Coking coal blending optimization method based on improved AOA algorithm
By using an improved AOA algorithm and a coking coal blending optimization model, a forecasting model and an optimization model were constructed, which solved the problems of resource waste and increased costs in coking coal blending and achieved a balanced optimization of coke quality and economy.
Patent Information
- Application Number
- CN202411235473.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-29
AI Technical Summary
Existing coking coal blending optimization methods rely on manual experience, leading to resource waste and increased production costs. Traditional optimization methods are prone to getting stuck in local optima, failing to effectively consider economic indicators, and are unable to accurately calculate coke quality and cost.
An improved AOA algorithm is adopted, combined with a coking coal blending optimization model, to construct a prediction model for the quality indicators of blended coal and coke. The effect of coal combination is predicted by multiple linear regression and BP neural network, and the proportion of single coal is optimized. A multi-group cooperative evolution mechanism and differential evolution strategy are adopted to avoid local optima and realize automated decision-making.
While ensuring coke quality, resources should be rationally allocated to reduce production costs, improve coke quality, and achieve precise optimization of single-type coal blending ratios.
Smart Images

Figure CN119047705A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of metallurgy, and in particular relates to a coking coal blending optimization method based on an improved AOA algorithm. Background Art
[0002] Coking coal blending refers to mixing different single coals in appropriate proportions to form blended coal, which is then loaded into the coke oven for coking. Blended coal is a necessary raw material for producing coke, and its cost accounts for 80% of the total production cost. 70% of the quality of coke depends on the quality of blended coal. Therefore, it is an issue worthy of attention to select the best coal blending plan and blend the most ideal raw materials for the furnace to improve the quality of coke and reduce production costs.
[0003] However, in the current production process, the proportion of blended coal mostly relies on manual experience. Based on the quality indicators of individual coals and the relevant characteristics of blended coals, the rough proportion of each individual coal is manually calculated to ensure the quality of coke, and accurate quantitative calculations cannot be performed. At the same time, the proportion of individual coals is finally determined through repeated corrections through small coke oven experiments, and the calculation and verification cycle is long. In the actual production process, in order to ensure the quality of coke, workers mostly use high-quality coal when blending coal, while the proportion of single coal with poor coking properties and low prices is small, which cannot ensure the reasonable allocation of resources, resulting in waste of resources and increased production costs. With the advancement of science and technology, more and more researchers are applying information and intelligent technologies to coal blending optimization problems. Usually, manual operation experience is implanted into the control method. When making decision optimization, only the quality of coke is considered, and the impact of economic indicators on actual production is not fully considered.
[0004] The coking coal blending optimization problem is a typical combinatorial optimization problem, which has the characteristics of nonlinearity, non-differentiability, and multi-modality. Traditional optimization methods are prone to fall into local optimality when solving it, resulting in premature convergence, and the optimization effect is not ideal. Summary of the invention
[0005] The purpose of the present invention is to provide a coking coal blending optimization method based on an improved AOA algorithm, aiming to solve the technical problems existing in the prior art identified in the background technology.
[0006] The present invention is implemented in this way: a coking coal blending optimization method based on an improved AOA algorithm, the method comprising:
[0007] S1. Collect and organize production sample data, wherein the sample data includes coke oven production data and small coke oven experimental data;
[0008] S2. Preprocess the collected production sample data, perform correlation analysis on the preprocessed data, and obtain feature data with a correlation coefficient greater than 0.1;
[0009] S3. Based on the acquired characteristic data, respectively construct a blended coal quality index prediction model and a coke quality index prediction model, wherein the blended coal quality index prediction model is used to predict the quality index of the blended coal and analyze the potential risks of different coal combinations, and the coke quality index prediction model is used to predict the quality index of coke and provide data feedback for coke production;
[0010] S4. Based on the prediction model of quality index of blended coal and the prediction model of quality index of coke, a coking coal blending optimization model is constructed to simulate the effect of different coal combinations and make automated decisions on the coal blending process;
[0011] S5. Use the improved AOA algorithm to optimize and solve the coking coal blending optimization model, and output the optimal single coal type and proportion.
[0012] As a further solution of the present invention, the production sample data includes: type, unit price, adhesiveness index, ash content, volatile matter, sulfur content of a single coal, ash content, sulfur content, crushing strength, abrasion resistance, reactivity index, post-reaction strength, and process operation parameters of coke.
[0013] As a further solution of the present invention, in the blended coal quality index prediction model, the blended coal quality indexes are blended coal ash content, volatile matter, sulfur content and caking index. According to the blended coal quality indexes, a multivariate linear regression equation is constructed, and its mathematical expression is:
[0014] y=β0+β1·x1+β2·x2+...+β p ·x p ;
[0015] Where y represents the quality index of the blended coal; β0, β1, β2, ..., β p represents the fitting coefficients; x1, x2, …, x p It represents the forecast model input features determined by the data analysis in step S2;
[0016] The coke quality index prediction model is constructed by using a BP neural network. The coke quality index prediction model includes a three-layer network structure, namely an input layer, a hidden layer and an output layer.
[0017] The input of the coke quality index prediction model is characteristic data with a correlation coefficient greater than 0.1, and the process operation parameters are taken into account. The output is coke quality indicators, including coke sulfur content, ash content, crushing strength, abrasion resistance, reactivity index and post-reaction strength.
[0018] As a further solution of the present invention, the coking coal blending optimization model has the decision variables of the proportion of each single coal type and the objective function of minimizing the blending coal cost. The mathematical expression is as follows:
[0019]
[0020] Where C represents the cost of blended coal; x i represents the ratio of the i-th single coal; P i represents the unit price of the i-th type of coal; n represents the number of types of single coal.
[0021] The constraints are the quality constraints of the blended coal, the quality constraints of the coke and the proportion of the added quantity.
[0022] As a further solution of the present invention, the steps for obtaining the quality constraints of the blended coal are:
[0023] The blended coal quality prediction model is called to predict the blended coal quality. Its quality index satisfies the inequality constraint, and its mathematical expression is as follows:
[0024] I j ≤W j ≤U j ;
[0025] Where W j I represents the quality index of the blended coal, j represents ash content, sulfur content, volatile matter, and cohesiveness index; j and U j They respectively represent the lower limit and upper limit of the blended coal quality index j.
[0026] As a further solution of the present invention, the coke quality constraint obtaining step is:
[0027] The coke quality prediction model is called to predict the coke quality under the current single coal ratio condition. Its quality index satisfies the inequality constraint, and its mathematical expression is as follows:
[0028] I k ≤W k ≤U k ;
[0029] Where W k It represents the coke quality index k, where k represents ash content, sulfur content, crushing strength, abrasion resistance, reactivity index, and post-reaction strength; I k and U k They respectively represent the lower limit and upper limit of the coke quality index k.
[0030] As a further solution of the present invention, the steps of obtaining the proportion addition amount constraint are:
[0031] The sum of the proportions of a single type of coal is 100%, and its mathematical expression is as follows:
[0032]
[0033] At the same time, the single coal ratio meets the upper and lower limit constraints, and the mathematical expression is as follows:
[0034] I i ≤x i ≤U i ;
[0035] In the formula, I i with U i They respectively represent the lower limit and upper limit of the proportion of the i-th single coal.
[0036] As a further solution of the present invention, the improved AOA algorithm has the following process steps:
[0037] S5.1. Initialize the current number of iterations t=1 and the maximum number of iterations T;
[0038] Initialize the population with a population size of N, use the reverse learning strategy to generate N individuals, calculate the fitness value, and select the top N individuals to form the initial population:
[0039]
[0040] Where, L i with U i They respectively represent the lower and upper limits of the search range corresponding to the i-th individual.
[0041] Each individual in the population represents a potential solution vector, and its solution vector represents the set composed of the proportions of each single coal:
[0042] S5.2, randomly divide the entire population into nei subgroups, each of which is of equal size, sort each subgroup in ascending order based on fitness value, select the local optimal individuals in each subgroup to form an elite subgroup, introduce the idea of differential evolution, and evolve the elite individuals;
[0043] The other individuals in each subgroup adopt a hierarchical evolution strategy using a pyramid structure. Individuals at the same level do not affect each other, and the population is reorganized every 5 generations, resetting the record times flag = 0.
[0044] S5.3, when t≤T, the population evolves iteratively;
[0045] For individuals in the elite subgroup, since they have good position information, the information exchange between them should be enhanced to avoid falling into the local optimum. Therefore, the differential evolution idea is introduced, and the probability p is selected. According to the choice of probability p, the fusion Cauchy operator strategy and the DE / current-to-rand / 1 strategy are used for mutation. Specifically:
[0046] In the early stage of iteration, the probability of elite individuals using the fusion Cauchy operator strategy for mutation is greater than the probability of using the DE / current-to-rand / 1 strategy for mutation;
[0047] In the process of iteration, as the selection probability p increases linearly, the probability of individuals using the DE / current-to-rand / 1 strategy for mutation is greater than the probability of using the fusion Cauchy operator strategy for mutation, and a mutation vector is generated. The mathematical expression is shown in the following formula:
[0048]
[0049] In the formula, v i represents the variation vector of the ith individual; F represents the coefficient of variation; x r1 、x r2 and x r3 They represent individuals randomly selected from the elite subgroup, and r1, r2, r3 are not equal to i; rand represents a random number between [0,1]; p represents the selection probability, p = 0.6-0.2·(Tt) / T; Cauchy(0,1) represents the standard Cauchy distribution;
[0050] Constrain the mutation vector, perform crossover operations, and generate experimental vectors:
[0051]
[0052] In the formula, u i,j represents the jth dimension of the ith individual in the experimental vector; rand represents a random number between [0,1]; CR represents the crossover coefficient; randi(1,D) represents a number randomly selected in [1,D], and D represents the dimension;
[0053] Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection.
[0054] S5.4. For each subgroup, divide it into levels according to the proportion and carry out individual evolution layer by layer;
[0055] Calculate the mathematical optimizer probability MOP, the calculation formula is:
[0056]
[0057] In the formula, α is a sensitive parameter;
[0058] For local development of individuals, the mathematical expression is as follows:
[0059]
[0060] Where nbestj represents an individual randomly selected from a set, and the set consists of individuals in the upper level of the individual's level; μ represents a control parameter;
[0061] Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection.
[0062] S5.5. Determine whether to execute population recombination. When flag<5, flag=flag+1; otherwise, execute population recombination.
[0063] S5.6. Determine whether to end. If the current number of iterations t is less than the maximum number of iterations T, then t=t+1, and update the global optimal individual, and return to step S5.3; otherwise, output the optimal individual position and fitness value.
[0064] The beneficial effects of the present invention are:
[0065] The present invention provides a method for optimizing coking coal blending based on an improved AOA algorithm. According to the relevant characteristics of individual coals, under the premise of satisfying the quality of blended coal and coke, a blended coal quality index prediction model is constructed based on historical data and individual coal quality indicators by adopting a multivariate linear regression method. The influence of process operating parameters on coke quality is further considered, and a coke quality index prediction model is constructed based on a BP neural network algorithm. The method aims to minimize production costs and determine the optimal proportion of each individual coal, thereby achieving the purpose of rationally allocating resources, reducing production costs, and improving coke quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 A flowchart of a coking coal blending optimization method based on an improved AOA algorithm provided in an embodiment of the present invention;
[0067] Figure 2 The improved AOA algorithm flow chart provided by the embodiment of the present invention. DETAILED DESCRIPTION
[0068] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0069] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.
[0070] Figure 1 The flowchart of the coking coal blending optimization method based on the improved AOA algorithm (Improved Arithmetic Optimization Algorithm, IAOA) provided in the embodiment of the present invention is as follows: Figure 1 As shown, the method includes:
[0071] S1. Collect and organize production sample data, wherein the sample data includes coke oven production data and small coke oven experimental data;
[0072] S2. Preprocess the collected production sample data, perform correlation analysis on the preprocessed data, and obtain feature data with a correlation coefficient greater than 0.1;
[0073] S3. Based on the acquired characteristic data, respectively construct a blended coal quality index prediction model and a coke quality index prediction model, wherein the blended coal quality index prediction model is used to predict the quality index of the blended coal and analyze the potential risks of different coal combinations, and the coke quality index prediction model is used to predict the quality index of coke and provide data feedback for coke production;
[0074] S4. Based on the prediction model of quality index of blended coal and the prediction model of quality index of coke, a coking coal blending optimization model is constructed to simulate the effect of different coal combinations and make automated decisions on the coal blending process;
[0075] S5. Use the improved AOA algorithm to optimize and solve the coking coal blending optimization model, and output the optimal single coal type and proportion.
[0076] Based on data analysis, the present invention constructs a prediction model for the quality index of blended coal and a prediction model for the quality index of coke. First, the production sample data is processed for missing values and outliers to reduce the impact of abnormal data; then, the data is normalized to eliminate the impact of dimensions between different features; finally, a correlation analysis is performed, and features with a correlation coefficient greater than 0.1 are selected as input features of the prediction model for the quality index of blended coal and the prediction model for the quality index of coke. For the prediction model for the quality index of blended coal, a multivariate linear regression method is used to construct a regression equation for different features and the quality index of blended coal; for the prediction model for the quality index of coke, the process operating parameters are included in the model input, and based on the BP neural network, a prediction model for different features and the quality index of coke is constructed. The above two prediction models improve the accuracy of model prediction and lay a good foundation for better solving the problem of coal blending optimization.
[0077] The present invention constructs a coking coal blending optimization model based on the blending coal quality index prediction model and the coke quality index prediction model, aiming to determine the optimal proportion of each single type of coal with the production cost minimized under the premise of meeting the blending coal quality index and the coke quality index, and proposes an IAOA algorithm to optimize and solve it, determine the optimal proportion of each single type of coal, and provide important theoretical guidance and technical support for optimizing resource allocation, reducing production costs, and improving coke quality. In the algorithm, a multi-group collaborative evolution mechanism is proposed, the population is grouped according to a certain number, and based on the fitness value, the local optimal individuals are selected to form an elite subgroup, and the other individuals are hierarchically evolved according to the pyramid structure; at the same time, in order to further enhance the communication and cooperation of the population, avoid premature convergence, and better balance exploration and development, the population is reorganized every 5 generations. Specifically, the individuals in the elite subgroup have better position information, and by enhancing the information exchange between them, the population diversity is increased to avoid falling into the local optimum; while the other individuals in the subgroup adopt a layer-by-layer evolution strategy based on the pyramid structure to strengthen the guiding role of the upper individuals on the lower individuals, thereby increasing the search efficiency and optimization accuracy. Therefore, for elite individuals, the idea of differential evolution is introduced. Through parameter judgment, it is determined that individuals adopt the DE / current-to-rand / 1 strategy or introduce the Cauchy operator strategy for individual mutation to enhance the communication and cooperation between them and avoid falling into the local optimum. For other individuals in the subgroup, a layer-by-layer evolution strategy is adopted to strengthen the guiding role of upper-level individuals on lower-level individuals to enhance the search efficiency of lower-level individuals. As upper-level individuals evolve, they carry more valuable information that is beneficial to individual evolution. Therefore, the local development strategy in the standard AOA algorithm is adjusted. When the upper-level individuals complete their evolution, their latest position information is used for local development, thereby increasing the search efficiency and optimization accuracy of lower-level individuals.
[0078] In the present invention, the production sample data includes: single coal type, unit price, caking index, ash, volatile matter, sulfur content, coke ash, sulfur content, crushing strength, abrasion resistance, reactivity index, post-reaction strength, and process operation parameters.
[0079] In the present invention, in the blended coal quality index prediction model, the blended coal quality indexes are the blended coal ash content, volatile matter, sulfur content and cohesiveness index. According to the blended coal quality indexes, a multivariate linear regression equation is constructed, and its mathematical expression is:
[0080] y=β0+β1·x1+β2·x2+...+β p ·x p ;
[0081] Where y represents the quality index of the blended coal; β0, β1, β2, ..., β p represents the fitting coefficients; x1, x2, …, xp It represents the forecast model input features determined by the data analysis in step S2;
[0082] The coke quality index prediction model is constructed by using a BP neural network. The coke quality index prediction model includes a three-layer network structure, namely an input layer, a hidden layer and an output layer.
[0083] The input of the coke quality index prediction model is characteristic data with a correlation coefficient greater than 0.1, and the process operation parameters are taken into account. The output is coke quality indicators, including coke sulfur content, ash content, crushing strength, abrasion resistance, reactivity index and post-reaction strength.
[0084] In the present invention, the coking coal blending optimization model, the decision variables are the proportions of each single type of coal, the objective function is to minimize the blending coal cost, the mathematical expression is as follows:
[0085]
[0086] Where C represents the cost of blended coal; x i represents the ratio of the i-th single coal; P i represents the unit price of the i-th type of coal; n represents the number of types of single coal.
[0087] The constraints are the quality constraints of the blended coal, the quality constraints of the coke and the proportion of the added quantity.
[0088] In the present invention, the steps for obtaining the quality constraints of blended coal are:
[0089] The blended coal quality prediction model is called to predict the blended coal quality, including ash content, sulfur content, volatile matter, cohesiveness index, etc. Its quality index satisfies the inequality constraint, and its mathematical expression is as follows:
[0090] I j ≤W j ≤U j ;
[0091] Where W j I represents the quality index of the blended coal, j represents ash content, sulfur content, volatile matter, and cohesiveness index; j and U j They respectively represent the lower limit and upper limit of the blended coal quality index j.
[0092] In the present invention, the steps of obtaining the coke quality constraint are:
[0093] The coke quality prediction model is called to predict the coke quality under the current single coal ratio conditions, including ash content, sulfur content, crushing strength, wear resistance, reactivity index, post-reaction strength, etc. Its quality indicators meet the inequality constraints, and its mathematical expression is as follows:
[0094] I k ≤W k ≤U k ;
[0095] Where W k It represents the coke quality index k, where k represents ash content, sulfur content, crushing strength, abrasion resistance, reactivity index, and post-reaction strength; I k and U k They respectively represent the lower limit and upper limit of the coke quality index k.
[0096] In the present invention, the steps for obtaining the ratio addition amount constraint are:
[0097] The sum of the proportions of a single type of coal is 100%, and its mathematical expression is as follows:
[0098]
[0099] At the same time, the single coal ratio meets the upper and lower limit constraints, and the mathematical expression is as follows:
[0100] I i ≤x i ≤U i ;
[0101] In the formula, I i with U i They respectively represent the lower limit and upper limit of the proportion of the i-th single coal.
[0102] Figure 2 The improved AOA algorithm flow chart provided by the embodiment of the present invention is as follows: Figure 2 As shown, the improved AOA algorithm has the following process steps:
[0103] S5.1. Initialize the current number of iterations t=1 and the maximum number of iterations T;
[0104] Initialize the population with a population size of N, use the reverse learning strategy to generate N individuals, calculate the fitness value, and select the top N individuals to form the initial population;
[0105] Each individual in the population represents a potential solution vector, and its solution vector represents the set composed of the proportions of each single coal:
[0106]
[0107] Where, L i with U i They respectively represent the lower and upper limits of the search range corresponding to the i-th individual.
[0108] S5.2, randomly divide the entire population into nei subgroups, each of which is of equal size, sort each subgroup in ascending order based on fitness value, select the local optimal individuals in each subgroup to form an elite subgroup, introduce the idea of differential evolution, and evolve the elite individuals;
[0109] The other individuals in each subgroup adopt a hierarchical evolution strategy using a pyramid structure. Individuals at the same level do not affect each other, and the population is reorganized every 5 generations, resetting the record times flag = 0.
[0110] S5.3, when t≤T, the population evolves iteratively;
[0111] For individuals in the elite subgroup, since they have good position information, the information exchange between them should be enhanced to avoid falling into the local optimum. Therefore, the differential evolution idea is introduced, and the probability p is selected. According to the choice of probability p, the fusion Cauchy operator strategy and the DE / current-to-rand / 1 strategy are used for mutation. Specifically:
[0112] In the iteration process, the mutation strategy is adaptively selected by selecting probability p. In the early stage of iteration, the selection probability p is 0.4, so most individuals in the population use the fusion Cauchy operator strategy for mutation; as the iteration progresses, the selection probability p increases linearly to 0.6, so most individuals in the population use the DE / current-to-rand / 1 strategy for mutation.
[0113] In the early stage of iteration, the probability of elite individuals using the fusion Cauchy operator strategy for mutation is greater than the probability of using the DE / current-to-rand / 1 strategy for mutation;
[0114] In the process of iteration, as the selection probability p increases linearly, the probability of individuals using the DE / current-to-rand / 1 strategy for mutation is greater than the probability of using the fusion Cauchy operator strategy for mutation, and a mutation vector is generated. The mathematical expression is shown in the following formula:
[0115]
[0116] In the formula, v i represents the variation vector of the ith individual; F represents the coefficient of variation; x r1 、x r2 and x r3 They represent individuals randomly selected from the elite subgroup, and r1, r2, r3 are not equal to i; rand represents a random number between [0,1]; p represents the selection probability, p = 0.6-0.2·(Tt) / T; Cauchy(0,1) represents the standard Cauchy distribution;
[0117] Constrain the mutation vector, perform crossover operations, and generate experimental vectors:
[0118]
[0119] In the formula, u i,j represents the jth dimension of the ith individual in the experimental vector; rand represents a random number between [0,1]; CR represents the crossover coefficient; randi(1,D) represents a number randomly selected in [1,D], and D represents the dimension;
[0120] Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection;
[0121] S5.4. For each subgroup, divide it into levels according to the proportion and carry out individual evolution layer by layer;
[0122] Calculate the mathematical optimizer probability MOP, the calculation formula is:
[0123]
[0124] In the formula, α is a sensitive parameter, and its value is 5;
[0125] For local development of individuals, the mathematical expression is as follows:
[0126]
[0127] Where nbest j represents an individual randomly selected from a set, and the set consists of individuals in the upper level of the individual's level; μ represents a control parameter, and its value is 0.499;
[0128] Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection.
[0129] S5.5. Determine whether to execute population recombination. When flag<5, flag=flag+1; otherwise, execute population recombination.
[0130] S5.6. Determine whether to end. If the current number of iterations t is less than the maximum number of iterations T, then t=t+1, and update the global optimal individual, and return to step S5.3; otherwise, output the optimal individual position and fitness value.
[0131] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.
[0132] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
[0133] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0134] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.
[0135] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for optimizing coking coal blending based on an improved AOA algorithm, characterized in that: The method comprises: S1. Collect and organize production sample data, wherein the sample data includes coke oven production data and small coke oven experimental data; S2. Preprocess the collected production sample data, perform correlation analysis on the preprocessed data, and obtain feature data with a correlation coefficient greater than 0.1; S3. Based on the acquired characteristic data, respectively construct a blended coal quality index prediction model and a coke quality index prediction model, wherein the blended coal quality index prediction model is used to predict the quality index of the blended coal and analyze the potential risks of different coal combinations, and the coke quality index prediction model is used to predict the quality index of coke and provide data feedback for coke production; S4. Based on the prediction model of quality index of blended coal and the prediction model of quality index of coke, a coking coal blending optimization model is constructed to simulate the effect of different coal combinations and make automated decisions on the coal blending process; S5. Use the improved AOA algorithm to optimize and solve the coking coal blending optimization model, and output the optimal single coal type and proportion.
2. The method according to claim 1, characterized in that The production sample data include: type, unit price, caking index, ash content, volatile matter, sulfur content of a single coal, ash content, sulfur content, crushing strength, abrasion resistance, reactivity index, post-reaction strength, and process operation parameters of coke.
3. The method according to claim 2, characterized in that In the blended coal quality index prediction model, the blended coal quality indexes are the blended coal ash content, volatile matter, sulfur content and cohesiveness index. According to the quality indexes of a single coal, a multivariate linear regression equation is constructed, and its mathematical expression is: y=β0+β1·x1+β2·x2+...+β p ·x p ; Where y represents the quality index of the blended coal; β0, β1, β2, ..., β p represents the fitting coefficients; x1, x2, …, x p It represents the forecast model input features determined by the data analysis in step S2; The coke quality index prediction model is constructed by using a BP neural network. The coke quality index prediction model includes a three-layer network structure, namely an input layer, a hidden layer and an output layer. The input of the coke quality index prediction model is characteristic data with a correlation coefficient greater than 0.1, and the process operation parameters are taken into account. The output is coke quality indicators, including coke sulfur content, ash content, crushing strength, abrasion resistance, reactivity index and post-reaction strength.
4. The method according to claim 1, characterized in that: The decision variable of the coking coal blending optimization model is the proportion of each single coal type, and the objective function is to minimize the blending coal cost. The mathematical expression is as follows: Where C represents the cost of blended coal; x i represents the ratio of the i-th single coal; P i represents the unit price of the i-th type of coal; n represents the number of types of single coal; The constraints are the quality constraints of the blended coal, the quality constraints of the coke and the proportion of the added quantity.
5. The method according to claim 4, characterized in that The steps for obtaining the quality constraints of the blended coal are as follows: The blended coal quality prediction model is called to predict the blended coal quality. Its quality index satisfies the inequality constraint, and its mathematical expression is as follows: I j ≤W j ≤U j ; Where W j I represents the quality index of the blended coal, j represents ash content, sulfur content, volatile matter, and cohesiveness index; j and U j They respectively represent the lower limit and upper limit of the blended coal quality index j.
6. The method according to claim 5, characterized in that The steps for obtaining the coke quality constraint are: The coke quality prediction model is called to predict the coke quality under the current single coal ratio condition. Its quality index satisfies the inequality constraint, and its mathematical expression is as follows: I k ≤W k ≤U k ; Where W k It represents the coke quality index k, where k represents ash content, sulfur content, crushing strength, abrasion resistance, reactivity index, and post-reaction strength; I k and U k They respectively represent the lower limit and upper limit of the coke quality index k.
7. The method according to claim 6, characterized in that The steps for obtaining the proportion addition amount constraint are as follows: The sum of the proportions of a single type of coal is 100%, and its mathematical expression is as follows: At the same time, the single coal ratio meets the upper and lower limit constraints, and the mathematical expression is as follows: I i ≤x i ≤U i ; In the formula, I i with U i They respectively represent the lower limit and upper limit of the proportion of the i-th single coal.
8. The method according to claim 7, characterized in that The improved AOA algorithm has the following process steps: S5.
1. Initialize the current number of iterations t=1 and the maximum number of iterations T; Initialize the population with a population size of N, use the reverse learning strategy to generate N individuals, calculate the fitness value, and select the top N individuals to form the initial population; Each individual in the population represents a potential solution vector, and its solution vector represents the set composed of the proportions of each single coal: Where, L i with U i Respectively represent the lower limit and upper limit of the search range corresponding to the i-th individual; S5.2, randomly divide the entire population into nei subgroups, each of which is of equal size, sort each subgroup in ascending order based on fitness value, select the local optimal individuals in each subgroup to form an elite subgroup, introduce the idea of differential evolution, and evolve the elite individuals; The other individuals in each subgroup adopt a pyramid structure and a hierarchical evolution strategy. Individuals in the same level do not affect each other, and the population is reorganized every 5 generations, and the record number flag is reset to 0; S5.3, when t≤T, the population evolves iteratively; According to the choice of probability p, the fusion Cauchy operator strategy and the DE / current-to-rand / 1 strategy are used for mutation. Specifically: In the early stage of iteration, the probability of elite individuals using the fusion Cauchy operator strategy for mutation is greater than the probability of using the DE / current-to-rand / 1 strategy for mutation; In the process of iteration, as the selection probability p increases linearly, the probability of individuals using the DE / current-to-rand / 1 strategy for mutation is greater than the probability of using the fusion Cauchy operator strategy for mutation, and a mutation vector is generated. The mathematical expression is shown in the following formula: In the formula, v i represents the variation vector of the ith individual; F represents the coefficient of variation; x r1 、x r2 and x r3 They represent individuals randomly selected from the elite subgroup, and r1, r2, r3 are not equal to i; rand represents a random number between [0,1]; p represents the selection probability, p = 0.6-0.2·(Tt) / T; Cauchy(0,1) represents the standard Cauchy distribution; Constrain the mutation vector, perform crossover operations, and generate experimental vectors: In the formula, u i,j represents the jth dimension of the ith individual in the experimental vector; rand represents a random number between [0,1]; CR represents the crossover coefficient; randi(1,D) represents a number randomly selected in [1,D], and D represents the dimension; Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection; S5.
4. For each subgroup, divide it into levels according to the proportion and carry out individual evolution layer by layer; Calculate the mathematical optimizer probability MOP, the calculation formula is: In the formula, α is a sensitive parameter; For local development of individuals, the mathematical expression is as follows: Where nbest j represents an individual randomly selected from a set, and the set consists of individuals in the upper level of the individual's level; μ represents a control parameter; Constrain the updated position, calculate the fitness value, and update the current individual position and fitness value based on greedy selection; S5.5, determine whether to perform population reorganization, when flag<5, then flag=flag+1; Otherwise, perform population reorganization; S5.
6. Determine whether to end. If the current number of iterations t is less than the maximum number of iterations T, then t=t+1, and update the global optimal individual, and return to step S5.3; otherwise, output the optimal individual position and fitness value.
Citation Information
Cited By
Accurate coking coal blending method and system
CN120258261A
Coal blending planning method based on geometric constraint of coal rock reflectivity distribution curve
CN122022604A
