A method for optimizing a formulation of carbon for aluminum, an electronic device, and a medium

Abstract

The application provides a formula optimization method for carbon for aluminum, an electronic device and a medium, and through collection of a CT image of the carbon for aluminum, accurate three-dimensional shapes of internal particles of the carbon material for aluminum can be obtained, formula calculation is performed in combination with a closest packing theory and a particle size distribution of actual raw materials, calculation and optimization of the formula of the carbon for aluminum are realized, scientificity and rationality of the formula of the carbon for aluminum are improved, and trial and error costs are reduced.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of material design, and particularly relates to a formula optimization method of carbon for aluminum, an electronic device and a medium. BACKGROUND

[0002] The carbon for aluminum is a core component of an aluminum electrolysis cell, and the quality of the carbon for aluminum directly affects the current efficiency and service life of the electrolysis cell. The degree of internal defects of the cathode carbon is one of important factors affecting the quality of the cathode carbon, and determines the mechanical strength, electrical conductivity, thermal conductivity and many other important properties of the carbon electrode. The shape and particle size ratio of raw materials directly affect the defect rate of the carbon product, and the performance of the carbon block can be optimized by using a reasonable formula to make the particles in the carbon block contact more closely.

[0003] At present, the bulk density experiment or the method based on experience is usually used to screen the high-quality formula of the carbon for aluminum. However, these methods have high blindness, lack of theoretical guidance, and have problems such as complex shape of raw materials, unknown stacking structure, and difficulty in optimization. It is difficult to select the appropriate amount and ratio of raw materials for the carbon for aluminum, and the trial and error cost is high. SUMMARY

[0004] The application aims at the deficiencies in the prior art, and provides a formula optimization method of carbon for aluminum, an electronic device and a medium, so as to realize the formula optimization of the carbon for aluminum and reduce the trial and error cost.

[0005] In order to achieve the above-mentioned purpose, the technical scheme adopted by the application is as follows:

[0006] A formula optimization method of carbon for aluminum, comprising the following steps:

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

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

[0009] S3: separating out the aggregate phase image, the asphalt phase image and the pore phase image in the preprocessed CT image of the carbon for aluminum by using a threshold segmentation method; and extracting the fractal dimension of the aggregate phase image;

[0010] S4: constructing an ideal particle size distribution model according to the fractal dimension and the particle size of the aggregate, and the ideal particle size distribution model is as follows:

[0011]

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

[0013] S5: using a particle swarm optimization algorithm 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 the carbon for aluminum.

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

[0015] Further, the implementation process of S2 comprises:

[0016] The CT image of the carbon for aluminum is sequentially subjected to shape clipping processing, noise reduction filtering processing and interface clarity enhancement processing.

[0017] The noise points and artifacts are reduced to reduce the error caused by the result analysis.

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

[0019] The aggregate phase image is covered with a three-dimensional grid;

[0020] The three-dimensional grid is divided into a plurality of cubes, and the side length of the cube is ε;

[0021] The value of ε is gradually reduced, and the reduction amplitude is For each side length value, the number of cubes containing aggregate particles N(ε) is counted;

[0022] The fractal dimension F is calculated, and the expression of the fractal dimension F is as follows:

[0023]

[0024] Further, the implementation process of S5 comprises:

[0025] A1: randomly generating the initial position and speed of the particle swarm, and each particle represents a target formula;

[0026] A2: calculating the objective function value of each particle, and the expression of the objective function MSE(w) is as follows:

[0027]

[0028] A3: updating the velocity and position of the particle, if the objective function value of the current particle is less than the objective function value of the optimal position of the particle, updating 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, updating the global optimal position; the global optimal position is the position corresponding to the minimum value of the objective function value of all particles;

[0029] A4: repeating steps A2-A3, iteratively optimizing the carbon formula for aluminum under the constraint condition to obtain the optimized formula of the carbon for aluminum; the expression of the constraint condition is as follows:

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

[0031] Wherein, 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 are the baseline formula.

[0032] Based on the same inventive concept, the application also provides an electronic device, comprising:

[0033] One or more processors;

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

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

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

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

[0038] Figure 1 It is a schematic diagram of the formula optimization method of the carbon for aluminum of the application;

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

[0040] Figure 3(a) is a comparison chart of the resistivity of the carbon for aluminum prepared by each formula;

[0041] Figure 3(b) is a comparison chart of the density of the carbon for aluminum prepared by each formula;

[0042] Figure 3(c) is a comparison chart of the bending strength of the carbon for aluminum prepared by each formula;

[0043] Figure 3(d) is a comparison chart of the compressive strength of the carbon for aluminum prepared by each formula;

[0044] Figure 4 Figure 4 is a comparison chart of the CT images of the carbon for aluminum prepared by the optimized and non-optimized formulas. DETAILED DESCRIPTION

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

[0046] EMBODIMENT

[0047] As Figure 1 The formula optimization method of the carbon for aluminum in the 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] Firstly, a high-resolution CT scanning device is used to scan the carbon for aluminum sample to obtain the shape parameters of the particles in the carbon. The shape parameters reflect the shape complexity of the carbon aggregate and its occurrence state in the multi-aggregate-asphalt system.

[0049] Secondly, according to the obtained shape parameters, according to the fractal theory and the packing relationship between large and small particles, the ideal particle size distribution that can achieve the closest packing of each level of raw material under the given shape condition can be derived.

[0050] Subsequently, the particle size distribution of each level of raw material needs to be tested. For this continuous particle size distribution system, the relationship between the particle size and the number of particles in the range of the maximum particle size and the minimum particle size needs to be obtained.

[0051] Based on the established packing model and the particle size test data, the target formula is calculated by the particle swarm optimization algorithm. The optimization target is to minimize the difference 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 obtained target formula is prepared according to the carbon preparation process, and the prepared carbon material sample is detected for physical and chemical indexes; meanwhile, CT scanning detection is performed thereon to obtain the particle accumulation condition inside the carbon material under the formula, and indexes such as defect rate and defect shape are obtained, and further feedback is provided for optimization of the formula.

[0053] 1. CT-based particle parameter acquisition

[0054] Firstly, CT image acquisition is performed on the carbon electrode material for aluminum by using a high-resolution X-ray CT device (including but not limited to Nano CT, Micro CT, etc.). The carbon electrode material includes but is not limited to prebaked anode for aluminum, carbon cathode for aluminum, etc.

[0055] The shape of the acquired CT image is cut to make the size of the scanning body in a reasonable range (2000mm 3 -10000mm 3 ). Subsequently, noise reduction filtering processing is performed on the image to reduce the error caused by noise and artifacts to the result analysis. In this embodiment, a non-local mean filtering method is used to remove the CT image noise, and the non-local mean filtering method is a prior art.

[0056] The carbon material is composed of aggregate (petroleum coke, graphite, etc.), binder (pitch) and pore. Since the aggregate and the binder in the carbon material are fully mixed, and the binder is mixed with a large amount of powder, an image enhancement means is used to reduce the interface tolerance effect. In this embodiment, a de-sharpening mask method is used to enhance the interface definition, and the de-sharpening mask method is a prior art.

[0057] Further, a threshold segmentation method is used to separate the aggregate, pitch and pore in the carbon CT image. In this embodiment, the threshold segmentation method used is a watershed algorithm, and the watershed algorithm 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 a box counting method. The shape parameter extracted is a fractal dimension. The basic flow of extracting the fractal dimension by using the box counting method is as follows: a three-dimensional grid is used to cover the aggregate phase CT image. The three-dimensional grid is divided into smaller cubes (boxes), and the edge length of the small cubes is ε, and the initial value is the edge length of the aggregate phase CT image; ε is gradually reduced, and the reduction amplitude is For each box size, the number of boxes containing aggregate particles N(ε) is counted; finally, the fractal dimension F is calculated using the following formula:

[0059]

[0060] 2. Ideal particle size distribution calculation

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

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

[0063]

[0064] N max is the number of maximum size particles; define the density function n(D) as a function representing the change in the number of particles per unit size interval:

[0065]

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

[0067]

[0068] where D min <D' < D.

[0069] The integral of equation 4 is calculated to 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 the single particle. The final ideal grain size distribution model can be expressed as:

[0074]

[0075] where φ(<D) is the percentage of aggregate particles less than D to the total number of particles.

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

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

[0078] 3. Raw material particle size distribution test

[0079] Each type of aggregate particle used to prepare carbon materials is tested for particle size distribution to obtain its continuous particle size distribution characteristics. The aggregates of carbon materials include petroleum coke, graphite and the like of various particle sizes.

[0080] Specifically, for each type of aggregate, a vibrating screen is used to obtain its particle size distribution from 1 mm to 16 mm, and a laser particle size analyzer is used 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 device that can achieve the above testing effect can also be used.

[0081] 4. Target formula optimization

[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 use intelligent algorithm programming and adjustment to make the difference between the particle size distribution of the dry material after mixing of each level of aggregate (the particle size distribution of the mixed material when the target ratio is used) and the ideal particle size distribution as small as possible, so as to reduce the defects of carbon blocks caused by the non-tight packing 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 of the previous step; at the same time, a particle size distribution vector y should also 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 ] (Formula 12)

[0085]

[0086] Wherein, 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4-8respectively 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 stone tar particles with a particle size less than 1 mm, the particle size distribution of stone tar particles with a particle size between 1 mm and 2 mm, the particle size distribution of stone tar particles with a particle size between 2 mm and 4 mm, and the particle size distribution of stone tar particles with a particle size between 4 mm and 8 mm, x 粉子 is the particle size distribution of the fine particles, y j is the percentage of particles in the jth size fraction to the total number of particles, and m is the number of size fractions. The particle size distribution of each size fraction is as follows Figure 2 .

[0087] Specifically, if there are 8 size fractions from small to large, the percentage of particles in each size fraction to the total number of particles is calculated by formula 10 and formula 11, and the ideal particle size distribution vector is as follows:

[0088]

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

[0090]

[0091] where w is the target formula, X i，w is the particle size distribution of the aggregate using the target formula w, y is the ideal particle size distribution vector, and n is the dimension of the particle size distribution matrix X of each size fraction of aggregate.

[0092] The constraints of the optimization process are defined 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 to the ideal formula. The initial weights are the baseline formula, i.e., the formula before optimization, which is obtained by the bulk density experiment.

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

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

[0097] 2) Calculate the fitness value, i.e., the objective function value, for each particle. If the current particle's fitness value is better than its previous best position, update its best position; if the current particle's fitness value is better than the global best position, update the global best position. In each iteration, the global best position is obtained by comparing the fitness values ​​of the individual best 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 denoted as the individual optimal position of the particle, g is the global optimal position, w0 is the inertial 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 globally optimal individual is found or the maximum number of iterations is reached. The optimal individual obtained is the target recipe.

[0103] Specifically, f is set to 0.1, 0.15, and 0.2 respectively to obtain three optimized target formulations, referred to as Formulation 2, Formulation 3, and Formulation 4, while the baseline formulation is referred to as Formulation 1.

[0104] 5. Formulation experiments and CT scan verification

[0105] Carbon materials were prepared using the optimized target formulation. Specifically, the preparation process included, but was not limited to, batching, mixing, molding, and calcination. The above preparation process was repeated for formulations 1, 2, 3, and 4, respectively. The preparation of carbon materials is an existing technology.

[0106] The prepared carbon materials were subjected to physicochemical tests, specifically including but not limited to tests of bulk density, resistivity, compressive strength, and tensile strength; the results are as follows: Figure 3(a) , 3(b) As shown in 3(c) and 3(d), the carbon materials made from the optimized target formulation have better bulk density, resistivity, compressive strength and tensile strength than the baseline formulation before optimization.

[0107] Further, the prepared carbon material is collected again for CT image, and the internal structure difference of the carbon material prepared by the formula before and after optimization is analyzed. Including but not limited to the comparison of particle accumulation condition, porosity, pore shape, and further adjusting the proportion of each particle size in the formula according to the above data. The results are shown in Figure 4 Figure 4 The left side is the CT image of the carbon material prepared by formula 1, and the right side is the CT image of the carbon material prepared by formula 2. The particle size distribution in the carbon material prepared by the optimized formula is more reasonable, and the aggregate and asphalt are mixed more uniformly, so that the carbon block has fewer initial defects. And the lower defect rate means better electrical and thermal conductivity, and stronger anti-deformation ability, which further improves the service life of the carbon material.

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

[0109] One or more processors;

[0110] A memory having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the steps of the formula optimization method of the aluminum carbon.

[0111] In some implementations, the memory can be a high-speed random access memory (RAM: Random Access Memory), and can also include a non-volatile memory, for example, at least one disk memory.

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

[0113] The embodiment provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the formula optimization method of the aluminum carbon.

[0114] The content illustrated by the above embodiments should be understood as the embodiments only for more clearly illustrating the present application, and should not be used to limit the scope of the present application. After reading the present application, those skilled in the art can make various equivalent modifications of the present application, which all fall within the scope defined by the appended claims of the present application.​

Claims

1. A method for optimizing the formulation of carbon for aluminum, characterized in that, The method comprises the following steps: S1: obtaining a CT image of carbon for aluminum; obtaining a particle size distribution of each aggregate of the carbon for aluminum; S2: preprocessing the CT image of carbon for aluminum; S3: separating an aggregate phase image, an asphalt phase image and a pore phase image in the preprocessed CT image of carbon for aluminum by using a threshold segmentation method; and extracting a fractal dimension of the aggregate phase image; S4: constructing an ideal particle size distribution model according to the fractal dimension and the particle size of the aggregate, the ideal particle size distribution model being as follows: wherein φ (<D) is a percentage of aggregate particles having a size less than D with respect to the total number of particles, F is a fractal dimension, D is a particle size of the aggregate particles, D min <D < D max , D min is a minimum particle size of the aggregate particles, and D max is a maximum particle size of the aggregate particles. S5: minimizing a mean square error between the ideal particle size distribution and a mixed particle size distribution of each aggregate by using a particle swarm optimization algorithm, and iteratively optimizing a carbon formula for aluminum by using a constraint condition to obtain an optimized formula of the carbon for aluminum; The expression of the constraint condition is as follows: max(0, initial weights-f)≤w≤min(1, initial weights+f) Wherein, w is a target formula, f is an approximation factor, and initial weights is a benchmark formula.

2. The method for formulation optimization of carbon for aluminum as claimed in claim 1 wherein, The implementation process of S2 comprises: The CT image of carbon for aluminum is sequentially subjected to shape clipping processing, noise reduction filtering processing and interface clarity enhancement processing.

3. The method for formulation optimization of carbon for aluminum as claimed in claim 1 wherein, The implementation process of extracting the fractal dimension of the aggregate phase image comprises: The aggregate phase image is covered with a three-dimensional grid; The three-dimensional grid is divided into a plurality of cubes, and the side length of the cube is ε; The value of ε is gradually reduced, each time by a factor of For each edge length value, the number of cubes N(ε) containing aggregate particles is counted. The fractal dimension F is calculated, and the expression of the fractal dimension F is as follows:

4. The method for formulation optimization of carbon for aluminum as claimed in claim 1 wherein, The implementation process of S5 comprises: A1: randomly generating initial positions and velocities of a particle swarm, each particle representing a target formula; A2: calculating a target function value of each particle, and the expression of the target function MSE(w) is as follows: A3: updating the velocity and position of the particle, if the target function value of the current particle is less than the target function value of the optimal position thereof, updating the optimal position of the particle; if the target function value of the current particle is less than the target function value of the global optimal position, updating the global optimal position; the global optimal position is a position corresponding to the minimum value of the target function values of all particles; A4: repeating steps A2-A3, iteratively optimizing the carbon formula for aluminum by using a constraint condition to obtain an optimized formula of the carbon for aluminum; and the expression of the constraint condition is as follows: max(0, initial weights-f)≤w≤min(1, initial weights+f) where w is the target formulation, X i，w is the particle size distribution of each level of aggregate when the target formulation w is used, 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 baseline formulation.

5. An electronic device, comprising: The method comprises: One or more processors; A memory having one or more programs stored thereon, which, when executed by the one or more processors, cause the one or more processors to implement the steps of the method of any one of claims 1-4.

6. A computer-readable storage medium, characterized in that, The computer program is stored in the memory and is executed by the processor to implement the steps of the method of any one of claims 1-4.

Images