EP4656741A1

EP4656741A1 - Molten iron constituent estimation device, constituent estimation method, and desulfurization method

Info

Publication number: EP4656741A1
Application number: EP24766687.8A
Authority: EP
Inventors: Hironori Yoshida; Yoshie Nakai; Amon Takahashi; Koshiro Ejima
Original assignee: JFE Steel Corp
Current assignee: JFE Steel Corp
Priority date: 2023-03-08
Filing date: 2024-01-19
Publication date: 2025-12-03
Also published as: CN120813707A; JPWO2024185312A1; TW202442877A; TWI891267B; KR20250140574A; JP7786580B2; WO2024185312A1

Abstract

There are provided a hot metal constituent estimation device, a hot metal constituent estimation method, and a hot metal desulfurization method capable of accurately estimating, in desulfurization treatment of hot metal using a mechanical stirring facility, a constituent of the hot metal after the desulfurization treatment. A constituent estimation device (14) for estimating the S concentration in hot metal (3) in desulfurization treatment of the hot metal (3) using a mechanical stirring desulfurization facility (1) including: a data input unit (140) inputting current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment, the hot metal temperature being the temperature of the hot metal (3) measured using a noncontact temperature measuring device (12) after a start of stirring of the hot metal (3) and before a start of charging of a desulfurization flux (4); and a model calculation unit (142) estimating the S concentration in the hot metal (3) after the desulfurization treatment from a machine learning model indicating the relation among the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux (4), and a reaction rate constant in the desulfurization treatment based on past operation data in the desulfurization treatment and the current operation data.

Description

Technical Field

The present invention relates to a hot metal constituent estimation device, a hot metal constituent estimation method, and a hot metal desulfurization method.

Background Art

S contained in steel leads to hot shortness or deterioration of corrosion resistance of steel and a decrease in toughness and workability, and therefore a decrease in the S concentration in steel has been demanded. Further, in recent years, a demand for higher purity or higher cleanliness of steel has increased, requiring efforts to further reduce the sulfurization of steel, i.e., to accelerate desulfurization and suppress resulfurization in smelting treatment of molten iron.
Desulfurization treatment of molten iron is carried out in pretreatment step of hot metal tapped from a blast furnace or a secondary smelting step of molten steel. For the desulfurization treatment in the pretreatment step of hot metal, a KR process using a mechanical stirring facility has been widely applied and various methods have been proposed to efficiently perform the desulfurization treatment.
For example, PTL 1 proposes a constituent concentration arithmetic unit and a constituent concentration arithmetic method estimating the S concentration in hot metal after smelting treatment based on past operation data in the smelting treatment of molten iron. This method is a method including, assuming that a desulfurization reaction of hot metal follows a first-order reaction formula, estimating model parameters contained in the first-order reaction formula, such as a reaction rate constant, which are difficult to measure, using a constituent concentration estimation model from the past operation data and a current hot metal constituent before desulfurization treatment.

Citation List

Patent Literature

PTL 1: JP 2020-15959 A

Summary of Invention

Technical Problem

However, the method in PTL 1 estimates the S concentration in the hot metal based on only the current information before the desulfurization treatment, and does not disclose a method for controlling operating conditions using information obtained during the desulfurization treatment. Therefore, there is a possibility that the S concentration in the hot metal after the desulfurization treatment cannot be accurately estimated. For example, when unexpected changes, such as a sudden change in the hot metal temperature or a change in the inclusion state of a desulfurization flux and a slag into the hot metal, occur during the desulfurization treatment, the desulfurization capacity decreases, which leads to a desulfurization failure of the hot metal.
The present invention has been made in view of the above-described problems. It is an object of the present invention to provide a hot metal constituent estimation device, a hot metal constituent estimation method, and a hot metal desulfurization method capable of accurately estimating, in desulfurization treatment of hot metal using a mechanical stirring facility, the S concentration which is a constituent of the hot metal after the desulfurization treatment.

Solution to Problem

(1) One aspect of the present invention provides a hot metal constituent estimation device for estimating the S concentration in hot metal in desulfurization treatment of the hot metal using a mechanical stirring desulfurization facility,
the hot metal constituent estimation device including: a data input unit configured to input current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment, the hot metal temperature being the temperature of the hot metal measured using a noncontact temperature measuring device after a start of stirring of the hot metal and before a start of charging of a desulfurization flux; and a model calculation unit configured to estimate the S concentration in the hot metal after the desulfurization treatment from a machine learning model indicating the relation among the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and a reaction rate constant in the desulfurization treatment based on past operation data in the desulfurization treatment and the current operation data.
(2) The hot metal constituent estimation device in (1) above, the data input unit being configured to input at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and the reaction rate constant, the reaction rate constant being a constant in a desulfurization reaction formula represented by Formula (1), in past operation as the past operation data,
- the hot metal constituent estimation device further including: a model parameter decision unit configured to create the machine learning model based on the past operation data input by the data input unit. $[S] = {[S]}_{0} \cdot \exp \{K (t - t_{0})\}$
- In Formula (1), [S] denotes the S concentration (mass%) in the hot metal, [S]₀ denotes the S concentration (mass%) in the hot metal before a start of the desulfurization treatment, K denotes the reaction rate constant (s^-1), t denotes desulfurization treatment time (s), and t₀ denotes time (s) from the start of the stirring to the completion of inclusion of a slag into the hot metal.
(3) One aspect of the present invention provides a hot metal constituent estimation method for estimating the S concentration in hot metal in desulfurization treatment of the hot metal using a mechanical stirring desulfurization facility,
the hot metal constituent estimation method including: measuring the hot metal temperature during the desulfurization treatment, the hot metal temperature being the temperature of the hot metal after a start of stirring of the hot metal and before a start of charging of a desulfurization flux using a noncontact temperature measuring device; inputting current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment to be measured; and estimating the S concentration in the hot metal after the desulfurization treatment from a machine learning model indicating the relation among the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and a reaction rate constant in the desulfurization treatment based on past operation data in the desulfurization treatment and the current operation data.
(4) The hot metal constituent estimation method in (3) above, at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and the reaction rate constant, the reaction rate constant being a constant in a desulfurization reaction formula represented by Formula (1), in past operation being used as the past operation data,
- the hot metal constituent estimation method further including: creating the machine learning model based on the input past operation data. $[S] = {[S]}_{0} \cdot \exp \{K (t - t_{0})\}$
- In Formula (1), [S] denotes the S concentration (mass%) in the hot metal, [S]₀ denotes the S concentration (mass%) in the hot metal before a start of the desulfurization treatment, K denotes the reaction rate constant (s^-1), t denotes desulfurization treatment time (s), and t₀ denotes time (s) from the start of the stirring to the completion of inclusion of a slag into the hot metal.
(5) One aspect of the present invention provides a hot metal desulfurization method using a mechanical stirring desulfurization facility including: estimating the S concentration in the hot metal after the desulfurization treatment using the hot metal constituent estimation method according to (3) or (4) above during the desulfurization treatment.
(6) The hot metal desulfurization method in (5) above, further including: after the estimating the S concentration, determining whether the S concentration in the hot metal after the desulfurization treatment is in a predetermined range, in which, when the S concentration in the hot metal after the desulfurization treatment is not in the predetermined range, the charging amount of the desulfurization flux is changed, the estimating the S concentration is performed again, and the S concentration in the hot metal is estimated again.
(7) The hot metal desulfurization method according to (5) or (6) above, further including: after the estimating the S concentration, measuring the temperature of the bath surface of the hot metal; determining the inclusion state of a slag on the bath surface of the hot metal from the measurement result in the measuring the temperature of the bath surface of the hot metal; and charging a desulfurization flux into the hot metal when it is determined that the inclusion of the slag has completed in the determining the inclusion state of the slag.
(8) The hot metal desulfurization method according to any one of (5) to (7) above, further including:
measuring the temperature of the bath surface of the hot metal after the desulfurization flux has been charged; and
determining the inclusion state of the slag on the bath surface of the hot metal from the measurement result in the measuring the temperature of the bath surface of the hot metal, in which, when it is determined that the slag is properly included in the determining the inclusion state of the slag, stirring by an impeller of the mechanical stirring desulfurization facility is continued and when it is determined that the slag is not properly included in the determining the inclusion state of the slag, at least one of the rotational speed and the immersion depth of the impeller is changed.

Advantageous Effects of Invention

One aspect of the present invention can provide a hot metal constituent estimation device, a hot metal constituent estimation method, and a hot metal desulfurization method capable of accurately estimating, in desulfurization treatment of hot metal using a mechanical stirring facility, the constituent of the hot meatal after the desulfurization treatment.

Brief Description of Drawings

FIG. 1 is a configuration diagram illustrating a mechanical stirring desulfurization facility 1 according to one embodiment of the present invention;
FIG. 2 is a flowchart showing a hot metal desulfurization method according to one embodiment of the present invention;
FIG. 3 is a flowchart showing a hot metal constituent estimation method according to one embodiment of the present invention; and
FIG. 4 is a flowchart showing a method for determining the inclusion state of a slag after a start of charging of a desulfurization flux and changing stirring conditions.

Description of Embodiments

A detailed description below describes embodiments of the present invention with reference to the drawings. In the description of the drawings, the same or similar reference numerals are attached to the same or similar parts, and duplicated descriptions are omitted. The drawings are schematic and are sometimes different from the actual ones. The embodiments described below exemplify devices and methods for embodying the technical idea of the present invention. The technical idea of the present invention does not specify materials, structures, arrangement, and the like of constituent components to the materials, structures, arrangement, and the like described below. The technical idea of the present invention can be variously altered within the technical range defined by Claims.
First, the present inventors have decided the charging amount of a desulfurization flux based on past operation data and current operation data in hot metal desulfurization treatment using a mechanical stirring desulfurization facility, charged the desulfurization flux into hot metal, and investigated the desulfurization behavior of the hot metal. As a result, the present inventors have found that the S concentration can be accurately estimated by estimating the S concentration in the hot metal during the desulfurization treatment based on a machine learning model indicating the relation among the temperature of the hot metal, the charging amount of the desulfurization flux, and a reaction rate constant using the temperature of the hot metal before a start of the charging of the desulfurization flux and a first-order reaction formula. Then, the present inventors have found that, by the use of the estimation method, a desulfurization failure of the hot metal and the excessive charging of the desulfurization flux can be suppressed and the desulfurization treatment can be efficiently performed. The following description describes a constituent estimation device, a hot metal constituent estimation method, and a hot metal desulfurization method according to one embodiment of the present invention based on these findings.

A mechanical stirring desulfurization facility in one embodiment of the present invention is described. As illustrated in FIG. 1, a mechanical stirring desulfurization facility 1 is a desulfurization treatment facility (also referred to as a KR desulfurization treatment facility) performing desulfurization treatment of hot metal 3 stored in a hot metal ladle 2 by mechanical stirring. The mechanical stirring desulfurization facility 1 includes an impeller 10, a desulfurization flux charging device 11, a temperature measuring device 12, a control terminal 13, a constituent estimation device 14, and a determination device 15.
The impeller 10 is a mechanical stirring bar having a rotating shaft and a refractory stirring blade at the tip of the rotating shaft. The impeller 10 is configured to move up and down in the substantially vertical direction by a lifting device (not illustrated) and to be rotated by a rotating device (not illustrated), which contains a drive motor and a reduction gear, with a shaft having a stirring blade connected to the tip as the rotating shaft. The impeller 10 immerses the stirring blade at the tip in the hot metal 3 and rotates to mechanically stirrer the hot metal 3.
The desulfurization flux charging device 11 is a device charging the desulfurization flux 4 into the hot metal 3. The desulfurization flux 4 is not limited to any specific substance, and a lime-based solvent material, a mixture of a lime-based solvent material and aluminum oxide, and the like can be applied. A method for charging the desulfurization flux 4 is not restricted to a specific method, and a method for charging the desulfurization flux 4 in batches or continuously from above the hot metal ladle 2, a method for spraying the desulfurization flux 4 onto the surface of the hot metal 3 together with a carrier gas, and the like can be applied.
The temperature measuring device 12 is a noncontact thermometer arranged above the hot metal ladle 2 and measures the temperature of the bath surface of the hot metal 3 stored in the hot metal ladle 2. The temperature measuring device 12 is not particularly limited insofar as the temperature of the hot metal 3 or a slag 5 can be measured in a noncontact manner and, for example, a radiation thermometer, a thermography, and the like can be used. A method for measuring the temperature of the hot metal 3 by immersing a temperature measuring probe in the hot metal has a risk that the temperature measuring probe comes into contact with the impeller 10 or the slag 5 in a solid phase during the desulfurization treatment, i.e., in a case of a state in which the impeller 10 rotates. In this embodiment, the use of the noncontact thermometer as the temperature measuring device 12 can avoid such a contact risk and enables the measurement of the temperatures of the hot metal 3 exposed to the bath surface of the hot metal 3 and the slag 5 coating the bath surface of the hot metal 3.
The control terminal 13 controls the desulfurization treatment in the mechanical stirring desulfurization facility 1 by controlling the impeller 10 and the desulfurization flux charging device 11. Specifically, the control terminal 13 instructs the stirring conditions, such as the rotational speed and the immersion depth, which are the operating conditions of the impeller 10, to the impeller 10 and acquires the actual stirring conditions. The immersion depth of the impeller 10 is the depth with which the impeller 10 is immersed with respect to the bath surface position of the hot metal 3 indicated by the dotted line in the hot metal ladle 2 illustrated in FIG. 1. The control terminal 13 also instructs the charging amount and the charging timing of the desulfurization flux 4 to the desulfurization flux charging device 11 and acquires the actual charging amount and charging timing of the desulfurization flux 4. The control terminal 13 also instructs the desulfurization flux charging device 11 to charge the desulfurization flux 4 according to the charging amount of the desulfurization flux 4 acquired from the constituent estimation device 14, the determination result of the slag inclusion acquired from the determination device 15, or the like. Further, the control terminal 13 acquires the determination result of the slag inclusion from the determination device 15 and, as necessary, changes the stirring conditions during the desulfurization treatment. Further, the control terminal 13 stores past operation data. The past operation data is information on the hot metal 3 from before the desulfurization treatment to after the desulfurization treatment of the hot metal 3 and information on the running conditions of the mechanical stirring desulfurization facility 1 in past desulfurization treatment. The past operation data includes, for example, the constituent, the temperature, and the weight of the hot metal 3, the weight of a desulfurization slag, the charging amounts of the desulfurization flux 4 and the other auxiliary raw materials, the rotational speed, the immersion depth, and the use frequency of the impeller 10, desulfurization treatment time, time to the completion of the inclusion of the slag 5, and the use frequency of the hot metal ladle 2. The temperature of the hot metal 3 also includes the hot metal temperature during the desulfurization treatment described later besides the temperature before the desulfurization treatment. The time to the completion of the inclusion of the slag 5 is time from a start of the stirring (mechanical stirring) of the hot metal 3 by the impeller 10 to the completion of the inclusion of the slag 5 into the hot metal 3. The other auxiliary raw materials are auxiliary raw materials other than the desulfurization flux 4 to be charged into the hot metal 3, and include, for example, iron-containing dust, used refractory wastes, and the like generated in an iron making process.
The constituent estimation device 14 estimates at least the S concentration in the hot metal 3 as the constituent of the hot metal 3 after the desulfurization treatment based on the past and current operation data and the temperature measurement result of the hot metal 3 by the temperature measuring device 12. The constituent estimation device 14 has a data input unit 140, a model parameter decision unit 141, and a model calculation unit 142. The details of a constituent estimation method and each configuration of the constituent estimation device 14 are described later.
The determination device 15 is a device performing various determinations in the desulfurization treatment, and has a temperature distribution image creation unit 150, a slag inclusion determination unit 151, and an S concentration determination unit 152. The temperature distribution image creation unit 150 creates a temperature distribution image from the measurement result of the temperature measuring device 12. The slag inclusion determination unit 151 determines whether the slag 5 is included from the temperature distribution image created in the temperature distribution image creation unit 150. The S concentration determination unit 152 determines whether the estimation value of the S concentration in the hot metal 3 after the desulfurization treatment estimated in the constituent estimation device 14 is in a predetermined range described later. The details of a determination method and each configuration of the determination device 15 are described later.
The control terminal 13, the constituent estimation device 14, and the determination device 15 may be provided in the form of programming to be executed by a computer or a PLC.

The desulfurization method of the hot metal 3 according to this embodiment is described. In this embodiment, the desulfurization treatment is performed according to the flowchart illustrated in FIG. 2. In this embodiment, the hot metal 3 is tapped from a blast furnace, and is received by a hot metal holding/transporting vessel, such as a hot metal ladle or a torpedo car, after the tapping. Then, hot metal pretreatment, such as desiliconization or dephosphorization, is performed as necessary. Thereafter, the hot metal 3 is transferred to the hot metal ladle 2 as necessary, and transported to the mechanical stirring desulfurization facility 1 in a state of being stored in the hot metal ladle 2. Before the desulfurization treatment, the constituent, the weight, and the temperature of the hot metal 3 are measured, and the measurement results are recorded in the control terminal 13. Into the hot metal ladle 2, auxiliary raw materials, such as the desulfurization slag, which is a slag generated in the last desulfurization treatment, may be charged in advance as necessary.
First, the data input unit 140 of the constituent estimation device 14 acquires pre-desulfurization treatment data, which is information on the hot metal 3 before the desulfurization treatment, from the control terminal 13 (S100). The pre-desulfurization treatment data is an operating condition related to the desulfurization treatment before the desulfurization treatment measured in advance. The pre-desulfurization treatment data includes, for example, the constituent, the temperature, and the weight of the hot metal 3, the weight of the desulfurization slag, and the use frequency of the hot metal ladle 2. The pre-desulfurization treatment data also includes, as preset values (initial values), the charging amounts of the desulfurization flux 4 and the other auxiliary raw materials, the rotational speed, the immersion depth, and the use frequency of the impeller 10, the desulfurization treatment time, and the time to the completion of the inclusion of the slag 5. Further, it may be configured such that the pre-desulfurization treatment data is stored also on a host computer, a server, and the like which are not illustrated, and the data input unit 140 acquires the pre-desulfurization treatment data from the host computer, the server, and the like. Step S100 may be performed before the desulfurization treatment, e.g., after the tapping, after the hot metal pretreatment, or the like.
Subsequently, the control terminal 13 starts the desulfurization treatment by immersing the stirring blade of the impeller 10 in the hot metal 3 and rotating the impeller 10 (S102). For the stirring conditions, such as the rotational speed and the immersion depth, of the impeller 10, preset conditions are used.
The temperature measuring device 12 measures the temperature of the bath surface of the hot metal 3 (S104). Step S104 is performed after a predetermined time has elapsed from the start of the stirring of the impeller 10. The predetermined time is set as time for the hot metal 3 to be sufficiently exposed to the extent that the temperature can be measured in the bath surface (upper surface) of the hot metal 3, and is set as appropriate according to the amount of the slag 5, the shape/dimension of the stirring blade of the impeller 10, the stirring conditions, and the like. In mechanical stirring desulfurization treatment, the treatment is usually started in a state in which the bath surface of the hot metal 3 is coated with the slag 5, such as the desulfurization slag. Therefore, the stirring is performed by the impeller 10 and the predetermined time has passed, so that the hot metal 3 is exposed to the bath surface of the hot metal 3, enabling the measurement of the temperature of the hot metal 3. The temperature measurement result of the hot metal 3 is transmitted to and stored in the control terminal 13. The temperature of the hot metal 3 measured in step S104, i.e., the temperature of the hot metal 3 after the start of the stirring and before the start of the charging of the desulfurization flux 4 during the desulfurization treatment, is also referred to as the hot metal temperature during the desulfurization treatment.
Thereafter, the constituent estimation device 14 estimates the S concentration in the hot metal 3 after the desulfurization treatment (S106). The estimation of the constituent (S concentration) of the hot metal 3 by the constituent estimation device 14 is performed by the following method.

(Hot metal constituent estimation method)

In a constituent estimation method of the hot metal 3 according to this embodiment, the S concentration in the hot metal 3 after the desulfurization treatment is estimated as described above. In this embodiment, the S concentration in the hot metal 3 after the desulfurization treatment is estimated based on the temperature of the hot metal 3 measured by the temperature measuring device 12, the pre-desulfurization treatment data of the hot metal 3, and the part operation data. In the following description, the temperature of the hot metal 3 measured by the temperature measuring device 12 (hot metal temperature in desulfurization treatment) and the pre-desulfurization treatment data of the hot metal 3 are also collectively referred to as the current operation data.
In the constituent estimation method of the hot metal 3 according to this embodiment, the estimation of the constituent of the hot metal 3 is performed according to the flowchart illustrated in FIG. 3. First, the data input unit 140 acquires the current operation data and the past operation data from the control terminal 13 (S200).
Subsequently, the model parameter decision unit 141 creates the machine learning model indicating the relation among the temperature of the hot metal 3, the charging amount of the desulfurization flux 4, and a reaction rate constant K from the past operation data (S202). Herein, the reaction rate constant K is a constant in a desulfurization reaction formula represented by Formula (1) described below. $[S] = {[S]}_{0} \cdot \exp \{K (t - t_{0})\}$
In Formula (1), [S] denotes the S concentration (mass%) in the hot metal, [S]₀ denotes the S concentration (mass%) in the hot metal before the start of the desulfurization treatment, K denotes the reaction rate constant (s^-1), t denotes desulfurization treatment time (s), and t₀ denotes time (s) from the start of the stirring to the completion of inclusion of a slag into the hot metal.
When the inclusion of the slag 5 into the hot metal 3 is improper and the desulfurization flux 4 is charged in a state in which the surface of the hot metal 3 is coated with the slag, there is a risk that the desulfurization flux 4 is captured by the slag 5 before reacting with S in the hot metal 3, reducing the desulfurization capacity and causing a desulfurization failure of the hot metal 3. Therefore, it is important to create a model considering the time t₀ from the start of the stirring to the completion of the inclusion of the slag 5 into the hot metal 3. The time t₀ from the start of the stirring to the completion of the inclusion of the slag 5 into the hot metal 3 can be decided by continuously measuring the temperature of the bath surface of the hot metal 3 using the temperature measuring device 12 and determining the inclusion state of the slag 5 described later.
It is important to decide the reaction rate constant K in Formula (1) from the past operation data and the current operation data because it is difficult to actually measure the reaction rate constant K.
The type of the machine learning model is not particularly restricted and, for example, a model utilizing a neural network and the like can be applied. Thus, the reaction rate constant K can be accurately decided based on various kinds of operation data.
When the hot metal temperature is higher and the charging amount of the desulfurization flux 4 is larger, the desulfurization reaction of the hot metal 3 is more likely to proceed. Therefore, when the reaction rate constant K is decided, it is important to input the temperature of the hot metal 3 and the charging amount of the desulfurization flux 4 as explanatory variables.
After step S202, the model parameter decision unit 141 decides the reaction rate constant K from the machine learning model created in step S202, the hot metal temperature during the desulfurization treatment measured in step S104, and the pre-desulfurization treatment data of the hot metal 3 (S204). In step S204, it is preferable that the reaction rate constant K is decided using at least the initial value of the charging amount of the desulfurization flux 4, the weight of the desulfurization slag, and the charging amount of the other auxiliary raw materials among the pre-desulfurization treatment data of the hot metal 3.
Thereafter, the model calculation unit 142 estimates the S concentration in the hot metal 3 after the desulfurization treatment from the reaction rate constant K determined in step S204 and the pre-desulfurization treatment data of the hot metal 3 based on the desulfurization reaction formula represented by Formula (1) (S206). The estimated S concentration of the hot metal 3 after the desulfurization treatment is transmitted to the S concentration determination unit 152 of the determination device 15.
In the constituent estimation method of the hot metal 3 according to this embodiment, the S concentration in the hot metal 3 after the desulfurization treatment is estimated using the hot metal temperature during the desulfurization treatment measured in step S104. In the estimation of the S concentration in the hot metal 3 after the desulfurization treatment, it is important to accurately estimate the reaction rate constant K in the desulfurization treatment. A factor significantly influencing the reaction rate constant K includes the temperature of the hot metal 3. Therefore, the use of not the temperature measured before the desulfurization treatment but the actual temperature during the desulfurization treatment as the temperature of the hot metal 3 can enhance the estimation accuracy. Herein, in the conventional findings, it has been difficult to estimate the hot metal temperature during the desulfurization treatment based on the temperature before the desulfurization treatment due to various factors, such as the use state of the hot metal ladle 2 where the hot metal 3 is stored and the temperature and the charging amount of the desulfurization slag to be charged. In contrast thereto, this embodiment can accurately estimate the reaction rate constant K and also accurately estimate the S concentration in the hot metal 3 after the desulfurization treatment by measuring the temperature of the hot metal 3 during the desulfurization treatment in step S104 and using the measured hot metal temperature during the desulfurization treatment.
After step S106, the S concentration determination unit 152 determines whether the S concentration in the hot metal 3 after the desulfurization treatment estimated in step S106 is in a predetermined range (S108). The predetermined range of the S concentration is set as appropriate as a preferable S concentration after the desulfurization treatment according to the target S concentration of the hot metal 3. Specifically, the upper limit of the predetermined range of the S concentration is set from the viewpoint of the quality. On the other hand, the lower limit of the predetermined range of the S concentration is set as an acceptable value considering resulfurization or the like in subsequent smelting treatment. The determination result in step S108 is transmitted to the control terminal 13.
When it has been determined in step S108 that the S concentration in the hot metal 3 after the desulfurization treatment is not in the predetermined range, the control terminal 13 changes the charging amount (scheduled quantity) of the desulfurization flux 4 (S110). In step S110, when the estimated S concentration of the hot metal 3 after the desulfurization treatment is larger than the upper limit of the predetermined range, the charging amount of the desulfurization flux 4 is changed to be larger than the charging amount of the desulfurization flux 4 used in the estimation in immediately preceding step S106. On the other hand, when the estimated S concentration in the hot metal 3 after the desulfurization treatment is smaller than the lower limit of the predetermined range, the charging amount of the desulfurization flux 4 is changed to be smaller than the charging amount of the desulfurization flux 4 used in the estimation in immediately preceding step S106. More specifically, in step S108, the charging amount of the desulfurization flux 4 is changed such that the S concentration in the hot metal 3 after the desulfurization treatment is in the predetermined range. The adjustment amount of the desulfurization flux 4 in step S108 is not particularly limited. However, when the adjustment amount is excessively increased, there is a possibility that the S concentration after the desulfurization treatment is not in the predetermined range. On the other hand, when the adjustment amount is excessively small, repeated calculation time is prolonged, and therefore there is a possibility that it takes time to decide the charging amount of the desulfurization flux 4. Therefore, it is preferable that the adjustment amount is set to a proper adjustment amount according to the specification, the actual operation, and the like of the mechanical stirring desulfurization facility 1.
After step S110, the processing in and after step S106 is performed again. When the processing in step S106 is performed after step S110, the charging amount changed in step S110 is used as the charging amount of the desulfurization flux 4 to be used for the estimation of the S concentration in the hot metal 3.
More specifically, in a series of repeated processing in step S106 to S110, the charging amount of the desulfurization flux 4 is adjusted such that the estimated S concentration in the hot metal 3 after the desulfurization treatment is in the predetermined range. Thus, the charging amount of the desulfurization flux 4 is appropriately set. When the charging amount of the desulfurization flux 4 is small, the S concentration in the hot metal 3 after the desulfurization treatment becomes larger than the target upper limit value, so that the desulfurization treatment is required to be performed again or desulfurization treatment is required to be separately performed in the subsequent processing. On the other hand, when the charging amount of the desulfurization flux 4 is large, there is no problem of the quality of the hot metal 3 but the problem is that the production cost becomes higher.
When it has been determined in step S108 that the S concentration in the hot metal 3 after the desulfurization treatment is in the predetermined range, the control terminal 13 decides the charging amount of the desulfurization flux 4 used in the estimation of the S concentration in the hot metal 3 after the desulfurization treatment in immediately preceding step S106 as the charging amount of the desulfurization flux 4 when the desulfurization flux 4 is actually charged (S112).
After step S112, the temperature measuring device 12 measures the temperature of the bath surface of the hot metal 3 (S114). In step S114, the temperature measuring device 12 measures the temperature for a predetermined region of the bath surface of the hot metal 3 and transmits the measurement result to the determination device 15. The predetermined region of the bath surface includes a plurality of points of the bath surface of the hot metal 3, may be such a region that the inclusion of the slag 5 on the bath surface of the hot metal 3 described later can be confirmed, and is preferably a region occupying 80% or more of the bath surface area of the hot metal 3.
After step S114, the determination device 15 determines the slag inclusion state from the measurement result in step S114 and determines whether the inclusion of the slag 5 has completed to the extent that the desulfurization flux 4 can be charged (S116). In step S116, the determination device 15 determines the inclusion state of the slag 5 into the hot metal 3 based on a temperature difference between the hot metal 3 and the slag 5 on the bath surface of the hot metal 3 from the measurement result of the temperature of the bath surface of the hot metal 3. The temperature of the slag 5 is lower than that of the hot metal 3, and therefore the inclusion state of the slag can be determined by verifying the temperature distribution of the bath surface of the hot metal 3.
Specifically, the temperature distribution image creation unit 150 of the determination device 15 creates the temperature distribution image in the plurality of points or in the predetermined region in the bath surface of the hot metal 3 from the measurement result in step S114. Subsequently, the slag inclusion determination unit 151 calculates the total area of each of a hot metal exposed part and a slag coated part in the created temperature distribution image. At this time, a point where the measured temperature is equal to or larger than a threshold X (°C) is defined as the hot metal exposed part and a point where the measured temperature is less than the threshold X (°C) is defined as the slag coated part, and then a total area A_m of the hot metal exposed part and a total area A_s of the slag coated part are individually calculated. For the temperature of the slag coated part, the lower limit value may be further provided considering influence of the impeller 10, the hot metal ladle 2, and the like. Then, the slag inclusion determination unit 151 determines the inclusion state based on the area ratio A_m/A_s between the hot metal exposed part and the slag coated part. At this time, for example, a threshold Y₁ may be provided for the area ratio A_m/A_s, and the threshold Y₁ may be used to determine whether the inclusion of the slag 5 has completed. Although the threshold X and the threshold Y₁ can be set as desired, the threshold Y₁ is preferably set to 4.0 or larger.
When it has been determined in step S116 that the inclusion of the slag 5 has not completed, the processing in and after step S114 is performed again.
On the other hand, when it has been determined in step S116 that the inclusion of the slag 5 has completed, the control terminal 13 controls the desulfurization flux charging device 11 and charges the desulfurization flux 4 into the hot metal 3 with the charging amount decided in step S112 (S118). As described above, when the slag 5 is not sufficiently included, there is a possibility that the charged desulfurization flux 4 is captured by the slag 5, reducing the desulfurization capacity. However, the desulfurization failure can be prevented by determining the inclusion state of the slag 5 in step S116 and charging the desulfurization flux 4 in step S118 after the slag 5 has been sufficiently included. As a method for determining the inclusion of the slag 5, a method for determining the inclusion by visual confirmation of an operator is conceivable. However, this method causes variations in time until the inclusion is determined even in the same operating condition, because the determination criteria are personal determination criteria. This raises concerns about a decrease in the desulfurization capacity due to an inclusion failure of the slag 5 and the extension of the desulfurization treatment time due to a delay in the start timing of charging the desulfurization flux 4.
After step S118, i.e., after the charging of the desulfurization flux 4 has been started, the stirring by the impeller 10 is continued and the mechanical stirring is performed for a predetermined time according to the desulfurization treatment time, so that the desulfurization treatment ends. After the desulfurization treatment has ended, the slag 5 in the hot metal ladle 2 may be removed as post-treatment to suppress the S pick-up from the slag 5 after the desulfurization treatment into the hot metal 3. The hot metal 3 refined in such a step passes through primary smelting treatment in a converter or the like and secondary smelting treatment in an LF or RH vacuum degassing device or the like to become a steel material, such as a slab, by a continuous casting method, an ingot making/blooming method, or the like.
Herein, after step S118, i.e., after the charging of the desulfurization flux 4 has been started, the determination of the slag inclusion and the adjustment of the stirring conditions illustrated in FIG. 4 may be further performed. In the flowchart illustrated in FIG. 4, the temperature measuring device 12 first measures the temperature of the bath surface of the hot metal 3 (S300). The measurement in step S300 may be performed in the same manner as in step S114.
Subsequently, the determination device 15 determines the slag inclusion state from the measurement result in step S300 and determines whether the inclusion of the slag 5 is proper (S302). The determination in step S302 can be performed in the same manner as in the determination in step S116. More specifically, the temperature distribution image is created by the temperature distribution image creation unit 150 from the measurement result in step S300 and the inclusion state of the slag 5 is determined by the slag inclusion determination unit 151 from the area ratio A_m/A_s between the hot metal exposed part and the slag coated part in the created temperature distribution image. At this time, it is preferable for the slag inclusion determination unit 151 to set a value larger than the threshold Y₁ as a threshold Y₂ of the area ratio A_m/A_s.
When it has been determined in step S302 that the inclusion of the slag 5 is improper, the control terminal 13 acquires the determination result and changes the stirring conditions (S304). At this time, as the stirring conditions, at least one of the rotational speed and the immersion depth of the impeller 10 is changed such that the stirring conditions are such that the slag 5 is further included. In step S304, the stirring conditions are changed such that the slag 5 is further included. More specifically, when the rotational speed of the impeller 10 is changed, the stirring conditions are changed such that the rotational speed increases and when the immersion depth of the impeller 10 is changed, the stirring conditions are changed such that the immersion depth decreases.
The desulfurization flux 4 charged into the hot metal 3 reacts with the S in the hot metal 3 and then floats to the bath surface of the hot metal 3 to be captured by the slag 5. Therefore, the slag amount in the hot metal ladle 2 increases with an increase in the integrated charging amount of the desulfurization flux 4. Thus, even when the stirring conditions are held constant, the inclusion state of the slag 5 into the hot metal 3 sometimes becomes improper with an increase in the integrated charging amount of the desulfurization flux 4. For the control of the rotational speed of the impeller 10, the inclusion of the slag 5 into the hot metal 3 can be accelerated by increasing the rotational speed. However, there is a risk that an excessive increase in the rotational speed of the impeller 10 leads to a facility failure due to overloading of an impeller control device or accelerated refractory wear of the impeller 10. For the control of the immersion depth of the impeller 10, it is preferable to perform control such that the tip of a vortex formed on the surface of the hot metal 3 by the mechanical stirring is positioned at a height equal to or less than the height of the upper end of the impeller 10.
On the other hand, when it has been determined in step S302 that the inclusion of the slag 5 is proper, the determination of the slag inclusion and the adjustment of the stirring conditions end. In the treatment illustrated in FIG. 4, when it has been determined that the inclusion of the slag 5 is insufficient, the stirring conditions are changed such that the desulfurization efficiency increases. This can stably enhance the desulfurization efficiency.

As described above, the present invention is described with reference to the specific embodiment, but it is not intended to limit the invention by the description. Not only the disclosed embodiment but the other embodiments of the present invention including various modifications will be apparent to those skilled in the art by reference to the description of the present invention. Therefore, the embodiment of the invention described in Claims should be construed to cover embodiments including modifications thereof described in this specification alone or in combination.
For example, although the S concentration after the desulfurization treatment is estimated as the constituent estimation of the hot metal 3 in the above-described embodiment, the present invention is not limited to such an example. For example, the S concentration in the hot metal 3 during the desulfurization treatment after a predetermined time has elapsed may be estimated.
Although the step of creating the machine learning model in step S202 is performed after the start of the desulfurization treatment in the above-described embodiment, the present invention is not limited to such an example. The step in step S202 may be performed in advance before the desulfurization treatment is started.
Further, although the charging amount of the desulfurization flux 4 is changed such that the S concentration is in the predetermined range from the estimation value of the S concentration in the hot metal 3 after the desulfurization treatment in the above-described embodiment, the present invention is not limited to such an example. For example, the adjustment may be performed such that the estimation value of the S concentration in the hot metal 3 after the desulfurization treatment is in the predetermined range by changing the desulfurization treatment time while the charging amount of the desulfurization flux 4 is held constant. In an environment where there is a margin in the operation, e.g., a case where there is a margin of time due to a trouble or a case where the production capacity is lowered, it is conceivable that the desulfurization treatment time also has a margin. In such cases, there is a possibility that the desulfurization treatment time can be adjusted, and therefore there is a possibility that the charging amount of the desulfurization flux 4 can be reduced and the production cost can be reduced by changing the desulfurization treatment time while the charging amount of the desulfurization flux 4 is held constant.

(1) The hot metal constituent estimation device 14 according to one aspect of the present invention is the constituent estimation device 14 for estimating the S concentration in the hot metal 3 in the desulfurization treatment of the hot metal 3 using the mechanical stirring desulfurization facility 1,
the hot metal constituent estimation device 14 including:
- the data input unit 140 configured to input the current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment, the hot metal temperature being the temperature of the hot metal 3 measured using the noncontact temperature measuring device 12 after the start of the stirring of the hot metal 3 and before the start of the charging of the desulfurization flux 4; and
- the model calculation unit 142 configured to estimate the S concentration in the hot metal 3 after the desulfurization treatment from the machine learning model indicating the relation among the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux 4, and the reaction rate constant in the desulfurization treatment based on the past operation data in the desulfurization treatment and the current operation data.

According to the configuration in (1) above, the S concentration in the hot metal 3 after the desulfurization treatment can be accurately estimated because the machine learning model having high accuracy can be used by the use of the hot metal temperature during the desulfurization treatment actually measured during the desulfurization treatment. Thus, the occurrence of the desulfurization failure of the hot metal 3 can be suppressed in the desulfurization treatment in the mechanical stirring desulfurization facility 1, and the desulfurization treatment can be efficiently performed.
(2) The configuration of (1) above, the data input unit 140 being configured to input at least the temperature of the hot metal 3 measured using the noncontact temperature measuring device 12 before the start of the charging of the desulfurization flux 4, the charging amount of the desulfurization flux, and the reaction rate constant K, the reaction rate constant K being the constant in a desulfurization reaction formula represented by Formula (1), in past operation as the past operation data,
the configuration of (1) above, further including: the model parameter decision unit 141 configured to create the machine learning model based on the past operation data input by the data input unit 140.
According to the configuration in (2) above, the reaction rate constant K can be accurately estimated. Further, by considering the time t₀ from the start of the stirring to the completion of the inclusion of the slag 5 into the hot metal 3, the influence of the decrease in the desulfurization efficiency due to insufficient inclusion of the slag 5 can be excluded and the S concentration in the hot metal 3 after the desulfurization treatment can be accurately estimated.
(3) The hot metal constituent estimation method according to one aspect of the present invention is the constituent estimation method of the hot metal 3 for estimating the S concentration in the hot metal 3 in the desulfurization treatment of the hot metal 3 using the mechanical stirring desulfurization facility 1,
the hot metal constituent estimation method including:

measuring the hot metal temperature during the desulfurization treatment, the hot metal temperature being the temperature of the hot metal 3 after the start of the stirring of the hot metal and before the start of the charging of the desulfurization flux 4 using the noncontact temperature measuring device 12 (S104);
inputting the current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment to be measured (S200, S106); and
estimating the S concentration in the hot metal 3 after the desulfurization treatment from the machine learning model indicating the relation among the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and the reaction rate constant in the desulfurization treatment based on the past operation data in the desulfurization treatment and the current operation data (S206, S106).

According to the configuration in (3) above, the same effects as those in (1) above can be obtained.
(4) The configuration in (3) above,

at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux 4, and the reaction rate constant K, the reaction rate constant K being the constant in the desulfurization reaction formula represented by Formula (1), in past operation being used as the past operation data, and
in the inputting the operation data, at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux 4, and the reaction rate constant K, the reaction rate constant K being the constant in the desulfurization reaction formula represented by Formula (1), in past operation being input as the past operation data,
the configuration in (3) above, further including: creating the machine learning model based on the input past operation data (S202).

According to the configuration in (4) above, the same effects as those in (2) above can be obtained.
(5) The desulfurization method of the hot metal 3 according to one aspect of the present invention is the desulfurization method of the hot metal 3 using the mechanical stirring desulfurization facility 1,
the desulfurization method including:
estimating the S concentration in the hot metal 3 after the desulfurization treatment using the constituent estimation method of the hot metal 3 according to (3) or (4) above during the desulfurization treatment.
According to the configuration in (5) above, the same effects as those in (1) above can be obtained.
(6) The configuration in (5) above, further including: after the estimating the S concentration (S206, S106), determining whether the S concentration in the hot metal 3 after the desulfurization treatment is in the predetermined range (S108), in which
when the S concentration in the hot metal 3 after the desulfurization treatment is not in the predetermined range, the charging amount of the desulfurization flux 4 is changed, the estimating the S concentration (S110, S106) is performed again, and the S concentration in the hot metal 3 is estimated again.
According to the configuration in (6) above, the occurrence of the desulfurization failure can be suppressed. Further, the excessive charging of the desulfurization flux 4 can be prevented, and therefore the cost of the desulfurization treatment can be reduced and the extension of the desulfurization treatment time can also be suppressed.
(7) The configuration in (5) or (6) above, further including:

after the estimating the S concentration (S206, S106), measuring the temperature of the bath surface of the hot metal 3 (S114);
determining the inclusion state of the slag 5 on the bath surface of the hot metal 3 (S116) from the measurement result in the measuring the temperature of the bath surface of the hot metal 3 (S114); and
charging the desulfurization flux 4 into the hot metal 3 (S118) when it is determined that the inclusion of the slag 5 has completed in the determining the inclusion state of the slag 5 (S116).

According to the configuration in (7) above, the desulfurization flux 4 can be reliably charged after the slag 5 has been included, and therefore the decrease in the desulfurization efficiency can be prevented.
(8) The configuration according to any one of (5) to (7) above, further including:

measuring the temperature of the bath surface of the hot metal 3 after the desulfurization flux 4 has been charged (S300); and
determining the inclusion state of the slag 5 on the bath surface of the hot metal 3 from the measurement result in the measuring the temperature of the bath surface of the hot metal 3 (S300) (S302), in which
when it is determined that the slag 5 is properly included in the determining the inclusion state of the slag 5 (S302), the stirring by the impeller 10 of the mechanical stirring desulfurization facility 1 is continued and
when it is determined that the slag 5 is not properly included in the determining the inclusion state of the slag 5 (S302), at least one of the rotational speed and the immersion depth of the impeller 10 is changed (S304).

According to the configuration in (8) above, the inclusion state of the slag 5 can be properly controlled even when the integrated charging amount of the desulfurization flux 4 increases, and the decrease in the desulfurization efficiency can be prevented.

EXAMPLES

Examples performed by the present inventors are described. In Examples, the hot metal 3 tapped from a blast furnace was stored in the hot metal ladle 2, and then was subjected to desiliconization and dephosphorization treatment as necessary in an actual machine with a hot metal amount per charge of about 200 tons. Herein, the S concentration in the hot metal 3 after the desiliconization and dephosphorization treatment was 0.0300 mass% or less. Subsequently, the hot metal ladle 2 storing the hot metal 3 was transported to the mechanical stirring desulfurization facility 1, in which the desulfurization treatment was applied. At this time, the operation was performed while the stirring conditions of the impeller 10 and the charging conditions of the desulfurization flux 4 were being variously changed. As the desulfurization flux 4, calcined limestone having a particle size of 0.5 mm or less was used. As a method for charging the desulfurization flux 4, a method was used which includes spraying the desulfurization flux 4 onto the surface of the hot metal 3 together with a carrier gas.
In the operation of Examples, when the desulfurization flux 4 was charged into each hot metal 3 having a constituent and a temperature shown in Table 1, the S concentration in the hot metal 3 after the desulfurization treatment was estimated using the constituent estimation device 14 according to the above-described embodiment, and the charging amount of the desulfurization flux 4 was decided such that the estimated S concentration is equal to or less than the target value. When the S concentration in the hot metal 3 after the desulfurization treatment was estimated, the S concentration in the hot metal during the desulfurization treatment was estimated based on the machine learning model indicating the relation among the temperature of the hot metal 3, the charging amount of the desulfurization flux 4, and the reaction rate constant K and the first-order reaction equation using the past operation data and the current operation data. Then, the desulfurization treatment was performed by charging the desulfurization flux 4 into the hot metal 3 with the decided charging amount.
In Comparative Examples 1 to 3, the charging amount of the desulfurization flux 4 was decided, instead of using the machine learning model and the first-order reaction equation, using a multiple regression model indicating the relation among the temperature of the hot metal 3, the charging amount of the desulfurization flux 4, and the S concentration in the hot metal 3 after the desulfurization treatment constructed based on only the past operation data. In Comparative Examples 4 to 9, when the charging amount of the desulfurization flux 4 was decided in the same manner as in the above-described embodiment, the temperature of the hot metal 3 before the desulfurization treatment was used instead of the hot metal temperature during the desulfurization treatment.
In Examples 4 to 6, 10 to 12, the processing in steps S114, S116 was performed, the inclusion state of the slag 5 into the hot metal 3 was determined, and the charging of the desulfurization flux 4 was started after the completion of the inclusion of the slag 5 into the hot metal 3. In other Examples and Comparative Examples, the charging of the desulfurization flux 4 was started before the completion of the inclusion of the slag 5 into the hot metal 3.
In Examples 7 to 12, the processing in steps S300, S302, S304 was performed, the inclusion state of the slag 5 into the hot metal 3 after the start of the charging of the desulfurization flux 4 was determined, and the stirring conditions were changed when the inclusion of the slag 5 into the hot metal 3 was improper. The changed stirring conditions are shown in the control of the stirring conditions in Table 1.

Further, the results of investigating the obtained constituent of the hot metal 3 in the hot metal ladle 2 after the desulfurization treatment are shown together in Table 1. The contact risk in Table 1 is the contact risk between the temperature measuring probe and the solid-phase slag. When the noncontact thermometer is used, the contact risk is "None". When a contact thermometer is used and the temperature measuring probe comes into contact with the solid-phase slag, making it impossible to continuously use the thermometer three times or more, the contact risk is "High". The desulfurization rate in Table 1 expresses a difference between the S concentration in the hot metal 3 before the desulfurization treatment and the S concentration in the hot metal 3 after the desulfurization treatment as the percentage with respect to the S concentration in the hot metal 3 before the desulfurization treatment.

[Table 1]

	Temperature measurement method	Temperature measurement timing	Contact risk	Method for determining charging amount of desulfurization stuff	Hot metal temperature (°C)		S concentration in hot metal (mass%)		Desulfurization rate (%)	Charging amount of desulfurization flux (kg/t)	Timing of starting charging of desulfurization flux	Control of stirring conditions
	Temperature measurement method	Temperature measurement timing	Contact risk		Before start of stirring	After start of stirring	Before desulfurization treatment	After desulfurization treatment	Desulfurization rate (%)	Charging amount of desulfurization flux (kg/t)	Timing of starting charging of desulfurization flux	Rotational speed of impeller	Immersion depth of impeller
Ex. 1	Noncontact type	After start of stirring	None	Machine learning model + First-order reaction formula	-	1385	0.0286	0.0047	83.6	7.5	Before completion of slag inclusion	Not controlled	Not controlled
Ex. 2	Noncontact type	After start of stirring	None	Machine learning model + First-order reaction formula	-	1394	0.0294	0.0061	82.7	7.2	Before completion of slaa inclusion	Not controlled	Not controlled
Ex. 3	Noncontact type	After start of stirring	None	Machine learning model + First-order reaction formula	-	1404	0.0291	0.0063	81.8	7.0	Before completion of slaa inclusion	Not controlled	Not controlled
Ex. 4	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1399	0.0282	0.0021	92.6	7.8	After completion of slag inclusion	Not controlled	Not controlled
Ex. 5	Noncontact type	After start of stirring	None	Machine learning model + First-order reaction formula	-	1385	0.0287	0.0024	91.6	7.4	After completion of slaa inclusion	Not controlled	Not controlled
Ex. 6	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1401	0.0288	0.0028	90.3	7.3	After completion of slaa inclusion	Not controlled	Not controlled
Ex. 7	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1387	0.0281	0.0018	93.6	7.7	Before completion of slaa inclusion	Controlled	Not controlled
Ex. 8	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1406	0.0284	0.0022	92.3	7.3	Before completion of slaa inclusion	Not controlled	Controlled
Ex. 9	Noncontact type	After start of stirring	None	Machine learning model + First-order reaction formula	-	1411	0.0283	0.0024	91.5	6.8	Before completion of slaa inclusion	Controlled	Controlled
Ex. 10	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1387	0.0289	0.0009	96.9	8.0	After completion of slaa inclusion	Controlled	Not controlled
Ex. 11	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1406	0.0285	0.0011	96.1	7.4	After completion of slaa inclusion	Not controlled	Controlled
Ex. 12	Noncontact tvpe	After start of stirring	None	Machine learning model + First-order reaction formula	-	1411	0.0283	0.0015	94.7	7.1	After completion of slag inclusion	Controlled	Controlled
Comp. Ex. 1	Noncontact type	After start of stirring	None	Multiple regression model	-	1388	0.0287	0.0130	54.7	6.4	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 2	Noncontact tvpe	After start of stirring	None	Multiple regression model	-	1396	0.0290	0.0136	53.1	6.2	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 3	Noncontact tvpe	After start of stirring	None	Multiple regression model	-	1403	0.292	0.0140	52.1	5.8	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 4	Noncontact tvpe	Before start of stirring	None	Machine learning model + First-order reaction formula	1341	-	0.0284	0.0061	78.5	10.5	Before completion of slag inclusion	Not controlled	Not controlled
Comp. Ex. 5	Noncontact type	Before start of stirring	None	Machine learning model + First-order reaction formula	1343	-	0.0286	0.0063	78.0	10.2	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 6	Noncontact tvpe	Before start of stirring	None	Machine learning model + First-order reaction formula	1350	-	0.0291	0.0066	77.3	99	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 7	Contact type	Before start of stirring	High	Machine learning model + First-order reaction formula	1396	-	0.0288	0.0055	80.9	7.9	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 8	Contact type	Before start of stirring	High	Machine learning model + First-order reaction formula	1405	-	0.0292	0.0059	79.8	7.6	Before completion of slaa inclusion	Not controlled	Not controlled
Comp. Ex. 9	Contact type	Before start of stirring	High	Machine learning model + First-order reaction formula	1408	-	0.0293	0.0062	78.8	7.1	Before completion of slaa inclusion	Not controlled	Not controlled

It was able to be confirmed from the results shown in Table 1 that the desulfurization rate was 80% or more under the conditions (Examples 1 to 3) in which the S concentration in the hot metal 3 after the desulfurization treatment was estimated using the constituent estimation device 14 according to the above-described embodiment, and the charging amount of the desulfurization flux 4 was decided such that the estimated S concentration was equal to or less than the target value.
The conditions of Examples 4 to 6 are conditions in which the temperature during the desulfurization treatment is measured using the temperature measuring device 12, the inclusion state of the slag 5 is determined before the charging of the desulfurization flux 4, and the desulfurization flux 4 is charged after the completion of the inclusion of the slag 5 into the hot metal 3. It was able to be confirmed that the desulfurization effect is further improved as compared with that of Examples 1 to 3 and the desulfurization rate is 90% or more under the conditions.
The conditions of Example 7 to 9 are conditions in which the temperature during the desulfurization treatment is measured using the temperature measuring device 12, the inclusion state of the slag 5 is determined after the start of the charging of the desulfurization flux 4, and the stirring conditions are changed when the inclusion of the slag 5 into the hot metal 3 is improper. It was able to be confirmed that the desulfurization effect is further improved as compared with that of Examples 1 to 3 and the desulfurization rate is 90% or more also under the conditions.
The conditions of Examples 10 to 12 are conditions in which the temperature during the desulfurization treatment is measured using the temperature measuring device 12, the inclusion state of the slag 5 is determined before the charging of the desulfurization flux 4, and the inclusion state of the slag 5 is determined after the start of the charging of the desulfurization flux 4. Under the conditions, the desulfurization flux 4 is charged after the completion of the inclusion of the slag 5 into the hot metal 3, and the stirring conditions are changed when the inclusion of the slag 5 into the hot metal 3 is improper. It was able to be confirmed that the desulfurization effect is further improved as compared with that of Examples 1 to 3, the desulfurization rate is 90% or more, and the desulfurization rate is much higher than that of Examples 4 to 9 also under the conditions.
On the other hand, it was able to be confirmed that, under the conditions in which the charging amount of the desulfurization flux 4 was decided using the multiple regression model constructed based on only the past operation data (Comparative Examples 1 to 3), the charging amount of the desulfurization flux 4 was insufficient, and therefore the desulfurization rate is less than 60%.
It was able to be confirmed that, when the temperature of the hot metal 3 is measured using the noncontact temperature measuring device 12 before the start of the stirring (Comparative Example 4 to Comparative Example 6), the desulfurization rate is equivalent to the results shown in Examples 1 to 3 in Table 1, but the charging amount of the desulfurization flux 4 is excessive. This is because the temperature measuring device 12 actually measures the temperature of the slag 5 present on the bath surface of the hot metal 3, and the temperature measurement result of the hot metal 3 is lower than the actual temperature.
Further, it was able to be confirmed that, when the temperature of the hot metal 3 is measured using a contact thermometer before the start of the stirring (Comparative Examples 7 to 9), the desulfurization rate and the charging amount of the desulfurization flux 4 are equivalent to the results shown in Example 1 to 3, but the contact risk between the temperature measuring probe and the solid-phase slag is high. This shows that the continuous use of the contact thermometer is difficult.

Reference Signs List

1: mechanical stirring desulfurization facility
10: impeller
11: desulfurization flux charging device
12: temperature measuring device
13: control terminal
14: constituent estimation device
140: data input unit
141: model parameter decision unit
142: model calculation unit
15: determination device
150: temperature distribution image creation unit
151: slag inclusion determination unit
152: S concentration determination unit
2: hot metal ladle
3: hot metal
4: desulfurization flux
5: slag

Claims

A hot metal constituent estimation device for estimating an S concentration in hot metal in desulfurization treatment of the hot metal using a mechanical stirring desulfurization facility,
the hot metal constituent estimation device comprising:
a data input unit configured to input current operation data in the desulfurization treatment containing at least a hot metal temperature during the desulfurization treatment, the hot metal temperature being a temperature of the hot metal measured using a noncontact temperature measuring device after a start of stirring of the hot metal and before a start of charging of a desulfurization flux; and

a model calculation unit configured to estimate the S concentration in the hot metal after the desulfurization treatment from a machine learning model indicating a relation among the hot metal temperature during the desulfurization treatment, a charging amount of the desulfurization flux, and a reaction rate constant in the desulfurization treatment based on past operation data in the desulfurization treatment and the current operation data.
The hot metal constituent estimation device according to claim 1,
the data input unit being configured to input at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and the reaction rate constant, the reaction rate constant being a constant in a desulfurization reaction formula represented by Formula (1), in past operation as the past operation data,

the hot metal constituent estimation device further comprising:
a model parameter decision unit configured to create the machine learning model based on the past operation data input by the data input unit, $[S] = {[S]}_{0} \cdot \exp \{K (t - t_{0})\}$

wherein [S] denotes the S concentration (mass%) in the hot metal, [S]₀ denotes the S concentration (mass%) in the hot metal before a start of the desulfurization treatment, K denotes the reaction rate constant (s^-1), t denotes desulfurization treatment time (s), and t₀ denotes time (s) from the start of the stirring to completion of inclusion of a slag into the hot metal.
A hot metal constituent estimation method for estimating an S concentration in hot metal in desulfurization treatment of the hot metal using a mechanical stirring desulfurization facility,
the hot metal constituent estimation method comprising:
measuring a hot metal temperature during the desulfurization treatment, the hot metal temperature being a temperature of the hot metal after a start of stirring of the hot metal and before a start of charging of a desulfurization flux using a noncontact temperature measuring device;

inputting current operation data in the desulfurization treatment containing at least the hot metal temperature during the desulfurization treatment to be measured; and

estimating the S concentration in the hot metal after the desulfurization treatment from a machine learning model indicating a relation among the hot metal temperature during the desulfurization treatment, a charging amount of the desulfurization flux, and a reaction rate constant in the desulfurization treatment based on past operation data in the desulfurization treatment and the current operation data.
The hot metal constituent estimation method according to claim 3,
at least the hot metal temperature during the desulfurization treatment, the charging amount of the desulfurization flux, and the reaction rate constant, the reaction rate constant being a constant in a desulfurization reaction formula represented by Formula (1), in past operation being used as the past operation data,

the hot metal constituent estimation method further comprising:
creating the machine learning model based on the input past operation data, $[S] = {[S]}_{0} \cdot \exp \{K (t - t_{0})\}$

wherein [S] denotes the S concentration (mass%) in the hot metal, [S]₀ denotes the S concentration (mass%) in the hot metal before a start of the desulfurization treatment, K denotes the reaction rate constant (s^-1), t denotes desulfurization treatment time (s), and t₀ denotes time (s) from the start of the stirring to completion of inclusion of a slag into the hot metal.
A hot metal desulfurization method using a mechanical stirring desulfurization facility comprising:
estimating an S concentration in the hot metal after desulfurization treatment using the hot metal constituent estimation method according to claim 3 or 4 during the desulfurization treatment.
The hot metal desulfurization method according to claim 5, further comprising:
after the estimating the S concentration, determining whether the S concentration in the hot metal after the desulfurization treatment is in a predetermined range, wherein

when the S concentration in the hot metal after the desulfurization treatment is not in the predetermined range, a charging amount of the desulfurization flux is changed, the estimating the S concentration is performed again, and the S concentration in the hot metal is estimated again.
The hot metal desulfurization method according to claim 5 or 6, further comprising:
after the estimating the S concentration, measuring a temperature of a bath surface of the hot metal;

determining an inclusion state of a slag on the bath surface of the hot metal from a measurement result in the measuring the temperature of the bath surface of the hot metal; and

charging a desulfurization flux into the hot metal when it is determined that the inclusion of the slag has completed in the determining the inclusion state of the slag.
The hot metal desulfurization method according to any one of claims 5 to 7, further comprising:
measuring the temperature of the bath surface of the hot metal after the desulfurization flux has been charged; and

determining the inclusion state of the slag on the bath surface of the hot metal from the measurement result in the measuring the temperature of the bath surface of the hot metal, wherein

when it is determined that the slag is properly included in the determining the inclusion state of the slag, stirring by an impeller of the mechanical stirring desulfurization facility is continued and

when it is determined that the slag is not properly included in the determining the inclusion state of the slag, at least one of a rotational speed and an immersion depth of the impeller is changed.