Iron and steel smelting alloy proportion optimization method and device, electronic equipment and storage medium

Abstract

The invention provides a steel smelting alloy proportion optimization method and device, electronic equipment and a storage medium, and the method comprises the steps: constructing a yield dynamic correction and double-algorithm collaborative solving system, and continuously optimizing the alloy element yield through the comparison of the alloy element yield between heat times and the deviation correction in the yield dynamic correction, in the double-algorithm collaborative solution system, a plurality of groups of initial feasible solutions with optimal total cost are screened out through a genetic algorithm, and local accurate optimization is performed on each group of initial feasible solutions through simplex method refinement. The corresponding alloy adding amount scheme is selected as the optimal alloy matching scheme based on the lowest alloy total cost and / or the alloy adding amount balance in all the groups, global optimal solution of the alloy combination is achieved, the method has the characteristics of precision improvement and high adaptability, and when the alloy inventory fluctuation is smaller than or equal to 30% or the steel grade component standard is adjusted, the alloy combination is optimized. The initial solution can be screened again within one minute through genetic algorithm primary screening, and the one-time standard reaching rate of molten steel component adjustment is increased to 96% or above.

Description

Technical Field

[0001] This invention belongs to the field of steel smelting production control technology, specifically relating to a method, apparatus, electronic equipment, and storage medium for optimizing alloy proportions in steel smelting. Background Technology

[0002] The steelmaking process involves the addition of alloy materials. For example, in the LF furnace smelting process, alloy materials are the core auxiliary materials for adjusting the composition of molten steel, such as the content of elements like carbon, manganese, and silicon. Their cost accounts for 15%-20% of the production cost per ton of steel. Therefore, optimizing the alloy ratio and reducing unnecessary input of alloy materials are key ways to reduce costs. At the same time, in modern steel production with multiple varieties and batches, if alloy inventory fluctuates or steel grade standards are adjusted, manually recalculating the batching scheme would be too time-consuming and difficult to guarantee accuracy. Therefore, it is necessary to develop a method, device, electronic equipment, and storage medium for optimizing the alloy ratio in steelmaking to improve the optimization effect and adaptability of the alloy ratio.

[0003] In steelmaking alloy proportion optimization, the single simplex method is often used to solve alloy combinations. For example, the literature "Development and Application of Foundry Charge Proportioning Software Based on Constrained Optimization Algorithm" uses a two-stage simplex linear programming algorithm to solve the linear model and obtain the optimal solution based on constrained optimization algorithm. However, its main drawback is that the single simplex method depends on the selection of the initial basic feasible solution. If the initial solution is close to the local optimum region, it is easy to get trapped in local optima and cannot find the globally lowest cost solution. Although some other alloy calculation methods have been proposed in the existing technology, such as the smelting process disclosed in patent CN115936239A, the single simplex method is used to solve alloy combinations. The optimization method, device, equipment, and storage medium for batching are described. Based on the type of ingredients, the proportion of chemical elements, and the unit price of ingredients, an unconstrained objective function is obtained. The proportion of ingredients is determined by solving the unconstrained objective function using a genetic algorithm. However, alloy batching is a key link in steel production and requires rapid output of solutions. The genetic algorithm relies on a large number of iterations to approach the optimal solution. When faced with changes in production conditions, it may be necessary to readjust the algorithm parameters or even redesign the algorithm process, which cannot respond quickly. Furthermore, the randomness of the genetic algorithm may still cause the search to stop at a local optimum, affecting the continuity and stability of production.

[0004] Secondly, most production sites currently assume that the recovery rate of alloy elements after alloy material is added to molten steel is a fixed value, such as 100%, or a simple comparison. However, in actual production, elements may not be completely absorbed by the molten steel due to oxidation, volatilization, equipment issues, etc., resulting in fluctuations in the recovery rate of alloy elements. Although, for example, a real-time cost optimization control method for ladle refining alloying disclosed in patent CN119850056A calculates the optimal alloy ratio based on the dynamic average of the alloy element recovery rate and an optimization function, in the long run, relying on manual experience or simple averaging without dynamic optimization through inter-furnace deviation correction will still lead to low accuracy in the calculation of alloy material composition based on the alloy element recovery rate. Furthermore, multiple feedings and tests are required to control the composition of molten steel in the LF furnace, which can easily lead to excessive input of expensive alloy materials, affecting the compliance rate of molten steel composition and increasing production costs. It is also difficult to quickly adapt to production fluctuations. Summary of the Invention

[0005] The present invention aims to at least partially solve one of the above-mentioned technical problems. The present invention provides a method, apparatus, electronic equipment and storage medium for optimizing the alloy ratio in steel smelting, which can improve the first-time compliance rate of steel composition adjustment, enhance the alloy ratio optimization effect, realize the global optimal solution of alloy combination, ensure that the solution can be quickly adapted to production changes and effectively reduce costs.

[0006] The technical solution adopted by this invention to solve its technical problem is:

[0007] A method for optimizing alloy proportions in steel smelting, the method comprising:

[0008] Obtain alloy data, including the maximum on-site inventory of each alloy, and steel composition standards;

[0009] The initial screening using a genetic algorithm includes using the amount of each alloy added as the decision variable and the objective function being the lowest total cost of the alloy calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial individuals that satisfy the constraints through real-number encoding as an initial population. The fitness function is the reciprocal of the objective function. The initial population undergoes several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total cost of the alloy.

[0010] The simplex method refinement includes using the simplex method to perform local exact optimization on each set of initial feasible solutions to obtain the lowest total alloy cost corresponding to each set of initial feasible solutions. Based on the lowest total alloy cost among all sets and / or the balance of alloy addition, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme.

[0011] To ensure both the accuracy of the yield and the combination of global optimization and local refinement, a "dynamic yield correction + dual-algorithm collaborative solution" system can be constructed by combining dynamic yield correction with initial screening using a genetic algorithm and refinement using the simplex method. Dynamic yield correction continuously optimizes the accuracy of alloy element yield calculations through inter-furnace yield comparison and deviation correction, providing accurate basic data for alloy proportion calculations. In the preferred technical solution, the method further includes dynamic yield correction, which includes:

[0012] Establish a database to store the yield of each alloy element in each heat.

[0013] Collect current furnace production data and calculate the actual yield of the target alloying element accordingly;

[0014] Extract the target alloy element yield from the previous batch from the database and calculate the deviation between the yield and the actual yield of the target alloy element.

[0015] If a deviation threshold is set, and the obtained deviation is greater than the deviation threshold, the target alloy element yield for the current furnace in the database will not be updated; if the deviation is less than or equal to the deviation threshold, then... Correction, in the formula The corrected target alloy element yield. For correction factor, The target alloy element yield of the previous batch. The actual yield of the target alloying element, calculated according to... The recovery rates of each alloy element in the current furnace are updated in the database accordingly, and this process is repeated to dynamically optimize the recovery rates of each alloy element in each furnace.

[0016] To better provide the foundational data for continuous optimization of alloy proportion calculations, in the preferred technical solution, when setting constraints, the steel composition is calculated based on the amount of each alloy added and the yield of each alloy element in the most recent furnace from the database; preferably, when setting constraints, the steel mass, the initial content of each element in the steel before the addition of alloys, and the yield of each alloy element in the most recent furnace from the database are obtained, and the constraints include:

[0017] In the formula The first in the steel composition standard The lower limit of control for alloying elements, The first step before alloying in molten steel The content of various alloying elements, for The first alloy The amount of each alloy added, For the first The first alloy The content of each element For the most recent batch in the database The yield of alloying elements For the quality of molten steel, The first in the steel composition standard The upper limit of control for certain alloying elements.

[0018] To obtain the actual yield of each alloying element, in a preferred technical solution, the actual yield of the target alloying element is... The calculation formulas include: In the formula For the quality of molten steel, The content of the target element in the molten steel after the addition of the alloy. The target alloy element content in the molten steel before alloying. For the current furnace batch The first alloy The amount of each alloy added, for The first alloy The content of the target alloying element in the alloy.

[0019] In order to further eliminate abnormal interference and optimize the measurement accuracy during continuous correction, in the preferred technical solution, the deviation threshold is gradually reduced from the initial production stage to the stable production stage.

[0020] To increase the likelihood of retaining alloy proportion schemes with lower total alloy cost, in the preferred technical solution, the selection operation adopts a roulette wheel method, determining the selection probability based on the proportion of individual fitness values, and selecting multiple individuals with higher fitness values ​​to enter the next generation.

[0021] In order to generate a better alloy scheme by combining the advantages of different individuals, in a preferred technical solution, the crossover operation includes performing single-point crossover on selected individuals with a crossover probability of 0.7-0.8.

[0022] In order to introduce new changes, avoid the algorithm from getting stuck in local optima, and ensure that the constraints are still met after adjustment, in the preferred technical solution, the mutation operation includes randomly mutating the amount of a certain alloy added in the selected individual with a mutation probability of 0.05-0.1, and the mutation range is ≤10%.

[0023] To further avoid the algorithm getting trapped in local optima, in the preferred technical solution, a population diversity protection mechanism is set during the initial screening process of the genetic algorithm: including setting the number of iterations and a change rate threshold. When the change rate of the optimal fitness value in the first consecutive iteration is less than the change rate threshold, the mutation probability is increased and some individuals are regenerated. When the change rate of the optimal fitness value in the second consecutive iteration is less than the change rate threshold, the iteration stops.

[0024] To obtain the final optimal alloy mix scheme by performing local exact optimization using the simplex method based on the initial feasible solutions selected by the genetic algorithm, the linear programming problem is transformed into a standard form that can be handled by the simplex method. In the preferred technical solution, the local exact optimization of each set of initial feasible solutions using the simplex method includes:

[0025] By introducing slack variables, the constraints are transformed into standard form.

[0026] Combining the standard form constraints, the initial basic feasible solution is determined for each set of initial feasible solutions to obtain the initial basic feasible solution and the corresponding objective function value, and the test number is calculated accordingly.

[0027] Perform an optimality test: if all test numbers are ≥0, then the current basic feasible solution is the optimal solution; if there is a test number <0, then select the non-basic variable with the smallest test number as the input variable, determine the output variable according to the θ rule, perform a basis transformation, recalculate the test numbers, until all test numbers are ≥0, and obtain the lowest total alloy cost corresponding to each initial feasible solution.

[0028] To further ensure that the solutions input by the simplex method meet the actual production requirements, in the preferred technical solution, if a set of initial feasible solutions does not meet the constraints during the simplex method refinement, it is automatically fed back to the genetic algorithm model, and the crossover probability and mutation probability are adjusted before re-screening and outputting the initial feasible solutions.

[0029] To further reduce the difficulty of production operations, the preferred technical solution includes selecting the corresponding alloy addition amount scheme based on the total alloy cost and the balance of alloy addition amount across all groups:

[0030] Set a cost difference threshold, compare the total cost of alloys in all groups, and if the total cost difference of multiple alloy groups including the lowest total cost of alloys is less than the cost difference threshold, then prioritize the alloy addition scheme with the smallest ratio of the maximum alloy addition amount to the minimum alloy addition amount among the multiple alloy groups.

[0031] An alloy proportioning optimization device for steel smelting, comprising:

[0032] The acquisition module is used to acquire alloy data, including the maximum on-site inventory of each alloy and steel composition standards.

[0033] A dual-algorithm collaborative solution module connects to the acquisition module and is used to perform genetic algorithm initial screening and simplex method refinement. The genetic algorithm initial screening includes using the amount of each alloy added as the decision variable and the objective function being the lowest total alloy cost calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial populations that satisfy the constraints through real-number encoding. The fitness function is the reciprocal of the objective function. The initial populations undergo several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total alloy cost. The simplex method refinement includes using the simplex method to perform local exact optimization on each group of initial feasible solutions to obtain the lowest total alloy cost corresponding to each group of initial feasible solutions. Based on the lowest total alloy cost and / or the balance of alloy additions among all groups, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme.

[0034] An electronic device includes at least one processor and a memory; the memory stores at least one instruction, and when the at least one instruction is loaded and executed by at least one processor, the electronic device implements the steel smelting alloy ratio optimization method as described in any of the preceding claims.

[0035] A computer-readable storage medium storing at least one instruction that, when loaded and executed by a computer's processor, causes the computer to implement the steel smelting alloy proportion optimization method as described in any of the preceding claims.

[0036] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0037] (1) In view of the technical problems that the single simplex method is highly dependent on the initial solution and is prone to getting trapped in local optima when solving alloy combinations, and that the existing material optimization method based on genetic algorithm is difficult to adapt to production changes quickly, this invention first uses the genetic algorithm to adapt to production changes by combining constraints, and selects multiple high-quality initial feasible solutions in the global scope. Then, based on the initial feasible solutions selected by the genetic algorithm, the simplex method is used for local precise optimization to obtain the lowest total alloy cost corresponding to each initial feasible solution. Based on the lowest total alloy cost and / or the balance of alloy addition among all groups, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme. The two algorithms work together to achieve the global optimal solution of alloy combination, so as to improve the alloy ratio optimization effect, ensure that the scheme can quickly adapt to production changes and effectively reduce costs. The alloy cost reduction can be 8%-12% higher than that of traditional methods. It has the characteristics of strong adaptability. When the alloy inventory fluctuation is ≤30% or the steel composition standard is adjusted, the initial solution can be re-screened within 1 minute by the genetic algorithm, so as to provide a more efficient and accurate calculation path for alloy feeding control in steel smelting such as LF furnace.

[0038] (2) The present invention can further combine dynamic yield correction with genetic algorithm initial screening and simplex method fine correction to construct a "dynamic yield correction + dual algorithm collaborative solution" system. Dynamic yield correction continuously optimizes the accuracy of alloy element yield calculation by comparing and correcting the deviation between furnace cycles, providing accurate basic data for alloy ratio calculation. It not only ensures the accuracy of yield and improves the first-time compliance rate of molten steel composition adjustment, but also achieves the combination of global optimization and local refinement. It can reduce the alloy element yield calculation error from the traditional ±5% to within ±2%, and increase the first-time compliance rate of molten steel composition adjustment to over 96%. Attached Figure Description

[0039] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0040] Figure 1 This is a schematic diagram of the method flow of Embodiment 1 of the present invention.

[0041] Figure 2 This is a schematic diagram of the dynamic yield correction process in Embodiment 1 of the present invention.

[0042] Figure 3 This is a schematic diagram of the initial screening process of the genetic algorithm in Embodiment 1 of the present invention.

[0043] Figure 4 This is a schematic diagram of the simplex method refinement process in Embodiment 1 of the present invention. Detailed Implementation

[0044] The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the invention, and should not be construed as limiting the invention.

[0045] Example 1:

[0046] To address the technical problems in existing steelmaking alloy proportion optimization methods, such as the strong dependence on initial solutions and susceptibility to local optima when using a single simplex method to solve alloy combinations, and the difficulty of adapting existing batching optimization methods based on genetic algorithms to quickly adapt to production changes, this invention considers a dual-algorithm approach. It first uses a genetic algorithm to globally screen for high-quality initial solutions, and then uses the simplex method for local precise optimization. This dual-algorithm approach collaboratively achieves the global optimal solution for alloy combinations, thereby improving the alloy proportion optimization effect. Therefore, a preferred embodiment of the steelmaking alloy proportion optimization method described in this invention includes:

[0047] Obtain alloy data, including the maximum on-site inventory of each alloy, and steel composition standards;

[0048] The initial screening using a genetic algorithm involves using the amount of each alloy added as the decision variable and the objective function being the lowest total cost of the alloy calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial individuals that satisfy the constraints through real-number encoding as an initial population. The fitness function is the reciprocal of the objective function. The initial population undergoes several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total cost of the alloy. Addressing the issue of the simplex method's strong dependence on initial solutions, the above-mentioned genetic algorithm initial screening can adapt to production changes by setting constraints, screening multiple high-quality initial feasible solutions globally, laying the foundation for subsequent refinement.

[0049] The simplex method refinement includes using the simplex method to perform local precise optimization on each set of initial feasible solutions, obtaining the lowest total alloy cost corresponding to each set of initial feasible solutions, and selecting the corresponding alloy addition scheme as the optimal alloy ratio scheme based on the lowest total alloy cost among all groups and / or the balance of alloy addition. The above simplex method refinement process is based on the initial feasible solutions screened by the genetic algorithm, and completes local precise optimization through the simplex method, ultimately achieving the global optimal solution of the alloy combination, ensuring that the scheme can quickly adapt to production changes and effectively reduce costs.

[0050] Considering the adverse effects of low alloy element recovery accuracy on alloy composition optimization in existing technologies, and in order to ensure both recovery accuracy and improve the first-time compliance rate of molten steel composition adjustment, while also achieving a combination of global optimization and local refinement, a further approach can be taken: dynamically correcting the recovery rate, combining it with initial screening using a genetic algorithm and refinement using the simplex method. This will construct a "dynamic recovery rate correction + dual-algorithm collaborative solution" system. More specifically, for example... Figure 1 As shown, the above method may also include the following steps:

[0051] Step S1, Dynamic Correction of Yield: By comparing and correcting the yield of alloying elements between furnace cycles, the accuracy of alloying element yield calculation is continuously optimized, providing accurate basic data for alloy proportion calculation, such as... Figure 2 As shown, the method includes the following steps:

[0052] S101: Obtain alloy data, including alloy type, calculation method, etc. The database contains information on various alloys and the content of each alloying element within each alloy. A database is established to store the yield of each alloying element for each heat, corresponding to the steel grade. This database is dynamic and supports continuous optimization and precise proportioning.

[0053] S102. Actual yield calculation: Collect the production data of the current furnace and calculate the actual yield of the target alloying element accordingly.

[0054] For example, to obtain the actual yield of each alloying element, the current heat production data includes the molten steel quality, the target element content obtained from the molten steel composition after adding the alloy, the target alloying element content obtained from the molten steel composition before adding the alloy, and the current heat production data obtained from the new alloying scheme. The first alloy The amount of each alloy added, The first alloy The content of the target alloying element in the alloy, and the actual yield of the target alloying element. The calculation formulas include: (1), in equation (1) For the quality of molten steel, The content of the target element in the molten steel after the addition of the alloy. The target alloy element content in the molten steel before alloying. For the current furnace batch The first alloy The amount of each alloy added, for The first alloy The content of the target alloying element in the alloy.

[0055] S103. Yield Deviation Judgment and Correction: Extract the target alloy element yield from the previous batch in the database, calculate its deviation from the actual yield of the target alloy element, set a deviation threshold, and gradually reduce the deviation threshold from the initial production stage to the stable production stage. If the obtained deviation > the deviation threshold, the target alloy element yield in the current batch in the database will not be updated; if the deviation ≤ the deviation threshold, proceed according to... (2) Correction, in equation (2) The corrected target alloy element yield. For correction factor, The target alloy element yield of the previous batch. The actual yield of the target alloying element, calculated according to... Update the database accordingly to the recovery rate of each alloy element in the current furnace.

[0056] For example, determine whether the target alloying element yield for the steel grade exists in the database. If the target alloying element yield for the steel grade does not exist in the database, then the result obtained in step S102 is... The initial alloy element yields are stored in the database. If the target alloy element yield for that steel grade exists in the database, to ensure that the yields of each alloy element in the database continuously approach the actual values ​​and provide accurate basic data for alloy proportioning, fixed values ​​are no longer used as standards. Instead, the target alloy element yields stored in the previous batch are extracted from the database. Calculate the deviation .

[0057] Setting deviation thresholds: In the initial stage of production, to ensure data convergence, the deviation threshold can be relaxed to 5%; in the middle stage of production, the deviation threshold can be set to 3%, and narrowed to 2% in the stable production stage, to determine whether the yield of the target alloy element in the current furnace is within a reasonable fluctuation range; taking 3% as an example, if If the deviation is deemed too large, the purity of the raw materials will be investigated: for example, the actual alloy composition differs from the labeled composition; smelting temperature fluctuations: whether the smelting temperature deviates from the standard temperature by more than ±25℃, etc. In this case, the target alloy element yield for the current furnace will not be updated in the database to avoid abnormal erroneous data contaminating the calculation. If the deviation is within the allowable range and the data validity can be ensured, then the correction step is initiated, and the correction is performed according to formula (2) to obtain the updated target alloy element yield. And store it in the database, in formula (2) The correction factor can be between 0.1 and 0.5, depending on the large fluctuation in the yield of the target alloying element. Take the larger value to quickly match reality; minimize fluctuations. To maintain data stability, smaller values ​​are taken, and the yield of alloying elements is gradually converged by weighted updates of the yield.

[0058] S104. Cyclic Optimization: After each smelting cycle is completed, repeat the calculation, judgment, and correction process of S102~103 above to dynamically optimize the yield of each alloy element in each cycle, so that the yield of each alloy element in the database continuously approaches the actual value.

[0059] Compared to steel production, where factors such as alloy composition and furnace conditions are constantly changing, traditional fixed yields or simple comparisons cannot adapt to these fluctuations, leading to frequent deviations in batching from the target. The aforementioned dynamic yield correction can capture changes in real time, ensuring that the alloy yield matches actual production. In the long run, the database can effectively improve the accuracy of alloy element yield through continuous iteration. As the accuracy improves, alloy proportion calculations can more accurately match actual needs, thereby improving the accuracy of subsequent alloy proportion calculations. At the same time, it can reduce or even avoid secondary feeding, adapting to the smelting cycle, reducing costs, and ensuring the stability of molten steel composition. Statistics show that the calculation error of alloy element yield has been reduced from the traditional ±5% to within ±2%, and the first-time compliance rate of molten steel composition adjustment has increased to over 96%.

[0060] Step S2, Initial Screening using Genetic Algorithm: Addressing the issue of strong dependence on the initial solution in the simplex method, a genetic algorithm is used to screen multiple sets of high-quality initial feasible solutions globally, based on dynamic correction of the yield rate. This lays the foundation for subsequent refinement. Figure 3 As shown, the method includes the following steps:

[0061] S201: Obtain alloy data, including the maximum on-site inventory of each alloy and steel composition standards; for example, alloy data includes: Types of alloys, types and contents of alloying elements in each alloy, unit price of each alloy, and the first alloy. Maximum on-site inventory of a certain alloy; steel composition standards include the first [item] in the steel composition standards. The lower limit of control for alloying elements, the first in the steel composition standard The lower limit for controlling certain alloying elements.

[0062] Definition of decision variables and objective function: The decision variables are the amount of each alloy added, and the objective function is to minimize the total cost of the alloy calculated based on the amount of each alloy added.

[0063] For example, with The first alloy Amount of alloy added ( Let the decision variables be (e.g., 1, 2, ..., n); and the objective function be to minimize the total cost of the alloy. The objective function is: (3), in equation (3) For total cost, for The first alloy The amount of each alloy added, for The first alloy Unit price of a certain alloy.

[0064] S202. Constraint setting, including:

[0065] Steel composition constraints: The steel composition is calculated based on the amount of each alloy added and the yield of each alloy element from the most recent heat in the database, ensuring that the steel composition meets the steel grade composition standard. For example: by obtaining the steel mass, the initial content of each element in the steel before adding the alloy, and extracting the yield of each alloy element from the most recent heat in the database, then:

[0066] (4), in equation (4) The first in the steel composition standard The lower limit of control for alloying elements, The first step before alloying in molten steel The content of various alloying elements, for The first alloy The amount of each alloy added, For the first The first alloy The content of each element For the most recent batch in the database The yield of alloying elements For the quality of molten steel, The first in the steel composition standard The upper limit of the control of alloying elements; the constraint of steel composition ensures that the final content of each element in the steel is within the qualified range, so as to guarantee the quality of steel.

[0067] Inventory constraints: The amount of each alloy added must be less than or equal to the maximum on-site inventory level, for example: ,in For the first The maximum on-site inventory of each alloy is determined by inventory constraints, which take into account the actual inventory limitations of the alloy in production, ensuring the feasibility of the alloy proportioning scheme and production plan.

[0068] S203: Construct a genetic algorithm model, generate several initial individuals that satisfy the constraints through real-number encoding as an initial population, use the reciprocal of the objective function as the fitness function, and iterate the initial population through selection, crossover, and mutation operations several times. Combined with the fitness function, output several initial feasible solutions with the optimal total cost of the alloy. The selection operation adopts the roulette wheel method. The crossover operation includes single-point crossover of selected individuals with a crossover probability of 0.7-0.8. The mutation operation includes random mutation of a certain alloy addition amount in selected individuals with a mutation probability of 0.05-0.1, and the mutation amplitude is ≤10%. Set a population diversity protection mechanism: including setting the number of iterations and the change rate threshold. When the change rate of the optimal fitness value in consecutive iterations is less than the change rate threshold, increase the mutation probability and regenerate some individuals.

[0069] For example, the execution flow of a genetic algorithm includes:

[0070] S2031. Population initialization: Using real number encoding, each individual corresponds to a set of alloy addition amounts. An initial population of 50-80 individuals is generated as the initial population. Each initial individual is ensured to satisfy the constraints of step S202, providing a candidate pool of initial feasible solutions for the algorithm.

[0071] S2032. Fitness function calculation: The reciprocal of the objective function in step S201 is used as the fitness function. The higher the fitness value, the lower the total cost of the corresponding alloy ratio scheme, so as to quantify the quality of each individual alloy ratio scheme.

[0072] S2033 Selection Operation: Using the roulette wheel method, the selection probability is determined according to the proportion of individual fitness values. 30-40 individuals with higher fitness values ​​are selected to enter the next generation. This is to screen out the better individuals in the current population, so that alloy ratio schemes with higher fitness, i.e. lower total alloy cost, have a greater probability of being retained.

[0073] S2034. Crossover Operation: Perform single-point crossover on the selected individuals after the selection operation with a crossover probability of 0.7-0.8. For example, individual A... With individual B In the A new individual is generated after the intersection of the positions. By combining the advantages of different components through cross-operation, a better alloy ratio scheme can be generated.

[0074] S2035. Mutation Operation: Randomly adjust the amount of a certain alloy added in the individual from step S2034 with a mutation probability of 0.05-0.1, where the mutation range is ≤10%. For example, ... Adjusted to This leads to new changes, preventing the algorithm from getting stuck in local optima, avoiding the search for solutions in a small range and missing out on globally better solutions, while ensuring that the constraints are still met after the adjustment.

[0075] S2036, Iteration Termination and Screening: Set the population diversity protection mechanism to 15 iterations and the change rate threshold to 0.5%. Repeat steps S2032 to S2035 for 100-200 iterations. When the change rate of the optimal fitness value is <0.5% for the first 15 consecutive iterations, increase the mutation probability to 0.12-0.15 and regenerate some individuals for iteration to avoid the algorithm getting trapped in local optima. When the change rate of the optimal fitness value is <0.5% for the second 15 consecutive iterations, stop the iteration. Screen 5-8 groups of individuals with the highest fitness values ​​from the final population. That is, quickly screen and output several groups of high-quality alloy ratio schemes that meet the constraints and have the optimal total alloy cost from a large number of candidate schemes, so as to serve as the initial feasible solution of the simplex method.

[0076] Statistics show that when alloy inventory fluctuations are ≤30% or steel composition standards are adjusted, the genetic algorithm can re-screen the initial solution within 1 minute, ensuring that the solution can quickly adapt to production changes.

[0077] S3. Simplex Method Refinement: Based on multiple initial feasible solutions selected by the genetic algorithm, the simplex method is used to achieve local precise optimization, obtaining the final optimal alloy ratio scheme, such as... Figure 4 As shown, the method includes the following steps:

[0078] S301. Standard Form Transformation of Constraints: By introducing slack variables, the constraints are transformed into standard form, so that the linear programming problem can be transformed into a standard form that can be handled by the simplex method. For example:

[0079] Transform the constraints in step S202 into equality constraints. (5), where A is the constraint coefficient matrix. For a group of alloy additions, As slack variables, For variable vectors, This is the constraint limit vector.

[0080] S302. Determination of Initial Basic Feasible Solutions: Combining the constraints of the standard form, determine the initial basic feasible solutions for each set of initial feasible solutions to obtain the initial basic feasible solutions and their corresponding objective function values. For example:

[0081] Each initial feasible solution output by the genetic algorithm in step S2036 is used as the initial basic feasible solution of the simplex method. The coefficients of the corresponding basic variables are extracted from the constraint coefficient matrix A to form a feasible basis B. The left side of both sides of the constraint equation is multiplied by the inverse matrix of the feasible basis. Setting the non-basic variables to 0, we obtain the initial basic feasible solution and the corresponding objective function value. This provides a starting point for iterative optimization using the simplex method.

[0082] S302. Optimality Test: Calculate the test number based on the initial basic feasible solution and the corresponding objective function value. This is used to determine whether the current basic feasible solution is the optimal solution. If all test numbers satisfy the optimality condition (all test numbers are non-negative), then the current solution is the optimal solution; otherwise, iterative improvement is needed. For example:

[0083] According to the formula (6) Calculate the test number In equation (6) For the first The cost of each variable, i.e., the cost of the corresponding alloy. The objective function coefficient vector of the basic variables. Let be the inverse matrix of a feasible basis. For the constraint coefficient matrix, the first Column; if all test numbers If the current basic feasible solution is the optimal solution, then the test number exists. If so, a basis transformation is required.

[0084] S303, Basis Transformation Operation: For step S302, there exists a test number. The basic feasible solution is to select the non-basic variable with the smallest test number as the replacement variable, according to the θ rule: (7) Determine the variable to be replaced, in equation (7) Let be the inverse matrix of a feasible basis. For the constraint limit vector, For the constraint coefficient matrix, the first List, The first of the current basic feasible solutions The values ​​of the basic variables, for At the base The coordinate components are below; replace the corresponding columns in feasible basis B, recalculate the test numbers, and repeat until all test numbers are obtained. .

[0085] S304. Optimal Solution Screening: For the 5-8 initial feasible solutions output by the genetic algorithm in step S2036, the simplex method refinement process described above is executed according to steps S301-303. If a certain initial feasible solution does not meet the constraints, it is automatically fed back to the genetic algorithm model. After adjusting the crossover probability and mutation probability, the initial feasible solutions are re-screened and output to obtain the lowest total alloy cost corresponding to each initial feasible solution. Based on the lowest total alloy cost and / or the balance of alloy addition among all groups, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme, and smelting is performed accordingly. After each smelting is completed, the calculation, judgment, and correction process of S102-103 is repeated to dynamically optimize the yield of each alloy element in each smelting.

[0086] For example, the alloy addition scheme with the lowest total cost among all groups is selected as the optimal alloy ratio scheme.

[0087] Alternatively, for example, the appropriate alloy addition scheme can be selected based on the balance between the total alloy cost and the alloy addition amount across all groups:

[0088] A cost difference threshold of 0.3% is set. The total alloy cost of all groups is compared. If the total cost difference of multiple groups including the lowest total alloy cost is less than the cost difference threshold of 0.3%, the alloy addition scheme with the smallest ratio of the maximum alloy addition amount to the minimum alloy addition amount among the multiple groups is selected first. For example, if the ratio of the maximum alloy addition amount to the minimum alloy addition amount is less than 4, it indicates that the alloy addition amount is more balanced, which reduces the difficulty of production operation, such as avoiding the need for a large amount of one alloy to be added and only a very small amount of another, thus reducing the complexity of batching and adding materials. If the total cost difference of multiple groups including the lowest total alloy cost is greater than or equal to the cost difference threshold of 0.3%, the alloy addition scheme with the lowest total alloy cost among all groups is selected as the optimal alloy ratio scheme.

[0089] Compared to existing genetic algorithms that rely on numerous iterations to approach the optimal solution, which may require readjusting algorithm parameters or even redesigning the algorithm process when faced with changes in production conditions, resulting in slow response or the search stopping at a local optimum, the simplex method refinement mentioned above, as the second step in the collaborative solution of the two algorithms, starts with a high-quality initial feasible solution provided by the genetic algorithm. Through steps such as constraint transformation, determining the initial basic feasible solution, optimality testing, and basis transformation operations, it accurately searches within a local range. The quality of the solution can be judged by the test number, and a better solution can be obtained through iterative basis transformation. Finally, from multiple sets of optimal solutions, the final alloy ratio scheme can be selected by combining the total alloy cost and / or the amount of alloy added, ultimately obtaining the optimal alloy ratio scheme that satisfies all constraints and has the lowest cost or better production convenience.

[0090] The above method was applied to optimize the alloy ratio in LF furnace smelting of different steel grades. The alloy cost per ton of steel was calculated in the three months before and after the application. The average cost per ton of steel was 325,000 tons for each steel grade. The cost reduction per ton of alloy steel achieved by applying the above method to different steel grades is as follows:

[0091] Table 1. Cost reduction results per ton of steel for different steel grades

[0092]

[0093] It is evident that the above method overcomes the limitations of single algorithms in global optimization and local refinement, enabling the global optimal solution for alloy combinations. Statistical analysis shows that the alloy cost reduction is 8%-12% higher than traditional methods, and the first-time compliance rate for composition adjustments exceeds 96%. This ensures that alloy additions meet steel composition standards and inventory requirements while minimizing total alloy costs, thereby improving the alloy proportion optimization effect. It ensures that the solution can quickly adapt to production changes and effectively reduce costs, providing a more efficient and accurate calculation path for alloy feeding control in steel smelting, such as LF furnaces.

[0094] Example 2:

[0095] Based on the same inventive concept, the preferred embodiment of the steel defect periodicity determination device of the present invention corresponds to the steel smelting alloy ratio optimization method described in Example 1, including:

[0096] The acquisition module is used to acquire alloy data, including the maximum on-site inventory of each alloy and steel composition standards.

[0097] A dual-algorithm collaborative solution module connects to the acquisition module and is used to perform genetic algorithm initial screening and simplex method refinement. The genetic algorithm initial screening includes using the amount of each alloy added as the decision variable and the objective function being the lowest total alloy cost calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial populations that satisfy the constraints through real-number encoding. The fitness function is the reciprocal of the objective function. The initial populations undergo several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total alloy cost. The simplex method refinement includes using the simplex method to perform local exact optimization on each group of initial feasible solutions to obtain the lowest total alloy cost corresponding to each group of initial feasible solutions. Based on the lowest total alloy cost and / or the balance of alloy additions among all groups, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme.

[0098] To ensure both the accuracy of the yield and the combination of global optimization and local refinement, the acquisition module is further used to collect production data for the current furnace batch. The device also includes:

[0099] A database is used to establish and store the yield of each alloy element in each furnace.

[0100] The yield dynamic correction module is used to calculate the actual yield of the target alloying element based on the current furnace production data acquired by the acquisition module; extract the target alloying element yield of the previous furnace from the database and calculate its deviation from the actual yield of the target alloying element; set a deviation threshold; if the obtained deviation > the deviation threshold, the target alloying element yield of the current furnace in the database will not be updated this time; if the deviation ≤ the deviation threshold, it will be adjusted accordingly. Correction, by The recovery rates of each alloy element in the current furnace are updated in the database accordingly, and this process is repeated to dynamically optimize the recovery rates of each alloy element in each furnace.

[0101] The dual-algorithm collaborative solution module is also used to calculate the steel composition based on the amount of each alloy added and the yield of each alloy element in the most recent furnace in the database when setting constraints. In this way, the accuracy of the basic data is ensured by the yield dynamic correction module, and the optimal solution of the alloy combination is achieved by the dual-algorithm collaborative solution module.

[0102] Example 3:

[0103] Based on the same inventive concept, a preferred embodiment of the electronic device of the present invention is provided. The electronic device may vary considerably due to different configurations or performance. It includes at least one processor and a CPU memory, which can be located in a terminal. The memory stores at least one instruction. When at least one instruction is loaded and executed by at least one processor, the electronic device implements the steel smelting alloy ratio optimization method as described in Example 1.

[0104] Example 4:

[0105] Based on the same inventive concept, a preferred embodiment of the computer-readable storage medium of the present invention is provided, wherein the storage medium stores at least one instruction, which, when loaded and executed by the processor of a computer, enables the computer to implement the steel smelting alloy ratio optimization method as described in Example 1.

[0106] The detailed descriptions listed above are merely specific illustrations of feasible embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including ROM, RAM, SD card, magnetic tape, floppy disk, and optical data storage devices. In this technical solution, various physical quantities in the formula, such as the mass of molten steel and the amount of alloy added, are flexibly set by those skilled in the art according to actual production needs, as long as the internal logic of the formula is self-consistent. These are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention. It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Claims

1. A method for optimizing alloy proportions in iron and steel smelting, characterized in that, The methods include: Obtain alloy data, including the maximum on-site inventory of each alloy, and steel composition standards; The initial screening using a genetic algorithm includes using the amount of each alloy added as the decision variable and the objective function being the lowest total cost of the alloy calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial individuals that satisfy the constraints through real-number encoding as an initial population. The fitness function is the reciprocal of the objective function. The initial population undergoes several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total cost of the alloy. The simplex method refinement includes using the simplex method to perform local exact optimization on each set of initial feasible solutions to obtain the lowest total alloy cost corresponding to each set of initial feasible solutions. Based on the lowest total alloy cost among all sets and / or the balance of alloy addition, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme.

2. The method for optimizing alloy proportions in iron and steel smelting according to claim 1, characterized in that, The method also includes dynamic yield correction, which includes: Establish a database to store the yield of each alloy element in each heat. Collect current furnace production data and calculate the actual yield of the target alloying element accordingly; Extract the target alloy element yield from the previous batch from the database and calculate the deviation between the yield and the actual yield of the target alloy element. If a deviation threshold is set, and the obtained deviation is greater than the deviation threshold, the target alloy element yield for the current furnace in the database will not be updated; if the deviation is less than or equal to the deviation threshold, then... Correction, in the formula The corrected target alloy element yield. For correction factor, The target alloy element yield of the previous batch. The actual yield of the target alloying element, calculated according to... The corresponding recovery rate of each alloy element in the current furnace is updated in the database, and this process is repeated to dynamically optimize the recovery rate of each alloy element in each furnace. When setting constraints, the steel composition is calculated based on the amount of each alloy added and the yield of each alloy element in the most recent furnace from the database.

3. The method for optimizing alloy proportions in iron and steel smelting according to claim 2, characterized in that, Actual recovery rate of the target alloying element The calculation formulas include: In the formula For the quality of molten steel, The content of the target element in the molten steel after the addition of the alloy. The target alloy element content in the molten steel before alloying. For the current furnace batch The first alloy The amount of each alloy added, for The first alloy The content of the target alloying element in the alloy; from the initial stage of production to the stable production stage, the deviation threshold gradually decreases.

4. The method for optimizing alloy proportions in iron and steel smelting according to claim 1, characterized in that, The selection operation employs a roulette wheel method; the crossover operation includes performing single-point crossover on the selected individuals with a crossover probability of 0.7-0.8; the mutation operation includes randomly mutating the amount of a certain alloy added to the selected individuals with a mutation probability of 0.05-0.1, with a mutation range ≤10%. During the initial screening process of the genetic algorithm, a population diversity protection mechanism is set up, including setting the number of iterations and a change rate threshold. When the change rate of the optimal fitness value in the first consecutive iteration is less than the change rate threshold, the mutation probability is increased and some individuals are regenerated. When the change rate of the optimal fitness value in the second consecutive iteration is less than the change rate threshold, the iteration stops.

5. The method for optimizing alloy proportions in iron and steel smelting according to claim 1, characterized in that, The local exact optimization of each set of initial feasible solutions using the simplex method includes: By introducing slack variables, the constraints are transformed into standard form. Combining the standard form constraints, the initial basic feasible solution is determined for each set of initial feasible solutions to obtain the initial basic feasible solution and the corresponding objective function value, and the test number is calculated accordingly. Perform an optimality test: if all test numbers are ≥0, then the current basic feasible solution is the optimal solution; if there is a test number <0, then select the non-basic variable with the smallest test number as the input variable, determine the output variable according to the θ rule, perform a basis transformation, recalculate the test numbers, until all test numbers are ≥0, and obtain the lowest total alloy cost corresponding to each initial feasible solution.

6. The method for optimizing alloy proportions in iron and steel smelting according to claim 1, characterized in that, During the refinement of the simplex method, if a set of initial feasible solutions does not meet the constraints, it is automatically fed back to the genetic algorithm model, and the crossover probability and mutation probability are adjusted before the initial feasible solutions are re-selected and output.

7. The method for optimizing alloy proportions in iron and steel smelting according to claim 1, characterized in that, This includes selecting the appropriate alloy addition scheme based on the total alloy cost and the balance of alloy addition across all groups: Set a cost difference threshold, compare the total cost of alloys in all groups, and if the total cost difference of multiple alloy groups including the lowest total cost of alloys is less than the cost difference threshold, then prioritize the alloy addition scheme with the smallest ratio of the maximum alloy addition amount to the minimum alloy addition amount among the multiple alloy groups.

8. A device for optimizing alloy proportions in steel smelting, characterized in that, include: The acquisition module is used to acquire alloy data, including the maximum on-site inventory of each alloy and steel composition standards. A dual-algorithm collaborative solution module connects to the acquisition module and is used to perform genetic algorithm initial screening and simplex method refinement. The genetic algorithm initial screening includes using the amount of each alloy added as the decision variable and the objective function being the lowest total alloy cost calculated based on the amount of each alloy added. Constraints are set, including that the steel composition calculated based on the amount of each alloy added meets the steel grade composition standard, and the amount of each alloy added is less than or equal to the maximum on-site inventory. A genetic algorithm model is constructed, generating several initial populations that satisfy the constraints through real-number encoding. The fitness function is the reciprocal of the objective function. The initial populations undergo several iterations of selection, crossover, and mutation operations, and the fitness function is used to filter and output several initial feasible solutions with the optimal total alloy cost. The simplex method refinement includes using the simplex method to perform local exact optimization on each group of initial feasible solutions to obtain the lowest total alloy cost corresponding to each group of initial feasible solutions. Based on the lowest total alloy cost and / or the balance of alloy additions among all groups, the corresponding alloy addition scheme is selected as the optimal alloy ratio scheme.

9. An electronic device, characterized in that, The electronic device includes at least one processor and a memory; the memory stores at least one instruction, and when the at least one instruction is loaded and executed by at least one processor, the electronic device implements the steel smelting alloy proportion optimization method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, which, when loaded and executed by the computer's processor, enables the computer to implement the steel smelting alloy proportion optimization method as described in any one of claims 1 to 7.

Images