Preparation method of catalyst applied to cyanide cracking

Abstract

The invention relates to the technical field of calcined coke preparation, in particular to a preparation method of a catalyst applied to cyanide cracking, which comprises the following steps: screening and dehydrating petroleum coke, crushing and grading to obtain petroleum coke particles with qualified grade, stirring and mixing the petroleum coke particles and sodium hydroxide, adding water, stirring to prepare slurry, and drying to obtain the catalyst applied to cyanide cracking. Digesting and stirring the slurry, adding calcium hydroxide, continuing digesting to obtain a petroleum coke base material, collecting an SEM scanning electron microscope graph of the petroleum coke base material, obtaining an edge angle correction coefficient and layer edge significance of each pixel point, and obtaining the optimal calcination time length of the petroleum coke base material in combination with a neural network model. And calcining and cooling the petroleum coke base material by utilizing the optimal calcining time, and grinding and granulating to obtain the special catalyst for cyanide cracking. The invention aims to improve the desulfurization rate in the calcining process of the petroleum coke base material, so that the petroleum coke base material is calcined more sufficiently.

Description

Technical Field

[0001] The invention relates to the technical field of calcined coke preparation, and in particular to a method for preparing a catalyst used for cyanide cracking. Background Art

[0002] The quality of the cyanide cracking reaction plays a vital role in the yield of sodium cyanide and the final application of the product. Calcined coke acts as a solid heat carrier in the cracking reaction. Solid heat carrier is also a kind of catalyst. Calcined coke is usually used as a catalyst for cyanide cracking in industry. Sulfur has no chance to be discharged in the cyanide cracking reaction and will exist in cyanide. If the high sulfur content is not removed cleanly, it will affect the product quality, corrode the equipment and affect the environmental protection standards. Therefore, the sulfur content in the catalyst has a significant impact on the cyanide cracking reaction. High-quality catalysts can help enterprises improve production efficiency and benefits, so that enterprises can create greater value in the pharmaceutical, fertilizer, daily chemical and other industries.

[0003] The basic raw material of the catalyst is petroleum coke, but the petroleum coke used in actual process production has a high sulfur content, unstable properties, and a large fluctuation range. Even if the same preparation process is used, the quality of the catalyst will fluctuate, affecting the quality of the cyanide cracking reaction. At the same time, with the increase of calcination time, the desulfurization rate of petroleum coke is also increasing. However, when the calcination time is too long, it is impossible to react to produce harmful gases and increase the desulfurization rate of petroleum coke. There is a problem in the preparation process of special catalysts for cyanide cracking that the calcination time cannot be adaptively determined according to the properties of the raw materials. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a method for preparing a catalyst used for cyanide cracking to solve the existing problems.

[0005] The method for preparing a catalyst for cyanide cracking of the present invention adopts the following technical scheme: One embodiment of the present invention provides a method for preparing a catalyst for cyanide cracking, the method comprising the following steps: The preparation raw materials include petroleum coke, sodium hydroxide and calcium hydroxide; the petroleum coke is screened and dehydrated, and then crushed and graded to obtain petroleum coke particles of qualified grade; the petroleum coke particles and sodium hydroxide are stirred and mixed, and then water is added and stirred to form slurry; the slurry is steamed and stirred, and then calcium hydroxide is added, and the steaming is continued to obtain the petroleum coke base material; Adaptive adjustment of the calcination time of petroleum coke base material includes: Collect the SEM scanning electron microscope image of the petroleum coke base material; set the first neighborhood of each pixel point with each pixel point as the center; analyze the grayscale difference and position relationship between each non-central pixel point and the central pixel point in the first neighborhood to obtain the true coefficient of the same-layer edge pixel of each non-central pixel point; The edge correction coefficient of each pixel is obtained by combining the true coefficient of the edge pixels of the same layer of each non-central pixel and the clustering algorithm; The optimal calcination time of the petroleum coke base material is obtained by combining the edge correction coefficient of each pixel point and the neural network model; The petroleum coke base material is calcined and cooled using an optimal calcination time to obtain a calcined coke base material; the calcined coke base material is ground and granulated to obtain a special catalyst for cyanide cracking.

[0006] Preferably, the screening and dehydration comprises: The petroleum coke is screened through a coke screen, and the coke particles smaller than 10 mm are screened out and sent to a dehydration tank for dehydration to reduce the moisture content to below 10%.

[0007] Preferably, the crushing and classification comprises: The dehydrated petroleum coke is crushed to a size of 5 mm or less, and the crushed petroleum coke is graded through a vibrating screen. The petroleum coke particles with a particle size of 5 mm or less are qualified, and the petroleum coke particles with a particle size greater than 5 mm are unqualified. The unqualified petroleum coke particles are crushed and graded again until they reach the qualified grade.

[0008] Preferably, the step of mixing the petroleum coke particles with sodium hydroxide and then adding water comprises: The mass ratio of petroleum coke particles to sodium hydroxide is 2:1, and the mass ratio of petroleum coke particles to added water is 1:1.

[0009] Preferably, the slurry is boiled and stirred, calcium hydroxide is added, and the boiling is continued to obtain the petroleum coke base material, comprising: The cooking temperature is 200° C., the mass ratio of petroleum coke particles to calcium hydroxide is 2:1, the cooking time is 1 hour, and the continued cooking time is 1.2 hours.

[0010] Preferably, the process of determining the true coefficient of the same-layer edge pixel of each non-central pixel point includes: Calculate the absolute value of the difference between the grayscale values ​​of each non-central pixel and the central pixel in the first neighborhood of each pixel, determine it as the grayscale difference of each non-central pixel, determine the difference between the grayscale difference of any non-central pixel and other non-central pixels in the first neighborhood as the grayscale neighborhood difference, determine the sum of all the grayscale neighborhood differences of any non-central pixel as the neighborhood difference degree, forward map the neighborhood difference degree and reverse map the grayscale difference of any non-central pixel, and obtain the grayscale offset microfactor of any non-central pixel; A second neighborhood is set with each non-central pixel point in the first neighborhood as the center, wherein the second neighborhood is smaller than the first neighborhood, and all the pixels in the second neighborhood belong to the first neighborhood, and the pixel point corresponding to the maximum value of the grayscale offset micro-factor in the second neighborhood and the central pixel point of the second neighborhood are recorded as an off-center combination; all the pixels in the first neighborhood are taken as undirected graph nodes, and the undirected graph nodes corresponding to all the off-center combinations are connected, and the paths from each non-central pixel point in the first neighborhood to the central pixel point are obtained in combination with the graph search algorithm, which are recorded as off-center paths from each non-central pixel point to the central pixel point, and the grayscale offset micro-factors of all the pixels on each off-center path are arranged in path order as the path offset sequence of each non-central pixel point, and all the non-central pixels in the first neighborhood are divided into circles from the inside to the outside according to the distance from the central pixel point; For each of the off-center paths, the difference between the grayscale offset microfactor of any pixel point on the off-center path and the maximum grayscale offset microfactor of all pixel points on the circle where the any pixel point is located is calculated to determine the grayscale offset difference. The path off-center intensity of each non-center pixel point in the first neighborhood is positively correlated with the sum of the grayscale offset differences of all pixel points on the corresponding off-center path, and negatively correlated with the number of pixel points on the corresponding off-center path. The information entropy of all elements in the path offset sequence of each non-central pixel point in the first neighborhood of each pixel point is calculated, and the ratio of the path off-center intensity of each non-central pixel point in the first neighborhood to the information entropy is determined as the off-center confusion degree. The number of occurrences of the corresponding node of each non-central pixel point in the first neighborhood and the off-center confusion degree of each non-central pixel point are forward mapped to obtain the true coefficient of the same-layer edge pixel of each non-central pixel point in the first neighborhood.

[0011] Preferably, the process of determining the edge correction coefficient of each pixel point includes: Perform threshold segmentation on the true coefficients of the same-layer edge pixels of all non-central pixels in the first neighborhood of any pixel, record the non-central pixels whose true coefficients of the same-layer edge pixels are greater than or equal to the segmentation threshold as suspected same-layer edge pixels, connect all pixels on the off-center path of each suspected same-layer edge pixel in sequence according to the path order, calculate the slope of each connection line, and take the average of all connection line slopes as the suspected fiber extension slope of each suspected same-layer edge pixel; Cluster the suspected fiber extension slopes of all suspected same-layer edge pixels in the first neighborhood of any pixel point, record the cluster with the largest number of elements as the first-class representative cluster, and the cluster with the second largest number of elements as the second-class representative cluster. Count the number of other elements of each element in the first-class representative cluster within the preset neighborhood radius as the core density of each element, and record the element with the largest core density as the suspected direction core point; Calculate the extreme value of the suspected fiber extension slope in each cluster in the first neighborhood of each pixel point, calculate the sum of the reciprocals of the core direction deviation values ​​of all suspected same-layer edge pixel points in each cluster cluster, determine it as the direction deviation sum, and map the extreme value inversely and the direction deviation sum in direct proportion to obtain the directional stability factor of each cluster in the first neighborhood of each pixel point, wherein the core direction deviation value is the Euclidean distance between the suspected same-layer edge pixel point and the suspected direction core point; For each pixel point, the difference between the number of suspected same-layer edge pixel points of the first representative cluster and the second representative cluster in the first neighborhood is determined as the first same-layer edge difference value, the difference between the number of all suspected same-layer edge pixel points in the first neighborhood and the number of suspected same-layer edge pixel points in the first representative cluster is determined as the second same-layer edge difference value, the ratio of the first same-layer edge difference value to the second same-layer edge difference value is used as the same-layer edge ratio, the directional stability factor of the first representative cluster in the first neighborhood and the same-layer edge ratio are positively correlated and normalized to obtain the fiber extension direction consistency coefficient of each pixel point; The sum of the directional stability factors of the first representative cluster and the second representative cluster in the first neighborhood of each pixel point is determined as the inter-cluster stability sum value, the inter-cluster stability sum value is positively correlated and the first same-layer edge difference value is anti-correlated, and normalized to obtain the edge correction coefficient of each pixel point.

[0012] Preferably, the method of combining the angle correction coefficient of each pixel point and the neural network model to obtain the optimal calcination time of the petroleum coke base material includes: Calculate the multiplication result of the edge correction coefficient of each pixel point and the fiber extension direction consistency coefficient. If the edge correction coefficient of the pixel point is less than the preset edge threshold, the layer edge extension consistency index of the pixel point is the fiber extension direction consistency coefficient of the pixel point. Otherwise, the layer edge extension consistency index of the pixel point is the multiplication result of the pixel point. The layer edge significance of each pixel point is determined based on the gradient of each pixel point and the layer edge extension matching index, and the gradient in the Canny edge detection algorithm is replaced by the layer edge significance of each pixel point to obtain each edge pixel point in the SEM scanning electron microscope image, and the number of the edge pixel points is used as the layered contour intensity of the corresponding SEM scanning electron microscope image, and the layered contour intensity of a preset number of SEM scanning electron microscope images of the same batch of petroleum coke base materials is used as the input of the neural network model, and the output is the optimal calcination time of the same batch of petroleum coke base materials.

[0013] Preferably, the method of calcining the petroleum coke base material and then cooling it to obtain the calcined coke base material using the optimal calcination time comprises: The petroleum coke base material is crushed by a crusher and then put into a tank furnace for calcination, maintaining the optimal calcination time, wherein the calcination temperature is 1250°C and the pressure in the furnace is 150MPa during calcination. The calcined material is taken out and sent to a cooler to be cooled to room temperature to obtain a calcined coke base material.

[0014] Preferably, the calcined coke base material is ground and granulated to obtain a catalyst specifically for cyanide cracking, comprising: The calcined coke base material is ground by a grinding machine and mechanically granulated to prepare calcined coke with a diameter of 2 mm, thereby obtaining a catalyst specifically for cyanide cracking.

[0015] The present invention has at least the following beneficial effects: The invention constructs the true coefficient of the edge pixel on the same layer through the grayscale similarity between the adjacent pixels on the same layered contour line in the petroleum coke electron microscope image, which reflects the possibility that the non-central pixel and the central pixel in the first neighborhood belong to the same layered contour line. The beneficial effect lies in that the off-center path information of each non-central pixel in the first neighborhood is comprehensively considered, and it is avoided that the high-circle number itself is more likely to have a larger grayscale offset micro-factor than the low-circle number non-central pixel, and a larger tolerance is given to the high-circle number non-central pixel, so as to improve the accuracy of obtaining the true coefficient of the edge pixel on the same layer; based on the fiber extension between the adjacent pixels on the same layered contour line, the fiber extension between the adjacent pixels on the same layered contour line is more likely to have a larger grayscale offset micro-factor than the low-circle number non-central pixel, and the fiber extension between the adjacent pixels on the same layered contour line is more likely to have a larger grayscale offset micro-factor than the low-circle number non-central pixel, and the fiber extension between the adjacent pixels on the same layered contour line is more likely to have a larger grayscale offset micro-factor than the low-circle number non-central pixel, and the fiber extension between the adjacent pixels on the same layered contour line is more likely to have a larger grayscale offset micro-factor. The fiber extension between the adjacent pixels on the same layered contour line is more likely to have a larger grayscale offset micro-factor than the low-circle number non-central pixel, and the ... The similarity of the stretching direction is used to construct the layer edge extension matching index, which reflects the matching of the pixel point in the first neighborhood with the layered contour line in the petroleum coke electron microscope image, and the contour is extracted in combination with the gradient of the pixel point to obtain the optimal calcination time. The beneficial effect is to avoid the problem of unstable gradient differences in the petroleum coke electron microscope image with high magnification and the poor effect of the layered contour line obtained by the Canny edge detection algorithm, improve the accuracy of subsequent contour extraction, and adaptively obtain the optimal calcination time in the calcination process according to the micromorphology of the petroleum coke substrate, while ensuring sufficient calcination and increasing the desulfurization rate of petroleum coke, thereby improving the quality of the special catalyst for cyanide cracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 A flow chart of steps of a method for preparing a catalyst for cyanide cracking provided by one embodiment of the present invention; Figure 2 Obtain flow chart for calcination control indicators; Figure 3A detailed flow chart for the preparation of a special catalyst for cyanide cracking; Figure 4 Schematic diagram of the catalyst entering the cracking furnace. DETAILED DESCRIPTION

[0018] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the catalyst preparation method for cyanide cracking proposed by the present invention, its specific implementation method, structure, characteristics and effects in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0019] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0020] The specific scheme of the method for preparing the catalyst for cyanide decomposition provided by the present invention is described in detail below with reference to the accompanying drawings.

[0021] Example 1 Example 1 provides a method for preparing a catalyst for cyanide cracking. For the preparation flow chart, please refer to Figure 1 , the specific preparation process is as follows: (1) Screening and dehydration: The petroleum coke is screened through a coke screen, and the coke particles smaller than 10 mm are screened out and then transferred to a dehydration tank for dehydration to reduce the moisture content to below 10%; (2) Crushing and grading: Crushing the dehydrated petroleum coke into particles no larger than 5 mm, and grading the crushed petroleum coke through a vibrating screen. Specifically, the particles no larger than 5 mm are classified as qualified and then proceed to the subsequent steps. The particles larger than 5 mm are classified as unqualified and then crushed and graded again; (3) stirring and mixing: adding the crushed petroleum coke particles and sodium hydroxide into a stirrer for stirring and mixing, wherein the mass ratio of the petroleum coke particles to the sodium hydroxide is 2:1, and after the mixture is evenly mixed, adding water and continuing to stir to form a slurry, wherein the mass ratio of the petroleum coke particles to the added water is 1:1; (4) Cooking and stirring: The slurry is placed in a steamer for cooking, wherein the temperature is set to 200°C. During the cooking process of the slurry, it is necessary to continuously stir and mix. After the first cooking time, calcium hydroxide is added, and cooking is continued with stirring. The mass ratio of petroleum coke particles to calcium hydroxide is 2:1. The first cooking time is 1 hour. After the second cooking time, cooking is completed and the petroleum coke base material is obtained from the steamer. The second cooking time is 1.2 hours. After obtaining the petroleum coke base material, the calcination time of the subsequent steps of the petroleum coke base material is adaptively adjusted, specifically including: Step S001, obtaining an electron microscope image of the petroleum coke of the petroleum coke base material through a field emission scanning electron microscope (SEM).

[0022] From the same batch of petroleum coke base material obtained from the steamer, 50 portions of petroleum coke base material are randomly selected and subjected to field emission scanning electron microscope (SEM) to obtain scanning electron microscope images of the petroleum coke base material, which are recorded as petroleum coke electron microscope images, wherein the acceleration voltage (HV) is set to 10 kV, the working distance (WD) is set to 5.1 mm, and the magnification is set to 20000. The implementer can set the number of randomly selected portions of petroleum coke base material according to actual conditions, and this embodiment does not impose any restriction on this.

[0023] Step S002, setting the first neighborhood of each pixel point with each pixel point as the center, and obtaining the true coefficient of the same-layer edge pixel of each non-central pixel point according to the grayscale difference and position relationship between each non-central pixel point and the central pixel point in the first neighborhood.

[0024] Specifically, in this embodiment, the petroleum coke is screened and dehydrated, and then crushed and graded to obtain petroleum coke particles of qualified grade. The petroleum coke particles are mixed with sodium hydroxide, and then water is added to stir to form a slurry. The slurry is steamed and stirred, and then calcium hydroxide is added to continue steaming to obtain a petroleum coke base material. The SEM scanning electron microscope image of the petroleum coke base material is collected to obtain the edge correction coefficient and layer edge significance of each pixel point. The optimal calcination time of the petroleum coke base material is obtained by combining the neural network model. The petroleum coke base material is calcined and cooled using the optimal calcination time, and then ground and granulated to obtain a special catalyst for cyanide cracking. The calcination control index acquisition flow chart is as shown in Figure 2 As shown, the detailed flow chart of the preparation of the special catalyst for cyanide cracking is as follows Figure 3 The construction process of the layer edge extension matching index of each pixel point is as follows: In the cracking of cyanide, ammonia and light oil introduced into the cracking furnace in the production process blow the calcined coke to churn in the furnace. When the calcined coke is blown to a certain height, it will collide with the three-phase electrodes in the cracking furnace, converting the electrical energy in the three-phase electrodes into thermal energy of the calcined coke, playing a role as a carrier for regulating temperature. Figure 4 The schematic diagram of the catalyst entering the cracking furnace is shown.

[0025] In the actual preparation process of the special catalyst for cyanide cracking, due to the scattered purchase channels of petroleum coke, more manufacturers, and the increase in imported crude oil, the volatile content of petroleum coke is high, the sulfur content increases, and the fluctuation range between different batches of petroleum coke is large, which affects the parameters in the subsequent preparation process. At the same time, due to the increase of calcination time, after the CS bond in petroleum coke is destroyed, the internal sulfur-containing group of petroleum coke can be well exposed, and the desulfurization rate of petroleum coke is also constantly rising, but when the calcination time is too long, it is not enough to break the CS bond in more complex sulfur-containing compounds under the calcination time, and no harmful gas can be generated by reaction, so continuing to increase the holding time will not increase the desulfurization rate of petroleum coke, and reduce preparation efficiency. Therefore, the present embodiment is according to the properties of each batch of petroleum coke substrates, and the calcination time in the actual preparation process of the special catalyst for cyanide cracking is adaptively determined, so as to avoid the influence of the fluctuation of petroleum coke components on the special catalyst for cyanide cracking, and improve the quality of the catalyst.

[0026] Petroleum coke with high sulfur content has the characteristics of smooth surface, fine layered structure and high degree of order in microscopic morphology. This microscopic morphology increases in edges and corners, gradually becomes more obvious in layered structure, has a structure perpendicular to the fiber axis, and forms more gullies during the subsequent calcination process. The microscopic base material of petroleum coke base material mainly depends on the effect of cooking and stirring. Sodium hydroxide and calcium hydroxide can absorb the sulfur component of petroleum coke. At the same time, cooking treatment is first carried out, which is conducive to the improvement of the graphitization degree of petroleum coke.

[0027] In the electron microscope image of petroleum coke, the microscopic morphology of petroleum coke has a layered structure and a layered contour line. The more obvious the layered structure is, the higher the initial desulfurization rate of the petroleum coke after cooking, the higher the degree of graphitization, and the shorter the calcination time required for the calcination process. However, the gradient difference in the high-magnification electron microscope image of petroleum coke is unstable, and the effect of the layered contour line obtained by the Canny edge detection algorithm is poor. The grayscale values ​​between the pixels on the layered contour line and the adjacent pixels on the same layered contour line are similar, and have similar fiber extension directions.

[0028] As an embodiment of the present application, a first neighborhood of size 9×9 is constructed with each pixel as the center, wherein the implementer can set the size of the first neighborhood according to actual conditions, and the present application does not impose any restrictions thereon; the absolute value of the difference between the grayscale values ​​of each non-central pixel and the central pixel in the first neighborhood of each pixel is calculated as the grayscale difference of each non-central pixel, the difference between the grayscale difference of any non-central pixel and other non-central pixels in the first neighborhood is determined as the grayscale neighborhood difference, the sum of all the grayscale neighborhood differences of any non-central pixel is determined as the neighborhood difference, the neighborhood difference is forward mapped, and the grayscale difference of any non-central pixel is reversely mapped, to obtain the grayscale offset microfactor of any non-central pixel.

[0029] It should be understood that the forward mapping indicates that the neighborhood difference and the grayscale offset small factor have a positive correlation, which can be specifically a multiplication relationship, an addition relationship, or the power of an exponential function. The reverse mapping indicates that the grayscale difference and the grayscale offset small factor have an inverse correlation, which can be specifically a subtraction relationship, a division relationship, etc. The implementer can determine this based on actual applications, and this application does not impose any limitations here.

[0030] As an embodiment of the present application, the calculation method of the grayscale offset micro-factor of each non-center pixel in the first neighborhood of the pixel point can be: calculate the sum of the grayscale difference of each non-center pixel and the preset adjustment factor greater than 0, and use the ratio of the neighborhood difference of each non-center pixel to the sum as the grayscale offset micro-factor of each non-center pixel, wherein the preset adjustment factor greater than 0 can be set manually to avoid the denominator being 0. In this embodiment, the value is 0.01, and the implementer can set it according to the actual situation. This application does not impose any restrictions on this.

[0031] In the electron microscope image of petroleum coke, the pixels on the layered contour lines are basically directly adjacent to each other, that is, the pixels on the two layered contour lines are adjacent in the eight neighborhoods. Taking pixel i as an example, in the first neighborhood centered on pixel i, a second neighborhood of size 3×3 is constructed with each non-center pixel as the center, wherein the pixels in the second neighborhood are all pixels in the first neighborhood. Taking the jth non-center pixel in the first neighborhood as an example, the pixel with the largest grayscale offset microfactor in the second neighborhood of the jth non-center pixel is obtained, and it is recorded as an off-center combination with the center pixel of the second neighborhood, wherein, if there is a pixel i in the second neighborhood of the jth pixel, the combination of the jth pixel and the pixel i is directly recorded as an off-center combination. It should be noted that, under the premise that the size of the second neighborhood is smaller than the size of the first neighborhood, the implementer can determine the size of the second neighborhood according to the actual application, and this application does not limit it here.

[0032] All pixels in the first neighborhood of pixel i as the center are taken as undirected graph nodes, and the undirected graph nodes corresponding to all off-center combinations are connected to construct an off-center undirected graph of pixel i. Taking the jth non-center pixel as an example, the off-center undirected graph is used as the input of the Breadth First Search (BFS) algorithm, where the undirected graph node corresponding to the jth non-center pixel is taken as the starting point, and the undirected graph node corresponding to pixel i is taken as the end point. The output is each path from the jth non-center pixel to pixel i, and the minimum path is recorded as the off-center path from the jth non-center pixel to pixel i. In addition, according to the order in which all nodes appear in the off-center path, the grayscale offset factors of all nodes are combined into a path offset sequence. Secondly, in the first neighborhood of pixel point i, according to the straight-line distance between the non-central pixel point and pixel point i, the number of circles is divided from small to large and from inside to outside, and recorded as the 1st, 2nd, 3rd, and 4th circle pixels, for example, for the first neighborhood of pixel point i, the 3rd neighborhood pixel points of pixel point i are the 1st circle pixel points, the 4th neighborhood pixel points of pixel point i are the 2nd circle pixel points, the 5th neighborhood pixel points of pixel point i are the 3rd circle pixel points, and the 6th neighborhood pixel points of pixel point i are the 4th circle pixel points. The breadth-first search algorithm is an existing well-known technology, and this application will not be described in detail here.

[0033] It should be understood that this embodiment only proposes a graph search method, namely, the breadth-first search algorithm. Under the premise that the purpose of graph search can be achieved, as other implementation methods, implementers can also adopt other feasible graph search methods in the prior art, and this application does not impose any restrictions on this.

[0034] Based on the above analysis, the true coefficient of the same-layer edge pixel of each non-central pixel in the first neighborhood of each pixel can be obtained. The specific calculation process is: for each off-center path, the difference between the grayscale offset tiny factor of any pixel on the off-center path and the maximum value of the grayscale offset tiny factor of all pixels on the circle where the any pixel is located is calculated, and determined as the grayscale offset difference. The path off-center intensity of each non-central pixel in the first neighborhood is positively correlated with the sum of the grayscale offset differences of all pixels on the corresponding off-center path, and negatively correlated with the number of pixels on the corresponding off-center path; The information entropy of all elements in the path offset sequence of each non-central pixel point in the first neighborhood of each pixel point is calculated, and the ratio of the path off-center intensity of each non-central pixel point in the first neighborhood to the information entropy is determined as the off-center confusion degree. The number of occurrences of the corresponding node of each non-central pixel point in the first neighborhood and the off-center confusion degree of each non-central pixel point are forward mapped to obtain the true coefficient of the same-layer edge pixel of each non-central pixel point in the first neighborhood.

[0035] It can be understood that a positive correlation means that the dependent variable will increase as the independent variable increases, and the dependent variable will decrease as the independent variable decreases. The specific relationship can be a multiplication relationship, an addition relationship, or the power of an exponential function, which is determined by actual application and this application does not impose any special restrictions. A negative correlation means that the dependent variable will decrease as the independent variable increases, and the dependent variable will increase as the independent variable decreases. It can be a subtraction relationship, a division relationship, etc., which is determined by actual application.

[0036] As an embodiment of the present application, the path off-center intensity of each non-central pixel point in the first neighborhood can be expressed as follows: the grayscale offset difference of the pixel point on the off-center path is used as the exponent of an exponential function with a natural constant as the base, and the mean of the corresponding exponential function calculation results of all pixel points on the off-center path is calculated as the path off-center intensity of each non-central pixel point in the first neighborhood; In the first neighborhood, the true coefficient of the same-layer edge pixel of each non-central pixel can be expressed as follows: the product of the off-center disorder of the non-central pixel and the number of occurrences of the corresponding node of the non-central pixel in all off-center paths of the first neighborhood is taken as the true coefficient of the same-layer edge pixel of the non-central pixel; As other embodiments of the present application, the true coefficient of the same-layer edge pixels of each non-central pixel point can also be expressed as: the sum of the off-center chaos degree of the non-central pixel point and the number of times the corresponding nodes of the non-central pixel point appear in all off-center paths of the first neighborhood is used as the true coefficient of the same-layer edge pixels of the non-central pixel point.

[0037] The closer the grayscale offset tiny factor of the node in the off-center path is to the maximum value of the grayscale offset tiny factor in the circle to which it belongs, that is, the larger the grayscale offset difference is, the smaller the grayscale offset of the nodes in the off-center path may be from the central pixel point, the more the grayscale value of the node in the off-center path is biased towards the grayscale value of pixel point i, and the greater the intensity of the path off-center.

[0038] The greater the intensity of path eccentricity and the smaller the information entropy of the path offset sequence, that is, the greater the degree of eccentricity, the more the grayscale value of the node in the eccentric path is biased towards the grayscale value of pixel i, and the more stable the degree of bias, the more likely this eccentric path is to be part of the layered contour line in the petroleum coke scanning electron microscope. At the same time, when the number of times the j-th non-central pixel point appears in all eccentric paths is greater, it means that from the first neighborhood information, the j-th non-central pixel point is more likely to be on the same petroleum coke layered contour line as pixel point i, and the greater the true coefficient of the same-layer edge pixel.

[0039] Step S003, combining the true coefficients of the same-layer edge pixels of each non-central pixel and the clustering algorithm to obtain the edge correction coefficient of each pixel.

[0040] Taking pixel i as an example, the true coefficients of the same-layer edge pixels of all non-central pixels in the first neighborhood of pixel i are used as the input of the Otsu method and output as the segmentation threshold. The non-central pixels whose true coefficients of the same-layer edge pixels are greater than or equal to the segmentation threshold are recorded as suspected same-layer edge pixels.

[0041] The pixel points on the layered contour line have similar fiber extension directions with the adjacent pixel points on the same layered contour line. The eccentric path of each suspected same-layer edge pixel point is obtained, and the corresponding pixel points of the nodes in the eccentric path are connected in the order of appearance. The slope of each connection line is calculated, and the average of all connection line slopes is recorded as the suspected fiber extension slope of each suspected same-layer edge pixel point. Secondly, the suspected fiber extension slope of each suspected same-layer edge pixel point is used as the input of the DBSCAN clustering algorithm, where the neighborhood radius is 0.1 in this embodiment, and the minimum number of neighborhood points, that is, the MinPts parameter, is 2 in this embodiment. Each cluster cluster is output. The DBSCAN clustering algorithm is an existing well-known technology and will not be described in detail in this embodiment. Further, the number of data elements in each cluster cluster is counted, and the cluster cluster with the largest number of data elements is recorded as the first representative cluster, and the cluster cluster with the second largest number of data elements is recorded as the second representative cluster. The number of other data elements within a 0.1 neighborhood radius of each data element in a representative cluster is counted as the core density of each data element, and the data element with the largest core density is recorded as the suspected directional core point.

[0042] It should be noted that this embodiment only proposes a threshold segmentation method and a data clustering method, that is, the segmentation threshold is obtained by using the Otsu method, and clustering is performed using the DBSCAN clustering algorithm. On the premise that the purpose of threshold segmentation and data clustering can be achieved, as other implementation methods, the implementer can also adopt other feasible threshold segmentation methods and data clustering methods in the prior art, and this application does not impose any restrictions on this.

[0043] Based on the above analysis, the range values ​​of the suspected fiber extension slopes in each cluster in the first neighborhood of each pixel point are calculated, and the sum of the reciprocals of the core direction deviation values ​​of all suspected same-layer edge pixels in each cluster is calculated and determined as the sum of direction deviations. The range values ​​are inversely mapped and the sum of direction deviations are directly mapped to obtain the directional stability factor of each cluster in the first neighborhood of each pixel point, where the core direction deviation value is the Euclidean distance between the suspected same-layer edge pixel point and the suspected direction core point.

[0044] It should be noted that a positive proportional mapping indicates that there is a positive correlation between variables, which may be multiplication, addition, etc., and an inverse proportional mapping indicates that there is a negative correlation between variables, which may be subtraction, division, etc. The implementer may determine this based on the actual application, and this application does not impose any restrictions here.

[0045] In one embodiment of the present application, the specific calculation process of the directional stability factor of the z-type representative cluster in the first neighborhood of the pixel point i can be: the opposite number of the extreme difference value of the suspected fiber extension slope in the z-type representative cluster in the first neighborhood of the pixel point i is used as the exponent of the exponential function with the natural constant as the base, and the product of the calculation result of the exponential function and the sum of the directional deviations of the z-type representative cluster in the first neighborhood of the pixel point i is used as the directional stability factor of the z-type representative cluster in the first neighborhood of the pixel point i. Wherein, z belongs to one or two.

[0046] For each pixel point, the difference in the number of suspected same-layer edge pixels of the first representative cluster and the second representative cluster in the first neighborhood is determined as the first same-layer edge difference value, the difference in the number of all suspected same-layer edge pixels in the first neighborhood and the number of suspected same-layer edge pixels in the first representative cluster is determined as the second same-layer edge difference value, the ratio of the first same-layer edge difference value to the second same-layer edge difference value is used as the same-layer edge ratio, the directional stability factor of the first representative cluster in the first neighborhood and the same-layer edge ratio are positively correlated and normalized to obtain the fiber extension direction consistency coefficient of each pixel point.

[0047] It should be understood that a positive correlation mapping also indicates that there is a positive correlation between variables.

[0048] In one embodiment of the present application, the fiber extension direction consistency coefficient of pixel point i can be expressed as: calculating the product of the same-layer edge ratio of pixel point i and the directional stability factor of a representative cluster in the first neighborhood of pixel point i, and taking the normalized value of the product as the fiber extension direction consistency coefficient of pixel point i.

[0049] The sum of the directional stability factors of the first representative cluster and the second representative cluster in the first neighborhood of each pixel point is determined as the inter-cluster stability sum value, the inter-cluster stability sum value of the pixel point is positively correlated and the first same-layer edge difference value is anti-correlated, and normalized to obtain the angular correction coefficient of each pixel point.

[0050] It should be understood that the anti-correlation mapping indicates that there is a negative correlation between variables.

[0051] As an embodiment of the present application, the edge correction coefficient of pixel point i can be expressed as: calculating the ratio of the inter-cluster stable sum value of pixel point i to the first same-layer edge difference value as the edge correction coefficient of pixel point i.

[0052] The more the suspected same-layer edge pixels are concentrated in the first representative cluster and the fewer the number of suspected same-layer edge pixels in the second representative cluster, it means that for the first neighborhood, pixel i and the suspected same-layer edge pixel may have only one fiber extension direction. At the same time, the more stable the suspected fiber extension slope in the first representative cluster, that is, the larger the direction stability factor of the first representative cluster, the more accurate the first representative cluster is obtained and the more it can be approximately regarded as a fiber extension direction. Then, the more consistent the fiber extension direction of pixel i in the first neighborhood, the larger the fiber extension direction consistency coefficient, and the more likely pixel i is a pixel on the layered contour line of petroleum coke.

[0053] When the number of suspected same-layer edge pixels in the first representative cluster and the second representative cluster is closer, that is, the first same-layer edge difference is smaller, it means that the first neighborhood of pixel i is more likely to have two fiber extension directions. At the same time, when the suspected fiber extension slopes in the first representative cluster and the second representative cluster are more stable, that is, the larger the inter-cluster stability and value is, it means that the two representative clusters can each be approximately regarded as the fiber extension direction, then the pixel i is more likely to be the angular pixel in the layered contour line of petroleum coke, and the angular correction coefficient is larger.

[0054] Step S004, combining the edge correction coefficient of each pixel point and the neural network model to obtain the optimal calcination time of the petroleum coke base material.

[0055] The multiplication result of the edge correction coefficient of each pixel and the fiber extension direction consistency coefficient is calculated. If the edge correction coefficient of the pixel is less than the preset edge threshold, the layer edge extension consistency index of the pixel is the fiber extension direction consistency coefficient of the pixel, otherwise, the layer edge extension consistency index of the pixel is the multiplication result of the pixel. In one embodiment of the present application, the preset edge threshold is set to 0.6, and the implementer can set it according to the actual situation, and the present application does not limit it here.

[0056] When the range value of the suspected fiber extension slope in the z-type representative cluster is smaller, it means that the range of the suspected fiber extension slope of each suspected same-layer edge pixel point in the z-type representative cluster is smaller. At the same time, when the core direction deviation value is smaller, it means that each suspected same-layer edge pixel point in the z-type representative cluster is closer to the suspected direction core point in the cluster space, and the suspected fiber extension slope in the z-type representative cluster is more stable, then the direction stability factor is larger.

[0057] Taking pixel point i as an example, the greater the possibility that pixel point i is an edge in the layered contour line of petroleum coke, that is, the larger the edge correction coefficient, and the edge correction coefficient is greater than or equal to the preset edge threshold, it means that the fiber extension direction consistency of pixel point i is lower, and the more it needs to be corrected by the edge correction coefficient, the more consistent the fiber extension characteristics are with the pixel on the layered contour line of petroleum coke, the larger the layer edge extension consistency index, and the more likely pixel point i is a pixel on the layered contour line of petroleum coke.

[0058] At this point, the layer edge extension conformity index of each pixel point in the petroleum coke electron microscope image can be obtained. The larger the value, the more consistent the fiber extension feature is with the pixel point on the petroleum coke layered contour line in the first neighborhood, and the more likely it belongs to the petroleum coke layered contour line. The gradient of each pixel point in the petroleum coke electron microscope image is calculated, where the calculation of the gradient is an existing well-known technology, and the specific process is not described in detail. The layer edge significance of each pixel point in the petroleum coke electron microscope image is constructed, and the layer edge significance of each pixel point maintains a positive correlation with the layer edge extension conformity index and gradient of the pixel point.

[0059] As an embodiment of the present application, the layer edge saliency of the pixel point i may be the product of the layer edge extension coincidence index of the pixel point i and the gradient.

[0060] The layer edge significance of each pixel point in the petroleum coke electron microscope image is obtained. The greater the layer edge significance, the more likely it is to be a pixel point of the petroleum coke layered contour line. The layer edge significance is replaced by the gradient in the Canny edge detection to obtain each petroleum coke layered contour line in the petroleum coke electron microscope image. The Canny edge detection is an existing well-known technology and is not described in detail in this embodiment. The number of pixel points on the layered contour line of the petroleum coke is counted and recorded as the layered contour intensity. The greater the layered contour intensity, the more likely the petroleum coke electron microscope image is to have a significant layered structure, the more obvious the structure perpendicular to the fiber axis, and the less smooth the surface under the microscopic morphology. In terms of the microscopic morphology of the petroleum coke, the higher the initial desulfurization rate of the petroleum coke after cooking, the higher the degree of graphitization, and the shorter the calcination time required for the calcination process.

[0061] The layered contour intensity corresponding to each petroleum coke electron microscope image is obtained, and the difference between the layered contour intensity and the standard layered contour intensity in the preparation process of the special catalyst for cyanide cracking is calculated, and recorded as the layered contour standard deviation. Secondly, a CNN (Convolutional Neural Network) neural network model is used, with the mean square error MSE (Mean-square error) function as the loss function and the Adam algorithm as the optimization algorithm. Each petroleum coke electron microscope image and its corresponding layered contour standard deviation are used as the input of the network model, and the output is the optimal calcination time during the calcination of the petroleum coke substrate. The training process of the CNN neural network is an existing well-known technology and will not be described in detail.

[0062] Further, the petroleum coke base material is subsequently processed according to the following steps: (5) Calcination: After the petroleum coke base material is crushed again by a crusher, the crushed petroleum coke base material is put into a tank furnace for calcination, and the optimal calcination time is maintained, wherein the calcination temperature is 1250°C and the pressure in the furnace during calcination is 150MPa; (6) Cooling: The calcined material is taken out and sent to a cooler to be cooled to room temperature to obtain a calcined coke base material; (7) Grinding and granulation: The calcined coke base material is ground by a grinder and mechanically granulated to produce calcined coke with a diameter of 2 mm. The calcined coke is a special catalyst for cyanide cracking.

[0063] In summary, the embodiments of the present invention avoid the problem of unstable gradient differences in high-magnification petroleum coke electron microscope images and poor effect of layered contour lines obtained using the Canny edge detection algorithm, improve the accuracy of subsequent contour extraction, and adaptively obtain the optimal calcination time in the calcination process according to the microscopic morphology of the petroleum coke substrate, thereby ensuring sufficient calcination while increasing the desulfurization rate of the petroleum coke and improving the quality of the special catalyst for cyanide cracking.

[0064] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0065] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0066] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for preparing a catalyst for cyanide cracking, It is characterized in that The method comprises the following steps: The preparation raw materials include petroleum coke, sodium hydroxide and calcium hydroxide; the petroleum coke is screened and dehydrated, and then crushed and graded to obtain petroleum coke particles of qualified grade; the petroleum coke particles and sodium hydroxide are stirred and mixed, and then water is added and stirred to form slurry; the slurry is steamed and stirred, and then calcium hydroxide is added, and the steaming is continued to obtain the petroleum coke base material; Adaptive adjustment of the calcination time of petroleum coke base material includes: Collect the SEM scanning electron microscope image of the petroleum coke base material; set the first neighborhood of each pixel point with each pixel point as the center; analyze the grayscale difference and position relationship between each non-central pixel point and the central pixel point in the first neighborhood to obtain the true coefficient of the same-layer edge pixel of each non-central pixel point; The edge correction coefficient of each pixel is obtained by combining the true coefficient of the edge pixels of the same layer of each non-central pixel and the clustering algorithm; The optimal calcination time of the petroleum coke base material is obtained by combining the edge correction coefficient of each pixel point and the neural network model; The petroleum coke base material is calcined and cooled using an optimal calcination time to obtain a calcined coke base material; the calcined coke base material is ground and granulated to obtain a special catalyst for cyanide cracking.

2. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The screening and dehydration comprises: The petroleum coke is screened through a coke screen, and the coke particles smaller than 10 mm are screened out and sent to a dehydration tank for dehydration to reduce the moisture content to below 10%.

3. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The crushing and classification comprises: The dehydrated petroleum coke is crushed to a size of 5 mm or less, and the crushed petroleum coke is graded through a vibrating screen. The petroleum coke particles with a particle size of 5 mm or less are qualified, and the petroleum coke particles with a particle size greater than 5 mm are unqualified. The unqualified petroleum coke particles are crushed and graded again until they reach the qualified grade.

4. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The step of stirring and mixing petroleum coke particles and sodium hydroxide and then adding water comprises: The mass ratio of petroleum coke particles to sodium hydroxide is 2:1, and the mass ratio of petroleum coke particles to added water is 1:

1.

5. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The slurry is boiled and stirred, calcium hydroxide is added, and the boiling is continued to obtain a petroleum coke base material, comprising: The cooking temperature is 200° C., the mass ratio of petroleum coke particles to calcium hydroxide is 2:1, the cooking time is 1 hour, and the continued cooking time is 1.2 hours.

6. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The process of determining the true coefficients of the same-layer edge pixels of each non-central pixel point includes: Calculate the absolute value of the difference between the grayscale values ​​of each non-central pixel and the central pixel in the first neighborhood of each pixel, determine it as the grayscale difference of each non-central pixel, determine the difference between the grayscale difference of any non-central pixel and other non-central pixels in the first neighborhood as the grayscale neighborhood difference, determine the sum of all the grayscale neighborhood differences of any non-central pixel as the neighborhood difference degree, forward map the neighborhood difference degree and reverse map the grayscale difference of any non-central pixel, and obtain the grayscale offset microfactor of any non-central pixel; A second neighborhood is set with each non-central pixel point in the first neighborhood as the center, wherein the second neighborhood is smaller than the first neighborhood, and all pixels in the second neighborhood belong to the first neighborhood, and the pixel point corresponding to the maximum value of the grayscale offset micro-factor in the second neighborhood and the central pixel point of the second neighborhood are recorded as an off-center combination; All pixels in the first neighborhood are taken as undirected graph nodes, and the undirected graph nodes corresponding to all off-center combinations are connected. The paths from each non-center pixel in the first neighborhood to the center pixel are obtained by combining the graph search algorithm, which are recorded as off-center paths from each non-center pixel to the center pixel. The grayscale offset micro factors of all pixels on each off-center path are arranged in path order as the path offset sequence of each non-center pixel. All non-center pixels in the first neighborhood are divided into circles from inside to outside according to their distance from the center pixel. For each of the off-center paths, the difference between the grayscale offset microfactor of any pixel point on the off-center path and the maximum grayscale offset microfactor of all pixel points on the circle where the any pixel point is located is calculated to determine the grayscale offset difference. The path off-center intensity of each non-center pixel point in the first neighborhood is positively correlated with the sum of the grayscale offset differences of all pixel points on the corresponding off-center path, and negatively correlated with the number of pixel points on the corresponding off-center path. The information entropy of all elements in the path offset sequence of each non-central pixel point in the first neighborhood of each pixel point is calculated, and the ratio of the path off-center intensity of each non-central pixel point in the first neighborhood to the information entropy is determined as the off-center confusion degree. The number of occurrences of the corresponding node of each non-central pixel point in the first neighborhood and the off-center confusion degree of each non-central pixel point are forward mapped to obtain the true coefficient of the same-layer edge pixel of each non-central pixel point in the first neighborhood.

7. A method for preparing a catalyst for cyanide decomposition according to claim 6, It is characterized in that The process of determining the edge correction coefficient of each pixel point includes: Perform threshold segmentation on the true coefficients of the same-layer edge pixels of all non-central pixels in the first neighborhood of any pixel, record the non-central pixels whose true coefficients of the same-layer edge pixels are greater than or equal to the segmentation threshold as suspected same-layer edge pixels, connect all pixels on the off-center path of each suspected same-layer edge pixel in sequence according to the path order, calculate the slope of each connection line, and take the average of all connection line slopes as the suspected fiber extension slope of each suspected same-layer edge pixel; Cluster the suspected fiber extension slopes of all suspected same-layer edge pixels in the first neighborhood of any pixel point, record the cluster with the largest number of elements as the first-class representative cluster, and the cluster with the second largest number of elements as the second-class representative cluster. Count the number of other elements of each element in the first-class representative cluster within the preset neighborhood radius as the core density of each element, and record the element with the largest core density as the suspected direction core point; Calculate the extreme value of the suspected fiber extension slope in each cluster in the first neighborhood of each pixel point, calculate the sum of the reciprocals of the core direction deviation values ​​of all suspected same-layer edge pixel points in each cluster cluster, determine it as the direction deviation sum, and map the extreme value inversely and the direction deviation sum in direct proportion to obtain the directional stability factor of each cluster in the first neighborhood of each pixel point, wherein the core direction deviation value is the Euclidean distance between the suspected same-layer edge pixel point and the suspected direction core point; For each pixel point, the difference between the number of suspected same-layer edge pixel points of the first representative cluster and the second representative cluster in the first neighborhood is determined as the first same-layer edge difference value, the difference between the number of all suspected same-layer edge pixel points in the first neighborhood and the number of suspected same-layer edge pixel points in the first representative cluster is determined as the second same-layer edge difference value, the ratio of the first same-layer edge difference value to the second same-layer edge difference value is used as the same-layer edge ratio, the directional stability factor of the first representative cluster in the first neighborhood and the same-layer edge ratio are positively correlated and normalized to obtain the fiber extension direction consistency coefficient of each pixel point; The sum of the directional stability factors of the first representative cluster and the second representative cluster in the first neighborhood of each pixel point is determined as the inter-cluster stability sum value, the inter-cluster stability sum value is positively correlated and the first same-layer edge difference value is anti-correlated, and normalized to obtain the edge correction coefficient of each pixel point.

8. A method for preparing a catalyst for cyanide decomposition according to claim 7, It is characterized in that The method of combining the angle correction coefficient of each pixel point and the neural network model to obtain the optimal calcination time of the petroleum coke base material includes: Calculate the multiplication result of the edge correction coefficient of each pixel point and the fiber extension direction consistency coefficient. If the edge correction coefficient of the pixel point is less than the preset edge threshold, the layer edge extension consistency index of the pixel point is the fiber extension direction consistency coefficient of the pixel point. Otherwise, the layer edge extension consistency index of the pixel point is the multiplication result of the pixel point. The layer edge significance of each pixel point is determined based on the gradient of each pixel point and the layer edge extension matching index, and the gradient in the Canny edge detection algorithm is replaced by the layer edge significance of each pixel point to obtain each edge pixel point in the SEM scanning electron microscope image, and the number of the edge pixel points is used as the layered contour intensity of the corresponding SEM scanning electron microscope image, and the layered contour intensity of a preset number of SEM scanning electron microscope images of the same batch of petroleum coke base materials is used as the input of the neural network model, and the output is the optimal calcination time of the same batch of petroleum coke base materials.

9. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The method of calcining the petroleum coke base material using the optimal calcination time and then cooling it to obtain the calcined coke base material comprises: The petroleum coke base material is crushed by a crusher and then put into a tank furnace for calcination, maintaining the optimal calcination time, wherein the calcination temperature is 1250°C and the pressure in the furnace is 150MPa during calcination. The calcined material is taken out and sent to a cooler to be cooled to room temperature to obtain a calcined coke base material.

10. A method for preparing a catalyst for cyanide decomposition according to claim 1, It is characterized in that The calcined coke base material is ground and granulated to obtain a catalyst specifically for cyanide cracking, comprising: The calcined coke base material is ground by a grinding machine and mechanically granulated to prepare calcined coke with a diameter of 2 mm, thereby obtaining a catalyst specifically for cyanide cracking.

Images