JP7453532B2 - Sintered ore structure learning device, structure structure learning method, and structure structure learning program, and sintered ore manufacturing condition changing device, manufacturing condition changing method, and manufacturing condition changing program - Google Patents

Description

The present invention relates to a structure learning device, a structure learning method, and a structure learning program for learning the structure of sintered ore, as well as a manufacturing condition changing device, a manufacturing condition changing method, and a manufacturing condition changing program for manufacturing sintered ore. .

Sintered ore is a major raw material that accounts for 80% of the raw materials charged into blast furnaces. The quality of sintered ore is one of the factors that determines the operational status of a blast furnace, and is determined by the pore structure of sintered ore and the composition of minerals forming the matrix. In self-fusing sinter, which is currently widely used as a raw material for blast furnaces, the substrates are hematite (Fe ₂ O ₃ ), magnetite (Fe ₃ O ₄ ), calcium ferrite (CaO・Fe ₂ O ₃ ), and silicate slag (CaO・It has four mineral phases: SiO ₂ ). Quantitative data regarding the structural composition of sintered ore, such as the proportion of mineral phases and pores in sintered ore, diameter distribution, and shape, is useful in obtaining knowledge regarding quality control of sintered ore.

There is a known technology for quantifying the structure of sintered ore by analyzing cross-sectional images of sintered ore taken with imaging devices such as optical microscopes and scanning electron microscopes (SEM). For example, see Patent Documents 1 to 3). In the techniques described in Patent Documents 1 to 3, the structure of the sintered ore is estimated based on the brightness of pixels forming an image of the sintered ore. However, since the brightness of images differs depending on the imaging conditions and the imaging location of the sample, in the techniques described in Patent Documents 1 to 3, the estimated structure of the sintered ore differs from image to image, and the sintered ore It is not easy to estimate the texture of ores with high accuracy.

On the other hand, there is a known technology that uses image data representing an image as training data and uses a learning model that is machine learned using deep learning to identify various features included in an image with high accuracy ( For example, see Patent Document 4).

Japanese Patent Application Publication No. 2013-122403 Japanese Patent Application Publication No. 2014-215987 Japanese Patent Application Publication No. 2014-137344 JP 2019-200769 Publication

“Development and Application of Comprehensive Simulation Model for Sinter Manufacturing Process”, Kawaguchi et al., Tetsu-to-Hagane Vol. 73 (1987) No. 15

However, there is no known technique for estimating the structure of sintered ore using a machine-learned learning model. The inventors of the present invention have found that by using a cross-sectional image of sintered ore to machine-learn the structure of sintered ore in a learning model, it is possible to estimate the structure of sintered ore efficiently and with high accuracy. I discovered something.

An object of the present invention is to provide a structure learning device capable of estimating the structure of sintered ore efficiently and with high precision.

The present invention to solve such problems includes a sintered ore microstructure learning device, a microstructure learning method, and a microstructure learning program, as well as a manufacturing condition changing device, a manufacturing condition changing method, and a sintered ore manufacturing condition changing device shown below. The gist of this is a program for changing manufacturing conditions.
(1) an image data acquisition unit that acquires a plurality of image data each showing a plurality of cross-sectional images taken of a cross-section of sintered ore;
a learning data generation unit that generates a plurality of learning data by associating each of the plurality of regions of the cross-sectional image with hematite, magnetite, calcium ferrite, silicate slag, and pores;
a learning processing unit that causes a learning model to learn the relationship between the plurality of cross-sectional images and the structure of the sintered ore using the learning data;
A learning model output unit that outputs a learning model that estimates the structure of sintered ore from cross-sectional images of the cross-section of sintered ore, based on the relationship between the learned cross-sectional images and the structure of sintered ore. and,
A device for learning the structure of sintered ore, comprising:
(2) The learning data generation unit classifies hematite into at least two mineral phases of porous original hematite, dense original hematite, primary hematite, and secondary hematite, and classifies each of the multiple regions of the cross-sectional image. The sintered ore structure learning device according to (1), which is associated with.
(3) The learning data generation unit classifies calcium ferrite into acicular calcium ferrite and columnar calcium ferrite and associates the same with each of the plurality of regions of the cross-sectional image. Organizational structure learning device.
(4) The learning data generation unit classifies pores into pores that do not include cracks and cracks, and associates the pores with each of a plurality of regions of the cross-sectional image, according to any one of (1) to (3). A device for learning the organization structure of condensation.
(5) a learning model learned by the organizational structure learning device described in any one of (1) to (4);
a cross-sectional image data acquisition unit that acquires cross-sectional image data showing a cross-sectional image obtained by capturing a cross-section of the sintered ore;
a structure information acquisition unit that obtains structure structure information indicating the structure of the sintered ore that is output in response to inputting cross-sectional image data representing a cross-sectional image obtained by capturing a cross-section of the sintered ore into a learning model;
a quality determination unit that determines whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data was taken satisfies predetermined quality conditions, based on the tissue structure corresponding to the structure information;
a manufacturing condition change instruction unit that outputs a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore when it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions;
A device for changing manufacturing conditions for sintered ore, comprising:
(6) The quality judgment department:
a reducibility parameter calculation unit that calculates a reducibility parameter indicating the reducibility of the sintered ore from the structure structure corresponding to the structure structure information;
a reducibility determination unit that determines whether the reducibility of the sintered ore satisfies a predetermined reducibility condition based on changes in the reducibility parameter over time;
The sintered ore manufacturing condition changing device according to (5), which has:
(7) Obtaining a plurality of image data each showing a plurality of cross-sectional images of the cross-section of the sintered ore,
generating a plurality of learning data by associating each of the plurality of regions of the cross-sectional image with hematite, magnetite, calcium ferrite, silicate slag, and pores;
Using training data, a learning model learns the relationship between multiple cross-sectional images and the structure of sintered ore,
Outputs a learning model that estimates the structure of sintered ore from cross-sectional images of the cross-section of sintered ore, based on the relationship between the learned cross-sectional images and the structure of sintered ore.
A method for learning the structure of sintered ore, the method comprising:
(8) Obtaining a plurality of image data each showing a plurality of cross-sectional images of the cross-section of the sintered ore,
generating a plurality of learning data by associating each of the plurality of regions of the cross-sectional image with hematite, magnetite, calcium ferrite, silicate slag, and pores;
Using training data, a learning model learns the relationship between multiple cross-sectional images and the structure of sintered ore,
Outputs a learning model that estimates the structure of sintered ore from cross-sectional images of the cross-section of sintered ore, based on the relationship between the learned cross-sectional images and the structure of sintered ore.
A sintered ore structure learning program characterized by having a computer perform processing.
(9) Obtaining cross-sectional image data showing a cross-sectional image of the cross-section of the sintered ore,
Cross-sectional image data showing a cross-sectional image of a cross-section of sintered ore is associated with each of a plurality of regions of a plurality of cross-sectional images of a cross-section of sintered ore with hematite, magnetite, calcium ferrite, silicate slag, and pores. This figure shows the structure of sintered ore output by inputting it into a learning model that has learned the relationship between multiple cross-sectional images and the structure of sintered ore using multiple pieces of training data generated by Obtain organizational structure information,
Determining whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data was taken satisfies predetermined quality conditions based on the tissue structure corresponding to the structure structure information,
outputting a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore when it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions;
A method for changing manufacturing conditions of sintered ore, the method comprising:
(10) Obtaining cross-sectional image data showing a cross-sectional image of the cross-section of the sintered ore,
Cross-sectional image data showing a cross-sectional image of a cross-section of sintered ore is associated with each of a plurality of regions of a plurality of cross-sectional images of a cross-section of sintered ore with hematite, magnetite, calcium ferrite, silicate slag, and pores. This figure shows the structure of sintered ore output by inputting it into a learning model that has learned the relationship between multiple cross-sectional images and the structure of sintered ore using multiple pieces of training data generated by Obtain organizational structure information,
Determining whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data was taken satisfies predetermined quality conditions based on the tissue structure corresponding to the structure structure information,
outputting a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore when it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions;
A program for changing manufacturing conditions for sintered ore, characterized by causing a computer to execute processing.

The learning model generated by the structure learning device according to one embodiment can estimate the structure of sintered ore efficiently and with high precision.

It is a figure which shows an example of the cross-sectional image which imaged the cross section of sintered ore. (a) is a diagram showing a cross-sectional image of sintered ore containing hematite, magnetite, and silicate slag, and (b) is a diagram showing a threshold value for extracting hematite in the cross-sectional image shown in (a) by a conventional technique. (c) is a diagram showing a cross-sectional image different from the cross-sectional image shown in (a) containing hematite, magnetite and silicate slag, and (d) is a diagram showing a cross-sectional image different from the cross-sectional image shown in (b). It is a figure which shows the cross-sectional image which extracted hematite from the cross-sectional image shown in (c) using the determined threshold value. FIG. 1 is a diagram showing an organizational structure learning system including an organizational structure learning device according to an embodiment. 4 is a block diagram showing the organizational structure learning device shown in FIG. 3. FIG. 5 is a flowchart of an organizational structure learning process executed by the organizational structure learning device shown in FIG. 4. FIG. 5 is a diagram showing a process of associating each of a plurality of regions of a cross-sectional image with a mineral phase contained in sintered ore using a GUI displayed by the learning data generation unit shown in FIG. 4, in which (a) is a first state; , (b) shows the second state, (c) shows the third state, and (d) shows the fourth state. (a) is a diagram (part 1) showing a cross-sectional image of sintered ore, (b) is a diagram (part 2) showing a cross-sectional image of sintered ore, and (c) is a diagram (part 2) showing a cross-sectional image of sintered ore. (d) is a diagram (part 4) showing a cross-sectional image of sintered ore, (e) is a diagram (part 5) showing a cross-sectional image of sintered ore, (f) is a diagram (part 6) showing a cross-sectional image of sintered ore. FIG. 1 is a diagram showing a manufacturing condition changing system including a manufacturing condition changing device according to an embodiment. 9 is a block diagram showing the manufacturing condition changing device shown in FIG. 8. FIG. 10 is a flowchart of a manufacturing condition changing process executed by the manufacturing condition changing apparatus shown in FIG. 9. FIG. (a) is a diagram showing the source of a cross-sectional image used as learning data, and (b) is a diagram showing a cross-sectional image used as learning data. (a) is a cross-sectional image for evaluation (part 1), (b) is a cross-sectional image for evaluation (part 2), (c) is a cross-sectional image for evaluation (part 3), and (d ) is the cross-sectional image for evaluation (part 4), (e) is the output image of the learning model corresponding to the cross-sectional image shown in (a), and (f) corresponds to the cross-sectional image shown in (b). (g) is an output image of the learning model corresponding to the cross-sectional image shown in (c), and (h) is an output image of the learning model corresponding to the cross-sectional image shown in (d). be. (a) is a cross-sectional image for evaluation (part 5), (b) is a cross-sectional image for evaluation (part 6), (c) is a cross-sectional image for evaluation (part 7), and (d ) is the cross-sectional image for evaluation (part 8), (e) is the output image of the learning model corresponding to the cross-sectional image shown in (a), and (f) corresponds to the cross-sectional image shown in (b). (g) is an output image of the learning model corresponding to the cross-sectional image shown in (c), and (h) is an output image of the learning model corresponding to the cross-sectional image shown in (d). be.

With reference to the drawings below, a structure learning device, a structure learning method, and a structure learning program for learning the structure of sintered ore according to the present invention, a manufacturing condition changing device for manufacturing sintered ore, and a manufacturing condition changing device are described below. The method and manufacturing condition change program will be explained. However, the technical scope of the present invention is not limited to these embodiments.

(Overview of organizational structure learning device according to embodiment)
The inventors of the present invention extracted the problems of the conventional technology for estimating the structure of sintered ore, and investigated whether the extracted problems of the conventional technology can be solved by using a learning model. investigated.

FIG. 1 is a diagram showing an example of a cross-sectional image of a cross-section of sintered ore. In FIG. 1, arrow H indicates hematite, arrow M indicates magnetite, arrow F indicates calcium ferrite, arrow S indicates silicate slag, and arrow P indicates pores.

The brightness of pixels that display the four minerals that form sintered ore, hematite, magnetite, calcium ferrite, and silicate slag, as well as pores, is highest for hematite, and decreases in the order of magnetite, calcium ferrite, slag, and pores. .

Conventional technology for estimating the structure of sintered ore focuses on differences in brightness between the four minerals and pores that form sintered ore, and classifies the brightness of pixels included in cross-sectional images using a predetermined threshold. The four minerals and pores forming the sintered ore included in the cross-sectional image are classified.

However, in a cross-sectional image of sintered ore, the brightness of pixels included in the cross-sectional image changes depending on the type of mineral contained in the imaged sintered ore, the imaging conditions when the sintered ore was imaged, etc. It is not easy to accurately classify the four minerals and pores that form sintered ore.

FIG. 2(a) is a diagram showing a cross-sectional image of sintered ore containing hematite, magnetite, and calcium ferrite. FIG. 2(b) is a diagram showing a cross-sectional image in which a threshold value for extracting hematite was determined using a conventional technique in the cross-sectional image shown in FIG. 2(a). FIG. 2(c) is a diagram showing a cross-sectional image different from the cross-sectional image shown in FIG. 2(a), which includes hematite, magnetite, and calcium ferrite. FIG. 2(d) is a diagram showing a cross-sectional image in which hematite is extracted from the cross-sectional image shown in FIG. 2(c) using the threshold determined by the cross-sectional image shown in FIG. 2(b).

In the image shown in FIG. 2(b), areas where the pixel brightness is higher than a predetermined threshold brightness are colored red. The areas colored red correspond to areas determined to be hematite by a skilled worker.

On the other hand, when hematite is extracted from Fig. 2(c), which is different from the cross-sectional image shown in Fig. 2(a), using the threshold determined using Fig. 2(a), the cross-sectional image shown in Fig. 2(d) is extracted. "Bleeding" occurs in which the area around the hematite that is extracted as red is colored red in a mottled pattern.

Measures to reduce boundary errors by increasing the imaging magnification, measures to improve various imaging devices such as automatic focus and automatic light source control, and measures using image analysis software such as brightness difference correction and edge correction within the screen to prevent "bleeding" Measures to prevent this from occurring will be considered. Additionally, measures will be considered to improve the accuracy of area determination by correcting the brightness distribution of minerals. However, this fundamentally solves the problem of conventional technology that determines the area where minerals are placed based on pixel brightness, which is that the estimation error decreases due to changes in pixel brightness depending on imaging conditions, etc. No solution has been proposed.

Therefore, the inventors of the present invention efficiently and highly accurately estimate the structure of sintered ore by having a learning model machine learn the structure of sintered ore using cross-sectional images of sintered ore. I found out that it is possible. Furthermore, the inventors of the present invention have developed a method for controlling manufacturing conditions for manufacturing sintered ore using a learning model that machine-learns the structure of sintered ore using cross-sectional images of sintered ore. I found it. Hereinafter, with reference to the drawings, a sintered ore structure learning device, a structure structure learning method, a structure structure learning program, a sintered ore manufacturing condition changing device, and a sintered ore manufacturing condition discovered by the inventors of the present invention will be described. A modification method and a manufacturing condition modification program will be explained.

(Configuration and functions of organizational structure learning device according to embodiment)
FIG. 3 is a diagram showing an organizational structure learning system including an organizational structure learning device according to an embodiment.

The tissue structure learning system 100 includes an optical microscope 101, an imaging device 102, and a tissue structure learning device 1. The optical microscope 101 magnifies and observes a desired region of the sintered ore 105, which has been cut so that its cross section is visible and is placed on the stage 110, at a desired magnification, such as 500 times. The area of the sintered ore 105 magnified by the optical microscope 101 is imaged by the imaging device 102. The imaging device 102 includes an image detection element such as a CCD (Charge Coupled Device) image sensor, and a storage element that stores an image detected by the image detection element, and has an area of the sintered ore 105 magnified by the optical microscope 101. Take an image.

Image data showing a cross-sectional image of the sintered ore captured by the imaging device 102 is transmitted to the tissue structure learning device 1 via the communication wiring 103. In the tissue structure learning system 100, the imaging device 102 and the tissue structure learning device 1 are communicably connected via communication wiring 103. However, in the tissue structure learning system according to the embodiment, the imaging device 102 and the tissue structure learning device 1 do not need to be communicably connected. When the imaging device 102 and the tissue structure learning device 1 are not connected, image data showing a cross-sectional image of the sintered ore is stored in a removable storage medium in the imaging device 102 and the tissue structure learning device 1, and the image data is stored in the imaging device 102 and the tissue structure learning device 1. It is moved from the device 102 to the organizational structure learning device 1.

(Configuration and functions of organizational structure learning device according to embodiment)
FIG. 4 is a block diagram showing the organizational structure learning device 1. As shown in FIG.

The organizational structure learning device 1 includes a communication section 11 , a storage section 12 , an input section 13 , an output section 14 , a processing section 20 , and a learning model 30 . The communication section 11, the storage section 12, the input section 13, the output section 14, the processing section 20, and the learning model 30 are connected to each other via a bus 15. The structure learning device 1 uses image data showing a cross-sectional image of the sintered ore captured by the imaging device 102 to learn the relationship between the cross-sectional image of the sintered ore and the structure of the sintered ore using a learning model 30. Let them learn. In one example, the organizational structure learning device 1 is a personal computer.

The communication unit 11 has a wired communication interface circuit such as Ethernet (registered trademark). The communication unit 11 communicates with the imaging device 102, a server (not shown), etc. via the communication wiring 103.

The storage unit 12 includes, for example, at least one of a semiconductor storage device, a magnetic tape device, a magnetic disk device, or an optical disk device. The storage unit 12 stores operating system programs, driver programs, application programs, data, etc. used in processing by the processing unit 20. For example, the storage unit 12 uses image data showing a cross-sectional image of the sintered ore to perform a structure-structure learning process that causes the learning model 30 to learn the relationship between the cross-sectional image of the sintered ore and the structure of the sintered ore. It stores an organizational structure learning program and the like for causing the processing unit 20 to execute. The organizational structure learning program may be installed in the storage unit 12 from a computer-readable portable recording medium such as a CD-ROM or DVD-ROM using a known setup program or the like. Furthermore, the storage unit 12 stores various data used in the organizational structure learning process.

The input unit 13 may be any device that can input data, such as a touch panel or a keyboard. The operator can use the input unit 13 to input characters, numbers, symbols, and the like. When the input unit 13 is operated by a worker, it generates a signal corresponding to the operation. The generated signal is then supplied to the processing unit 20 as an instruction from the operator.

The output unit 14 may be any device that can display videos, images, etc., such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The output unit 14 displays a video according to the video data supplied from the processing unit 20, an image according to the image data, and the like. Further, the output unit 14 may be an output device that prints video, images, characters, etc. on a display medium such as paper.

The processing unit 20 includes one or more processors and their peripheral circuits. The processing unit 20 centrally controls the overall operation of the organizational structure learning device 1, and is, for example, a CPU. The processing unit 20 executes processing based on programs (driver programs, operating system programs, application programs, etc.) stored in the storage unit 12. Further, the processing unit 20 can execute multiple programs (application programs, etc.) in parallel.

The processing section 20 includes an image data acquisition section 21 , an image data correction section 22 , a learning data generation section 23 , a learning processing section 24 , and a learning model output section 25 . Each of these units is a functional module realized by a program executed by a processor included in the processing unit 20. Alternatively, each of these parts may be implemented in the organizational structure learning device 1 as firmware.

The learning model 30 is a learning model that learns the features of an abstracted image through supervised learning. For example, using known machine learning techniques such as deep learning, a cross-sectional image of sinter Learn the relationship with organizational structure. Deep learning is machine learning using a multilayer neural network consisting of an input layer, a hidden layer, and an output layer. Each of the feature vectors of the cross-sectional image of the sintered ore is input to each node of the input layer. The feature vector may be, for example, a partial image obtained by cutting out a part of the image. Further, each of the partial images used as a feature vector may be cut out so that a portion thereof overlaps with another partial image. Each node in the middle layer outputs the sum of the values obtained by multiplying each feature vector output from each node in the input layer by a weight, and the output layer outputs the sum of the values obtained by multiplying each feature vector output from each node in the middle layer by a weight. Outputs the sum of the values multiplied by the weights. The learning model 30 learns while adjusting each weight so that the difference between the output value from the output layer and each cross-sectional image of the photographed sintered ore becomes small. The learning model 30 is, for example, "ZEN Intellesis" manufactured by ZEISS.

(Organizational structure learning process by the organizational structure learning device according to the embodiment)
FIG. 5 is a flowchart of the organizational structure learning process executed by the organizational structure learning device 1. The organizational structure learning process shown in FIG. 5 is mainly executed by the processing section 20 in cooperation with each element of the organizational structure learning device 1 based on a program stored in the storage section 12 in advance.

First, the image data acquisition unit 21 acquires a plurality of image data each representing a plurality of cross-sectional images obtained by capturing a cross-section of sintered ore (S101). The plurality of cross-sectional images corresponding to the plurality of image data images acquired by the image data acquisition unit 21 are cross-sectional images of the sintered ore captured by the imaging device 102.

Next, the image data correction unit 22 performs a predetermined correction process on each of the plurality of image data acquired in the process of S101 (S102). The image data correction unit 22 corrects the brightness of the cross-sectional image so that, for example, the highest brightness among the brightness of pixels included in the cross-sectional image corresponding to the image data becomes a predetermined brightness. The image data correction unit 22 also adjusts the cross-sectional image so that the difference in brightness between the highest brightness and the lowest brightness among the brightness of pixels included in the cross-sectional image corresponding to the image data becomes a desired brightness difference. Correct the contrast.

Next, the learning data generation unit 23 generates a plurality of learning data by associating each of the plurality of regions of each cross-sectional image corresponding to the plurality of image data images with the mineral phase contained in the sintered ore ( S103). By displaying a GUI including a cross-sectional image of the sintered ore on the output unit 14, the learning data generation unit 23 converts each of the plurality of regions of the cross-sectional image into mineral phases contained in the sintered ore according to the operator's instructions. to relate to. Moreover, the area|region with several learning data generation parts 23 may be linked|related with a pore.

FIG. 6 is a diagram showing a process of associating each of a plurality of regions of a cross-sectional image with a mineral phase contained in sintered ore using a GUI displayed by the learning data generation unit 23. 6(a) shows the first state, FIG. 6(b) shows the second state, FIG. 6(c) shows the third state, and FIG. 6(d) shows the fourth state. show.

First, the learning data generation unit 23 displays a learning data generation GUI including a cross-sectional image on the output unit 14 (see FIG. 6(a)). Next, the operator specifies the region of mineral phase A on the cross-sectional image of the sintered ore included in the learning data generation GUI (see FIG. 6(b)). As shown by the arrow α in FIG. 6(b), the region of the mineral phase A is designated by the operator by designating a part of the region.

Next, the learning data generation unit 23 estimates a region similar to mineral phase A specified by the worker as a region of mineral phase A, and adds a region of mineral phase A other than that estimated by the worker. Generate a cross-sectional image. The learning data generation unit 23 displays a learning data generation GUI including the generated cross-sectional image on the output unit (see FIG. 6(c)). The region of mineral phase A is displayed so as to be visually distinguishable from other regions, for example, by being colored in a predetermined color.

Thereafter, the learning data generation unit 23 repeats the process of interactively specifying the region of mineral phase A with the worker until the region of mineral phase A in the cross-sectional image of the sintered ore matches the region desired by the worker. .

When the area of mineral phase A in the cross-sectional image of the sintered ore matches the area desired by the operator, the learning data generation unit 23 instructs the operator to specify an area of mineral phase B that is different from mineral phase A. Run interactively. The learning data generation unit 23 specifies the mineral phase area by performing the same process as the process of specifying the mineral phase A area for all the mineral phases included in the cross-sectional image. When the learning data generation unit 23 completes the process of specifying all the mineral phase regions included in the cross-sectional image, it performs the same process on other cross-sectional images in which no mineral phase region is specified.

The learning data generation unit 23 performs a process of specifying a mineral phase region for all the image data acquired in the process of S101 (see FIG. 6(d)). In the example shown in FIG. 6(d), the cross-sectional image includes three mineral phases: mineral phase A, mineral phase B, and mineral phase C. When the learning data generation unit 23 completes the process of specifying the region of the mineral phase for all the image data acquired in the process of S101, the process of S103 ends.

A plurality of pieces of learning data generated in the process of S103 were generated by associating each of a plurality of regions of a cross-sectional image of sintered ore with hematite, magnetite, calcium ferrite, silicate slag, and pores shown in FIG. Contains training data.

In addition, the plurality of learning data generated in the process of S103 are cross-sectional images of hematite classified into at least two mineral phases: original porous hematite, original dense hematite, patchy hematite, and skeletal hematite. may be associated with each of a plurality of areas. Mottled hematite is also called recrystallized hematite or primary hematite, and skeletal hematite is also called secondary hematite.

FIG. 7(a) is a diagram (part 1) showing a cross-sectional image of sintered ore, FIG. 7(b) is a diagram (part 2) showing a cross-sectional image of sintered ore, and FIG. 7(c) is a diagram showing a cross-sectional image of sintered ore (part 2). It is a figure (part 3) which shows the cross-sectional image of sintered ore, and FIG.7(d) is a figure (part 4) which shows the cross-sectional image of sintered ore. FIG. 7(e) is a diagram (part 5) showing a cross-sectional image of the sintered ore, and FIG. 7(f) is a diagram (part 6) showing a cross-sectional image of the sintered ore.

The original porous hematite is a mineral phase in which porous hematite ore or goethite ore remains unreacted, and is indicated by arrow A in FIG. 7(a). When the content of the source porous hematite increases, the reducibility of the sintered ore improves.

The original ore dense hematite is a mineral phase in which dense hematite ore remains unreacted, and is indicated by arrow B in FIGS. 7(b) and 7(c). As the content of dense hematite in the original ore increases, the reducibility of the sintered ore decreases.

Primary hematite is a mineral phase that remained unreacted due to the low temperature even though the melt was generated, and is indicated by arrow C in FIGS. 7(b) and 7(c). Secondary hematite is a mineral phase generated from the melt, and is indicated by arrow D in FIG. 7(d). When the content of secondary hematite increases, reduction and pulverization of the sintered ore is promoted.

In addition, the plurality of learning data generated in the process of S103 classify magnetite into at least two mineral phases of source porous magnetite, primary magnetite, and secondary magnetite, and classify each of the plurality of regions of the cross-sectional image. May be associated with

The original ore dense magnetite is a mineral phase in which dense magnetite ore remains unreacted, and is indicated by arrow E in FIG. 7(e). When the content of the source ore dense magnetite increases, the reducibility of the sintered ore decreases significantly.

Primary magnetite is a mineral phase that is taken into the surrounding melt after hematite is reduced, and then remains unreacted, and is indicated by arrow F in Figures 7(d) and 7(e). . As the content of primary magnetite increases, the reducibility of the sintered ore decreases. Secondary magnetite is a mineral phase generated from the melt under conditions of low oxygen partial pressure, and is indicated by arrow G in FIG. 7(f). As the secondary magnetite content increases, the reducibility of the sintered ore decreases.

Further, the plurality of pieces of learning data generated in the process of S103 may classify calcium ferrite into acicular calcium ferrite and columnar calcium ferrite and may be associated with each of the plurality of regions of the cross-sectional image.

Acicular calcium ferrite is a mineral phase formed at low temperatures and is indicated by arrow H in FIG. 7(b). As the content of acicular calcium ferrite increases, the reducibility of the sintered ore improves. Columnar calcium ferrite is a mineral phase formed at high temperatures and is indicated by arrow I in FIG. 7(d). Note that calcium ferrite is classified here into acicular calcium ferrite and columnar calcium ferrite, but calcium ferrite may be further classified into fine calcium ferrite.

In addition, the plurality of pieces of learning data generated in the process of S103 are divided into a plurality of cross-sectional images by classifying the pores into pores not including cracks, which are indicated by arrow J, and cracks, which are indicated by arrow K, in FIG. 7(c). It may be associated with each area. As the content of cracks increases, the strength of the sintered ore decreases.

Next, the learning processing unit 24 uses the learning data generated in the process of S103 to determine the relationship between the plurality of cross-sectional images corresponding to the plurality of image data acquired in the process of S101 and the structure of the sintered ore. The learning model 30 is made to learn the gender (S104). The learning model 30 executes a process of estimating the structure of the sintered ore with high accuracy from the cross-sectional image of the cross section of the sintered ore by learning the relationship between the cross-sectional image and the structure of the structure of the sintered ore. It becomes possible.

Then, the learning model output unit 25 outputs the learning model 30 subjected to machine learning in the process of S104 to, for example, a server (not shown) (S105). The learning model 30 is used to estimate the structure of sintered ore, for example, in a sintered ore manufacturing process at a steelworks.

(Operations and effects of the organizational structure learning device according to the embodiment)
In the microstructure learning device 1, the learning model 30 performs machine learning using cross-sectional images of the sintered ore containing hematite, magnetite, calcium ferrite, silicate slag, and pores as training data, so that the learning model 30 can learn the microstructural structure of the sintered ore. Can be estimated with high accuracy. Furthermore, after the learning model 30 is machine learned, there is no need to judge which mineral phase the region included in the cross-sectional image corresponds to, so the structure learning device 1 can estimate the structure of the sintered ore. It is possible to improve efficiency.

In addition, the organizational structure learning device 1 does not need to be equipped with special error correction means such as brightness correction processing realized by hardware or software, and the organizational structure learning device 1 is realized by general-purpose equipment such as a personal computer. It is possible.

In addition, since the tissue composition learning device 1 does not directly estimate the tissue composition based on the brightness of pixels, there is no need to increase the magnification when capturing a cross-sectional image of sintered ore, and the number of pixels in the cross-sectional image can be reduced. There's no need to make it expensive.

In addition, in the tissue structure learning device 1, the learning model 30 performs machine learning by classifying hematite into original porous hematite, original dense hematite, primary hematite, and secondary hematite. It can be estimated. For example, the learning model 30 can estimate the structure including the content of source porous hematite, which improves reducibility, and source dense hematite, which decreases reducibility. Information for estimation can be provided. Furthermore, the learning model 30 can estimate the structure including the content of secondary hematite that promotes the reduction and pulverization of sintered ore, so it can provide information for estimating the reduction and pulverization of sintered ore. can.

In addition, in the tissue structure learning device 1, the learning model 30 performs machine learning by classifying magnetite into original porous magnetite, primary magnetite, and secondary magnetite, so that it is possible to estimate the structure structure of finely classified magnetite. For example, the learning model 30 can estimate the structure including the content of the source dense magnetite, which significantly reduces the reducibility, and therefore can provide information for estimating the reducibility of sintered ore. . In addition, the learning model 30 can estimate the structure composition including the content of primary and secondary magnetite, which significantly reduces reducibility, and therefore provides further information for estimating reduction powdering of sintered ore. can do.

In addition, in the tissue structure learning device 1, the learning model 30 performs machine learning by classifying calcium ferrite into acicular calcium ferrite and columnar calcium ferrite, so that it is possible to estimate the structure of finely classified calcium ferrite. For example, the learning model 30 can estimate the structure including the content of acicular calcium ferrite that improves reducibility, and therefore can provide information for estimating the reducibility of sintered ore.

Furthermore, in the tissue configuration learning device 1, the learning model 30 performs machine learning by classifying pores into pores that do not include cracks and cracks, so that it is possible to estimate a tissue configuration in which pores are finely classified. For example, the learning model 30 can estimate the structure including the content of cracks that reduce the strength of the sintered ore, and therefore can provide information for estimating the strength of the sintered ore.

(Configuration and functions of manufacturing condition changing device according to embodiment)
FIG. 8 is a diagram showing a manufacturing condition changing system including a manufacturing condition changing device according to an embodiment.

The manufacturing condition changing system 200 includes an optical microscope 201, an imaging device 202, and a manufacturing condition changing device 2, and changes a desired area of sintered ore 205 placed on a stage 210 to a desired magnification of, for example, 500 times. Magnify and observe. The configurations and functions of the optical microscope 201 and the imaging device 202 are the same as those of the optical microscope 101 and the imaging device 102, so a detailed explanation will be omitted here. Image data showing a cross-sectional image of the sintered ore captured by the imaging device 202 is transmitted to the manufacturing condition changing device 2 via the communication wiring 203.

(Configuration and functions of manufacturing condition changing device according to embodiment)
FIG. 9 is a block diagram showing the manufacturing condition changing device 2. As shown in FIG.

The manufacturing condition changing device 2 includes a communication section 31, a storage section 32, an input section 33, an output section 34, a processing section 40, and a learning model 50. The communication section 31, the storage section 32, the input section 33, the output section 34, the processing section 40, and the learning model 50 are connected to each other via a bus 35. The manufacturing condition changing device 2 estimates the content of the mineral phase included in the cross-sectional image of the sintered ore captured by the imaging device 202 using a learning model 50 that has learned the relationship between the cross-sectional image and the structure structure. The manufacturing conditions of the sintered ore are changed based on the content of the mineral phase. In one example, the manufacturing condition changing device 2 is a personal computer.

The communication unit 31 has a wired communication interface circuit such as Ethernet (registered trademark). The communication unit 31 communicates with the imaging device 202, a server (not shown), etc. via the communication wiring 203.

The storage unit 32 includes, for example, at least one of a semiconductor storage device, a magnetic tape device, a magnetic disk device, or an optical disk device. The storage unit 32 stores operating system programs, driver programs, application programs, data, etc. used in processing by the processing unit 40. For example, the storage unit 32 stores a manufacturing condition change process for causing the processing unit 40 to change the manufacturing conditions of sintered ore based on the content of the mineral phase included in the cross-sectional image captured by the imaging device 202. Stores condition change programs, etc. The manufacturing condition change program may be installed in the storage unit 32 from a computer-readable portable recording medium such as a CD-ROM or DVD-ROM using a known setup program or the like. Furthermore, the storage unit 32 stores various data used in the manufacturing condition change process.

The input unit 33 may be any device that can input data, such as a touch panel or a keyboard. The operator can use the input unit 33 to input characters, numbers, symbols, and the like. When the input unit 33 is operated by a worker, it generates a signal corresponding to the operation. The generated signal is then supplied to the processing unit 40 as an instruction from the operator.

The output unit 34 may be any device that can display videos, images, etc., such as a liquid crystal display or an organic EL (Electro-Luminescence) display. The output unit 34 displays a video according to the video data supplied from the processing unit 40, an image according to the image data, and the like. Further, the output unit 34 may be an output device that prints video, images, characters, etc. on a display medium such as paper.

The processing unit 40 includes one or more processors and their peripheral circuits. The processing unit 40 controls the overall operation of the manufacturing condition changing device 2, and is, for example, a CPU. The processing unit 40 executes processing based on programs (driver programs, operating system programs, application programs, etc.) stored in the storage unit 32. Further, the processing unit 40 can execute multiple programs (application programs, etc.) in parallel.

The processing unit 40 includes a cross-sectional image data acquisition unit 41 , a tissue configuration information acquisition unit 42 , a reducibility parameter calculation unit 43 , a reducibility determination unit 44 , and a manufacturing condition change instruction unit 45 . Each of these units is a functional module realized by a program executed by a processor included in the processing unit 40. Alternatively, each of these parts may be implemented in the manufacturing condition changing device 2 as firmware.

The learning model 50 is a learning model learned by the tissue structure learning device 1, and indicates the structure of the sintered ore in response to input of cross-sectional image data showing a cross-sectional image obtained by capturing a cross-section of the sintered ore. Output organizational structure information. The structure information outputted by the learning model 50 is, for example, the content of mineral phases and pores included in the cross section of the sintered ore imaged by the imaging device 202.

The learning model 50 is machine learned using learning data generated by associating hematite, magnetite, calcium ferrite, silicate slag, and pores with each of a plurality of regions of a cross-sectional image of sintered ore. The learning data used for machine learning of the learning model 50 classifies hematite into original porous hematite, original dense hematite, primary hematite, and secondary hematite, and associates them with each of a plurality of regions of the cross-sectional image. Further, the learning data used for machine learning of the learning model 50 classifies magnetite into original porous magnetite, primary magnetite, and secondary magnetite and associates them with each of a plurality of regions of the cross-sectional image. Further, the learning data used for machine learning of the learning model 50 classifies calcium ferrite into acicular calcium ferrite and columnar calcium ferrite and associates them with each of a plurality of regions of the cross-sectional image. Further, the learning data used for machine learning of the learning model 50 classifies pores into pores that do not include cracks and cracks, and associates them with each of a plurality of regions of the cross-sectional image.

(Manufacturing condition changing process by manufacturing condition changing device according to embodiment)
FIG. 10 is a flowchart of the manufacturing condition changing process executed by the manufacturing condition changing device 2. As shown in FIG. The manufacturing condition changing process shown in FIG. 10 is mainly executed by the processing unit 40 in cooperation with each element of the manufacturing condition changing device 2 based on a program stored in the storage unit 32 in advance. The manufacturing condition change process shown in FIG. 10 is executed, for example, three times a day, every eight hours.

The cross-sectional image data acquisition unit 41 acquires cross-sectional image data representing a cross-sectional image obtained by capturing a cross-section of the sintered ore (S201). The cross-sectional image corresponding to the image data image acquired by the cross-sectional image data acquisition unit 41 is a cross-sectional image of the sintered ore captured by the imaging device 202.

Next, the structure information acquisition unit 42 obtains structure structure information indicating the structure of the sintered ore, which is output in response to inputting cross-sectional image data indicating a cross-sectional image obtained by capturing a cross-section of the sintered ore into the learning model 50. (S202).

Next, the reducibility parameter calculation unit 43 calculates the reducibility, which is a reducibility parameter indicating the reducibility of the sintered ore, from the structure structure corresponding to the structure structure information acquired in the process of S402 using equation (1). (S203). The reducibility parameter calculation unit 43 stores the reducibility RI (JIS) calculated using equation (1) in the storage unit 32.

Formula (1) is the reducibility model described in Non-Patent Document 1, and RI (JIS) indicates the reducibility specified in M8713:2009 of Japanese Industrial Standards (JIS). . Further, M _i indicates the amount (%) of each mineral phase, ε is the product porosity (%) specified in JIS Z2501:2000, and X _i indicates the reducibility index for each mineral.

Next, the reducibility determination unit 44 determines whether the reducibility of the sinter satisfies a predetermined reducibility condition based on the change over time in the reducibility RI (JIS) calculated in the process of S203. (S204). First, the reducibility determination unit 44 compares the reducibility RI (JIS) calculated in the previous manufacturing condition change process eight hours ago with the reducibility RI (JIS) calculated in the current manufacturing condition change process. Compare. When the reducibility RI (JIS) in the current manufacturing condition change process is higher than the reducibility RI (JIS) in the previous manufacturing condition change process, the reducibility determination unit 44 determines the reducibility of the sintered ore. It is determined that the reducibility satisfies the reducibility condition (S204-YES).

The reducibility determination unit 44 determines when the reducibility RI (JIS) calculated in the current manufacturing condition change process is lower than the reducibility RI (JIS) calculated in the previous manufacturing condition change process. , the difference is calculated as the amount of reduction in reducibility. The reducibility determination unit 44 determines that the reducibility of the sintered ore whose cross-sectional image has been taken satisfies the reducibility condition when the amount of decrease in reducibility is smaller than a predetermined decrease amount threshold. It is determined that (S204-YES). When the amount of decrease in reducibility is larger than a predetermined decrease amount threshold, the reducibility determining unit 44 determines that the sintered ore whose cross-sectional image has been taken does not satisfy the reducibility condition (S204 -NO).

When it is determined that the quality of the sintered ore does not satisfy the predetermined quality conditions (S204-NO), the manufacturing condition change instruction unit 45 issues a manufacturing condition change indicating that the manufacturing conditions of the sintered ore are to be changed. Output instructions.

The manufacturing condition change instruction may include various information. For example, the instructions may include an instruction to reduce the SiO ₂ concentration contained in iron ore in the blended raw materials, and an instruction to increase the content of dense iron ore. Further, the instruction to change the manufacturing conditions may include an instruction to accelerate the production of acicular calcium ferrite and increase the content of hematite by firing at a low temperature and for a short time.

When the structure structure information acquired in the process of S202 indicates that the original ore porous hematite is decreasing, the manufacturing condition change instruction may include sintered ore manufacturing conditions that increase the original ore porous hematite. . Furthermore, when the structure information acquired in the process of S202 indicates that the original ore dense hematite is increasing, the manufacturing condition change instruction may include sintered ore manufacturing conditions that reduce the original ore dense hematite. .

Further, when the structure structure information acquired in the process of S202 indicates that the original ore porous magnetite is increasing, the manufacturing condition change instruction includes the sintered ore manufacturing conditions that reduce the original ore porous magnetite. But that's fine. Further, when the structure information acquired in the process of S202 indicates that primary magnetite is increasing, the manufacturing condition change instruction may include sintered ore manufacturing conditions that reduce primary magnetite. Further, when the structure information acquired in the process of S202 indicates that secondary magnetite is increasing, the manufacturing condition change instruction may include sintered ore manufacturing conditions that reduce secondary magnetite.

Furthermore, when the structure structure information acquired in the process of S202 indicates that acicular calcium ferrite is decreasing, the manufacturing condition change instruction may include sintered ore manufacturing conditions that increase acicular calcium ferrite. .

(Operations and effects of the manufacturing condition changing device according to the embodiment)
When the manufacturing condition changing device 2 determines that the quality of the sintered ore does not satisfy the predetermined quality conditions based on the structure structure information output from the learning model 50 learned by the structure structure learning device 1, A manufacturing condition change instruction indicating that the manufacturing conditions of sintered ore are to be changed is output. The manufacturing condition changing device 2 outputs an instruction to change the manufacturing conditions of sintered ore based on the structure information output from the learning model 50, so that the manufacturing conditions can be changed according to the mineral phase contained in the sintered ore. Able to give appropriate instructions.

In addition, the manufacturing condition changing device 2 performs sintering based on the structure structure information output from the learning model 50 that performs machine learning by classifying hematite into original porous hematite, original dense hematite, primary hematite, and secondary hematite. Determine the quality of ore. The manufacturing condition changing device 2 judges the quality of the sintered ore based on the structure composition information output from the learning model 50, and therefore changes the manufacturing conditions according to the change in the content of the source porous hematite and the source dense hematite. can be instructed to change.

In addition, the manufacturing condition changing device 2 determines the quality of the sintered ore based on the structure structure information output from the learning model 50 that performs machine learning by classifying magnetite into source porous magnetite, primary magnetite, and secondary magnetite. do. Since the manufacturing condition changing device 2 judges the quality of the sintered ore based on the structure composition information output from the learning model 50, the manufacturing condition changing device 2 judges the quality of the sintered ore based on the structure composition information output from the learning model 50. Can instruct changes in manufacturing conditions.

In addition, the manufacturing condition changing device 2 judges the quality of the sintered ore based on the structure structure information output from the learning model 50 that performs machine learning by classifying calcium ferrite into acicular calcium ferrite and columnar calcium ferrite. . Since the manufacturing condition changing device 2 determines the quality of the sintered ore based on the structure information output from the learning model 50, it can instruct changes in the manufacturing conditions according to changes in the content of acicular calcium ferrite.

(Modified example of manufacturing condition changing device according to embodiment)
The manufacturing condition changing device 2 determines whether the quality of the sintered ore whose cross-sectional image corresponding to the cross-sectional image data has been taken through the processes of S203 and S204 satisfies predetermined quality conditions. The determination processing of such a manufacturing condition changing device is not limited to this. The manufacturing condition changing device according to the embodiment may include a quality determination unit that determines whether the quality of the sintered ore whose cross-sectional image corresponding to the cross-sectional image data is taken satisfies predetermined quality conditions. Bye.

For example, the manufacturing condition changing device 2 calculates the reducibility RI (JIS) using formula (1) in the process of S203, but the manufacturing condition changing device according to the embodiment calculates the reducibility RI (JIS) using formula (1). The reducibility RI may be calculated using a formula.

Further, the manufacturing condition changing device according to the embodiment may determine whether or not qualities other than the reducibility RI, such as the reduction powdering index RDI and the strength SI, satisfy predetermined quality conditions.

When the manufacturing condition changing device according to the embodiment determines whether the reduced pulverization index RDI satisfies predetermined quality conditions, the manufacturing condition changing instruction changes the manufacturing conditions of sintered ore to reduce secondary hematite. May include.

Furthermore, when the manufacturing condition changing device according to the embodiment determines whether the strength SI satisfies predetermined quality conditions, the manufacturing condition changing instruction may include manufacturing conditions for sintered ore that reduce cracks. .

Multiple regions of 12 cross-sectional images of sintered ore are associated with the mineral phase, composite structure, and pore structure to generate multiple pieces of learning data, and the generated learning data are used as training data. A learning model was used to learn the relationship between cross-sectional images and the structure of sintered ore. The learning model used was "ZEN Intellesis" manufactured by ZEISS.

The learning data used as training data was obtained by classifying multiple regions of a cross-sectional image of sintered ore taken at 500x magnification into eight types of mineral phases, composite structures, and pores. The mineral phases include primary hematite, secondary hematite, original ore dense hematite, primary magnetite, secondary magnetite, columnar calcium ferrite, acicular calcium ferrite, and silicate slag. The composite structures include composite structures of acicular calcium ferrite and silicate slag.

FIG. 11(a) is a diagram showing the source of the cross-sectional image used as the learning data, and FIG. 11(b) is a diagram showing the cross-sectional image used as the learning data. Each of FIGS. 11(a) and 11(b) includes 12 cross-sectional images arranged in 3 rows and 4 columns, and the image shown in FIG. 11(a) and the image shown in FIG. 11(b) are Corresponding images are placed at the same location. For example, the image placed at the upper left corner in FIG. 11(a) corresponds to the image placed at the upper left corner in FIG. 11(b).

In FIGS. 11(a) and 11(b), primary hematite is indicated by 60, secondary hematite is indicated by 61, original dense hematite is indicated by 62, and primary magnetite is indicated by 63. , secondary magnetite is designated by 64. Further, columnar calcium ferrite is indicated by 65, acicular calcium ferrite is indicated by 66, silicate slag is indicated by 67, a composite structure of columnar calcium ferrite and silicate slag is indicated by 68, and pores are indicated by 68. It is indicated by the reference numeral 69.

In Figure 11(b), primary hematite is colored orange, secondary hematite is colored gray, source dense hematite is colored red, primary magnetite is colored olive, and secondary magnetite is colored blue. colored. In addition, columnar calcium ferrite is colored yellow, acicular calcium ferrite is colored light blue, silicate slag is colored green, the composite structure of columnar calcium ferrite and silicate slag is colored yellow-green, and pores are colored purple. be done.

A learning model that was machine learned using the cross-sectional images shown in FIG. 11 as training data was evaluated using 300 cross-sectional images of sintered ore. The 300 cross-sectional images used to evaluate the learning model were prepared by capturing cross-sectional images at 10 locations from each of the 30 sintered ores.

12 and 13 are diagrams showing a cross-sectional image for evaluation and a cross-sectional image output by a learning model corresponding to the cross-sectional image for evaluation. The cross-sectional images shown in FIGS. 12(a) to 12(d) and 13(a) to 13(d) are images used as cross-sectional images for evaluation. 12(e) to 12(h) and 13(e) to 13(h) correspond to the cross-sectional images shown in FIGS. 12(a) to 12(d) and 13(a) to 13(d), respectively. The output image of the learning model is shown below.

The cross-sectional images shown in FIGS. 12(a) to 12(d) and 13(a) to 13(d) are cross-sectional images of sintered ore taken at a magnification of 500 times. The learning model, as shown in Figures 12(a) to 12(d) and 13(a) to 13(d), colors the structure of the sintered ore in the same color as the structure shown in Figure 11(b). and output it.

As shown in FIGS. 12 and 13, the cross-sectional images output by the learning model have no "bleeding" at all compared to the images generated by the conventional technique shown in FIG. The learning device can prevent the occurrence of "bleeding", which is a problem with conventional techniques.

Furthermore, the tissue structure learning device according to the embodiment is capable of classifying hematite into original porous hematite, original dense hematite, primary hematite, and secondary hematite, which was not easy to classify using conventional techniques. Table 1 shows the results of classifying hematite in five sintered ores into original porous hematite, original dense hematite, primary hematite, and secondary hematite.

As shown in Table 1, the structure learning device according to the embodiment can classify hematite into four mineral phases: original porous hematite, original dense hematite, primary hematite, and secondary hematite.

Further, the tissue structure learning device according to the embodiment can classify calcium ferrite into acicular calcium ferrite and columnar calcium ferrite, which was not easy to classify using conventional techniques. Table 2 shows the results of classifying calcium ferrite into acicular calcium ferrite and columnar calcium ferrite in five sintered ores.

As shown in Table 2, the tissue structure learning device according to the embodiment can classify calcium ferrite into acicular calcium ferrite and columnar calcium ferrite.

1 Tissue configuration learning device 2 Manufacturing condition changing device 20, 40 Processing unit 21 Image data acquisition unit 22 Image data correction unit 23 Learning data generation unit 24 Learning processing unit 25 Learning model output unit 30, 50 Learning model 41 Cross-sectional image data acquisition Section 42 Tissue composition information acquisition section 43 Reducibility parameter calculation section 44 Reducibility determination section 45 Manufacturing condition change instruction section 100 Tissue composition learning system 101 Optical microscope 102 Imaging device

Claims (10)

an image data acquisition unit that acquires a plurality of image data each showing a plurality of cross-sectional images taken of a cross-section of the sintered ore;
The brightness of the cross-sectional image is corrected so that the highest brightness among the brightnesses included in each of the plurality of regions of the cross-sectional image becomes a predetermined brightness, and the brightness is the highest among the brightness of pixels included in the cross-sectional image. By correcting the contrast of the cross-sectional image so that the difference in brightness between the brightness and the lowest brightness becomes a desired brightness difference, and by coloring each of the plurality of regions of the cross-sectional image, hematite, magnetite, calcium ferrite and a learning data generation unit that generates a plurality of learning data in association with the silicate slag and the pores;
a learning processing unit that causes a learning model to learn the relationship between the plurality of cross-sectional images and the structure of the sintered ore using the learning data;
Learning to output a learning model that estimates the structure of the sintered ore from a cross-sectional image of the cross section of the sintered ore, based on the learned relationship between the plurality of cross-sectional images and the structure of the sintered ore. a model output section ;
the learning model is deep learning;
A device for learning the structure structure of sintered ore. The learning data generation unit classifies the hematite into at least two mineral phases of porous original hematite, dense original hematite, primary hematite, and secondary hematite, and classifies each of the plurality of regions of the cross-sectional image. The sintered ore structure learning device according to claim 1, which is associated with. The structure of the sintered ore according to claim 1 or 2, wherein the learning data generation unit classifies the calcium ferrite into acicular calcium ferrite and columnar calcium ferrite and associates the same with each of the plurality of regions of the cross-sectional image. learning device. The sintered ore structure learning device according to claim 1, wherein the learning data generation unit classifies the pores into pores not including cracks and cracks, and associates the pores with each of a plurality of regions of the cross-sectional image. A learning model learned by the organizational structure learning device according to any one of claims 1 to 4,
a cross-sectional image data acquisition unit that acquires cross-sectional image data showing a cross-sectional image obtained by capturing a cross-section of the sintered ore;
a structure information acquisition unit that obtains structure structure information indicating the structure of the sintered ore that is output in response to inputting cross-sectional image data representing a cross-sectional image of a cross-section of the sintered ore into the learning model;
a quality determination unit that determines whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data is taken satisfies predetermined quality conditions, based on the tissue structure corresponding to the structure structure information; ,
a manufacturing condition change instruction unit that outputs a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore when it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions;
A device for changing manufacturing conditions for sintered ore, comprising: The quality judgment department is
a reducibility parameter calculating unit that calculates a reducibility parameter indicating the reducibility of the sintered ore from the structure structure corresponding to the structure structure information;
a reducibility determining unit that determines whether the reducibility of the sintered ore satisfies a predetermined reducibility condition based on the change in the reducibility parameter over time;
The sintered ore production condition changing device according to claim 5, comprising: Obtain multiple image data each showing multiple cross-sectional images of the cross-section of the sintered ore,
The brightness of the cross-sectional image is corrected so that the highest brightness among the brightnesses included in each of the plurality of regions of the cross-sectional image becomes a predetermined brightness, and the brightness is the highest among the brightness of pixels included in the cross-sectional image. By correcting the contrast of the cross-sectional image so that the difference in brightness between the brightness and the lowest brightness becomes a desired brightness difference, and by coloring each of the plurality of regions of the cross-sectional image, hematite, magnetite, calcium ferrite and generate a plurality of learning data in association with silicate slag and pores,
causing a learning model to learn the relationship between the plurality of cross-sectional images and the structure of the sintered ore using the learning data;
outputting a learning model that estimates the structure of the sintered ore from a cross-sectional image of the cross section of the sintered ore based on the learned relationship between the plurality of cross-sectional images and the structure of the sintered ore; including that
the learning model is deep learning;
A method for learning the structure of sintered ore. Obtain multiple image data each showing multiple cross-sectional images of the cross-section of the sintered ore,
The brightness of the cross-sectional image is corrected so that the highest brightness among the brightnesses included in each of the plurality of regions of the cross-sectional image becomes a predetermined brightness, and the brightness is the highest among the brightness of pixels included in the cross-sectional image. By correcting the contrast of the cross-sectional image so that the difference in brightness between the brightness and the lowest brightness becomes a desired brightness difference, and by coloring each of the plurality of regions of the cross-sectional image, hematite, magnetite, calcium ferrite and generate a plurality of learning data in association with silicate slag and pores,
causing a learning model to learn the relationship between the plurality of cross-sectional images and the structure of the sintered ore using the learning data;
outputting a learning model that estimates the structure of the sintered ore from a cross-sectional image of the cross section of the sintered ore based on the learned relationship between the plurality of cross-sectional images and the structure of the sintered ore;
Let the computer carry out the process,
the learning model is deep learning;
A learning program on the structure of sintered ore. Obtain cross-sectional image data showing a cross-sectional image of the cross-section of the sintered ore,
Cross-sectional image data showing a cross-sectional image of a cross-section of sintered ore is determined by determining the highest brightness among the brightnesses included in each of a plurality of regions of a plurality of cross-sectional images of cross-sections of sintered ore. In addition to correcting the brightness of the cross-sectional image so that By correcting the contrast and coloring each of the plurality of regions of the cross-sectional images , the plurality of cross-sectional images are created using the plurality of learning data generated in association with hematite, magnetite, calcium ferrite, silicate slag, and pores. and obtaining tissue structure information indicating the structure of the sintered ore output in response to inputting the relationship between the structure and the structure of the sintered ore into a learning model that has learned,
Determining whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data was taken satisfies predetermined quality conditions based on the tissue structure corresponding to the structure structure information,
When it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions, outputting a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore ,
the learning model is deep learning;
A method for changing manufacturing conditions of sintered ore, characterized by: Obtain cross-sectional image data showing a cross-sectional image of the cross-section of the sintered ore,
Cross-sectional image data showing a cross-sectional image of a cross-section of sintered ore is determined by determining the highest brightness among the brightnesses included in each of a plurality of regions of a plurality of cross-sectional images of cross-sections of sintered ore. In addition to correcting the brightness of the cross-sectional image so that By correcting the contrast and coloring each of the plurality of regions of the cross-sectional images , the plurality of cross-sectional images are created using the plurality of learning data generated in association with hematite, magnetite, calcium ferrite, silicate slag, and pores. and obtaining tissue structure information indicating the structure of the sintered ore output in response to inputting the relationship between the structure and the structure of the sintered ore into a learning model that has learned,
Determining whether the quality of the sintered ore in which the cross-sectional image corresponding to the cross-sectional image data was taken satisfies predetermined quality conditions based on the tissue structure corresponding to the structure structure information,
outputting a manufacturing condition change instruction indicating to change the manufacturing conditions of the sintered ore when it is determined that the quality of the sintered ore does not satisfy predetermined quality conditions;
Let the computer carry out the process,
the learning model is deep learning;
A program for changing manufacturing conditions for sintered ore.

Images