Model generation method, granulation method for powdered raw material, and manufacturing method for sintered ore

Abstract

To provide a model creating method, a method for granulating powder raw materials, and a method for producing sintered ore, wherein: when granulating powder raw materials using a pan-type pelletizer, a liquid for obtaining liquid cross-linking strength is injected under optimal conditions, it is possible to stably produce the granulated material having predetermined strength and particle size, with small variations in these properties.SOLUTION: A model creating method is applied to a method for granulating powder raw materials, which includes using a pan-type pelletizer rotating with the pan inclined, injecting the powder raw materials into the pan, and simultaneously injecting a liquid for obtaining liquid cross-linking strength from one site or plural sites. This model creating method creates a model for predicting the strength and / or particle size of the granulated material by performing machine learning by incorporating the liquid injection position as an explanatory variable with the strength and / or particle size of the granulated material as a target variable on the basis of the evaluation results of the strength and / or particle size of the granulated material for the liquid injection position.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a model generating method, a granulating method for powdered ore raw material, and a manufacturing method for sintered ore. [Background technology]

[0002] Pan-type pelletizers are widely used as granulators for powdered and granular materials. In these pan-type pelletizers, the pan is set at a predetermined angle, and the powdered and granular materials are tumbled inside the pan that rotates in this inclined state to granulate them.

[0003] A liquid (water or an aqueous solution containing a binder) is added to the powdered and granular raw materials in the pan, and this liquid exerts a liquid bridging force, which helps to bond the powder and granules together. During the granulation process, the powder and granules come into contact with each other as they roll down from the top of the pan, and are compressed together as they are lifted along the side walls of the pan. This process is repeated to produce strong granules.

[0004] In general, in the granulation process of powdered and granular materials, smaller powder particles adhere to larger core particles, and granulation proceeds. In a continuous pan-type pelletizer, when the powdered and granular materials are granulated to a certain size according to the inclination angle and rotation speed of the pan, the granules that tend to be distributed on the surface spill out from the edge of the pan, and a product (granules) with a uniform particle size is collected.

[0005] Pan-type pelletizers are widely used in food manufacturing and the preliminary processing of industrial raw materials, for example, in the steel industry, where they are used to granulate sintered raw materials (sintered raw materials made of powdered ore as the main raw material and limestone-based powdered raw materials as secondary raw materials) for producing sintered ore for steelmaking. In order to obtain granulated materials with a certain strength and particle size (particle size, size) using this pan-type pelletizer, it is important that the liquid that exerts the liquid bridging force is evenly distributed over the powdered raw materials in the pan.

[0006] Conventionally, when granulating sinter raw materials for sinter ore production, water is added (sprayed) to the sinter raw materials in the pan using a watering hose. At that time, there is no specific watering position, and water is added to several places within the range where the watering hose can be routed so that the water is distributed as evenly as possible to the sinter raw materials in the pan.

[0007] On the other hand, Patent Document 1 proposes a method for producing carbonaceous core granulated particles having a two-layer structure in which the periphery of the carbonaceous core is covered with a mixed powder of iron ore powder and lime-containing raw material, in which water is selectively sprayed onto the area inside the granulator where the carbonaceous core appears on the surface and rolls. [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Patent No. 6683155 Summary of the Invention [Problem to be solved by the invention]

[0009] The method proposed in Patent Document 1 is an effective water addition method for producing carbonaceous composite granulated particles with a two-layer structure, but the target of granulation is limited. In granulation of powdered granular materials using a pan-type pelletizer, it is considered important at which position in the pan to add the liquid in order to distribute the liquid (moisture) evenly throughout the powdered granular materials and obtain high-quality granulated materials with a certain strength and particle size. However, the conventional techniques such as Patent Document 1 do not propose the optimal position for adding the liquid, such as where to place a watering hose.

[0010] The present invention has been made in consideration of the above, and aims to provide a model generation method, a granulation method for powdered ore, and a manufacturing method for sintered ore, which are capable of stably obtaining granulated material having predetermined strength and particle size and little variation in strength and particle size by adding liquid under optimal conditions to obtain a liquid bridging force when granulating powdered ore raw materials using a pan-type pelletizer. [Means for solving the problem]

[0011] In order to solve the above-mentioned problems and achieve the objective, the model generation method of the present invention uses a pan-type pelletizer in which the pan rotates at an inclination, and in this method powdered granular raw material is fed into the pan, and at the same time, liquid for obtaining a liquid bridging force is fed from one or more locations to granulate the powdered granular raw material.Based on the evaluation results of at least one of the strength and particle size of the granulated product relative to the liquid feeding position, the liquid feeding position is included as an explanatory variable, and machine learning is performed using at least one of the strength and particle size of the granulated product as an objective variable to generate a model that predicts at least one of the strength and particle size of the granulated product.

[0012] In addition, in the model generation method according to the present invention, in the above invention, successive approximation optimization is performed as the machine learning.

[0013] In the model generation method according to the present invention, in the above invention, the successive approximation optimization is Bayesian optimization.

[0014] In order to solve the above-mentioned problems and achieve the objective, the granulation method of powdered or granular raw materials of the present invention uses a model generated by the above-mentioned model generation method to predict the liquid injection position, and by injecting liquid from the predicted liquid injection position, at least one of the strength and particle size of the granulated material is controlled within a desired range, thereby granulating the powdered or granular raw materials.

[0015] In order to solve the above-mentioned problems and achieve the object, the model generation method of the present invention uses a pan-type pelletizer in which the pan rotates with an inclined state, and feeds powdered or granular raw material into the pan, and at the same time feeds liquid from one or more locations to obtain a liquid bridging force, to granulate the powdered or granular raw material.Based on the evaluation results of at least one of the strength and particle size of the granulated product for the liquid feeding position, the liquid feeding position is included as an explanatory variable, and at least one of the strength and particle size of the granulated product is used as an objective variable, and successive approximation optimization is performed to determine the liquid feeding position at which at least one of the strength and particle size of the granulated product falls within a desired range, and the powdered or granular raw material is granulated.

[0016] Further, in the method for granulating powdered or granular raw material according to the present invention, in the above-mentioned invention, the powdered or granular raw material is a sintered raw material for producing sintered ore.

[0017] In order to solve the above-mentioned problems and achieve the object, the method for producing sintered ore of the present invention granulates sintered raw material for producing sintered ore by the above-mentioned method for granulating powdered ore raw material, and sinters the granulated material to produce sintered ore. Effect of the Invention

[0018] According to the model generation method, granulation method for powdered ore raw material, and sintered ore manufacturing method of the present invention, when granulating powdered ore raw material using a pan-type pelletizer, by adding liquid under optimal conditions to obtain a liquid bridging force, it is possible to stably obtain granulated material having predetermined strength and particle size and small variation in strength and particle size. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a schematic diagram showing an example of a pan-type pelletizer to which the present invention can be applied, in which (a) is a perspective view showing the arrangement of a pan and a scraper relative to the pan, and (b) is a side view of the pan. [Diagram 2] FIG. 2 is a schematic diagram showing another example of a pan-type pelletizer to which the present invention can be applied, and is a plan view showing the arrangement of a pan and a scraper relative to the pan. [Diagram 3]FIG. 3 is a schematic diagram showing another example of a pan-type pelletizer to which the present invention can be applied, and is a plan view showing the arrangement of a pan and a scraper relative to the pan. [Figure 4] FIG. 4 is a perspective view showing the results of analyzing the granulation process by the pan-type pelletizer of FIG. 1 using the discrete element method. [Diagram 5] FIG. 5 is a schematic diagram showing the movement of powdered granular raw material in a pan during the granulation process using a pan-type pelletizer, where (a) is an explanatory diagram showing the movement of the bottom layer of the powdered granular raw material layer, (b) is an explanatory diagram showing the movement of the middle layer of the powdered granular raw material layer, and (c) is an explanatory diagram showing the movement of the surface layer of the powdered granular raw material layer. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] An embodiment of a model generating method, a granulating method for powdered ore raw material, and a manufacturing method for sintered ore according to the present invention will be described with reference to the drawings. Note that the components in the following embodiments include those that are easily replaceable by a person skilled in the art, or those that are substantially the same.

[0021] In the granulation method of powdered ore according to the present embodiment, a pan-type pelletizer with a rotating inclined pan is used, and the powdered ore raw material is fed into the pan, and at the same time, liquid for obtaining a liquid bridging force is fed from one or more points to granulate the powdered ore raw material. In the production method of sintered ore according to the present embodiment, sintered raw material for sintered ore production is granulated by the above-mentioned granulation method of powdered ore raw material, and the granulated material is sintered to produce sintered ore. In the model generation method according to the present embodiment, a model (prediction model, trained model) for predicting at least one of the strength and grain size of the granulated material is generated in the above-mentioned granulation method of powdered ore raw material.

[0022] The present inventors have investigated the effect of the injection position of the liquid (liquid for obtaining liquid bridging force, hereinafter simply referred to as "liquid") that contributes to adhesion between powder and granular materials in the granulation of powder and granular materials using a pan-type pelletizer on the strength and uniformity of the granulated material. As a result of detailed investigation, they have found a method for determining the optimal liquid injection position.

[0023] 1 shows a schematic diagram of an example of a pan-type pelletizer to which the present invention can be applied, in which (a) is a perspective view showing the arrangement of a pan 1 and a scraper 2 relative to it, and (b) is a side view of the pan 1. In the following description, the bottom surface of the pan refers to the circular plate surface (upper surface) on which the powdered raw material rolls to be granulated.

[0024] Pan 1 comprises a circular bottom plate portion 10 and a side wall portion 11 surrounding this bottom plate portion 10. Pan 1 is usually held in an upright (inclined) position at a certain angle θ (see FIG. 1(b)) from the horizontal, and rotates in this inclined position. In the example of FIG. 1, pan 1 rotates counterclockwise as viewed in the figure, as indicated by the arrow. Pan 1 shown in FIGS. 2 to 4, which will be described later, also rotates counterclockwise as viewed in the figure.

[0025] The scraper 2 is a plate-shaped member. The scraper 2 has the function of scraping off powdered raw materials adhering to the side walls and bottom surface of the pan 1. Furthermore, the scraper 2 has the function of guiding the powdered raw materials that have moved above the inclined bottom surface of the pan downwards, and regulating (limiting) the rolling area of ​​the powdered raw materials on the bottom surface of the pan. Without this scraper 2, the powdered raw materials would rotate while still adhering to the side walls etc. of the pan 1, reducing the rolling effect of the raw materials rolling down from the top of the pan 1. Therefore, although the scraper 2 is not essential, it is used in many cases.

[0026] Scraper 2 is supported by a support arm (not shown) and is arranged so that one end 20 thereof is close to the side wall surface at top A of the inclined pan bottom surface or its periphery (top A of pan 1 in the example of FIG. 1). Scraper 2 is arranged so that its longitudinal direction is inclined toward the side where the upper half of the pan rotates (the rotation direction of the upper half of the pan shown by the arrow in FIG. 1) with respect to the pan radial direction based on one end 20. Scraper 2 shown in FIG. 1 has a length approximately the same as the radius of pan 1, and is arranged so as to be inclined, for example, by about φ30° from a perpendicular line g on the pan bottom surface that passes through one end 20.

[0027] 2 and 3 are schematic plan views showing another example of a pan-type pelletizer to which the present invention can be applied, showing the arrangement of a pan 1 and a scraper 2 relative to the pan 1. In the pan-type pelletizer shown in Fig. 2, the scraper 2 has a length of about 1 / 3 the diameter of the pan 1. Moreover, one end 20 of the scraper 2 is located on the periphery of the top A of the inclined pan bottom, and is arranged in a state inclined by, for example, about φ40° from a perpendicular line g on the pan bottom that passes through the one end 20.

[0028] On the other hand, in the pan-type pelletizer shown in Figure 3, the scraper 2 is composed of multiple scraper members 2a (two scraper members 2a in the example of Figure 3) that are parallel to each other and spaced apart in the vertical direction of the inclined pan 1.

[0029] Each scraper member 2a is a short member having a length of about 1 / 4 to 1 / 3 of the radius of the pan 1. The uppermost scraper member 2a among the scraper members 2a has one end 20a located on the periphery of the top A of the inclined pan bottom. The uppermost scraper member 2a is also arranged with its longitudinal direction inclined toward the side where the upper half of the pan rotates (the direction of rotation of the upper half of the pan shown by the arrow in FIG. 3) with respect to the pan radial direction based on one end 20a.

[0030] The uppermost scraper member 2a shown in Fig. 3 is arranged at an inclination of, for example, about φ15° from a perpendicular line g on the pan bottom surface passing through one end portion 20a of the scraper member 2a. The other scraper members 2a below it are also arranged at the same inclination as the upper scraper member 2a. The scraper 2 may be composed of three or more scraper members 2a.

[0031] Pan-type pelletizers may be operated in batch mode, but industrially they are generally operated continuously, with the powdered raw material and liquid being continuously fed into pan 1 from their respective feeding means. Binders to promote granulation are mixed into the powdered raw material in the hope of thickening and hardening it, and various inorganic or organic solid or liquid binders are used depending on the application. For example, when granulating sintered raw material (a mixture of fine ore and auxiliary materials) for sintered ore production, quicklime is generally mixed into the sintered raw material at about 1 to 2 mass%.

[0032] The liquid is added in an amount that effectively exerts a liquid bridging force on the powdered raw material. When granulating raw material for sintering, the adhesiveness is high at about 8 to 12 mass%, and adhesiveness decreases outside this range, so the liquid is generally added in this range.

[0033] In order to spray the liquid evenly, it is desirable to disperse it by spray, but when granulating the raw material for sintering, the spray nozzles quickly become clogged, causing problems. Therefore, water is added (sprayed) to the raw material for sintering in several places within the range where the spray hose can be routed.

[0034] Here, in order to obtain high-quality granulated material having a certain strength and particle size, it is important to decide at what position in the pan 1 the liquid is poured, and it is necessary to optimize the liquid pouring position. This process of optimizing the liquid pouring position requires trial experiments. In this trial experiment, for example, trials are performed multiple times to evaluate at least one of the strength and particle size of the granulated material while changing the liquid pouring position. Then, by performing machine learning using the liquid pouring position as an explanatory variable and at least one of the strength and particle size of the granulated material as an objective variable, it becomes possible to control at least one of the strength and particle size of the granulated material to a desired range.

[0035] Trial experiments and machine learning for optimizing the liquid injection position are mainly performed by an information processing device consisting of a general-purpose computer such as a workstation or a personal computer. This information processing device includes a calculation means realized by a CPU (Central Processing Unit) or the like, a storage means realized by various recording media, an input / output means, and the like.

[0036] For example, in Figure 1, facing directly onto the bottom of the pan, with center p as the origin, x coordinates are taken horizontally to the right and y coordinates are taken vertically upward, then the x(Sn) and y(Sn) coordinates of the n liquid injection positions Sn are the explanatory variables.

[0037] The feed position R of the powdered and granular raw materials (sintering raw materials) (hereinafter referred to as the "raw material feed position") is often determined by the equipment specifications, but if there is a possibility that the raw material feed position will be changed, each x(R) coordinate and y(R) coordinate of the raw material feed position R may be added to the explanatory variables. Furthermore, if operational conditions such as the rotation speed and tilt angle θ of the pan 1, the particle size distribution of the powdered and granular raw materials, the type and amount of binder added, the liquid addition ratio, and the specifications of the scraper 2 are changed, they may be added to the explanatory variables.

[0038] The strength and particle size of the granulated material when granulated under each condition are investigated and used as the objective variables. For example, a fixed amount of samples of the granulated material are taken after a certain time has elapsed since the start of operation of the pan-type pelletizer. Regarding the strength of the granulated material, multiple samples of the granulated material are extracted and subjected to a uniaxial compression test to examine the crushing strength P. For example, the uniaxial compression test is performed using a testing machine specified by "JIS A1216", and the crushing strength P is calculated from the ultimate load F0 using the following formula (1).

[0039]

number

[0040] In the above formula (1), π is the circumference ratio and A is the particle size of the granulated product. The arithmetic mean of the crushing strength P is calculated, and the standard deviation (the positive square root of the sum of the squares of the differences between each data value and the mean divided by the total number of data n) can be used as the variance.

[0041] Regarding the particle size of the granulated product, sieving is performed, and the average particle size can be calculated by taking the reference particle size as the median value of each sieve opening and weighting it by their mass ratio. The standard deviation, which indicates the variation, can be calculated as the positive square root of the sum of the squares of the differences between each reference particle size and the average particle size, weighted by the mass ratio, divided by the sum of the mass ratios (100%).

[0042] Only one objective variable may be set, or multiple variables may be set. The strength of the granulated material is generally considered to be the better, but it must be strong enough to prevent cracking or powdering during transportation in the subsequent process. When granulating raw sintered material to produce sintered ore for steelmaking, it is generally considered that a particle size of around 8 mm is good, and a crushing strength of around 0.5 MPa is required.

[0043] As described above, explanatory variables and objective variables are taken and optimization calculations are performed. As a method of optimization calculations, various models can be used, such as evolutionary calculation methods (evolutionary algorithms) represented by genetic algorithms that mimic the process of biological evolution, and successive approximation methods that generate approximate functions based on finite sampling data and obtain a solution by gradually increasing the accuracy.

[0044] Evolutionary computing requires a large number of trials, which tends to increase the cost and computational time required for experiments, so it can be difficult to implement on large facilities. On the other hand, successive approximation methods have fewer search points, making them a good method for high-speed optimization. One example of this successive approximation method is Bayesian optimization. Bayesian optimization is a technique that estimates the probability of the existence of unknown data between known data from a trial experiment, conducts the next trial experiment to reduce this uncertainty, estimates the shape of the overall function, and searches for the optimal value (minimum or maximum value).

[0045] In the above, an example of how to obtain explanatory variables and objective variables from an actual trial experiment is shown, but a virtual trial experiment using numerical analysis may be used instead. For example, the discrete element method (DEM) can be used for the numerical analysis of granulation. In this DEM, the actual granulation process is simulated by continuously feeding powdered raw materials and particles as a liquid into a model of a pan-type pelletizer, and an analysis is performed in which the particles (granulated material) spill out from the edge of the pan 1. In this case, it is possible to analyze actual granulation by feeding the powdered raw materials as fine powder, but since this requires a very high calculation cost, it is more realistic to feed the powdered raw materials of a size equivalent to the granulated material.

[0046] As explanatory variables, in addition to the liquid charging position S and the raw material charging position R, operational conditions such as the rotation speed and tilt angle θ of the pan 1, the particle size distribution of the powdered raw materials, the type and amount of binder added, the liquid addition ratio, and the specifications of the scraper 2 can be added as necessary.

[0047] As the objective variable, for example, the amount and frequency of contact of each ingredient with liquid can be tallied, and a model in which the liquid disappears can be used, and finally the average amount of liquid contact of the granulated material that spills over the edge of pan 1 and is collected can be obtained, as well as its variance.

[0048] Figure 4 shows an example of the results of a numerical analysis using the discrete element method of the behavior of powdered raw material and liquid in a pan-type pelletizer. The movement of the powdered raw material in Figure 4 is shown typically in Figure 5 (a) to (c). In Figure 5, (a) shows the movement of bottom layer 30 of powdered raw material layer 3, (b) shows the movement of middle layer 31 of the powdered raw material layer, and (c) shows the movement of surface layer 32 of the powdered raw material layer. In addition, P30, P31, and P32 in the figure indicate the centers of rotational vortexes in each layer.

[0049] As is generally known, the vortex of the powdered raw material becomes smaller from the bottom layer 30 to the top layer 32, and the center of the vortex shifts from the right side of the pan center to the lower right as viewed in the figure. In Figure 4, the raw material particles are shaded according to the amount of liquid that has come into contact with the raw material. As shown in the figure, the shade of the raw material particles differs depending on the position where the liquid is poured in, and it was possible to evaluate the difference in the contact history of the liquid.

[0050] In the above description, an example has been described in which optimization calculations are directly performed on data obtained by real or virtual trial experiments, but a model may be created in advance by machine learning, and optimization calculations such as Bayesian optimization may be performed on the created model. In this case, various machine learning techniques such as neural networks, support vector machines, and random forests may be used to create the model.

[0051] The method for granulating powdered or granular raw material according to this embodiment can be carried out in the following three ways. (1) A model is not created in advance, and optimization calculations (e.g., sequential approximation optimization) are performed directly. (2) Create a model in advance and determine the liquid injection location that falls within the desired strength and particle size range. (3) Create a model in advance, then perform optimization calculations to determine the liquid injection position.

[0052] In (1) above, experiments and analyses are carried out to estimate the probability of the existence of unknown data between known data, and the next trial experiment is carried out to reduce the uncertainty. Then, the shape of the overall function is estimated, and optimal values ​​(minimum and maximum values) are searched for. This method has the advantage that it does not require the preparation of a huge amount of data, since a model is not created in advance.

[0053] In the above (2), experiments and analyses are conducted to generate a model based on the relationship between explanatory variables (liquid injection position) and objective variables (strength and particle size). Next, the strength and particle size of the granulated material are predicted by inputting the liquid injection position into the generated model. Then, various liquid injection positions are input into the model to determine the liquid injection position that will bring the strength and particle size of the granulated material into the desired range.

[0054] In the above (3), experiments and analyses are carried out to generate a model based on the relationship between explanatory variables (liquid injection position) and objective variables (strength and particle size). Next, the liquid injection position is input to the generated model to predict the strength and particle size of the granulated material. Then, optimization calculations are carried out to determine the liquid injection position that will bring the strength and particle size of the granulated material into the desired range.

[0055] Furthermore, in this embodiment, the method of feeding the powdered raw material to the raw material feeding position R is arbitrary and is not particularly limited. Examples of methods that can be used to feed the powdered raw material to the raw material feeding position R include a feeding conveyor such as a belt conveyor, and a feeding chute equipped with a fixed amount dispensing mechanism. Furthermore, the method of feeding the liquid to the liquid feeding position S is also arbitrary and is not particularly limited. Examples of methods that can be used to feed the liquid to the liquid feeding position S include a method of feeding (sprinkling) the liquid with a hose, and a method of feeding (sprinkling) the liquid with a sprinkling means such as a sprinkling nozzle.

[0056] In addition, in this embodiment, there is no particular limitation on the object to be granulated, and various powdered raw materials can be granulated, but it is particularly suitable for granulating sintered raw materials for sintered ore production in the ironmaking field.

[0057] In the granulation of sintered raw material for sintered ore production, it is important to obtain granulated material (pseudo particles) having a predetermined strength and particle size and small variation in strength and particle size in order to produce high-quality sintered ore. According to the granulation method of powdered granular raw material of this embodiment, such granulated material can be stably obtained. Moreover, the granulated material of the sintered raw material for sintered ore production granulated by the granulation method of powdered granular raw material of this embodiment is loaded into, for example, a sintering machine (usually a Dwight Lloyd type sintering machine) and sintered to produce sintered ore as an iron source for blast furnaces. EXAMPLES

[0058] An embodiment of the granulation method for powdered or granular raw materials according to the present invention will be described with reference to Fig. 4. In this embodiment, in order to determine the optimal liquid injection position for liquid that contributes to adhesion in granulation using a pan-type pelletizer, a virtual granulation experiment was performed using a pan-type pelletizer by numerical analysis using the discrete element method.

[0059] In this example, a pan-type pelletizer with the configuration shown in Fig. 1 was used. The diameter of the pan was 0.24 m, the height of the side wall was 0.07 m, the length of the scraper was 0.12 m, the inclination angle of the scraper was φ30°, the inclination angle of the pan was θ51°, and the operation was performed at a rotation speed of 39 rpm.

[0060] The raw material particles were assumed to be granulated material for sintering, and had a diameter of 10 mm. The liquid was set as a particle, and in order to reduce the analysis load, the diameter was set to 2 mm, which is larger than the actual diameter. The center p of the pan bottom was set as the origin, and the coordinate axis was set in the direction of Figure 1, and the raw material input position R was set to a 50 mm square with the center px at "x(R) = -0.041, y(R) = -0.015". The sinter raw material was continuously input at 260 g / s to the raw material input position R. In this embodiment, the raw material input position R is fixed to one location, but the raw material input position R may be multiple locations and added to the explanatory variables.

[0061] The liquid was added from one point in a 30 mm square, and was added continuously at 8.2 g / s. The number of times each ingredient came into contact with the liquid was tallied, and liquid that had come into contact once was allowed to disappear. Finally, the variance (standard deviation σ) in the number of times the granulated particles came into contact with the liquid, which were collected after spilling over the edge of the pan between 40 and 80 seconds after the start of operation, was used as the objective variable.

[0062] As an example, the results of a numerical analysis using the discrete element method when the center py of the liquid injection position S is "x(s) = -0.015, y(S) = 0.01" are shown in Figure 4. In this figure, liquid particles are shown in black, and the more times the raw material particles come into contact with the liquid, the darker the color.

[0063] The search range for the optimal value of the liquid injection position was set to a realistic range of "x(S) = -0.05 to 0.05, y(S) = -0.085 to 0.085", with discrete values ​​in increments of 0.005. In this case, the candidates for the liquid injection position are a total of 735 (21 x 35) candidates.

[0064] Bayesian estimation was used as a sequential approximation optimization method for the optimization calculation. As shown in Table 1, four conditions on the boundaries of the search range and two conditions near the raw material input position were selected and used as initial values ​​in a preliminary experiment. After that, the search using five experimental points was repeated three times. The results are shown in Table 2.

[0065] [Table 1] [Table 2]

[0066] In the optimization calculation, six preliminary experiments and a total of 15 main experiments were performed, and it was predicted that the optimal value was "σ = 14.14 when x(S) = -0.045, y(S) = -0.045." When compared with the results of a separate manual experiment under all conditions, it was confirmed that "x(S) = -0.045, y(S) = -0.045, σ = 14.14" was the optimal value. In other words, out of a total of 735 (21 x 35) candidates, the optimal value was found with just 21 experiments. In this way, it was demonstrated that the present invention is an effective method for searching for the optimal liquid injection position.

[0067] According to the model generation method, granulation method of powdered ore raw material, and sintered ore manufacturing method of the present invention described above, when granulating powdered ore raw material using a pan-type pelletizer, by adding liquid under optimal conditions to obtain a liquid bridging force, it is possible to stably obtain granulated material having predetermined strength and particle size and small variation in strength and particle size.

[0068] The model generating method, the granulating method of powdered raw material, and the manufacturing method of sintered ore according to the present invention have been specifically described above using the mode for carrying out the invention and examples, but the gist of the present invention is not limited to these descriptions and must be broadly interpreted based on the claims. Furthermore, it goes without saying that various changes and modifications based on these descriptions are also included in the gist of the present invention. [Explanation of symbols]

[0069] 1. Bread 2 Scraper 2a Scraper member 3 Powder material layer 10 Bottom plate part 20,20a One end 30 bottom layer 31 Middle Class 32 Surface layer A top g Perpendicular P30, P31, P32 Rotating vortex center of raw material

Claims

1. A method for pelletizing powdered or granular raw materials by using a pan-type pelletizer in which the pan rotates in an inclined state, feeding powdered or granular raw materials into the pan, and simultaneously feeding a liquid for obtaining a liquid bridging force from one or more points, comprising the steps of: Based on an evaluation result of at least one of the strength and particle size of the granulated product relative to the liquid introduction position, machine learning is performed using the liquid introduction position and the raw material introduction position as explanatory variables and at least one of the strength and particle size of the granulated product as a target variable to generate a model that predicts at least one of the strength and particle size of the granulated product. Model generation method.

2. The model generation method according to claim 1 , wherein the machine learning is performed by sequential approximation optimization.

3. The method of claim 2 , wherein the iterative optimization is Bayesian optimization.

4. A method for predicting a liquid injection position using a model generated by the model generation method according to any one of claims 1 to 3, A method for granulating powdered or granular raw materials, in which at least one of the strength and particle size of the granulated material is controlled within a desired range by feeding liquid from a predicted liquid feeding position, thereby granulating the powdered or granular raw material.

5. A method for pelletizing powdered or granular raw materials by using a pan-type pelletizer in which the pan rotates in an inclined state, feeding powdered or granular raw materials into the pan, and simultaneously feeding a liquid for obtaining a liquid bridging force from one or more points, comprising the steps of: A method for granulating powdered and granular raw materials, which includes a liquid injection position and a raw material injection position as explanatory variables based on an evaluation result of at least one of the strength and particle size of the granulated material in relation to the liquid injection position, performs sequential approximation optimization using at least one of the strength and particle size of the granulated material as an objective variable, determines a liquid injection position where at least one of the strength and particle size of the granulated material falls within a desired range, and granulates the powdered and granular raw material.

6. 5. The method for granulating powdered ore according to claim 4, wherein the powdered ore is a sintered material for producing sintered ore.

7. A method for producing sintered ore, comprising granulating raw material for sintering by the method for granulating powdered ore as claimed in claim 6, and sintering the granulated material to produce sintered ore.

8. A method for granulating powdered raw material as described in claim 5, wherein the powdered raw material is a sintered raw material for producing sintered ore.

9. A method for producing sintered ore, which comprises granulating sintered raw materials for producing sintered ore by the granulation method for powdered and granular raw materials described in claim 8, and sintering the granulated material to produce sintered ore.

Images