Formula optimization method of carbon for aluminum, electronic equipment and medium

Abstract

The invention provides a formula optimization method of carbon for aluminum, electronic equipment and a medium, accurate three-dimensional shapes of particles in a carbon material for aluminum can be obtained by acquiring a CT image of the carbon for aluminum, and formula calculation is performed by combining the closest packing theory and the particle size distribution of actual raw materials, so that the formula optimization accuracy is improved. The calculation and optimization of the formula of the carbon for aluminum are realized, the scientificity and rationality of the formula of the carbon for aluminum are improved, and the trial and error cost is reduced.

Description

Technical Field

[0001] The invention belongs to the field of material design, and in particular relates to a formula optimization method for carbon for aluminum, electronic equipment and a medium. Background Art

[0002] Carbon for aluminum is the core component of aluminum electrolytic cells, and its quality directly affects the current efficiency and life of the electrolytic cells. The degree of internal defects in cathode carbon is one of the important factors affecting its quality, which determines many important properties of carbon electrodes, such as mechanical strength, electrical conductivity, and thermal conductivity. The shape and particle size ratio of the raw materials directly affect the defect rate of carbon products. By using a reasonable formula to make the particles inside the carbon block more closely contacted, better carbon block performance can be achieved.

[0003] At present, high-quality formulas of aluminum carbon are usually screened by bulk density experiments or experience-based methods. However, these methods are highly blind and lack theoretical guidance. There are problems such as complex raw material shapes, unclear stacking structures, and difficulty in optimization. It is difficult to select the appropriate amount and ratio of aluminum carbon raw materials, and the trial and error cost is high. Summary of the invention

[0004] The purpose of the present invention is to address the deficiencies in the prior art and to provide a method for optimizing the formula of carbon for aluminum, an electronic device and a medium, so as to achieve the optimization of the formula of carbon for aluminum and reduce the cost of trial and error.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A method for optimizing the formulation of aluminum carbon comprises the following steps:

[0007] S1: Obtain a CT image of carbon for aluminum; obtain the particle size distribution of each aggregate of carbon for aluminum;

[0008] S2: preprocessing the aluminum carbon CT image;

[0009] S3: using a threshold segmentation method to separate the aggregate phase image, the asphalt phase image, and the pore phase image from the pre-processed aluminum carbon CT image; extracting the fractal dimension of the aggregate phase image;

[0010] S4: construct an ideal particle size distribution model according to the fractal dimension and the aggregate particle size, and the ideal particle size distribution model is shown in the following formula:

[0011]

[0012] Where, φ(<D) is the percentage of aggregate particles smaller than D in the total number of particles, F is the fractal dimension, D is the particle size of the aggregate particles, and D min <D<D max , Dmin is the minimum particle size of aggregate particles, D max is the maximum particle size of aggregate particles;

[0013] S5: The particle swarm optimization algorithm is used to minimize the mean square error between the ideal particle size distribution and the mixed particle size distribution of each aggregate to obtain the optimized formula of carbon for aluminum.

[0014] The present invention can obtain the accurate three-dimensional shape of the internal particles of the carbon material for aluminum by collecting CT images of the carbon for aluminum, and perform formula calculation by combining the closest packing theory with the particle size distribution of the actual raw materials, thereby realizing the calculation and optimization of the carbon formula for aluminum, improving the scientificity and rationality of the carbon formula for aluminum, and reducing the cost of trial and error.

[0015] Furthermore, the implementation process of S2 includes:

[0016] The aluminum carbon CT image is subjected to shape cropping, noise reduction filtering and interface definition enhancement processing in sequence.

[0017] Reduce the errors caused by noise and artifacts in result analysis.

[0018] Further, the implementation process of extracting the fractal dimension of the aggregate phase image includes:

[0019] overlaying the aggregate phase image with a three-dimensional grid;

[0020] Divide the three-dimensional grid into a plurality of cubes, each of which has a side length of ε;

[0021] Gradually reduce the value of ε, each reduction is For each edge length value, count the number of cubes N(ε) that contain aggregate particles;

[0022] Calculate the fractal dimension. The expression of the fractal dimension F is as follows:

[0023]

[0024] Furthermore, the implementation process of S5 includes:

[0025] A1: Randomly generate the initial position and velocity of the particle swarm, and each particle represents a target recipe;

[0026] A2: Calculate the objective function value of each particle. The expression of the objective function MSE(w) is as follows:

[0027]

[0028] A3: Update the speed and position of the particle. If the objective function value of the current particle is less than the objective function value of its optimal position, update the optimal position of the particle; if the objective function value of the current particle is less than the objective function value of the global optimal position, update the global optimal position; the global optimal position is the position corresponding to the minimum value of the objective function values ​​of all particles;

[0029] A4: Repeat steps A2-A3, and iteratively optimize the carbon formula for aluminum using constraint conditions to obtain an optimized carbon formula for aluminum; the constraint condition is expressed as follows:

[0030] max(0,initial weights-f)≤w≤min(1,initial weights+f)

[0031] Among them, w is the target formula, X i，w is the particle size distribution of each level of aggregate when the target formula w is adopted, n is the dimension of the particle size distribution matrix of each level of aggregate, y is the ideal particle size distribution vector, f is the approximation factor, and initial weights is the benchmark formula.

[0032] Based on the same inventive concept, the present invention also provides an electronic device, including:

[0033] one or more processors;

[0034] A memory having one or more programs stored thereon, which, when executed by the one or more processors, enables the one or more processors to implement the steps of a method for optimizing the formulation of carbon for aluminum.

[0035] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for optimizing the formula of carbon for aluminum.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention can obtain the accurate three-dimensional shape of the internal particles of the carbon material for aluminum by collecting CT images of the carbon for aluminum, and perform formula calculation by combining the closest packing theory with the particle size distribution of the actual raw materials, thereby realizing the calculation and optimization of the carbon formula for aluminum, improving the scientificity and rationality of the carbon formula for aluminum, and reducing the cost of trial and error. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of a method for optimizing the formulation of carbon for aluminum according to the present invention;

[0039] Figure 2 is the particle size distribution diagram of each level of aggregate;

[0040] FIG3( a ) is a comparison of the resistivity of carbon for aluminum prepared by various formulations;

[0041] FIG3( b ) is a density comparison diagram of aluminum carbon prepared by various formulations;

[0042] FIG3(c) is a comparison of the flexural strength of aluminum carbon prepared by various formulations;

[0043] FIG3( d ) is a comparison of the compressive strength of aluminum carbon prepared by various formulations;

[0044] Figure 4 Comparison of CT images of carbon for aluminum preparation before and after optimization. DETAILED DESCRIPTION

[0045] The present invention will be described in detail below in conjunction with the embodiments. It should be noted that the embodiments and features of the embodiments of the present invention can be combined with each other without conflict. For the convenience of description, if the words "upper", "lower", "left" and "right" appear in the following, they only indicate that the upper, lower, left and right directions are consistent with the drawings themselves, and do not limit the structure.

[0046] Example

[0047] like Figure 1 The formula optimization method of aluminum carbon in this embodiment includes five steps: CT-based particle parameter acquisition, ideal particle size distribution calculation, raw material particle size distribution detection, target formula optimization, formula experiment and CT scanning verification.

[0048] First, a high-resolution CT scanning device was used to scan the carbon sample for aluminum to obtain the shape parameters of the carbon particles. The shape parameters reflect the shape complexity of the carbon aggregate and its occurrence state in the multi-aggregate-asphalt system.

[0049] Secondly, based on the acquired shape parameters, according to the fractal theory and the stacking relationship between large and small particles, it is possible to deduce the ideal particle size distribution for the most dense stacking of each level of raw materials under given shape conditions.

[0050] Subsequently, it is necessary to carry out particle size distribution tests on each level of raw materials. For this continuous particle size distribution system, it is necessary to obtain the relationship between the particle size and the number of particles within the maximum particle size and minimum particle size range.

[0051] Based on the established stacking model and particle size test data, the target formula is calculated by particle swarm optimization algorithm. The optimization goal is to minimize the gap between the particle size distribution of the mixed dry material and the ideal particle size distribution calculated by the ideal particle size distribution model.

[0052] Finally, the target formula is used to prepare samples according to the carbon preparation process, and the prepared carbon material samples are tested for physical and chemical indicators. At the same time, a CT scan is performed to obtain the particle accumulation condition inside the carbon material under the formula, and to obtain indicators such as defect rate and defect shape, so as to provide further feedback for formula optimization.

[0053] 1. CT-based particle parameter acquisition

[0054] First, a high-resolution X-ray CT device (including but not limited to Nano CT, Micro CT, etc.) is used to collect CT images of the carbon electrode material for aluminum. The carbon electrode material includes but is not limited to a prebaked anode for aluminum, a carbon cathode for aluminum, etc.

[0055] The acquired CT images are cropped to make the scan volume size within a reasonable range (2000mm 3 –10000mm 3 ). The image is then subjected to noise reduction filtering to reduce the error caused by noise points and artifacts in the result analysis. In this embodiment, a non-local mean filtering method is used to remove CT image noise, and the non-local mean filtering method is a prior art.

[0056] Carbon materials are composed of three phases: aggregate (petroleum coke, graphite, etc.), binder (asphalt), and pores. Since the aggregate and binder in the carbon material are fully mixed, and the binder is mixed with a large amount of powder, image enhancement means are used to reduce the interface tolerance effect. In this embodiment, the interface clarity is enhanced by the unsharp mask method, which is a prior art.

[0057] Furthermore, the three phases of aggregate, asphalt and pores in the carbon CT image are separated by a threshold segmentation method. In this embodiment, the threshold segmentation method used is a watershed algorithm, which is a prior art.

[0058] For the extracted aggregate phase, an image processing algorithm is used to extract its shape parameters. Specifically, the image processing algorithm used is the box counting method. The extracted shape parameter is the fractal dimension. The basic process of extracting the fractal dimension by the box counting method is: Use a three-dimensional grid to cover the aggregate phase CT image. Divide the three-dimensional grid into smaller cubes (boxes), the side length of these small cubes is ε, and the initial value is the side length of the aggregate phase CT image; gradually reduce the value of ε, and the reduction amplitude each time is For each box size, count the number of boxes containing aggregate particles N(ε); finally, calculate the fractal dimension F using the following formula:

[0059]

[0060] 2. Calculation of ideal particle size distribution

[0061] The derivation process of the ideal particle size distribution model is as follows: Assume that the maximum particle size in the aggregate system is D max , the minimum particle size is D m i n , any intermediate particle size is D, D m i n <D<D max ; The ratio of the size of adjacent smaller particles to that of larger particles is r (0<r<1); The ratio of the number of adjacent smaller particles to that of larger particles is k (0<k<1); k=r -F , F is the fractal dimension, which is obtained from the real structure of carbon materials using the box counting method in the previous step.

[0062] According to the assumption, the number of particles N(D) of any size D is:

[0063]

[0064] N max is the number of particles of the maximum size; the density function is defined as n(D), which represents the change in the number of particles per unit size interval:

[0065]

[0066] The cumulative number of particles N(<D) smaller than a certain size D is:

[0067]

[0068] Among them, D min <D′<D.

[0069] Calculating the integral of equation 4, we obtain:

[0070]

[0071] The cumulative volume V(<D) less than a certain size D can be expressed as:

[0072]

[0073] v(D') is the volume of a single particle, which depends on the size D' of a single particle. The final ideal particle size distribution model can be expressed by the following formula:

[0074]

[0075] Wherein, φ(<D) is the percentage of aggregate particles smaller than D in the total number of particles.

[0076] Assume that D1 and D2 are two adjacent size levels, and D1 < D2. Then the percentage P of the particles with particle sizes between D1 and D2 in the total number of particles is:

[0077] P = φ(<D2) - φ(<D1) (Equation 11)

[0078] 3. Detection of raw material particle size distribution

[0079] Perform particle size distribution tests on each type of aggregate particle required for the preparation of carbon materials to obtain its continuous particle size distribution characteristics. The aggregates of carbon materials include petroleum coke, graphite, etc. of each particle size grade.

[0080] Specifically, for each type of aggregate, use a vibrating sieve to obtain its particle size distribution from 1 mm to 16 mm; use a laser particle size analyzer to obtain its particle size distribution from 1 μm to 1 mm. According to the above method, the continuous particle size distribution of various aggregates from 1 to 16000 μm can be obtained. Any other equipment that can achieve the above test results can also be used.

[0081] 4. Optimization of the target formulation

[0082] Based on the establishment of the above ideal particle size distribution model and the particle size test results of each aggregate, the optimization calculation of the target ratio can be carried out. The principle of the calculation is to plan and adjust through intelligent algorithms to minimize the gap between the particle size distribution of the dry material after mixing various aggregates (the particle size distribution of the mixture when using the target ratio) and the ideal particle size distribution, thereby reducing the carbon block defects caused by the loose stacking of aggregates.

[0083] First, a particle size distribution matrix X of each level of aggregate should be constructed, which is obtained from the particle size test in the previous step; at the same time, a particle size distribution vector y also needs to be established, which is calculated from the ideal particle size distribution model:

[0084] X = [x 石墨L1 , x 石墨1-2 , x 石墨2-4 , x 石墨4-8 , x 石油焦L1 , x 石油焦1-2 , x 石油焦2-4 , x 石油焦4-8 , x 粉子 (Equation 12)

[0085]

[0086] Among them, x 石墨L1 , x 石墨1-2 , x 石墨2-4 , x 石墨4-8 , x 石油焦L1 , x 石油焦1-2 , x 石油焦2-4 , x 石油焦4-8They respectively represent the particle size distribution of graphite particles with a particle size less than 1 mm, the particle size distribution of graphite particles with a particle size between 1 mm and 2 mm, the particle size distribution of graphite particles with a particle size between 2 mm and 4 mm, the particle size distribution of graphite particles with a particle size between 4 mm and 8 mm, the particle size distribution of tar particles with a particle size less than 1 mm, the particle size distribution of tar particles with a particle size between 1 mm and 2 mm, the particle size distribution of tar particles with a particle size between 2 mm and 4 mm, and the particle size distribution of tar particles with a particle size between 4 mm and 8 mm, and x 粉子 is the particle size distribution of the powder particles, y j is the percentage of particles of the jth particle size in the total number of particles, and m is the number of particle sizes. The particle size distribution of each level of aggregate is as follows: Figure 2 .

[0087] Specifically, if there are 8 particle sizes from small to large, the percentage of particles in each particle size to the total number of particles can be calculated using equations 10 and 11, and the following ideal particle size distribution vector can be obtained:

[0088]

[0089] Then, the objective function of the optimization process is defined as minimizing the mean squared error (MSE):

[0090]

[0091] Among them, w is the target formula, X i，w is the particle size distribution of aggregate when the target formula w is adopted, y is the ideal particle size distribution vector, and n is the dimension of the particle size distribution matrix X of each level of aggregate.

[0092] The constraints that define the optimization process are as follows:

[0093] max(0,initial weights-f)≤w≤min(1,initial weights+f)

[0094] f is the approximation factor. The larger the approximation factor, the closer the target formula is to the ideal formula. Initial weights is the benchmark formula, that is, the formula before optimization. The benchmark formula is obtained from the bulk density experiment.

[0095] The above optimization process uses an intelligent algorithm to optimize the target formula. Specifically, a particle swarm optimization algorithm is used. The optimization steps of this method are:

[0096] 1) Randomly generate the initial position and velocity of the particle swarm, and each particle represents a target recipe.

[0097] 2) Calculate the fitness value of each particle, that is, the objective function value. If the fitness value of the current particle is better than the optimal position it has encountered before, update its optimal position; if the fitness value of the current particle is better than the global optimal position, update the global optimal position. In each iteration, the global optimal position is obtained by comparing the fitness values ​​of the individual optimal positions of all particles.

[0098] 3) Update speed v i and position x i Defined as:

[0099] v i (t+1)=w0v i (t)+c1r1(p i -x i (t))+c2r2(gx i (t)) (Equation 15)

[0100] x i (t+1)=x i (t)+v i (t+1) (Equation 16)

[0101] where p i is the individual optimal position of the particle, g is the global optimal position, w0 is the inertia weight, c1 and c2 are learning factors, and r1 and r2 are random numbers between 0 and 1.

[0102] Repeat steps 2)-3) until the global optimal individual is found or the maximum number of iterations is reached. The optimal individual obtained is the target formula.

[0103] Specifically, f is set to 0.1, 0.15, and 0.2, respectively, to obtain three sets of optimized target recipes, referred to as Recipe 2, Recipe 3, and Recipe 4, and the baseline recipe is referred to as Recipe 1.

[0104] 5. Formula experiment and CT scan verification

[0105] The optimized target formula is used to prepare the carbon material. Specifically, the preparation process includes but is not limited to batching, kneading, molding, roasting and other processes, and the above preparation process is repeated for formula 1, formula 2, formula 3, and formula 4 respectively. The preparation of the carbon material is a prior art.

[0106] The prepared carbon material is tested for physical and chemical indicators, specifically, including but not limited to the testing of bulk density, resistivity, compressive strength, and tensile strength; the results are as follows Figure 3(a) , 3(b) As shown in Figures 3(c) and 3(d), the volume density, resistivity, compressive strength and tensile strength of the carbon material made from the optimized target formula are better than those of the baseline formula before optimization.

[0107] Furthermore, CT images of the prepared carbon materials were collected again to analyze the differences in the internal structure of the carbon materials prepared by the optimized formula before and after optimization. This includes but is not limited to the comparison of particle stacking, porosity, and pore shape, and the proportion of each particle size in the formula was further fine-tuned based on the above data. The results are as follows: Figure 4 As shown, Figure 4 The left side shows the CT image of the carbon material prepared by formula 1, and the right side shows the CT image of the carbon material prepared by formula 2. The particle size distribution of the carbon material prepared by the optimized formula is more reasonable, and the aggregate and asphalt are mixed more evenly, so that the carbon block has fewer initial defects. A lower defect rate means better electrical and thermal conductivity, as well as stronger deformation resistance, which will further increase the service life of the carbon material.

[0108] This embodiment provides an electronic device, including:

[0109] one or more processors;

[0110] A memory having one or more programs stored thereon, which, when executed by one or more processors, enables the one or more processors to implement the steps of a method for optimizing the formulation of carbon for aluminum.

[0111] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.

[0112] In some other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors of various types, which are not limited herein.

[0113] This embodiment provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for optimizing the formulation of carbon for aluminum.

[0114] The contents explained in the above embodiments should be understood as these embodiments are only used to more clearly illustrate the present invention, and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

Claims

1. A method for optimizing the formulation of carbon for aluminum, characterized in that: The following steps are involved: S1: Obtain a CT image of carbon for aluminum; obtain the particle size distribution of each aggregate of carbon for aluminum; S2: preprocessing the aluminum carbon CT image; S3: using a threshold segmentation method to separate the aggregate phase image, the asphalt phase image, and the pore phase image from the pre-processed aluminum carbon CT image; extracting the fractal dimension of the aggregate phase image; S4: construct an ideal particle size distribution model according to the fractal dimension and the aggregate particle size, and the ideal particle size distribution model is shown in the following formula: Where, φ(<D) is the percentage of aggregate particles smaller than D in the total number of particles, F is the fractal dimension, D is the particle size of the aggregate particles, and D min <D<D max , D min is the minimum particle size of aggregate particles, D max is the maximum particle size of aggregate particles; S5: The particle swarm optimization algorithm is used to minimize the mean square error between the ideal particle size distribution and the mixed particle size distribution of each aggregate to obtain the optimized formula of carbon for aluminum.

2. The method for optimizing the formula of carbon for aluminum according to claim 1, characterized in that: The implementation process of S2 includes: The aluminum carbon CT image is subjected to shape cropping, noise reduction filtering and interface definition enhancement processing in sequence.

3. The method for optimizing the formula of carbon for aluminum according to claim 1, characterized in that: The implementation process of extracting the fractal dimension of the aggregate phase image includes: overlaying the aggregate phase image with a three-dimensional grid; Divide the three-dimensional grid into a plurality of cubes, each of which has a side length of ε; Gradually reduce the value of ε, each reduction is For each edge length value, count the number of cubes N(ε) that contain aggregate particles; Calculate the fractal dimension. The expression of the fractal dimension F is as follows:

4. The method for optimizing the formula of carbon for aluminum according to claim 1, characterized in that: The implementation process of S5 includes: A1: Randomly generate the initial position and velocity of the particle swarm, and each particle represents a target recipe; A2: Calculate the objective function value of each particle. The expression of the objective function MSE(w) is as follows: A3: Update the speed and position of the particle. If the objective function value of the current particle is less than the objective function value of its optimal position, update the optimal position of the particle; if the objective function value of the current particle is less than the objective function value of the global optimal position, update the global optimal position; the global optimal position is the position corresponding to the minimum value of the objective function values ​​of all particles; A4: Repeat steps A2-A3, and iteratively optimize the carbon formula for aluminum using constraint conditions to obtain an optimized carbon formula for aluminum; the constraint condition is expressed as follows: max(0,initialweights-f)≤w≤min(1,initialweights+f) Among them, w is the target formula, X i，w is the particle size distribution of each level of aggregate when the target formula w is adopted, n is the dimension of the particle size distribution matrix of each level of aggregate, y is the ideal particle size distribution vector, f is the approximation factor, and initial weights is the benchmark formula.

5. An electronic device, characterized in that: include: one or more processors; A memory having one or more programs stored thereon, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the steps of the method according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that: The computer program is stored therein, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images