Smelting burdening scheme generation method based on genetic algorithm and application

Abstract

The invention discloses a smelting ingredient generation method based on a genetic algorithm and application. According to the method, a smelting burdening scheme which meets user requirements and is lowest in cost is obtained through optimization through an improved genetic algorithm according to the smelting target weight, the smelting element target proportion and the upper and lower limits of the smelting element target proportion. The method comprises the following steps: firstly, selecting a material with the highest purity corresponding to each element as an initial parent, and taking each kilogram of material as a gene; secondly, performing gene replacement for more than 10000 times in each round of iteration to generate a new offspring; 10 filial generations which have the optimal genetic algorithm fitness function score and conform to the smelting element target proportion are reserved in the filial generations, and the filial generations serve as new parents to be optimized. The process is repeated for more than 1000 rounds, and finally a smelting burdening scheme with the lowest price is obtained.

Description

Technical Field

[0001] The invention relates to a method and application of generating a smelting batching scheme based on a genetic algorithm, which is applicable to industries such as cast iron, cast steel and metallurgy, and belongs to the technical field of batching scheme generation. Background Art

[0002] In industries such as cast iron, cast steel, and metallurgy, calculating the batching scheme is an important link, which is directly related to the performance, quality, and production efficiency of castings. During the smelting process, the precise proportion of the smelting ingredients not only affects the chemical composition, physical properties, and microstructure of the metal, but also determines the mechanical strength, corrosion resistance, and service life of the casting. Therefore, optimizing the smelting batching generation method is of great significance for improving the quality of castings, reducing the scrap rate, improving production efficiency, and saving energy and reducing emissions.

[0003] The traditional method of generating smelting ingredients mainly relies on the experience and judgment of engineers and operators, and achieves the expected smelting effect through repeated trial and error and adjustment of raw material ratios. This method is not only time-consuming and labor-intensive, but also difficult to ensure the consistency and stability of each batch, which can easily lead to problems such as large fluctuations in casting quality and unstable performance. Therefore, the traditional empirical batch generation method has been difficult to meet the production needs of high-precision and high-performance castings. The other type mainly studies the relationship between melting temperature and product quality, monitors the melting temperature in real time, and adjusts it in time when the temperature is abnormal. However, this method cannot quickly make the temperature reach the target melting temperature, resulting in large temperature adjustment errors. At the same time, it is impossible to optimize the raw material ratio from the root, and it is difficult to effectively reduce costs. Summary of the invention

[0004] Purpose of the invention: In view of the above technical problems, the first purpose of the present invention is to provide a method for generating a smelting batching scheme based on a genetic algorithm. The second purpose of the present invention is to apply the method for generating a smelting batching scheme based on a genetic algorithm in cast iron, cast steel or metallurgy.

[0005] Technical solution: The method for generating a smelting batching scheme based on a genetic algorithm of the present invention is characterized by comprising the following steps:

[0006] (1) Input the target weight of smelting and the target smelting element composition ratio triplet;

[0007] (2) Setting up a material database, with each kilogram of material in the database as a gene, and all genes forming a gene pool;

[0008] (3) The material with the highest purity corresponding to the target smelting element is used as the initial parent material of the genetic algorithm, and other genes in the gene pool are randomly replaced and optimized according to the fitness function to obtain offspring, and the 10 offspring with the best genetic algorithm fitness function scores are retained;

[0009] (4) The 10 retained offspring are used as new parent generations, and each parent generation is randomly replaced and optimized according to step (3). The optimal offspring in the last round is the final smelting batching scheme.

[0010] Furthermore, in step (1), the target smelting element composition ratio triplet is: smelting element target ratio (mid), smelting element ratio upper limit (max), smelting element ratio lower limit (min), wherein smelting element target ratio = (smelting element ratio upper limit + smelting element ratio lower limit) / 2, that is, mid = (max+min) / 2.

[0011] Furthermore, in step (2), the background material database includes five attributes: "material name", "material inventory weight", "material batch", "element composition ratio" and "material unit price".

[0012] Furthermore, in step (2), each kilogram of material in the database is taken as a gene, each gene contains the five attributes, and all the data in the database form a gene pool of the genetic algorithm.

[0013] Furthermore, in step (3), the target smelting elements are divided into common elements and uncommon trace elements.

[0014] Furthermore, the commonly used elements may be C, Si, Mn, Fe, Cr, Mo, Ni, Cu and / or N.

[0015] Furthermore, uncommon trace elements include P, S, Pb and the like.

[0016] Furthermore, the initial parent substitute materials corresponding to the commonly used elements are relatively fixed, among which the initial parent substitute material corresponding to the C element is the carburizer, the initial parent substitute material corresponding to the Si element is crystalline silicon, the initial parent substitute material corresponding to the Mn element is electrolytic manganese, the initial parent substitute material corresponding to the Fe element is pure iron, the initial parent substitute material corresponding to the Cr element is metallic chromium, the initial parent substitute material corresponding to the Mo element is ferromolybdenum, the initial parent substitute material corresponding to the Ni element is electrolytic nickel, the initial parent substitute material corresponding to the Cu element is electrolytic copper, and the initial parent substitute material corresponding to the N element is ferrochromium nitride. It should be noted that Fe is the main element in all smelting tasks, and almost all alloys contain Fe, but the user does not need to specifically specify the target proportion of Fe, because except for the specified elements, the remaining part can be Fe by default.

[0017] Furthermore, trace elements such as P, S, and Pb are very common trace elements. They are contained in most alloys containing multiple elements, and the proportion of the elements is very trace. The lower limit of the target proportion is usually 0, so these elements do not need to specify a special initial parent material as the initial parent. Other alloys containing multiple elements will bring these trace elements into the plan when they are replaced. If special circumstances require, the material containing the corresponding element and the highest purity will be selected in the database as the initial parent material. Different databases will be different.

[0018] Furthermore, in step (3), the purity corresponding to the target smelting element is the mass percentage of the target smelting element in all elements of the material after analysis by a spectrometer.

[0019] Furthermore, in step (3), if the initial parent generation weight does not meet the input target weight, pure iron is used as a supplementary material to make the initial parent generation reach the target weight.

[0020] Furthermore, in step (3), the fitness function is as follows:

[0021]

[0022] Among them, x i Indicates the weight of material i in the batching scheme, X i Indicates the inventory weight of material i, p i represents the price of material i, a ij Indicates the composition ratio of the jth smelting element of material i, N j Indicates the lower limit of the proportion of smelting element j, M j Indicates the upper limit of the proportion of smelting element j.

[0023] Furthermore, when the fitness function value of the generated batching scheme is -1, it means that the scheme is deleted.

[0024] Furthermore, in step (3), random replacement optimization is performed more than 10,000 times.

[0025] Furthermore, in step (3), the uncommon trace elements are included in most alloys containing multiple elements, and the uncommon trace elements account for a trace amount. When the lower limit of the target proportion is 0, the uncommon trace elements do not specify a specific material as the initial parent generation.

[0026] Furthermore, in step (4), random replacement optimization is performed more than 1000 times.

[0027] The invention discloses an application of the method for generating a smelting batching scheme based on a genetic algorithm in cast iron, cast steel or metallurgy.

[0028] The method of the present invention optimizes the smelting batching scheme that meets user needs and has the lowest cost according to the target weight of smelting, the target proportion of smelting elements and their upper and lower limits through an improved genetic algorithm. First, the material with the highest purity corresponding to each element is selected as the initial parent generation, and each kilogram of material is used as a gene. Secondly, more than 10,000 gene replacements are performed in each round of iteration to generate new offspring. The 10 offspring with the best genetic algorithm fitness function scores that meet the target proportion of smelting elements are retained and optimized as new parents. This process is repeated for more than 1,000 rounds, and finally the smelting batching scheme with the lowest price is obtained.

[0029] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0030] (1) The present invention inputs the target weight of smelting, the target proportion of smelting elements, and the upper and lower limits of the target proportion of smelting elements. The upper and lower limits method can make the genetic algorithm have sufficient optimization space.

[0031] (2) The technical solution adopted by the method of the present invention forms a gene pool of a genetic algorithm with the data in the material database, wherein each kilogram of material in the gene pool is used as a gene, and each gene contains 5 material properties. Secondly, the material with the highest purity corresponding to the target smelting element is used as the initial parent material. The definition of the initial parent material can ensure that the initial parent that meets the element ratio requirements will be obtained. After multiple rounds of iterations, the final batching scheme is obtained and returned to the user for display.

[0032] (3) This method can automatically generate a low-cost smelting batching plan and accurately control the smelting component ratio and target quality, which not only saves time and cost but also achieves the goal of lean production. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 4 is a flow chart of a genetic algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0035] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0036] Example 1

[0037] like Figure 1 As shown, the present invention provides a method for generating a smelting batching scheme based on a genetic algorithm, comprising the following steps:

[0038] S1. Input the target weight of molten steel of 4100 kg and the target smelting element composition ratio triplet: smelting element target ratio (mid), upper limit of each smelting element ratio (max), lower limit of each smelting element ratio (min), where mid = (max + min) / 2, and the target elements are C, Mn, P, Cr, Ni, Mo, Si (Fe is the main element in all smelting tasks, and almost all alloys contain Fe. Users do not need to specify the target ratio of Fe, because except for the specified elements, the rest can be Fe by default). The target smelting element composition ratio (the mass percentage of all elements in the material occupied by the target element) triplet table is shown in Table 1.

[0039] Table 1 Target smelting element composition ratio triple table

[0040] project C Mn P Cr Ni Mo Si Lower limit value (%) 0 0.5 0 22.5 4.4 1.6 0.19 Upper limit value (%) 0.01 1.1 0.05 30.1 5.6 2.1 0.51 Target value (%) 0.05 0.8 0.0025 26.3 5.0 1.85 0.35

[0041] S2. Set up the material database, including 5 attributes: "Material Name", "Material Inventory", "Material Batch", "Material Unit Price (Yuan)" and "Element Composition Ratio (Mass Percentage, %)". In this database, the mass is set to 8000Kg by default, and users can change it according to actual conditions. The details are shown in Table 2 below:

[0042]

[0043]

[0044] S3, the materials with the highest percentage corresponding to C, Si, Mn, Fe, Cr, Mo, Ni, Cu and N elements are used as the initial parent materials of the genetic algorithm, wherein the initial parent material corresponding to the C element is a carburizer, the initial parent material corresponding to the Si element is crystalline silicon, the initial parent material corresponding to the Mn element is electrolytic manganese, the initial parent material corresponding to the Cr element is metallic chromium, the initial parent material corresponding to the Mo element is ferromolybdenum, and the initial parent material corresponding to the Ni element is electrolytic nickel. The remaining trace elements such as P, S, Pb are very useful elements. They are contained in most alloys containing multiple elements, and the element percentage is very trace. The lower limit of the target percentage is usually 0, so these elements do not need to specify a special initial parent material as the initial parent. When other alloys containing multiple elements are replaced into the scheme, these trace elements will be brought in. If special circumstances require, the material containing the corresponding element and the highest purity is selected in the database as the initial parent material. Different databases will be different. For example, in the data provided in this embodiment, the initial parent material corresponding to P can select 11 steel scraps. Each kilogram of material in the material database is taken as a gene. Each gene contains the above five attributes. All the data in the database form the gene pool of the genetic algorithm. Each gene is randomly replaced and optimized with the genes in the gene pool, that is, 1-5 genes are randomly selected from the gene pool to replace the random 1-5 genes in the parent generation.

[0045] S4. Perform 10,000 random gene replacement optimizations on each parent generation to obtain 10,000 new offspring, and retain the 10 offspring with the best genetic algorithm fitness function scores. The fitness function used by the genetic algorithm is as follows:

[0046] f(x1,x2,x3,...,x n )

[0047]

[0048] Among them, x i Indicates the weight of material i in the batching scheme, X i Indicates the inventory weight of material i, p i represents the price of material i, a ij Indicates the composition ratio of the jth smelting element of material i, N j Indicates the lower limit of the proportion of smelting element j, M j Indicates the upper limit of the proportion of smelting element j. When the fitness function value of the generated batching scheme is -1, it means that the scheme is deleted.

[0049] S5. The 10 retained offspring are used as new parents. Each new parent is subjected to 1000 rounds of random replacement optimization according to the optimization steps of S4. The best offspring in the last round is used as the final batching scheme. The final batching scheme is shown in Table 3.

[0050] Table 3 Molten steel batching scheme based on genetic algorithm

[0051]

[0052]

[0053] Comparative Example 1

[0054] The materials used are the same as those in Example 1. In the prior art, experts obtain the formulation scheme based on their work experience. The specific steps are as follows:

[0055] 1. Determine the target: First determine the target weight of the smelting and the required element composition ratio.

[0056] 2. Select materials: Based on the target and available material inventory, select a range of materials that may be used for smelting from the material database.

[0057] 3. Preliminary calculation: Based on historical data and experience, make a preliminary estimate of the amount of each material required.

[0058] 4. Composition adjustment: According to the chemical composition ratio of the materials, adjust the amount of various materials to ensure that the chemical composition of the final smelting material meets the target requirements.

[0059] 5. Cost estimation: While adjusting the material proportions, consider the cost of the materials and try to find the lowest cost solution while meeting the ingredient requirements.

[0060] 6. Empirical judgment: Based on past experience, the initially calculated solution is fine-tuned to take into account factors that are difficult to quantify, such as material interactions during the smelting process, temperature control, etc.

[0061] 7. Solution optimization: After multiple rounds of adjustment and optimization, the best material combination is found.

[0062] 8. Final decision: After comprehensively considering ingredients, cost, production efficiency and risks, the final ingredient plan is determined.

[0063] The batching scheme generated based on expert experience is shown in Table 4.

[0064] Table 4 Molten steel batching scheme generated based on traditional expert experience

[0065]

[0066]

[0067] The weight, total price, unit price and solution calculation time of the molten steel batching scheme (Table 1) generated by the genetic algorithm proposed in Example 1 and the batching scheme (Table 2) obtained by experts based on experience in Comparative Example 1 are shown in Table 5:

[0068] Table 5 Comparison of molten steel batching schemes based on genetic algorithm and expert experience

[0069] method Total weight (kg) Total price (yuan) Unit price (kg / yuan) Solution calculation time Example 1 4100 33309.5 8.12 13s Comparative Example 1 4580.4 61502.68 13.43 20m

[0070] According to the comparison in Table 3, it can be concluded that the method for generating smelting ingredients based on genetic algorithm proposed in the present invention can generate a solution close to the target weight and at a lower price. At the same time, the calculation time of the method for generating smelting ingredients based on genetic algorithm of the present invention is much lower than the traditional expert-based experience estimation method in comparative example 1. Therefore, the method proposed in the present invention can improve production efficiency and effectively reduce production costs.

Claims

1. A method for generating a smelting batching scheme based on a genetic algorithm, characterized in that: The following steps are involved: (1) Input the target weight of smelting and the target smelting element composition ratio triplet; (2) Setting up a material database, with each kilogram of material in the database as a gene, and all genes forming a gene pool; (3) The material with the highest purity corresponding to each target smelting element is used as the initial parent material of the genetic algorithm, and other genes in the gene pool are randomly replaced and optimized according to the fitness function to obtain offspring, and the 10 offspring with the best genetic algorithm fitness function scores are retained; (4) The 10 retained offspring are used as new parent generations, and each parent generation is randomly replaced and optimized according to step (3). The optimal offspring in the last round is the final smelting batching scheme.

2. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 1, characterized in that: In step (1), the target smelting element composition ratio triplet is: smelting element target ratio, smelting element ratio upper limit, smelting element ratio lower limit, wherein smelting element target ratio = (smelting element ratio upper limit + smelting element ratio lower limit) / 2.

3. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 1, characterized in that: In step (2), the material database includes five attributes: "material name", "material inventory weight", "material batch", "element composition ratio" and "material unit price".

4. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 3, characterized in that: In step (2), each kilogram of material in the database is taken as a gene, each gene contains the five attributes, and all the data in the database form a gene pool of the genetic algorithm.

5. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 1, characterized in that: In step (3), if the initial parent generation weight does not meet the input target weight, pure iron is used as a supplementary material to make the initial parent generation reach the target weight.

6. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 1, characterized in that: In step (3), the fitness function is as follows: Among them, x i Indicates the weight of material i in the batching plan, X i Indicates the inventory weight of material i, p i represents the price of material i, a ij Indicates the composition ratio of the jth smelting element of material i, N j Indicates the lower limit of the proportion of smelting element j, M j Indicates the upper limit of the proportion of smelting element j.

7. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 6, characterized in that: When the fitness function value of the generated batching scheme is -1, the scheme is deleted.

8. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 1, characterized in that: In step (3), the target smelting elements are divided into common elements and uncommon trace elements, wherein the common elements are C, Si, Mn, Fe, Cr, Mo, Ni, Cu and / or N, and the uncommon trace elements include P, S, Pb. The random replacement optimization is performed more than 10,000 times. In step (4), the random replacement optimization is performed more than 1,000 times.

9. The method for generating a smelting batching scheme based on a genetic algorithm according to claim 8, characterized in that: In step (3), the uncommon trace elements are included in most alloys containing multiple elements, and the uncommon trace elements account for a trace amount. When the lower limit of the target proportion is 0, the uncommon trace elements do not specify a specific material as the initial parent generation.

10. Application of the method for generating a smelting batching scheme based on a genetic algorithm according to any one of claims 1 to 9 in cast iron, cast steel or metallurgy.

Images