Molten metal cleaning device and molten metal cleaning method

The molten metal purification device facilitates easy rotor insertion and real-time adjustment of stirring conditions using image analysis and machine learning, improving maintainability and purification efficiency.

JP7729312B2Active Publication Date: 2025-08-26TOYOTA JIDOSHA KK
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
JP2022175369
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Filing Date
2022-11-01
Publication Date
2025-08-26
Estimated Expiration
2042-11-01

AI Technical Summary

Technical Problem

Existing molten metal purification devices lack efficient mechanisms for easily inserting and adjusting the rotor within the molten metal, and for optimizing the stirring conditions to enhance purification efficiency.

Method used

A molten metal purification device equipped with a rotor that stirs and releases inert gas, a transport unit for inserting the rotor, first and second photographing units for image capture and detection, and an adjustment unit that adjusts rotor speed and gas flow rate based on image analysis and machine learning, to improve maintainability and purification efficiency.

Benefits of technology

Enables easy and efficient insertion of the rotor into the molten metal, and allows for real-time adjustment of stirring conditions based on bubble state and history, enhancing the purification process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007729312000001
    Figure 0007729312000001
  • Figure 0007729312000002
    Figure 0007729312000002
  • Figure 0007729312000003
    Figure 0007729312000003
Patent Text Reader

Abstract

To simply purify molten metal in a molten metal purification apparatus having a conveyance mechanism which can convey a rotor in and out of molten metal.SOLUTION: A molten metal purification apparatus for purifying molten metal in a container has: a rotor for stirring molten metal by being rotated in molten metal and releasing an inert gas into the molten metal; a conveyance part which is configured so as to insert the rotor into the molten metal by relatively moving the rotor to the container; a first imaging unit for imaging a first image until the rotor is inserted into the molten metal; a first detection unit for detecting positional information about a relative position between the rotor and the molten metal based on the first image; a conveyance control unit for inserting the rotor into the molten metal by controlling the conveyance part based on the positional information; a second imaging unit for imaging a second image including the molten metal during the stirring of the molten metal; a second detection unit for detecting the stirring state of the molten metal based on the second image; and an adjustment unit for adjusting at least one of the rotational speed of the rotor or the flow rate of the inert gas based on the stirring state.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical Field]

[0001] The present disclosure relates to a molten metal cleaning apparatus and a molten metal cleaning method. [Background technology]

[0002] Regarding a molten metal purification device for purifying molten metal, Patent Document 1 discloses a technology in which a rotor of a rotating rotor that stirs the molten metal and releases an inert gas is moved along the peripheral portion of the bottom surface of a pumping chamber that holds the molten metal. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2010-84213 Summary of the Invention [Problem to be solved by the invention]

[0004] A molten metal purification apparatus may be provided with a transfer mechanism capable of transferring a rotor into and out of the molten metal, for example, in order to improve maintainability of the rotor. In a molten metal purification apparatus having such a transfer mechanism, a technology for easily purifying the molten metal has been desired. [Means for solving the problem]

[0005] The present disclosure can be realized in the following forms.

[0006] (1) According to one aspect of the present disclosure, there is provided a molten metal purification device for purifying molten metal contained in a container, the molten metal purification device including: a rotor that rotates within the molten metal to stir the molten metal and release an inert gas into the molten metal; a transport unit configured to transport the rotor relative to the container so as to insert the rotor into the molten metal; a first photographing unit that takes first images of the rotor until it is inserted into the molten metal; a first detection unit that detects position information regarding the relative position of the rotor and the molten metal based on the first images; a transport control unit that controls the transport unit based on the position information to insert the rotor into the molten metal; a second photographing unit that takes second images including the molten metal while the molten metal is being stirred; a second detection unit that detects a stirring state regarding the stirring of the molten metal based on the second images; and an adjustment unit that adjusts at least one of a rotation speed of the rotor and a flow rate of the inert gas based on the stirring state. In this configuration, the rotor can be inserted into the molten metal by the transport unit based on the position information detected from the first image. Furthermore, the rotation speed of the rotor and the flow rate of the inert gas can be adjusted based on the stirring state detected from the second image. Therefore, the molten metal can be easily purified in a molten metal purification device equipped with a transport unit. (2) In the above embodiment, the second detection unit may detect the state of bubbles as the stirring state, which is represented by at least one of the size of bubbles contained in the molten metal and the number of the bubbles. The adjustment unit may perform at least one of adjusting the rotation speed and the flow rate to reduce the size of the bubbles when the size of the bubbles is larger than a predetermined reference size, or adjusting the rotation speed and the flow rate to increase the number of bubbles when the number of the bubbles is smaller than a predetermined reference number. With this embodiment, the rotation speed of the rotor and the flow rate of the inert gas can be appropriately adjusted based on the state of the bubbles detected based on the second image. This allows the molten metal to be effectively purified. (3) In the above embodiment, at least one of the rotation speed and the flow rate may be adjusted based on a history of the state of the bubbles and a history of at least one of the rotation speed and the flow rate. In this embodiment, the rotation speed of the rotor and the flow rate of the inert gas can be appropriately adjusted based on the history.

[0007] (4) In the above aspect, the adjustment unit may adjust at least one of the rotation speed and the flow rate using a machine learning model generated by machine learning based on the history of the state and the history of at least one of the rotation speed and the flow rate. According to this aspect, the rotation speed of the rotor and the flow rate of the inert gas can be appropriately adjusted using the results of machine learning.

[0008] The present disclosure can be realized in various forms other than the above-described form of the molten metal purification device, for example, a molten metal purification method. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing a schematic configuration of a molten metal purification device. [Figure 2] 1 is a flowchart of a molten metal cleaning process. [Figure 3] FIG. 2 is an explanatory diagram showing an example of a first image. [Figure 4] FIG. 10 is an explanatory diagram showing an example of a second image. [Figure 5] FIG. 10 is a diagram illustrating the state of bubbles. DETAILED DESCRIPTION OF THE INVENTION

[0010] A. First embodiment: FIG. 1 is a diagram showing a schematic configuration of a molten metal purification apparatus 100 in this embodiment. In FIG. 1, arrows are shown along the X, Y, and Z directions, which are orthogonal to each other. The X, Y, and Z directions are directions along the X, Y, and Z axes, which are three spatial axes that are orthogonal to each other, and each direction includes both a direction on one side of the X, Y, and Z axes and the opposite direction. The X and Y axes are axes along a horizontal plane, and the Z axis is an axis along a vertical line. Hereinafter, the +Z direction will also be referred to as "up" and the -Z direction will also be referred to as "down."

[0011] The molten metal purification device 100 includes a stirring unit 110 having a rotor 111 , a photographing unit 125 , a transport unit 130 , a preheating unit 140 , and a control unit 200 .

[0012] The molten metal purification apparatus 100 rotates a rotor 111 in the molten metal MM contained in a container Ct to stir the molten metal MM, and also releases an inert gas from the rotor 111 into the molten metal MM, thereby generating bubbles in the molten metal MM. The bubbles generated in the molten metal MM absorb impurities such as hydrogen gas and metal oxides in the molten metal MM and rise to the surface of the molten metal MM. This purifies the molten metal MM. The container Ct in this embodiment has an opening Op that has a substantially circular cross section when viewed vertically, and contains the molten metal MM in the opening Op. The inert gas may be, for example, argon or nitrogen.

[0013] The stirring unit 110 has a rotor 111, a shaft 112, a support unit 113, and a rotation drive unit 114. The stirring unit 110 may have, for example, a baffle plate for suppressing swirling of the molten metal MM due to the rotation of the rotor 111.

[0014] The shaft 112 has a hollow pipe shape that is long in the axial direction. The shaft 112 is rotatably supported by a support 113. The shaft 112 is disposed so that its axis AX is aligned vertically. The rotor 111 is provided at the lower end of the shaft 112. In this embodiment, the rotor 111 is integrally formed with the shaft 112. The rotor 111 is provided with a gas release hole 115 for releasing an inert gas. The gas release hole 115 may be provided, for example, on the lower surface of the rotor 111 or on a side surface of the rotor 111. The rotation drive unit 114 is configured, for example, by a motor, and rotates the shaft 112 to rotate the rotor 111 about the axis AX. The rotation drive unit 114 is controlled by an adjustment unit 260 of the control unit 200, which will be described later.

[0015] In this embodiment, the inert gas is introduced from the gas supply unit 120 into the hollow portion of the shaft 112 through the gas supply hole 116 formed in the support unit 113 and then released into the molten metal MM through the gas release hole 115. The gas supply unit 120 is, for example, configured by a cylinder that stores the inert gas, and supplies the inert gas to the gas supply hole 116 through a pipe connecting the gas supply unit 120 and the gas supply hole 116. In FIG. 1, the flow of the inert gas supply is indicated by a dashed line. The flow rate of the inert gas released from the gas release hole 115 is adjusted, for example, by changing the aperture of a valve provided in the support unit 113 under the control of the control unit 200. Hereinafter, the flow rate of the inert gas released from the rotor 111 will also be simply referred to as the "gas flow rate."

[0016] The transfer unit 130 is configured to transfer the rotor 111 relative to the container Ct, thereby moving the rotor 111 relative to the container Ct, so that the rotor 111 can be inserted into the molten metal MM. The transfer unit 130 in this embodiment transfers the rotor 111 relative to the fixed container Ct. Hereinafter, the transfer of the rotor 111 relative to the container Ct will also be simply referred to as "transfer of the rotor 111." Furthermore, the relative movement of the rotor 111 with respect to the container Ct will also be simply referred to as "movement of the rotor 111."

[0017] The conveying unit 130 in this embodiment includes a rail 131 and a crane 132. The rail 131 is disposed along the X direction so as to pass above the container Ct and above the preheating unit 140. The crane 132 is configured as a hoist crane and is attached to the rail 131. The crane 132 suspends and fixes the support unit 113 of the agitator 110. The crane 132 includes a first conveying mechanism 133 and a second conveying mechanism 134. The first conveying mechanism 133 conveys the crane 132 along the rail 131 by driving a trolley with the rotation of a motor. This conveys the rotor 111 along the X direction. The second conveying mechanism 134 drives a pulley with the rotation of a motor to unwind and rewind a wire connected to the support unit 113, thereby conveying the support unit 113 up and down. This conveys the rotor 111 along the Z direction. The first transport mechanism 133 and the second transport mechanism 134 are controlled by a transport control unit 255, which will be described later.

[0018] The photographing unit 125 is configured as a camera that photographs an image. In this embodiment, the photographing unit 125 is disposed on the +X direction side of the support unit 113 and is fixed to the support unit 113. Therefore, the photographing unit 125 is transported together with the agitation unit 110 to the container Ct by the transport unit 130. In this embodiment, the photographing unit 125 photographs an image facing downward.

[0019] In this embodiment, the photographing unit 125 functions as a first photographing unit that photographs a first image while the rotor 111 is being inserted into the molten metal MM, and a second photographing unit that photographs a second image while the molten metal MM is being stirred. The first image is used to detect position information, which will be described later. In this embodiment, the photographing unit 125 photographs, as the first image, an image including the molten metal MM and the container Ct, or an image including the periphery of the container Ct, for example. The second image is an image including the molten metal MM.

[0020] The preheating unit 140 is configured, for example, with a burner. Before the rotor 111 is inserted into the molten metal MM, the preheating unit 140 preheats the rotor 111 and the shaft 112 to a temperature of, for example, 200°C or higher and 300°C or lower under the control of the control unit 200. In this embodiment, the preheating unit 140 preheats the rotor 111 and the shaft 112 while the rotor 111 is located at position PH, which is horizontally spaced apart from the container Ct. Position PH is located on the +X direction side of the container Ct. Preheating the rotor 111 and the shaft 112 can prevent water droplets adhering to the rotor 111 and the shaft 112 from being mixed into the molten metal MM when the rotor 111 is inserted into the molten metal MM. Furthermore, in this embodiment, because position PH is horizontally spaced apart from the container Ct, it is also possible to prevent water droplets from being mixed into the molten metal MM during preheating.

[0021] The control unit 200 includes a computer 205 and a PLC (Programmable Logic Controller) 250 .

[0022] The computer 205 includes a CPU, a storage unit, and an input / output interface that inputs and outputs signals from and to the outside. In this embodiment, the storage unit included in the computer 205 stores a machine learning model 206, which will be described later. The CPU included in the computer 205 executes a program stored in the storage unit, causing the computer 205 to perform various functions, such as the functions of the first detection unit 210 and the second detection unit 215. The first detection unit 210 detects position information regarding the relative position between the rotor 111 and the molten metal MM based on the first image. The second detection unit 215 detects a stirring state that indicates a state regarding the stirring of the molten metal MM based on the second image.

[0023] The PLC 250 includes a CPU, a storage unit, and an input / output interface for inputting and outputting signals from and to the outside. The CPU included in the PLC 250 executes programs stored in the storage unit, causing the PLC 250 to perform various functions, such as the functions of the transfer control unit 255 and the adjustment unit 260. The transfer control unit 255 controls the transfer unit 130 based on position information detected by the second detection unit 215, thereby inserting the rotor 111 into the molten metal MM. The adjustment unit 260 adjusts the rotation speed of the rotor 111 and the gas flow rate based on the stirring state detected by the first detection unit 210. Hereinafter, the rotation speed of the rotor 111 will also be referred to as the "rotor rotation speed." In other embodiments, for example, the various functions of the PLC 250 may be realized by a computer.

[0024] 2 is a flowchart of a molten metal purification process illustrating the molten metal purification method according to this embodiment. The molten metal purification process is executed, for example, when the molten metal MM is contained in the container Ct and the rotor 111 is positioned at a standby position outside the container Ct, and a predetermined start operation is performed by the user on the control unit 200. In step S105, with the rotor 111 positioned at position PH, the control unit 200 controls the preheating unit 140 to preheat the rotor 111 and the shaft 112.

[0025] In steps S110 and S115, the control unit 200 executes an insertion process. The insertion process refers to a process of executing an insertion step in which the rotor 111 is inserted into the molten metal MM by conveying the rotor 111 based on the position information.

[0026] In the insertion process of this embodiment, the control unit 200 first executes a horizontal position adjustment process in step S110. In the horizontal position adjustment process, the control unit 200 causes the photographing unit 125 to photograph a first image and detects position information based on the first image, while controlling the first transport mechanism 133 based on the position information, thereby positioning the rotor 111 directly above the molten metal MM.

[0027] FIG. 3 is an explanatory diagram showing an example of a first image. FIG. 3 shows images Pc1, Pc2, and Pc3 as examples of first images captured during the horizontal position adjustment process, and images Pc4 and Pc5 as examples of first images captured during the height adjustment process described below. In FIG. 3, the molten metal MM included in the first image is hatched with a halftone dot pattern. Image Pc1 represents an image captured while the center position C1 of the rotor 111 is located at position P1 on the +X direction side of the container Ct. In this embodiment, when the center position C1 is located at position P1, the container Ct is located outside the field of view of the imaging unit 125. Therefore, the container Ct and the molten metal MM are not included in image Pc1. Image Pc3 represents an image captured when the center position C1 and the center CP of the molten metal MM overlap. Image Pc2 represents an image captured while the center position C1 is located at position P2 between positions P1 and C2. Image Pc2 is an image captured later than image Pc1, and image Pc3 is an image captured later than image Pc2. In this embodiment, the first image actually includes the shaft 112 and the rotor 111, but these are omitted from FIG.

[0028] In this embodiment, the control unit 200 performs image recognition on the first image during the insertion process and detects position information based on the results of the image recognition. More specifically, the control unit 200 detects the molten metal MM by binarizing the first image and then recognizing the outline of the molten metal MM contained in the binarized first image through image recognition. The control unit 200 then detects position information based on the detection results of the molten metal MM. For example, when detecting position information based on image Pc1, the control unit 200 detects information indicating that the horizontal distance between the rotor 111 and the container Ct is very large based on the fact that the molten metal MM is not recognized in image Pc1. Furthermore, when detecting position information based on images Pc2 and Pc3, the control unit 200 detects the distance between the center position C2 of the molten metal MM detected by image recognition on images Pc2 and Pc3 and the center position C1 of the rotor 111 as position information.

[0029] In this embodiment, in the horizontal position adjustment process, the control unit 200 repeatedly captures a first image, detects position information, and moves the rotor 111 along the X direction until the center position C1 coincides with the center CP, as in image Pc3 in FIG. 3 . For example, when image Pc1 is captured as the first image, the control unit 200 transports the rotor 111 a predetermined distance in the −X direction based on the position information from image Pc1, and then captures the first image and detects the position information again. The control unit 200 repeats these processes until the molten metal MM is included in the first image. Thereafter, when image Pc2 is captured as the first image, the control unit 200 transports the rotor 111 in the −X direction based on the position information from image Pc2. The range of the center CP is set, for example, as a range that allows the rotor 111 to be inserted into the molten metal MM while preventing contact between the rotor 111 and the container Ct. Note that, when the rotor 111 is transported by the crane 132 as in this embodiment, for example, the first image in the horizontal position adjustment process may be captured after the transport of the rotor 111 has stopped and the amplitude of the swing caused by the horizontal movement of the rotor 111 has become equal to or smaller than a predetermined amplitude. In this case, the amplitude of the swing of the rotor 111 may be detected, for example, based on the image captured by the first imaging unit.

[0030] In step S115 of FIG. 2, the control unit 200 executes a height adjustment process. In the height adjustment process, the control unit 200 causes the photographing unit 125 to photograph a first image and detects position information based on the first image, while controlling the second transport mechanism 134 based on the position information to insert the rotor 111 into the molten metal MM. Image Pc4 shown in FIG. 3 represents an image captured while the rotor 111 was positioned directly above the molten metal MM and lower than when image Pc3 was captured. Image Pc5 was captured after image Pc4 and represents an image captured with the rotor 111 inserted into the molten metal MM.

[0031] In this embodiment, the control unit 200 repeatedly captures a first image, detects position information, and moves the rotor 111 along the Z direction during the height adjustment process until the rotor 111 is inserted into the molten metal MM. For example, when detecting position information based on image Pc4, the control unit 200 detects the vertical distance between the rotor 111 and the container Ct as position information based on the diameter of the outline of the molten metal MM included in image Pc4. Then, based on the detected distance, the control unit 200 moves the rotor 111 downward, for example, and again captures a first image and detects position information.

[0032] 2, the control unit 200 executes the stirring process. The stirring process refers to a process of executing a stirring step in which an inert gas is released into the molten metal MM while stirring the molten metal MM with the rotor 111.

[0033] In the stirring process of this embodiment, the control unit 200 first starts cleaning the molten metal MM in step S120. More specifically, in step S120, the control unit 200 sets the rotor rotation speed and gas flow rate to predetermined initial values, respectively, and starts stirring the molten metal MM and releasing the inert gas. In step S125, the control unit 200 captures a second image. In order to stabilize the state of bubbles contained in the molten metal MM before capturing the second image, the control unit 200 preferably executes step S125, for example, several seconds after executing step S120 or step S135 (described later). In step S130, the control unit 200 detects the stirring state based on the second image captured in step S125.

[0034] FIG. 4 is an explanatory diagram showing an example of the second image. FIG. 4 shows an image Pc6 containing bubbles Bb contained in the molten metal MM as an example of the second image. In FIG. 4, the molten metal MM is hatched with a halftone dot pattern, as in FIG. 3. In this embodiment, in step S130 of FIG. 2, the control unit 200 detects the state of bubbles as the stirring state, which is represented by at least one of the size and number of bubbles contained in the molten metal MM. Generally, the smaller the size of each bubble and the greater the number of bubbles, the larger the area of ​​the gas-liquid interface between the molten metal MM and the bubbles, and the more easily impurities in the molten metal MM are captured by the bubbles.

[0035] In this embodiment, in step S130, the control unit 200 recognizes the bubbles Bb contained in the image Pc6, for example, by performing image recognition, and detects the size and number of the bubbles based on the recognition results. More specifically, the control unit 200 recognizes the bubbles Bb by, for example, binarizing the image Pc6 and then recognizing the outlines of the bubbles Bb contained in the binarized image Pc6 through image recognition. Then, the control unit 200 measures the size and number of the recognized bubbles Bb. The size of the bubbles Bb is defined as an arithmetic mean value calculated by, for example, dividing the sum of the maximum diameters of the bubbles Bb by the number of bubbles Bb.

[0036] FIG. 5 is a diagram illustrating the state of bubbles in this embodiment. FIG. 5 shows a schematic graph with the bubble size on the horizontal axis and the number of bubbles on the vertical axis. In this embodiment, if the bubble size is equal to or smaller than a predetermined first size S1 and the number of bubbles is equal to or larger than a predetermined first number N1, the control unit 200 detects the bubble state as State A in step S130. If the bubble size is equal to or smaller than a predetermined second size S2 and the number of bubbles is equal to or larger than a predetermined second number N2, and the conditions for State A are not met, the control unit 200 detects the bubble state as State B. If the bubble size is larger than the second size S2 or the number of bubbles is less than the second number N2, the control unit 200 detects the bubble state as State C.

[0037] In step S135, the control unit 200 executes an adjustment process. The adjustment process refers to a process of executing an adjustment step that adjusts at least one of the rotor rotation speed and the gas flow rate based on the detected stirring state. Hereinafter, adjusting at least one of the rotor rotation speed and the gas flow rate is also referred to as "adjusting the stirring conditions." In step S135 of this embodiment, the control unit 200 first determines whether the state of the bubbles detected in step S130 satisfies a predetermined judgment condition related to at least one of the bubble size and the number of bubbles. In this embodiment, the control unit 200 determines that the state of the bubbles satisfies the judgment condition if the bubble size is equal to or smaller than a predetermined reference size and the number of bubbles is equal to or greater than a predetermined reference number. In this embodiment, the reference size is a first size S1. The reference number is a first number N1. In other words, the control unit 200 determines that the judgment condition is satisfied when the state of the bubbles is State A, and determines that the judgment condition is not satisfied when the state of the bubbles is State B or State C. If it is determined that the bubble state satisfies the determination condition, the control unit 200 maintains the rotor rotation speed and gas flow rate without changing them. If it is determined that the bubble state does not satisfy the determination condition, the control unit 200 adjusts the stirring conditions to increase the number of bubbles. In this case, both the rotor rotation speed and the gas flow rate may be changed, or one may be fixed and only the other may be changed. Note that, since the number and size of bubbles are negatively correlated, the number of bubbles can also be increased by adjusting the stirring conditions to reduce the size of the bubbles.

[0038] In this embodiment, the control unit 200 adjusts the mixing conditions based on the history of the bubble state and the history of at least one of the rotation speed and the flow rate. Hereinafter, the history of at least one of the rotation speed and the flow rate is also referred to as the mixing condition history. More specifically, the control unit 200 adjusts the mixing conditions using a machine learning model 206 generated by machine learning based on the bubble state history and the mixing condition history. The machine learning model 206 is generated, for example, as a model that learns, through reinforcement learning, the optimal adjustment amounts of the rotor rotation speed and the gas flow rate to increase the number of bubbles based on the bubble state history and the mixing condition history. Note that the learning algorithm for generating the machine learning model 206 is not particularly limited, and in other embodiments, it may be, for example, supervised learning or unsupervised learning.

[0039] In step S140, the control unit 200 records the results of adjusting the mixing conditions. For example, in step S140, the control unit 200 associates the bubble state before executing the adjustment process in step S135 with the rotor rotation speed and gas flow rate before executing the adjustment process and records them in the storage unit of the computer 205, and also associates the bubble state after executing the adjustment process with the rotor rotation speed and gas flow rate after executing the adjustment process and records them in the storage unit. Furthermore, for example, the change in the bubble state due to the adjustment process may be associated and recorded with the change in the rotor rotation speed and gas flow rate in the adjustment process. This updates the bubble state history and the mixing condition history. Thereafter, the control unit 200 updates the machine learning model 206 based on the updated bubble state history and mixing condition history.

[0040] In step S145, the control unit 200 determines whether or not to terminate the purification of the molten metal MM. For example, when the time elapsed since the start of step S120 is equal to or longer than a predetermined stirring time at the time step S145 is executed, the control unit 200 determines to terminate the purification of the molten metal MM. In this case, the control unit 200 may vary the length of the stirring time based on the state detected during the stirring process. For example, the control unit 200 may shorten the stirring time each time state A is detected as a bubble state during the stirring process, and may lengthen the stirring time each time state C is detected. If the control unit 200 determines not to terminate the purification of the molten metal MM, it returns the process to step S125 again. If the control unit 200 determines to terminate the stirring of the molten metal MM, the control unit 200 stops the rotation of the rotor 111 and the release of the inert gas in step S150, thereby stopping the stirring of the molten metal MM.

[0041] The molten metal purification apparatus 100 of the present embodiment described above includes a first photographing unit that obtains a first image by photographing the rotor 111 before it is inserted into the molten metal MM, a first detection unit 210 that detects position information based on the first image, and a transfer control unit 255 that controls the transfer unit 130 based on the position information to insert the rotor 111 into the molten metal MM.The apparatus also includes a second photographing unit that obtains a second image by photographing the molten metal MM while the molten metal MM is being stirred, a second detection unit 215 that detects the stirring state based on the second image, and an adjustment unit 260 that adjusts the stirring conditions based on the stirring state.

[0042] This allows the conveying unit 130 to insert the rotor 111 into the molten metal MM based on position information detected from the first image. Therefore, not only can the rotor 111 be automatically inserted into the molten metal MM, but the rotor 111 can be more easily inserted into the desired position in the molten metal MM compared to, for example, a configuration in which the rotor 111 is inserted into the molten metal MM by controlling the position of the rotor 111 without using the first image. In particular, in this embodiment, the control unit 200 detects position information based on the detection results of the molten metal MM in the first image. Therefore, even if, for example, the shape or position of the container Ct is changed, the rotor 111 can be easily inserted into the desired position in the molten metal MM. Furthermore, in this embodiment, the rotor rotation speed and gas flow rate can be adjusted based on the stirring state detected from the second image. Therefore, not only can the rotor rotation speed and gas flow rate be automatically adjusted after the rotor 111 is inserted into the molten metal MM, but the rotor rotation speed and gas flow rate can be adjusted according to the actual stirring state without relying on the operator's skill level. In this way, in the molten metal purification device 100 equipped with the transfer section 130, the molten metal MM can be purified simply and easily.

[0043] In this embodiment, the first detection unit 210 detects the state of bubbles contained in the molten metal MM as the stirring state, and if the state of the bubbles does not satisfy a predetermined condition, the adjustment unit 260 adjusts the stirring conditions to increase the number of bubbles. This makes it possible to appropriately adjust the rotor rotation speed and gas flow rate based on the state of the bubbles detected by image recognition of the second image. As a result, the molten metal MM can be effectively purified.

[0044] Furthermore, in this embodiment, the adjustment unit 260 adjusts the stirring conditions based on the history of the bubble state and the history of the stirring conditions. Therefore, the rotor rotation speed and the gas flow rate can be appropriately adjusted based on the history. As described above, when the adjustment unit 260 adjusts only one of the rotor rotation speed and the gas flow rate as the stirring conditions, the history of the stirring conditions may be the history of either the rotor rotation speed or the gas flow rate. For example, when the adjustment unit 260 adjusts only the rotor rotation speed as the stirring condition, the history of the stirring conditions may be the history of only the rotor rotation speed.

[0045] Furthermore, in this embodiment, the adjustment unit 260 adjusts the mixing conditions using the machine learning model 206 generated by machine learning based on the history of the bubble state and the history of the mixing conditions. Therefore, the rotor rotation speed and the gas flow rate can be appropriately adjusted using the results of the machine learning.

[0046] B. Other Embodiments: (B1) In the above embodiment, the conveying unit 130 includes a rail 131 and a crane 132, and the rotor 111 is moved by the crane 132. However, the conveying unit 130 does not have to be configured in this manner, and may be configured, for example, as an automated guided forklift (AGF) that moves while supporting the agitating unit 110. Furthermore, the conveying unit 130 may be configured, for example, to convey the container Ct relative to the rotor 111. Furthermore, for example, the conveying unit 130 may be configured to convey both the container Ct and the rotor 111.

[0047] (B2) In the above embodiment, the control unit 200 detects the position information based on the result of image recognition of the molten metal MM included in the first image. However, this detection of the position information is not required. For example, when detecting the position information, the control unit 200 may perform image recognition of the container Ct included in the first image, or may perform image recognition of a marker for detecting the position of the container Ct or the molten metal MM. Furthermore, if the transport unit 130 is configured to transport the container Ct relative to the rotor 111, the control unit 200 may perform image recognition of the rotor 111, for example. Furthermore, the control unit 200 may detect the position information based on the first image using, for example, a machine learning model that learns the relationship between the image captured by the first image capture unit and the position information through machine learning such as supervised learning. By using such a machine learning model to detect the position information, it is possible to suppress a decrease in the detection accuracy of the position information due to the influence of impurities (e.g., metal oxides) attached to the container Ct.

[0048] (B3) In the above embodiment, the photographing unit 125 functions as both the first photographing unit and the second photographing unit, and therefore the photographing unit 125 must be positioned so that it can capture both the first image and the second image. Alternatively, the first photographing unit and the second photographing unit may be provided separately. In this case, the first photographing unit may be positioned so that it can capture the first image, and the second photographing unit may be positioned so that it can capture the second image. This increases the degree of freedom in the placement of the first photographing unit and the second photographing unit. For example, if a first photographing unit separate from the second photographing unit positioned so that it can face the molten metal MM is positioned so that it can face the rotor 111, an image including the rotor 111 can be easily captured as the first image. Furthermore, if an image including a marker for detecting the position of the container Ct or the molten metal MM is captured as the first image, the degree of freedom in the placement of the marker can be increased.

[0049] (B4) In the above embodiment, the state of bubbles is represented by both the size and number of bubbles, but it may be represented by either one of them. In this case, the determination condition only needs to be a condition related to either the size or the number of bubbles. Furthermore, although the second detection unit 215 detects the state of bubbles as the stirring state, it is not necessary to detect the state of bubbles. For example, the flow state of the molten metal MM may be detected as the stirring state based on a plurality of second images.

[0050] (B5) In the above embodiment, the control unit 200 adjusts the rotor rotation speed and the gas flow rate using the machine learning model 206. However, the control unit 200 does not have to adjust the rotor rotation speed and the gas flow rate using the machine learning model 206. For example, the control unit 200 may adjust the rotor rotation speed and the gas flow rate by referring to data recording the relationship between the bubble state history and the rotor rotation speed and the gas flow rate history required to achieve that bubble state, based on the bubble state detected based on the second image. This data may be, for example, data created based on the results of a prior experiment or simulation of the relationship between the bubble state and the rotor rotation speed and the gas flow rate, and corrected based on the history for a predetermined period from the past to the present. This allows the rotor rotation speed and the gas flow rate to be adjusted, taking into account, for example, the impact of seasonal changes in the temperature and humidity of the installation location of the molten metal purification apparatus 100 on purification. This data may also be data corrected based on the temperature and humidity of the installation location of the molten metal purification apparatus 100.

[0051] (B6) In the above embodiment, the control unit 200 adjusts the stirring conditions based on the history of the bubble state and the history of the stirring conditions. However, the control unit 200 does not have to adjust the stirring conditions in this way. For example, the control unit 200 may adjust the stirring conditions based on the results of a prior experiment or simulation of the relationship between the bubble state and the rotor rotation speed or gas flow rate.

[0052] The present disclosure is not limited to the above-described embodiments and can be realized in various configurations without departing from the spirit thereof. For example, the technical features in the embodiments corresponding to the technical features in each aspect described in the Summary of the Invention section can be appropriately replaced or combined to solve some or all of the above-described problems or achieve some or all of the above-described effects. Furthermore, if a technical feature is not described as essential in this specification, it can be appropriately deleted. [Explanation of symbols]

[0053] 100...molten metal purification device, 110...stirring section, 111...rotor, 112...shaft, 113...support section, 114...rotation drive section, 115...gas discharge hole, 116...gas supply hole, 120...gas supply section, 125...photographing section, 130...transport section, 131...rail, 132...crane, 133...first transport mechanism, 134...second transport mechanism, 140...preheating section, 200...control section, 205...computer, 206...machine learning model, 210...first detection section, 215...second detection section, 255...transport control section, 260...adjustment section

Claims

1. A molten metal cleaning device for cleaning molten metal contained in a container, comprising: a rotor that rotates within the molten metal to agitate the molten metal and releases an inert gas into the molten metal; a conveying unit configured to convey the rotor relative to the container so that the rotor can be inserted into the molten metal; a first photographing unit that photographs a first image until the rotor is inserted into the molten metal; a first detection unit that detects position information regarding a relative position between the rotor and the molten metal based on the first image; a transfer control unit that controls the transfer unit based on the position information to insert the rotor into the molten metal; a second photographing unit that photographs a second image including the molten metal while the molten metal is being stirred; a second detection unit that detects a stirring state regarding stirring of the molten metal based on the second image; an adjusting unit that adjusts at least one of the rotation speed of the rotor and the flow rate of the inert gas based on the stirring state.

2. The molten metal cleaning apparatus according to claim 1, the second detection unit detects, as the stirring state, a state of bubbles represented by at least one of a size of bubbles contained in the molten metal and a number of the bubbles; The molten metal cleaning device, wherein the adjustment unit adjusts at least one of the rotation speed and the flow rate so as to increase the number of particles when the state does not satisfy a predetermined condition.

3. The molten metal cleaning apparatus according to claim 2, The molten metal cleaning device adjusts at least one of the rotation speed and the flow rate based on the history of the state and the history of at least one of the rotation speed and the flow rate.

4. The molten metal cleaning apparatus according to claim 3, The adjustment unit adjusts at least one of the rotation speed and the flow rate using a machine learning model generated by machine learning based on the history of the state and the history of at least one of the rotation speed and the flow rate.

5. A molten metal cleaning method for cleaning molten metal contained in a vessel using a rotor, comprising the steps of: taking a first image while the rotor is inserted into the molten metal; detecting position information relating to a relative position between the rotor and the molten metal based on the first image; inserting the rotor into the molten metal by transporting the rotor relative to the container based on the position information; a step of rotating the rotor in the molten metal to stir the molten metal and releasing an inert gas from the rotor into the molten metal; taking a second image including the molten metal while stirring the molten metal; detecting a stirring state representing a state related to stirring of the molten metal based on the second image; and adjusting at least one of the rotation speed of the rotor and the flow rate of the inert gas based on the stirring state.

Citation Information

Patent Citations

  • JP1980124443U

  • Rotary device for dispersing gases for treatment of liquid metal baths

    JP2004511661A

  • Method and device for removing inclusion in molten metal

    JP2005014047A

  • Method and apparatus for cleaning molten metal

    JP2010084213A

  • Gas injection jig for molten metal processing

    JP2022147875A