Particle size determination device, particle size determination method, and manufacturing method of granulated substance

Abstract

To provide a particle size determination device, a particle size determination method, and a manufacturing method of a granulated substance, which can accurately determine the particle size of a particulate substance.SOLUTION: A particle size determination device (10) for determining the particle size of a particulate substance includes: an acquisition part (131) for acquiring an image of the particulate substance, which is taken by an imaging device; a poor visibility determination part (133) for determining whether the image has poor visibility; and a particle size determination part (134) for determining the particle size of the particulate substance by using the image determined not to have the poor visibility.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present disclosure relates to a particle size determination device, a particle size determination method, and a method for manufacturing a granulated product.The present disclosure particularly relates to a particle size determination device, a particle size determination method, and a method for manufacturing a granulated product that determine the particle size of a granulated product granulated by a granulator. [Background technology]

[0002] When producing granulated materials using a granulator, it is necessary to measure and manage the particle size of the granulated materials. As a method for measuring the particle size of granular materials such as granulated materials, for example, Patent Document 1 discloses an average particle size measurement method in which an image of the granular materials is taken and the taken image is subjected to image processing to determine the average particle size of the granulated materials. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 2-264845 Summary of the Invention [Problem to be solved by the invention]

[0004] Patent Document 1 describes that the average particle size of granulated material can be measured by processing images of the granulated material taken inside a plate-type tumbling granulator. However, dust can be generated around the granulator due to the flying material during granulation, and this dust can cause poor visibility. In images taken under such poor visibility conditions, it is difficult to distinguish the granulated material in the image, which can easily lead to errors in determining the particle size of the granulated material.

[0005] In view of the above circumstances, an object of the present disclosure is to provide a particle size determination device, a particle size determination method, and a method for manufacturing a granulated material that can accurately determine the particle size of a granular material. [Means for solving the problem]

[0006] (1) A particle size determination device according to an embodiment of the present disclosure, A particle size determination device for determining the particle size of a granular object, an acquisition unit that acquires an image of the granular object captured by an imaging device; a visual field defect determination unit that determines whether the image has a visual field defect; and a particle size determination unit that determines the particle size of the granular matter using an image that has been determined not to have a poor field of view.

[0007] (2) As one embodiment of the present disclosure, in (1), The particle size determination unit determines the particle size level in the image using a particle size detection model that has learned the particle size level of the granular matter.

[0008] (3) As an embodiment of the present disclosure, in (1) or (2), The visual field defect determination unit determines the dust generation state in the image, and determines that the visual field is not defective if there is no dust generation.

[0009] (4) As an embodiment of the present disclosure, in (3), The visual field defect determination unit determines the dust generation state in the image using a visual field defect detection model that has been trained on the dust generation state.

[0010] (5) As an embodiment of the present disclosure, in (3) or (4), The granules are granulated by a granulator, The poor visibility determination unit determines the dust generation situation around the granulator, The particle size determining unit determines the particle size of the granulated material using an image determined to be free of dust.

[0011] (6) A particle size determination method according to an embodiment of the present disclosure includes: A particle size determination method for determining the particle size of a granular object, comprising: an acquisition step of acquiring an image of the granular matter photographed by an imaging device; a visual field defect determination step of determining whether the image has a visual field defect; and a particle size determination step of determining the particle size of the granular matter using an image determined not to have a poor field of view.

[0012] (7) A steel sheet manufacturing equipment according to an embodiment of the present disclosure includes: (5) A step of determining the particle size level of the granulated product using a particle size determination device. [Effects of the Invention]

[0013] According to the present disclosure, it is possible to provide a particle size determination device, a particle size determination method, and a method for manufacturing a granulated material, which are capable of accurately determining the particle size of a granular material. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a diagram illustrating an example of the configuration of a particle size determining device according to an embodiment of the present disclosure. [Figure 2] FIG. 2 is a schematic diagram showing a granulated material production facility equipped with a particle size determining device according to an embodiment of the present disclosure. [Figure 3] FIG. 3 is a diagram illustrating an example of learning data for the visual field defect detection model. [Figure 4] FIG. 4 is a diagram illustrating an example of an image used as learning data for a particle size detection model. [Figure 5] FIG. 5 is a diagram illustrating an example of learning data for the granularity detection model. DETAILED DESCRIPTION OF THE INVENTION

[0015] A particle size determination device 10 (see FIG. 1), a particle size determination method, and a method for producing a granulated product according to one embodiment of the present disclosure will be described below with reference to the drawings. In each drawing, identical or corresponding parts are designated by the same reference numerals. In the description of this embodiment, the description of identical or corresponding parts will be omitted or simplified as appropriate.

[0016] Fig. 1 is a diagram showing an example of the configuration of a particle size determining device 10 according to this embodiment. Fig. 2 is a schematic diagram showing a granulated material manufacturing facility equipped with the particle size determining device 10.

[0017] <Device configuration> The particle size determination device 10 according to this embodiment determines the particle size of granular matter. The particle size determination device 10 includes a communication unit 11, a storage unit 12, and a control unit 13. The control unit 13 includes an acquisition unit 131, a model generation unit 132, a visual field defect determination unit 133, a particle size determination unit 134, and an output unit 135. The particle size determination device 10 may have a hardware configuration such as a computer. The computer may be a server computer or a portable computer such as a laptop or tablet. Details of the components of the particle size determination device 10 will be described later.

[0018] The granular material that is the subject of particle size determination is not limited to granulated materials, but an example is granulated material granulated by granulator 20. Furthermore, the granulator 20 is not limited to a specific type of machine, but an example is pan pelletizer equipment. In this embodiment, the granular material that is the subject of particle size determination will be described as granulated material that is granulated by pan pelletizer equipment and discharged from the pan pelletizer equipment.

[0019] In the granulated material manufacturing facility shown in FIG. 2, predetermined amounts of iron-containing raw material (e.g., iron ore), CaO-containing raw material (e.g., quicklime), and coke are dispensed from each blending tank and mixed to form a sintered raw material. The sintered raw material, to which granulation water has been added as moisture, is charged into a granulator 20, where it is granulated. The granulated particles (i.e., granulated material) are transported to, for example, a sintering machine, where they are sintered to form sintered ore. Here, the amounts of iron-containing raw material, CaO-containing raw material, and coke dispensed from each blending tank, the amount of moisture added, and the operation of the granulator 20 are controlled by a control device. The control device may be, for example, a computer.

[0020] The granulated material manufacturing equipment also includes a photographing device. In this embodiment, the photographing device is a camera installed around the granulator 20. The photographing device takes images of the granulated material discharged from the granulator 20. Here, a floodlight that irradiates the granulated material with light may be installed near the photographing device. By irradiating the granulated material with light from the floodlight, the photographing device can photograph the granulated material even in a dark room.

[0021] The photographing device is connected to a network. Images of granular matter photographed by the photographing device are transmitted to the particle size determining device 10 via the network. The network may be, for example, a LAN (Local Area Network) or the Internet. Not only the photographing device and particle size determining device 10, but also the control device are connected to the network, allowing information to be transmitted and received between these devices. The particle size determining device 10 and the photographing device may constitute a particle size determining system. The particle size determining system may also be configured with another device (for example, a control device).

[0022] The components of the particle size determination device 10 will be described in detail below. The communication unit 11 is configured to include one or more communication modules connected to a network. The communication unit 11 may include a communication module compatible with mobile communication standards such as 4G (4th Generation) and 5G (5th Generation). The communication unit 11 may include a communication module compatible with a wired or wireless LAN standard.

[0023] The storage unit 12 is one or more memories. The memory may be, for example, a semiconductor memory, a magnetic memory, an optical memory, or the like, but is not limited to these, and may be any memory. The storage unit 12 is, for example, built into the particle size judgment device 10, but may also be configured to be accessed from outside by the particle size judgment device 10 via any interface.

[0024] The storage unit 12 stores various data used in various calculations performed by the control unit 13. The storage unit 12 may also store results and intermediate data of various calculations performed by the control unit 13. The storage unit 12 may include a database, which will be described later.

[0025] The control unit 13 is one or more processors. The processor may be, for example, a general-purpose processor or a dedicated processor specialized for a specific process, but is not limited to these and may be any processor. The control unit 13 controls the overall operation of the particle size determination device 10.

[0026] Here, the particle size judgment device 10 may have the following software configuration: One or more programs used to control the operation of the particle size judgment device 10 are stored in the storage unit 12. When the programs stored in the storage unit 12 are read by the processor of the control unit 13, they cause the control unit 13 to function as an acquisition unit 131, a model generation unit 132, a visual field defect judgment unit 133, a particle size judgment unit 134, and an output unit 135.

[0027] The acquisition unit 131 acquires an image of a granular object captured by an imaging device.

[0028] The model generation unit 132 generates a visual field defect detection model used in the judgment of the visual field defect judgment unit 133 and a particle size detection model used in the judgment of the particle size judgment unit 134. The model generation unit 132 stores the generated visual field defect detection model and particle size detection model in the storage unit 12. The model generation unit 132 generates the visual field defect detection model and particle size detection model and stores them in the storage unit 12 before the particle size judgment device 10 executes the process of determining the particle size of the granular matter.

[0029] The poor visibility determination unit 133 determines whether the image acquired by the acquisition unit 131 has poor visibility. Poor visibility means that granular matter is difficult to see in the image (or the environment is such that granular matter is difficult to see), but as an example, it can correspond to the presence of dust. In this embodiment, whether or not visibility is poor corresponds to the presence or absence of dust. Generally, the granulator 20 rotates and mixes materials with water to granulate the granulated material. This rotation can cause some of the material to become airborne, generating dust. If the imaging device captures the granulated material in a state where dust is present, the visibility of the granulated material in the image becomes poor, making it impossible to accurately determine the particle size.

[0030] In this embodiment, the poor visibility determination unit 133 determines the dust generation status around the granulator 20 in the image. That is, the poor visibility determination unit 133 determines that the visibility is poor when dust is present. Furthermore, the poor visibility determination unit 133 determines that the visibility is not poor (good visibility) when dust is not present. Here, whether the visibility is poor is not limited to the presence or absence of dust, and may correspond, as another example, to the presence or absence of smoke or whether the ambient luminance (brightness) is lower or higher than a predetermined luminance.

[0031] The particle size determination unit 134 determines the particle size of the granular matter from the image acquired by the acquisition unit 131. The particle size determination unit 134 determines the particle size of the granular matter using an image that has been determined by the visual field defect determination unit 133 not to have a poor field of view. In other words, the images used in determining the particle size of the granular matter are only images that have been determined not to have a poor field of view. In this embodiment, the particle size determination unit 134 determines the particle size of the granular matter using an image that has been determined by the visual field defect determination unit 133 not to have dust generation. Furthermore, in this embodiment, the particle size determination unit 134 performs a determination on the granular matter in the image so as to classify the granular matter into a plurality of levels for each particle size (hereinafter referred to as "particle size levels").

[0032] The output unit 135 outputs the determination result, including the particle size of the granular material determined by the particle size determination unit 134, to another device. In the example of Fig. 2, the output unit 135 outputs the determination result to a control device. The control device can control the operation of the granulator 20, etc., based on the particle size level determined by the particle size determination device 10, so as to obtain granular material of the target particle size.

[0033] <Particle size determination method> The particle size judgment device 10 according to this embodiment executes the processes of a particle size judgment method described below. The particle size judgment method mainly includes an acquisition step, a visual field defect judgment step, and a particle size judgment step.

[0034] (Acquisition process) In the acquisition step, an image of the granular material is acquired by the photographing device. When the granular material is a granulated material as in this embodiment, it is preferable that the photographing device photographs the granulated material before and after it is discharged from the granulator 20. When the granulator 20 is a pan pelletizer, it is preferable that the photographing device photographs the lower outer periphery, close to the discharge point. Using the images photographed in this way, it is possible to determine the size of the granulated material that is discharged from the granulator 20 and transported to the next equipment (e.g., a sintering machine).

[0035] (Visual field defect determination process) In the poor field of view determination process, it is determined whether or not the captured image of the granular object has poor field of view. In this embodiment, the image is processed by the poor field of view determination unit 133 to determine whether or not dust is present in the image. Here, known techniques can be used for the image processing. For example, the poor field of view determination unit 133 can determine whether or not the field of view is poor using a poor field of view detection model that has been trained to determine whether the field of view is good or poor. In this embodiment, as described above, whether or not the field of view is poor corresponds to the presence or absence of dust. Therefore, the presence or absence of dust is determined using a poor field of view detection model (trained model) that has been trained to determine the presence or absence of dust. In other words, in this embodiment, if dust is present, it is determined that the field of view is poor.

[0036] Specifically, the model generation unit 132 uses images of the granulator 20, such as that shown in FIG. 3, as training data to generate a visual field defect detection model through machine learning before the visual field defect determination process is performed. The images serving as training data are labeled in advance, for example, by an operator with a label of "1" corresponding to "poor visual field" or "0" corresponding to "no visual field defect (good visual field)." For example, (0-a) and (0-b) in FIG. 3 are images of good visual field with no granular matter and good visual field with granular matter, respectively. Furthermore, for example, (1-a) and (1-b) in FIG. 3 are images of poor visual field with little dust and images of poor visual field with heavy dust, respectively. As an example, 1,090 images of good visual field and 113 images of poor visual field were prepared as training data, and machine learning was performed. At this time, 842 images of good visual field and 75 images of poor visual field were separately prepared as verification data to evaluate the generated visual field defect detection model. The model generation unit 132 can use the learning data to generate a poor visibility detection model that receives a captured image as input and outputs the dust generation status around the granulator 20 (information on the presence or absence of dust generation).

[0037] Image classification using machine learning can be used to generate a visual field defect detection model. As described above, the dust conditions of previously captured learning images are classified into the presence or absence of dust to generate learning data (teacher image data). Then, the learning data is trained using a convolutional neural network to generate a visual field defect detection model. When a captured image is input into the visual field defect detection model generated in this way, identification information regarding the presence or absence of dust is added and output. This makes it possible to identify which output images are classified as "with dust" and which are classified as "without dust." Here, the identification information may be added to the entire captured image, or may be added to the region of the image to be identified.

[0038] In this embodiment, the visual field defect determination unit 133 determines the dust state in the image using a visual field defect detection model that has learned the dust state, and determines that there is no visual field defect if there is no dust. Here, images captured by the imaging device are sequentially stored in a database that can be accessed by the particle size determination device 10. In this embodiment, the database is stored in the storage unit 12. The model generation unit 132 may appropriately perform processing to update the visual field defect detection model using the images stored in the database.

[0039] (Particle size determination process) In the particle size determination step, the particle size of the granular matter is determined using an image that is determined not to have a poor field of view in the poor field of view determination step. In this embodiment, the image is processed by the particle size determination unit 134, and the particle size of the granular matter in the image is determined. Here, known techniques can be used for the image processing. For example, the particle size determination unit 134 determines the particle size level in the image using a particle size detection model (trained model) that has been trained to learn the particle size levels of granular matter.

[0040] Specifically, the model generation unit 132 uses images of the granulator 20 as shown in Figures 4 and 5 as learning data to generate a particle size detection model by machine learning before the particle size determination process is performed. Here, Figure 4 is a diagram illustrating an example of an image used as learning data for the particle size detection model, and an image cut out to a size of 300 pixels x 300 pixels is used as one piece of learning data. Furthermore, the numbers in parentheses are coordinates (in pixels) indicating the position of the upper left corner of the multiple cut-out images. By cutting out multiple images not only of the area where the granulated material is discharged from the granulator 20 but also before and after that and using them as learning data, it is possible to average out uneven distribution due to particle size and ensure the number of data required for machine learning.

[0041] FIG. 5 is a diagram illustrating training data for the particle size detection model. In this embodiment, images serving as training data are labeled in advance by, for example, an operator with labels (0 to 4 in the example of FIG. 5) corresponding to the particle size level of the granulated material discharged from the granulator 20. The particle size of the granulated material is identified using area information (size in pixels) in the image. In the example of FIG. 5, an image in which it is determined that no granular material is visible is labeled with a "0" corresponding to "no granular material." Furthermore, if granular material with a radius of 2 pixels or less accounts for 50% or more of the area of ​​all granular material, it is labeled with a "1" corresponding to "powder-like." Furthermore, if granular material with a radius of 3 to 9 pixels accounts for 50% or more of the area of ​​all granular material, it is labeled with a "2" corresponding to "appropriate size." Furthermore, if granular material with a radius of 10 to 16 pixels accounts for 50% or more of the area of ​​all granular material, it is labeled with a "3" corresponding to "maximum acceptable range." Furthermore, if the area ratio of granular particles with a radius of 17 pixels or more to the total granular particles is 50% or more, the particle is labeled "4," corresponding to "excessive size." As an example, machine learning was performed using training data consisting of 4,203 images labeled with 0, 641 images labeled with 1, 2,341 images labeled with 2, 30 images labeled with 3, and 13 images labeled with 4. At this time, validation data was prepared separately to evaluate the generated particle size detection model. As an example, validation data consisting of 1,539 images labeled with 0, 1,348 images labeled with 1, 2,553 images labeled with 2, 48 images labeled with 3, and 14 images labeled with 4 was prepared. The model generation unit 132 can use the learning data to generate a particle size detection model that takes the cut-out image as input and outputs a label category (any of 0 to 4) corresponding to the particle size level of the granulated material discharged from the granulator 20.

[0042] Image classification using machine learning can be used to generate a particle size detection model. As described above, the particle size levels of granular objects in a previously captured learning image are classified using the criteria described above (e.g., granular objects with a radius of 2 pixels or less account for 50% or more of the total area of ​​all granular objects), to generate learning data (teacher image data). The learning data is then trained using a convolutional neural network, thereby generating a particle size detection model. When a captured image is input into the particle size detection model generated in this way, identification information related to the particle size level (i.e., a label classification corresponding to the particle size level) is assigned and output. This makes it possible to determine the particle size of the granular objects.

[0043] In this embodiment, the particle size determination unit 134 determines the particle size level in an image using a particle size detection model that has been trained to determine the particle size level of granular matter. As described above, images captured by the imaging device are sequentially stored in a database accessible by the particle size determination device 10. The model generation unit 132 may appropriately update the particle size detection model using the images stored in the database.

[0044] <Method of manufacturing granules> In the granulated material manufacturing equipment shown in FIG. 2, the particle size of the granulated material can be managed using the particle size determination device 10. That is, the granulated material manufacturing method performed by such manufacturing equipment may include a step of determining the particle size level of the granulated material using the particle size determination device 10. For example, the control device acquires information on the particle size level of the granulated material determined by the particle size determination device 10, and based on the information on the particle size level of the granulated material, controls the amount of sinter raw material to be cut, the amount of water to be added, or the operation of the granulator 20, thereby enabling the control device to take action such as changing the granulation conditions. For example, if the determined particle size level of the granulated material is smaller than the target (appropriate size), the control device may take action such as increasing the amount of water to be added. Furthermore, for example, the control device may change the granulation conditions only if the determined particle size level of the granulated material is neither the target (appropriate size) nor within the maximum allowable range.

[0045] As described above, the particle size determination device 10, particle size determination method, and granular material manufacturing method of this embodiment determine the particle size of granular materials using images that are not poorly viewed, and therefore can accurately determine the particle size of granular materials.

[0046] In other words, according to this embodiment, first it is determined whether the image has poor field of view or not, and then the particle size determination is performed using only images that are determined not to have poor field of view, so that the particle size can be determined more accurately than in a configuration in which the particle size determination is performed using images that also include images with poor field of view.

[0047] Furthermore, in this embodiment, image classification using machine learning is used, which allows for more accurate particle size determination than when other image processing methods (such as binarization) are used. For example, with binarization techniques, it is difficult to determine the particle size of granular matter by dividing it into three or more categories. Image classification using machine learning does not have any restrictions on the number of categories, and particle size determination can be performed in any desired category.

[0048] Although the embodiments of the present disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art would easily be able to make various modifications or alterations based on the present disclosure. Therefore, it should be noted that these modifications and alterations are included within the scope of the present disclosure. For example, the functions included in each component or step can be rearranged so as not to cause logical inconsistencies, and multiple components or steps can be combined or divided into one. The embodiments of the present disclosure can also be realized as a program executed by a processor included in an apparatus or a storage medium on which a program is recorded. It should be understood that these are also included within the scope of the present disclosure.

[0049] The model generation unit 132 generates the visual field defect detection model and the granularity detection model, and may generate them independently or with a relationship between them. The relationship may be, for example, the sharing of at least a portion of the learning data.

[0050] Furthermore, the particle size judgment device 10 may not be a single device, but may be composed of multiple devices located in multiple locations and capable of transmitting and receiving data to and from each other via a network. In other words, multiple devices connected via a network may function as the particle size judgment device 10 shown in FIG. 1 as a whole. Therefore, for example, the particle size judgment device 10 may be composed of a single computer or multiple computers connected via a network in terms of hardware configuration. When composed of multiple computers, the storage unit 12 may be a shared memory accessible by each computer. Furthermore, the particle size judgment device 10 may not be configured to include the model generation unit 132. In this case, the particle size judgment device 10 may acquire a visual field defect detection model and a particle size detection model generated by another device via a network and store them in the storage unit 12. Here, the other device may be, for example, a model generation device configured by another computer. [Explanation of symbols]

[0051] 10 Particle size determination device 11 Communications Department 12 Storage section 13 Control Unit 20 Granulator 131 Acquisition Department 132 Model Generation Unit 133 Poor visual field determination section 134 Particle size determination section 135 Output section

Claims

1. A particle size determination device for determining the particle size of a granular object, an acquisition unit that acquires an image of the granular object captured by an imaging device; a visual field defect determination unit that determines whether the image has a visual field defect; and a particle size determination unit that determines the particle size of the granular matter using an image that has been determined not to have poor visibility.

2. The particle size determination device according to claim 1 , wherein the particle size determination unit determines the particle size level in the image using a particle size detection model that has learned the particle size level of the granular matter.

3. 3. The particle size judgment device according to claim 1, wherein the visual field defect judgment unit judges a dust generation state in the image, and judges that there is no visual field defect when there is no dust generation.

4. The particle size judgment device according to claim 3 , wherein the visual field defect judgment unit judges the dust generation state in the image using a visual field defect detection model that has learned the dust generation state.

5. The granules are granulated by a granulator, The poor visibility determination unit determines the dust generation situation around the granulator, The particle size determining device according to claim 3 , wherein the particle size determining unit determines the particle size of the granulated material using an image determined to be free of dust.

6. A particle size determination method for determining the particle size of a granular object, comprising: an acquisition step of acquiring an image of the granular matter photographed by an imaging device; a visual field defect determination step of determining whether the image has a visual field defect; and a particle size determination step of determining the particle size of the granular matter using an image determined not to have a poor field of view.

7. A method for producing a granulated product, comprising a step of determining the particle size level of the granulated product using the particle size determining device according to claim 5.

Images