Method for evaluating the porosity structure of coke, method for estimating the characteristic values ​​of coke, method for manufacturing coke, program for evaluating the porosity structure of coke, and program for estimating the characteristic values ​​of coke

Abstract

This invention provides a method for evaluating the porosity structure of coke, which can be used to estimate the properties of coke. [Solution] The method for evaluating the pore structure of coke according to the present invention is characterized by including a persistent diagram calculation step of calculating a persistent diagram from an image including the pores of coke. The method for evaluating the pore structure of coke according to the present invention may further include an image acquisition step of acquiring an image, a vectorization step of vectorizing the persistent diagram calculated in the persistent diagram calculation step, and a dimensionality reduction step of reducing the dimensions from the persistent diagram vectorized in the vectorization step to obtain variables of the persistent diagram.

Description

Technical Field

[0001] The present invention relates to a method for evaluating the pore structure of coke, a method for estimating characteristic values of coke, a method for manufacturing coke, a program for evaluating the pore structure of coke, and a program for estimating characteristic values of coke.

Background Art

[0002] Coke for blast furnaces is used as a reducing agent for reducing iron ore in blast furnaces. However, since it causes clogging when it collapses into pieces in the furnace, high-strength coke is required. In addition, in the furnace, coke generates CO gas by the solution loss reaction (C (coke) + CO2 = 2CO (carbon monoxide)), and the generated CO gas becomes a reducing gas and reduces the iron ore. However, if the solution loss reaction is slow, it takes a significantly long time for the reduction reaction of iron ore, so the productivity of hot metal deteriorates.

[0003] On the other hand, if the solution loss reaction is too fast, generally, as the solution loss reaction progresses, the coke strength decreases, so the coke becomes powdery in the blast furnace, which is a factor in reducing the air permeability in the blast furnace. Therefore, a coke material with an appropriate reaction rate of the gasification reaction including the solution loss reaction (carbon dissolution reaction) is required.

[0004] Generally, the strength of coke is organized by DI (Drum Index), and the reactivity of coke is organized by CRI (Coke Reactivity Index). It is considered that the higher the DI value, the higher the coke strength, and the higher the CRI value, the better the reactivity with carbon dioxide (CO2) gas.

[0005] Here, the measurement method for DI is defined by the JIS standard (JIS standard K2151:2004, Cokes - Test methods), and the measurement method for CRI is defined by the ISO standard (ISO 18894:2018). Generally, coke used in blast furnaces is a porous material with 40-70% pores, and these indicators (coke strength, coke reactivity) are known to strongly depend on the porosity structure of the coke.

[0006] For example, Non-Patent Document 1 reports on the influence of pore size and shape on coke strength. According to Non-Patent Document 1, it is considered important to appropriately evaluate the pore structure of coke in order to control coke strength or coke reactivity.

[0007] On the other hand, several evaluation methods have been proposed for the porosity structure of coke. For example, the above-mentioned JIS standard (K2151:2004, Coke - Test Methods) specifies a method for measuring the porosity of coke using the Archimedes process.

[0008] Non-patent document 2 discloses a method for extracting pores by photographing coke with a stereomicroscope and evaluating the circularity of those pores.

[0009] Patent Document 1 proposes a method for visualizing the pores of coke using a microimaging method with a nuclear magnetic resonance spectrometer.

[0010] Patent document 2 proposes a method for visualizing the pores of coke using NMR gas imaging. [Prior art documents] [Patent Documents]

[0011] [Patent Document 1] Japanese Patent Publication No. 2004-77427 [Patent Document 2] Japanese Patent Publication No. 2006-38519 [Non-patent literature]

[0012] [Non-Patent Document 1] Saito et al., Effects of random pore shape, arrangement, and non-contact grain boundaries on coke strength, Iron and Steel, 100 (2014), 140. [Non-Patent Document 2] Hayashi et al., Evaluation of Coal Softening, Melting, and Expansion Behavior Based on Coke Pore Structure Analysis, Iron and Steel, 100 (2014), 118. [Overview of the project] [Problems that the invention aims to solve]

[0013] However, it was found that even using the conventional technologies disclosed above, such as the porosity of coke, the shape evaluation of pores within coke, and methods for visualizing pores within coke, it is difficult to estimate the properties of coke, such as coke strength and coke reactivity.

[0014] This invention has been made in view of the above problems, and aims to provide a method for evaluating the porosity structure of coke that can estimate the properties of coke. [Means for solving the problem]

[0015] The gist of the present invention, which solves the above problems, is as follows.

[0016] 1. A method for evaluating the structure of pores contained in coke, A method for evaluating the pore structure of coke, including a persistent diagram calculation step for calculating a persistent diagram from an image of coke containing pores.

[0017] 2. An image acquisition step in which the aforementioned image is acquired, A vectorization step which vectorizes the persistent diagram calculated in the persistent diagram calculation step, A dimensionality reduction step of obtaining variables of the persistent diagram by reducing the dimension from the persistent diagram vectorized in the vectorization step; The method for evaluating the pore structure of coke according to item 1, further comprising .

[0018] 3. By using the variables obtained by the method for evaluating the pore structure of coke according to item 2 as explanatory variables and inputting the variables of the coke for which the characteristic value is to be estimated into the prediction model obtained with the characteristic value of coke as the target variable, the characteristic value of the coke for which the characteristic value is to be estimated is estimated. A method for estimating the characteristic value of coke, including a characteristic value estimation step.

[0019] 4. A method for manufacturing coke, wherein the manufacturing conditions of coke are changed based on the characteristic value of coke estimated by the method for estimating the characteristic value of coke according to item 3.

[0020] 5. The method for manufacturing coke according to item 4, wherein the change in the manufacturing conditions is a change in the coal blending ratio.

[0021] 6. A program for evaluating the pore structure of coke for causing a computer to execute the method for evaluating the pore structure of coke according to item 1 or 2.

[0022] 7. A program for estimating the characteristic value of coke for causing a computer to execute the method for estimating the characteristic value of coke according to item 3.

Advantages of the Invention

[0023] According to the present invention, it is possible to provide a method for evaluating the pore structure of coke capable of estimating the characteristics of coke.

Brief Description of the Drawings

[0024] [Figure 1] It is a flowchart of a preferred example of the method for evaluating the pore structure of coke according to the present invention. [Figure 2] It is a schematic diagram showing the procedure of the persistent diagram calculation step. [Figure 3] This figure shows an example of a persistent diagram. [Figure 4] This figure shows an example of a vectorized persistent diagram. [Figure 5] (a) A microstructure photograph of the coke used in the example, (b) a binarized image of Figure (a), and (c) a zero-order persistent diagram calculated from Figure (b). [Figure 6] This figure shows the relationship between the principal component P1 and the normalized CRI value for the 0th-order persistent diagram calculated in the example. Each normalized CRI value is calculated by dividing by the average value of the CRI values ​​of all samples. [Figure 7] This figure shows the relationship between normalized porosity and normalized CRI value for Comparative Example 1. Note that each normalized CRI value is calculated by dividing by the average CRI value of all samples. [Figure 8] This figure shows the relationship between normalized circularity and normalized CRI value for Comparative Example 2. Each normalized CRI value is calculated by dividing by the average CRI value of all samples. [Modes for carrying out the invention]

[0025] (Method for evaluating the porosity structure of coke) Embodiments of the present invention will be described below with reference to the drawings. The method for evaluating the pore structure of coke according to the present invention includes a persistent diagram calculation step of calculating a persistent diagram from an image including the pores of coke.

[0026] The inventors diligently considered ways to solve the above problems. They hypothesized that coke contains a complex distribution of pores with diverse shapes, and that before coke fractures, cracks occur in the pores and these cracks connect. In other words, the inventors realized that in order to accurately estimate the strength of coke, it is important to evaluate not only the porosity or shape of the coke, but also the pore structure, such as the size of the pores, the distance between each pore, and the degree of connectivity.

[0027] Furthermore, regarding the reactivity of coke, the inventors found that, considering the process by which CO2 gas flows in through the complexly distributed pores and reacts with the coke substrate, it is important to evaluate the pore structure of coke, just as it is to evaluate the strength of coke.

[0028] The inventors then diligently studied methods for evaluating the pore structure of coke and conceived that the pore structure of coke, specifically the size of the pores and the distance between them, could be evaluated using the persistent homology method, thus completing the present invention.

[0029] Figure 1 shows a flowchart of a preferred example of a method for evaluating the pore structure of coke according to the present invention. Each step will be described below.

[0030] [Image acquisition process] First, in the image acquisition process, an image including the pores of the coke (hereinafter simply referred to as the "coke image") is acquired.

[0031] In order to calculate a persistent diagram of the coke pore structure in a later process, a coke image for evaluation is necessary. The coke image can be an image in which only the pores are extracted, visualizing the pores.

[0032] The visualization of stomata described above is not particularly limited to any method that can visualize stomata, but it has a significant impact on the DI value and CRI value. 1 ~10 4Images that can visualize pores with a size of μm are preferred. Methods for obtaining such images include, for example, X-ray CT images and microscope images (optical microscope images, scanning electron microscope images, etc.). In addition, since the size of pores in coke generally varies, it is preferable to evaluate at least 100 pores. For example, if a sufficient number of pores cannot be visualized in one field of view taken with an optical microscope, it is preferable to evaluate at least 100 pores by taking multiple fields of view.

[0033] First, we will explain a method for obtaining coke images by taking X-ray CT (Computed Tomography) images (hereinafter simply referred to as "CT images") of coke in a block of about 20 mm in size. In CT images of coke, pores generally appear black and show a different contrast from the coke matrix. Therefore, it is preferable to obtain images of pores extracted from the coke using an image analysis method. The image analysis method is not particularly limited; for example, an image of coke pores extracted may be obtained by binarizing the brightness values. Alternatively, an image of coke pores extracted may be obtained using a machine learning image analysis method. The machine learning image analysis method here is not particularly limited; for example, pores can be extracted using image analysis with a random forest model or image analysis with a CNN (Comvolutional Neural Network) model.

[0034] Next, a method for obtaining coke images using an optical microscope will be described. First, the coke is cut, then the resulting cross-section of the coke is polished, and then a coke image is obtained using an optical microscope. Alternatively, pores may be extracted from the obtained coke image using an image analysis method and used as the coke image.

[0035] Next, a method for acquiring images containing stomata using a scanning electron microscope will be described. Since the field of view is significantly limited with a scanning electron microscope, it is preferable to acquire multiple images while shifting the field of view and then combine the images using a tiling function to obtain a coke image. Alternatively, stomata may be extracted from the obtained coke image using an image analysis method to obtain a coke image.

[0036] As mentioned above, scanning electron microscopes have a narrow field of view, so extracting a large number of pores and obtaining average information requires a lot of work time. Therefore, it is preferable to acquire coke images using an optical microscope or X-ray CT, which have a wide field of view.

[0037] [Persistent Diagram Acquisition Process] Next, in the persistent diagram calculation process, a persistent diagram is calculated from the image of the coke including its pores.

[0038] As mentioned above, it is thought that the process leading to coke fracture involves the formation and connection of cracks in the pores. Therefore, in order to accurately estimate the strength of coke, it is considered important to evaluate the pore structure, such as the size of the coke pores, the distance between each pore, and the degree of connection.

[0039] In the coke images obtained during the image acquisition process described above, the pores are distributed in a complex three-dimensional manner, making it difficult to understand the pore structure from the coke images alone. Therefore, in this invention, the pore structure of coke is evaluated using a persistent diagram obtained using persistent homology, one of the topological analysis methods.

[0040] An overview of the procedure for calculating persistent diagrams is described in Hiraoka, Obayashi, Akagi, Artificial Intelligence, "Persistent Homology and Material Structure Analysis," 34 (2019), 330, etc. Below, an outline of the persistent homology method will be explained using Figure 2.

[0041] First, the equivalent circular diameter of each pore is calculated in the coke image. Figure 2(a) shows a schematic diagram where circles with the calculated equivalent circular diameters are placed at the center of each pore and plotted as black dots. Next, as shown in Figure 2(b), a virtual circle (virtual circle) is set with each black dot as its center, and the radius r1 of the virtual circle is gradually increased from the radius of the black dot (equivalent circular diameter of the pore). Then, as shown in Figure 2(c), the radius r2 when the virtual circles come into contact is taken as death in the 0th order persistent diagram. Also, as shown in Figure 2(c), the radius r3 when a virtual hole (virtual hole) A is created by the contact of the virtual circles is taken as birth in the 1st order persistent diagram. If the virtual circles are made larger, the virtual hole A disappears as shown in Figure 2(d), and the radius r4 at this point is taken as death in the 1st order persistent diagram. On the other hand, if the virtual circles are made smaller and the virtual circles disappear, the radius r0 at which point is taken as birth in the 0th order persistent diagram.

[0042] Through this procedure, the distance between adjacent pores is reflected in the birth and death values ​​of the zero-order persistent diagram, and the size of pore A surrounded by the pores is reflected in the birth and death values ​​of the first-order persistent diagram. The resulting birth-death-pair aggregates the information about the pore structure in the original coke image. By using a persistent diagram plotting the obtained birth-death-pair with birth values ​​on the x-axis and death values ​​on the y-axis, the information about the pore structure of the coke can be visually represented and quantitatively evaluated.

[0043] In Figure 2, the shape of each pore is approximated by a circle, and the equivalent circular diameter of each pore is calculated. However, instead of circles, the center coordinates of each pore can be given as points, and the persistent diagram can be calculated by performing the process shown in Figure 2 on these points. However, since the size of the pores contained in coke is not uniform, it is preferable to approximate each pore with a circle and give the equivalent circular diameter as shown in Figure 2, rather than using points. This is because the size of the pores can be evaluated using the birth value (absolute value of r0) of the 0th order persistent diagram, and calculations can be performed efficiently.

[0044] Furthermore, the shapes of the pores in coke are not actually circular, but rather diverse. Therefore, it is preferable to calculate a persistent diagram that focuses on the diverse shapes of each pore itself, as this allows for the evaluation of the pore shapes as well. In this case, the shape of each pore is enlarged (i.e., enlarged so that the enlarged shape is similar to the original shape) and the persistent diagram is calculated. However, this method requires significant computational resources, so the calculation method should be determined based on the available computational resources and the required accuracy.

[0045] Figure 3(a) shows an example of a zero-order persistent diagram of a coke image obtained using the procedure described above. The normalized birth value on the horizontal axis and the normalized death value on the vertical axis are normalized by the average of the absolute values ​​of the birth and death values, respectively. The zero-order persistent diagram obtained in this way shows information about the one-dimensional local arrangement of coke pores, while the first-order persistent diagram shows information about the two-dimensional local arrangement of coke pores. The pore structure of coke can be evaluated from these persistent diagrams.

[0046] For example, by calculating a persistent diagram for coke whose DI value is known in advance, the relationship between the pore structure of the coke and the DI value can be understood. Furthermore, for coke whose DI value is unknown, a persistent diagram can be calculated and compared with a persistent diagram of which the relationship with the DI value is known in advance to estimate the DI value. The same evaluation can be performed for the CRI value of coke.

[0047] In this way, the pore structure of coke can be evaluated by calculating a persistent diagram from a coke image. Furthermore, by performing a vectorization process and a dimensionality reduction process on the persistent diagram, the variables (principal components) of the pore structure can be extracted.

[0048] [Vectorization process] In the vectorization process, the persistent diagram obtained in the persistent diagram calculation process is vectorized.

[0049] The vectorization described above involves first dividing the persistent diagram into meshes of a desired size, calculating the number of plot points in each mesh, and representing this as a contour plot. In this process, it is preferable to divide the mesh into 1-pixel sections to avoid losing detailed information. Figure 3(b) is a contour plot obtained by dividing the 0th-order persistent diagram shown in Figure 3(a) into 1-pixel meshes and calculating the number of plot points in each 1-pixel-sized region. In Figure 3(b), darker colors indicate a larger number of plot points.

[0050] Next, the obtained contour plots are vectorized. Specifically, the number of plot points in each region of Figure 3(b) is arranged one-dimensionally. This makes it possible to capture the overall average characteristics of the stomatal structure. Here, in the persistent diagram, diagonal components where the birth value and death value are close often represent noise when stomata are extracted from the coke image, and the diagonal components are often meaningless information. Therefore, it is preferable to set the diagonal components to 0 and vectorize by weighting the points further away from the diagonal components more heavily. Note that there are no particular limitations on how the regions are arranged one-dimensionally.

[0051] Figure 4 is a plot of the persistent diagram obtained from the contour plot shown in Figure 3(b), where the elements further away from the diagonal are weighted more heavily. By converting the two-dimensional persistent diagram shown in Figure 3(a) into the one-dimensional vector shown in Figure 4, it becomes possible to utilize it in subsequent dimensionality reduction processes.

[0052] In the above explanation, we described the vectorization of a zero-order persistent diagram. However, since a first-order persistent diagram can consider a wider range of pore structures compared to a zero-order persistent diagram, it is preferable to vectorize the first-order persistent diagram using the same procedure. Ultimately, it is preferable to adopt variables that can accurately explain the characteristics of coke.

[0053] [Dimension reduction process] In the dimensionality reduction process, the variables of the persistent diagram obtained are obtained by reducing the dimensionality of the persistent diagram that was vectorized in the vectorization process.

[0054] The persistent diagram vectorized in the vectorization process is a vector with, for example, tens of thousands or more dimensions, making it difficult to obtain a correlation with the characteristic values ​​of coke in its current form. Therefore, dimensionality reduction is performed on the vectorized persistent diagram.

[0055] The method of dimensionality reduction is not particularly limited; for example, dimensionality reduction can be performed using principal component analysis or t-SNE (t-Distributed Stochastic Neighbor Embedding).

[0056] If a pore structure with a specific birth-death pair strongly influences the CRI value of coke, the dimensionality may be reduced by extracting only the points with that specific birth-death pair and removing the other components.

[0057] Furthermore, when estimating the characteristic values ​​of coke using the coke characteristic value estimation method according to the present invention, which will be described later, it is preferable to adopt the method that provides the highest prediction accuracy. In addition, variables with reduced dimensions can be obtained from two diagrams, a zero-order persistent diagram and a first-order persistent diagram. In the estimation process described later, it is sufficient to adopt the variable calculated from the persistent diagram with the highest prediction accuracy as the explanatory variable. If using variables calculated from both persistent diagrams as explanatory variables improves the prediction accuracy, then variables calculated from both persistent diagrams may also be used as explanatory variables.

[0058] It is preferable to reduce the number of elements from tens of thousands or more to 1-2 dimensions by using the methods described above. In three or more dimensions, it is often difficult to construct a predictive model in the estimation of characteristic values ​​described later, and by using the 1-2 dimensional numerical information obtained by this procedure, it becomes possible to investigate the correlation with important characteristic values ​​(indicators) of coke, such as DI or CRI. However, if the prediction accuracy is improved by using variables of three or more dimensions, then it is acceptable to use variables of three or more dimensions.

[0059] (Method for estimating the characteristic values ​​of coke) The coke characteristic value estimation method according to the present invention is characterized by including a characteristic value estimation step in which the variables obtained by the above-described coke porosity structure evaluation method are used as explanatory variables, and the characteristic values ​​of the coke are used as the objective variable, and the variables of the coke to be predicted are input into a prediction model to estimate the characteristic values ​​of the coke to be predicted.

[0060] <Coke property values> The characteristic values ​​of coke are not particularly limited as long as they are characteristics affected by the pores of the coke. For example, since the pores of coke have a significant effect on the coke strength (DI) or coke reactivity (CRI), the characteristic values ​​of coke are preferably coke strength or coke reactivity.

[0061] [Characteristic value estimation process] In the characteristic value estimation process, the variables obtained from the evaluation method for the porosity structure of coke are used as explanatory variables, and the characteristic values ​​of the coke are used as the objective variable. By inputting the variables of the coke whose characteristic values ​​are to be predicted into the prediction model, the characteristic values ​​of the coke whose characteristic values ​​are to be predicted are estimated.

[0062] <Predictive Model> The predictive model can be a machine learning model obtained through machine learning, or a regression model obtained through regression analysis. An example of a regression model is the following equation. Coke property value = α1 × P1 + α2 Here, the coefficients α1 and α2 should be determined in such a way that the error between the measured value and the estimated value is small, and they can be modified as appropriate according to the experimental results.

[0063] (Method of producing coke) The coke manufacturing method according to the present invention is characterized by changing the coke manufacturing conditions based on the coke characteristic values ​​estimated by the coke characteristic value estimation method described above.

[0064] The manufacturing conditions that can be changed are not particularly limited; for example, the coal-to-coke blending ratio can be altered based on estimated characteristic values.

[0065] Here, the modification of coke production conditions based on the estimated coke characteristics corresponds to the "coke production method" in this invention. Furthermore, the modification of coke production conditions includes a change in the coal blending ratio based on the estimated coke characteristics.

[0066] (Program for evaluating the porosity structure of coke) The coke pore structure evaluation program according to the present invention is a program that causes a computer to execute the coke pore structure evaluation method according to the present invention described above. This program can be stored in the memory of the computer, and the CPU in the computer can appropriately read the program describing the processing content for executing each processing content from the memory and execute each step.

[0067] Furthermore, a program describing this processing can be distributed, for example, by selling, transferring, or lending portable recording media such as Blu-ray®, DVD, or CD-ROM. In addition, such a program can be distributed by storing it in the memory of a server on a network and transferring it from the server to other computers via the network.

[0068] Furthermore, a computer executing such a program can, for example, store a program recorded on a portable recording medium or a program transferred from a server in its own memory. Another embodiment of this program is that the computer may directly read the program from the portable recording medium and execute processing according to that program, or it may sequentially execute processing according to the received program each time a program is transferred to the computer from a server.

[0069] (Program for estimating the characteristics of coke) The coke characteristic value estimation program according to the present invention is a program that causes a computer to execute the coke characteristic value estimation method according to the present invention described above. This program can be executed in the same way as the coke pore structure evaluation program described above, and therefore no further explanation is provided. [Examples]

[0070] The present invention will be described below with reference to examples. However, the present invention is not limited to the following examples.

[0071] (Example 1) A method for evaluating the porosity structure of coke will be explained using examples. First, test pieces of approximately 20 × 20 mm in size were cut from several coke samples with different CRI values, which are an indicator of coke reaction rate. The cut surfaces were mirror-polished to obtain the observation surfaces. The coke was prepared by annealing coal blended to have normalized vitrinite mean reflectance (Ro) values ​​of -1.28, -0.53, -0.151, 0.226, and 1.73. Some of the coke was also prepared by pressing and compressing the coal before annealing. The normalized Ro values ​​here were normalized using the mean and standard deviation of the Ro values ​​of all samples.

[0072] Next, an optical microscope was used to observe the observation surface of each specimen across multiple fields of view, and a tiling function was used to acquire an image of the entire observation surface of the specimen. From the obtained coke image, a machine learning-based image analysis method was used to extract the pores of the coke and acquire a coke image (image acquisition process).

[0073] Next, a 0th-order persistent diagram was calculated from the obtained coke image using the persistent homology method (persistent diagram calculation process). In this process, all coke pores were considered to be circular, and the equivalent diameter of the circle was calculated to obtain the persistent diagram. A portion of the acquired coke image, a coke pore image, and a 0th-order persistent diagram are shown in Figures 5(a) to (c), respectively. In Figure 5, the birth value and death value of the 0th-order persistent diagram are shown normalized by the average of the absolute values ​​of the birth value and the average of the absolute values ​​of the death value, respectively.

[0074] Subsequently, the zero-order persistent diagram was converted into a one-dimensional vector (vectorization process). During this process, diagonal elements where the birth and death values ​​were equal were weighted to zero, while elements further from the diagonal were weighted more heavily. The resulting zero-order persistent diagram had approximately 20,000 dimensions.

[0075] Next, the dimensionality was reduced to two dimensions by performing principal component analysis (dimensionality reduction process). The resulting variables become parameters representing the pore structure of the coke.

[0076] (Example 2) Next, we will explain the method for predicting the characteristic values ​​of coke using an example. First, we evaluated the CRI, an indicator of the reaction rate of the coke used in Example 1. The CRI value was measured according to the procedure of JIS standard K2151:2004.

[0077] Figure 6 shows the relationship between the variables obtained in Example 1 (principal component P1 of the 0th-order persistent diagram) and the CRI value of coke. Here, the normalized CRI value, obtained by dividing by the average CRI value of all test specimens, was used. As shown in Figure 6, there is a correlation between the principal component P1 of the 0th-order persistent diagram and the normalized CRI value, and it was found that the normalized CRI value can be predicted using the following equation (1). Normalized CRI value = 0.194 × P1 + 0.881 (1)

[0078] By associating the above-mentioned main component P1 with the coke manufacturing conditions, it is possible to change the conditions for producing coke with a specific CRI value. For example, the coal blending ratio of the coke can be changed.

[0079] For comparison, Figure 7 shows the relationship between coke porosity and the normalized CRI value. Figure 8 shows the relationship between coke roundness and the normalized CRI value, which Non-Patent Literature 2 stated were correlated with coke fracture strength. From Figures 7 and 8, it can be seen that coke porosity and coke roundness cannot explain the reactivity of coke.

[0080] Thus, the present invention makes it possible to predict the pore structure of coke in order to achieve ideal coke characteristics, and is expected to shorten the development period for coke with good characteristics. [Industrial applicability]

[0081] According to the present invention, a method for evaluating the pore structure of coke that can estimate the properties of coke can be provided.

Claims

1. A method for evaluating the structure of pores contained in coke, A method for evaluating the pore structure of coke, including a persistent diagram calculation step for calculating a persistent diagram from an image of coke containing pores.

2. The image acquisition step for acquiring the aforementioned image, A vectorization step which vectorizes the persistent diagram calculated in the persistent diagram calculation step, The vectorization process includes a dimensionality reduction step, which reduces the dimensionality of the vectorized persistent diagram to obtain the variables of the persistent diagram, A method for evaluating the porosity structure of coke according to claim 1, further comprising:

3. A method for estimating the characteristic values ​​of coke, comprising a characteristic value estimation step of estimating the characteristic values ​​of coke by inputting the variables of the coke to be estimated into a predictive model obtained by using the variables obtained by the coke porosity evaluation method of coke described in claim 2 as explanatory variables and the characteristic values ​​of coke as the objective variable.

4. A method for producing coke, comprising changing the coke production conditions based on the coke characteristic values ​​estimated by the coke characteristic value estimation method described in claim 3.

5. The method for producing coke according to claim 4, wherein the change in the manufacturing conditions is a change in the coal blending ratio.

6. A coke pore structure evaluation program for causing a computer to perform the coke pore structure evaluation method described in claim 1 or 2.

7. A coke characteristic value estimation program for causing a computer to perform the coke characteristic value estimation method described in claim 3.

Images