively update the genome of each chromosome in the initial population using a genetic algorithm, and at each update, the weighted mutation rate of each chromosome is used as the mutation rate at the next update, until a preset update termination condition is met to obtain a target control parameter, so as to achieve desulfurization control; The weighted mutation rate of any chromosome at any update is obtained by calculating the potential factor of the chromosome: , For the Updated at The growth rate of chromosome fitness, For the Updated at the same time as The number of chromosomes whose chromosome distance is less than the preset distance threshold. For the Updated at The fitness of a chromosome, For the Updated at Chromosomes and The distance between chromosomes, is the number of chromosomes, is the normalization function; An initial mutation rate is preset, and the initial mutation rate is weighted by using a weight to obtain a weighted mutation rate, wherein the weight is negatively correlated with the potential factor and the chromosome search efficiency.

2. According to the artificial intelligence-based wet flue gas desulfurization automation control system of claim 1, it is characterized in that: The weights satisfy the following relationship: ； In the formula, For the Updated at The weight of the initial mutation rate of each chromosome; For the Updated at Potential factors of chromosomes; For the Updated at The search efficiency of each chromosome.

3. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 1 is characterized in that: The search efficiency of the chromosome satisfies the following relationship: ； In the formula, For the Updated at The search efficiency of each chromosome; For the Updated at The fitness of each chromosome; To preset hyperparameters; For the Updated at The average of the reciprocals of the differences between the genes corresponding to each chromosome and the local optimal chromosome; is the absolute value symbol; is the normalization function.

4. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 3 is characterized in that: The method for obtaining the local optimal chromosome comprises: In any update, the chromosomes whose distance to any chromosome is less than the distance threshold are regarded as similar chromosomes; A quantity threshold is preset, and if the number of the similar chromosomes is greater than the quantity threshold, the chromosome with the highest fitness among the similar chromosomes is taken as the local optimal chromosome.

5. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 1 is characterized in that: The method for obtaining the growth degree of the chromosome fitness includes: The difference between the fitness of any chromosome at any historical update and the fitness at the last update is calculated, and all differences greater than zero are multiplied to obtain the degree of increase in the fitness of any chromosome.

6. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 1 is characterized in that: The initial population of the genetic algorithm is constructed within the control parameter range, including: A plurality of control parameters are randomly selected within the control parameter range, and each of the control parameters is used as a genome to construct a plurality of chromosomes, and a set consisting of all chromosomes is used as the initial population.

7. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 1 is characterized in that: The method for obtaining the fitness includes: Calculating the desulfurization efficiency and the desulfurization cost when desulfurization is performed under any control parameter, performing weighted summation on the desulfurization efficiency and the desulfurization cost, and obtaining the fitness of the chromosome corresponding to any control parameter; The weight of the desulfurization efficiency is 1, and the weight of the desulfurization cost is -1.

8. The wet flue gas desulfurization automation control system based on artificial intelligence according to claim 1 is characterized in that: A control parameter includes limestone slurry flow rate, absorber pH, and air flow rate in the oxidation zone.

Images