JP7794166B2 - Sintered ore yield prediction method, sintered ore manufacturing equipment control method, sintered ore manufacturing method, yield prediction model generation method, and sintered ore yield prediction device - Google Patents

Description

The present invention relates to a method for predicting the yield of sintered ore, a method for controlling sintered ore manufacturing equipment, a method for manufacturing sintered ore, a method for creating a yield prediction model, and a device for predicting the yield of sintered ore.

Dwight-Lloyd sintering machines are used in the iron ore sintering process. In these machines, powdered iron ore is mixed with water, limestone, and carbonaceous material (solid fuel) such as coke breeze or anthracite, and the sinter raw material is granulated in a granulator. In a Dwight-Lloyd sintering machine, the sinter raw material is loaded onto a pallet via a transport conveyor, a feed hopper, and a chute. In a Dwight-Lloyd sintering machine, for example, sinter raw material with a thickness of 400 to 800 mm is loaded onto the pallet, and the loaded sinter raw material becomes the loading layer for the sintering machine.

The sintering machine is equipped with an ignition furnace equipped with combustion burners (ignition furnace burners) above the pallets, and a wind box (wind box) for air intake is installed below the pallets. When the carbonaceous material on the surface of the charging layer is ignited by the combustion burners in the ignition furnace, the carbonaceous material in the sintering raw material burns along the airflow from above to below created by the air intake by the wind box. The sintering reaction caused by this combustion gradually moves from the upper to lower layers of the charging layer as the pallets move forward.

In the ignition furnace, multiple ignition furnace burners are arranged across the width of the pallet. The ignition furnace burners spray flames at roughly the same position across the width of the sintering bed. The flames sprayed from the ignition furnace burners ignite (ignite) the carbonaceous material on the surface of the sintering bed. The limestone and some of the iron ore contained in the sintering bed then melt to form a molten liquid, which bonds the raw materials together to produce a sintered cake.

During the sintering process, it is necessary to properly control the maximum temperature and the high-temperature holding time during the sintering reaction in the sintering bed. This results in the formation of calcium ferrite, which has high strength and relatively high reducibility, in the sintered ore. For example, if the ignition temperature of the sintering bed is too low, the molten liquid will not be sufficiently generated in the sintered ore, weakening the bonds between the raw materials and reducing the strength of the sintered ore. In particular, the upper layer of the sintered ore is prone to cooling after ignition due to the airflow from the wind box. This cooling shortens the high-temperature holding time and reduces the sintered ore yield. On the other hand, if the ignition temperature of the sintered ore is too high, the calcium ferrite in the sintered ore may decompose into amorphous silicate and secondary hematite.

Amorphous silicates reduce the strength and reducibility of sintered ore, and secondary hematite makes it more susceptible to reduction and disintegration. Sintered ore with such low strength is finely divided when the sinter cake is crushed, and is re-sintered as return ore without being fed into the blast furnace. In other words, if the ignition temperature of the sintering bed is too high, the strength of the sintered ore will decrease, which will result in a decrease in sintered ore yield and a decrease in sintered ore production efficiency, which is a problem.

If such a decrease in sintered ore yield could be predicted in advance, it would be possible to prevent it by taking measures in advance. Patent Document 1 discloses a technique for predicting sintered ore yield by calculating a relational equation that expresses the correlation between the calorific value of the sintering raw materials fed into a sintering machine and the sintered ore yield, and then applying the measured value of the calorific value to this relational equation. In Patent Document 1, the measured value of the calorific value is determined as a function of the calorific value of limestone decomposition, the sensible heat of sintering, and the sensible heat of the exhaust gas, and the sensible heat of the exhaust gas is calculated using the exhaust gas temperature.

Patent Document 2 discloses a method for predicting the return ore generation ratio, which indicates the yield of sintered ore, by accumulating operational condition data required for the sintering process, including the return ore generation ratio for each past case, and predicting the return ore generation ratio based on the operational condition data for each case. Patent Document 2 uses the M gas flow rate, pallet speed, sintering completion point, lime ratio, coke ratio, raw material components, and average particle size as operational condition data required for the sintering process. Here, the sintering completion point is an indicator that represents the position at which sintering is completed when the charging layer on the pallet is ignited in the ignition furnace and sintering progresses from the surface downward as the pallet moves forward.

Patent Document 3 discloses a technology for preventing a decrease in the cold strength of sintered ore and improving sinter yield. According to Patent Document 3, a decrease in the cold strength of sintered ore can be prevented by adjusting the fuel flow rate of the ignition furnace burner so that the surface temperature near the pallet side plates measured at the outlet of the ignition furnace is higher than the temperature at the center of the pallet.

Japanese Patent Application Publication No. 8-13047 JP 2010-7992 A Japanese Patent Application Publication No. 2-153027

However, the above-mentioned conventional technology has the following problems. In Patent Document 1, the sinter ore yield is predicted using the calorific value. Because the calorific value is calculated using physical quantities of the sintering raw materials, such as the specific heat of limestone decomposition and the specific heat of sintering, the impact of fluctuations in operating conditions during the sintering process is not taken into account. Furthermore, although the exhaust gas temperature is used to calculate the exhaust gas sensible heat, this exhaust gas temperature is not an indicator that directly reflects the temperature history of the sintering bed during the sintering reaction. In other words, there is a problem in that it is not possible to distinguish between cases where the temperature of the sintering bed transported by the pallet cools quickly from a high temperature state and cases where a relatively low temperature state is maintained for a long period of time. Therefore, the technology disclosed in Patent Document 1 has the problem of not being able to predict the sinter ore yield with high accuracy.

In Patent Document 2, the return ore generation ratio is predicted using operational condition data required for the sintering process. However, while the operational condition data includes information on the composition of the sintering raw materials as well as information on the operational conditions of the ignition furnace, such as M gas flow rate, pallet speed, and the sintering completion point, these are not indicators that directly reflect the temperature history of the sintering bed during the sintering reaction, and therefore there is a problem with low accuracy in predicting the sinter ore yield. For this reason, the technology disclosed in Patent Document 2 has the problem of not being able to predict the sinter ore yield with high accuracy.

Patent Document 3 describes a technology for improving sinter yield by adjusting the fuel flow rate of the ignition furnace burner so that the surface temperature near the pallet side plates measured at the outlet of the ignition furnace is higher than the temperature at the center of the pallet. However, it is not a technology for predicting sinter yield. Furthermore, within the ignition furnace, sintering begins at least on the surface of the sintered ore immediately after the ignition furnace burner ignites the surface of the sintered ore bed. Therefore, since the temperature of the sintered ore bed surface has already dropped downstream of the ignition furnace, measuring the temperature of the sintered ore bed using a radiation thermometer poses the problem of not being able to obtain information regarding the maximum temperature on the sintered ore bed surface and the duration of time it is held at a high temperature.

The present invention has been made to solve the above-mentioned problems, and its purpose is to provide a sintered ore yield prediction method, a sintered ore yield prediction device, and a method for creating a yield prediction model that can accurately predict the yield of sintered ore in sintered ore production equipment that produces sintered ore using an ignition furnace that combusts carbonaceous material in a sintering bed. Another purpose of the present invention is to provide a sintered ore production equipment control method and a sintered ore production method that can improve the yield of sintered ore.

The means for solving the above problems are as follows.
[1] A method for predicting the yield of sintered ore in a sintering ore production facility that charges sintering raw materials including carbonaceous material onto an endless moving pallet to form a charging layer, and then produces sintered ore using an ignition furnace that combusts the carbonaceous material in the charging layer, wherein input data including the surface temperature of the charging layer inside the ignition furnace cover, measured from the outside of the ignition furnace cover through a measurement window of the ignition furnace cover with a radiation thermometer that measures a wavelength selected from a range of 0.5 μm to 4.0 μm, is input into a yield prediction model, and the yield of sintered ore is output to predict the yield of sintered ore.
[2] The method for predicting the yield of sintered ore according to [1], wherein the transmittance of the measurement window at the measurement wavelength is 50% or more.
[3] The method for predicting the yield of sintered ore according to [1] or [2], wherein the pressure above the sintering bed inside the ignition furnace cover is made negative.
[4] A method for predicting the yield of sintered ore according to any one of [1] to [3], wherein the surface of the charging bed, which is ignited by a flame sprayed from an ignition furnace burner provided in the ignition furnace, is defined as an ignition part, and the surface temperature includes the surface temperature of the ignition part.
[5] A method for controlling sintered ore production equipment, comprising: specifying a surface temperature at which the sintered ore yield predicted by the sintered ore yield prediction method according to any one of [1] to [4] will be equal to or greater than a target yield value; and setting the operating conditions of the ignition furnace so that the surface temperature of the charging bed becomes the specified surface temperature.
[6] A method for producing sintered ore, which produces sintered ore using the control method for sintered ore production equipment described in [5].
[7] A method for generating a yield prediction model that predicts the yield of sintered ore in a sintering facility that charges sintering raw materials including carbonaceous material onto an endless moving pallet to form a charging bed, and then produces sintered ore using an ignition furnace that combusts the carbonaceous material in the charging bed, the method comprising: acquiring a plurality of data sets each consisting of a pair of actual values of input data including the surface temperature of the charging bed inside the ignition furnace cover measured from the outside of the ignition furnace cover through a measurement window of the ignition furnace cover with a radiation thermometer having a measurement wavelength selected from a range of 0.5 μm to 4.0 μm; and an actual value of the yield of sintered ore; and generating a yield prediction model that uses the input data as input and outputs the yield of the sintered ore through machine learning using the acquired plurality of data sets as training data.
[8] A sintered ore yield prediction device that predicts the yield of sintered ore in a sintered ore production facility that produces sintered ore using an ignition furnace that burns the carbonaceous material in the charging layer after charging sintered raw materials including carbonaceous material onto an endless moving pallet to form a charging layer, the sintered ore yield prediction device having: a data acquisition unit that acquires input data including the surface temperature of the charging layer inside the ignition furnace cover, measured from the outside of the ignition furnace cover through a measurement window of the ignition furnace cover with a radiation thermometer that measures a wavelength selected from a range of 0.5 μm to 4.0 μm; and a sintered ore yield prediction unit that inputs the input data into a yield prediction model, outputs the sintered ore yield, and predicts the sintered ore yield.
[9] The sintered ore yield prediction device according to [8], wherein the transmittance of the measurement window at the measurement wavelength is 50% or more.

The method for generating a sinter yield prediction model and the method for predicting sinter yield according to the present invention enable accurate prediction of the sinter yield in sinter production equipment. Furthermore, the method for controlling sinter production equipment and the method for producing sinter according to the present invention enable sinter to be produced with a high yield.

FIG. 1 is a schematic diagram showing an example of a sintered ore manufacturing facility for manufacturing sintered ore. FIG. 2 is a schematic side cross-sectional view of the sintering raw material charging device and the ignition furnace. FIG. 3 is a schematic cross-sectional view of the ignition furnace. FIG. 4 is a schematic diagram showing one of a plurality of burner nozzles that constitute the ignition furnace burner. FIG. 5 is a schematic diagram showing an example of an ignition furnace having a measurement window provided in the side cover. FIG. 6 is a graph showing the transmittance of light including the infrared wavelength band. FIG. 7 is a schematic diagram showing the temperature measurement positions using a radiation thermometer. FIG. 8 is a graph showing the intensity distribution of infrared rays emitted from a flame. FIG. 9 is a graph showing the transmittance of light passing through carbon dioxide gas. FIG. 10 is a graph showing the transmittance when lead selenide (PbSe) and lead sulfide (PbS) are selected as the detection element of the radiation thermometer and quartz glass is used as the measurement window 86. FIG. 11 is a graph showing the light transmittance of water vapor. FIG. 12 is a diagram schematically illustrating an example of the configuration of a sintered ore yield prediction device 110. FIG. 13 is a graph showing the relationship between the flow rate of the fuel gas supplied to the ignition furnace burner and the surface temperature of the sintering bed S. FIG. 14 is a schematic diagram showing an example of a yield prediction model generated using a neural network. FIG. 15 is a graph showing the change in surface temperature in the sintering bed S over time from ignition. FIG. 16 is a graph showing the results of measuring the temperature distribution in the width direction of the sintering bed.

Embodiments of the present invention will now be described in detail with reference to the drawings. The following embodiments are intended to illustrate preferred examples of the present invention, and are not intended to be limiting in any way.

Figure 1 is a schematic diagram showing an example of a sinter ore production facility 140 for producing sinter ore. The sinter ore production facility 140 includes a blending tank 10, a drum mixer 16, a sintering machine 30, a crusher 50, a cooler 52, a screening device 54, a process computer 100, and a sinter ore yield prediction device 110. The blending tank 10 is composed of multiple surge hoppers 12, which are storage tanks. Each of the multiple surge hoppers 12 stores raw materials for sinter ore. In this embodiment, four surge hoppers 12 are installed, and each stores, for example, an iron-containing raw material, a CaO-containing raw material, an MgO-containing raw material, and a carbonaceous material.

The iron-containing raw materials used to make sintered ore include various iron ores such as hematite and magnetite. CaO-containing raw materials include limestone and quicklime. MgO-containing raw materials include dolomite and refined nickel slag. Carbonaceous materials include solid fuels such as coke breeze and anthracite. Iron-containing raw materials may include steelworks dust generated within the steelworks. Sintered ore raw materials may also include return ore, which is fine-grained sintered ore that does not meet the specified size.

A predetermined amount of each raw material stored in the blending tank 10 is dispensed and transported to the drum mixer 16 by the blended raw material transport conveyor 14. An appropriate amount of water is added to each raw material transported to the drum mixer 16, and the raw materials are granulated into pseudo-particles with an average particle size of, for example, 3.0 to 6.0 mm.

The drum mixer 16 is an example of a granulation device that mixes and granulates the raw materials to produce sinter raw material. Multiple granulation devices may be used, and a pelletizer may be used instead of the drum mixer 16. Multiple types of granulation devices may also be used, for example, both a drum mixer and a pelletizer. The sinter raw material 18 granulated by the drum mixer 16 is transported to the sinter machine 30 by the sinter raw material transport conveyor 20.

The sintering machine 30 is, for example, a downward suction-type Dwight Lloyd sintering machine. The sintering machine 30 includes a sintering raw material charging device 32, an endless moving pallet 34, an ignition furnace 36, a gaseous fuel supply device 38, and a wind box 40. The sintering raw material charging device 32 charges the sintering raw material 18 transported by the sintering raw material transport conveyor 20 onto the pallet 34 to form a charging bed S. The endless moving pallet 34 transports the sintering raw material in one direction (hereinafter sometimes referred to as the traveling direction of the pallet 34). The ignition furnace 36 is located above the pallet 34 and ignites (lights) the surface of the charging bed S. The gaseous fuel supply device 38 supplies gaseous fuel from above the charging bed S. The wind box 40 is located below the pallet 34 and generates a downward airflow by sucking air from inside the charging bed S.

Figure 2 is a schematic side cross-sectional view of the sintering raw material charging device 32 and ignition furnace 36. The sintering raw material charging device 32 has a feed hopper 60, a drum feeder 62, a charging gate 64, and a chute 66. The sintering raw material 18 in the feed hopper 60 is cut out by the drum feeder 62, slides down the chute 66, and charged onto the pallet 34. The sintering raw material 18 charged onto the pallet 34 forms a charging layer S of sintering raw material.

The charging gate 64 is a plate-shaped member arranged along the width of the chute 66. Typically, multiple charging gates 64 are arranged along the width of the chute 66, and the opening of the charging gates 64 is adjusted by an opening adjustment device (not shown). By adjusting the opening of the charging gates 64, the charging speed and charging state of the sinter raw material 18 charged onto the pallet 34 are controlled. The thickness of the charging layer S formed by charging onto the pallet 34 is adjusted by a baffle 70 when it is transported to the ignition furnace 36.

The ignition furnace 36 is a device that ignites the surface of the sintering bed S, the thickness of which is adjusted by the baffle 70, to cause a sintering reaction from the upper layer to the lower layer of the sintering bed S. The ignition furnace 36 has an ignition furnace burner 72 and an ignition furnace cover 74.

The ignition furnace burner 72 sprays a flame toward the surface of the charging bed S being charged into the ignition furnace 36, igniting the carbonaceous material on the surface of the charging bed S. Because the wind box 40 draws air downward from inside the charging bed S, when the carbonaceous material on the surface of the charging bed S is ignited, the carbonaceous material inside the charging bed S burns sequentially downward. This combustion of the carbonaceous material forms a molten zone in the sintering raw material, which then moves downward in the charging bed S. Because the charging bed S is continuously transported by the pallet 34, the flame from the ignition furnace burner 72 continuously ignites the surface of the charging bed S transported to the position of the ignition furnace burner. Then, as the pallet 34 moves, the sintering reaction progresses from the upper layer to the lower layer of the charging bed S, sintering the sintering raw material 18 and producing a sintered cake.

Referring again to Figure 1, a wind box 40 is provided below the pallet 34. It sucks air from inside the sintering bed S downward, generating an airflow that flows from above to below in the thickness direction of the sintering bed S. The wind box 40 is connected to the main duct 42, and the exhaust fan 46 sucks the gas from the main duct 42, causing the wind box 40 to suck air from inside the sintering bed S downward, generating an airflow that flows from above to below in the thickness direction of the sintering bed S. The exhaust fan 46 is, for example, a blower or pump. A dust collector 44 is provided between the main duct 42 and the exhaust fan 46, which removes dust and other contaminants from the gas exhausted from the main duct 42. The exhaust fan 46 is connected to a chimney 48, and the gas exhausted from the main duct 42 is discharged from the chimney 48 after harmful substances have been removed.

By installing a wind box 40 below the pallets 34 that pass through the ignition furnace 36, the sintering reaction of the sintering raw materials progresses from the upper layer to the lower layer of the charging bed S, allowing for efficient production of sintered cakes. In addition, a gaseous fuel supply device 38 is installed downstream of the ignition furnace 36 in the direction of travel of the pallets 34. By supplying gaseous fuel from the gaseous fuel supply device 38 into a hood installed above the charging bed S, the upper layer of the charging bed S, which is difficult to maintain at a high temperature, can be maintained at a high temperature. However, the gaseous fuel supply device 38 does not have to be installed in the sinter ore production facility 140.

Once the sinter cake is produced downstream of the ignition furnace 36, it is crushed by a crusher 50 to produce sintered ore. The sintered ore crushed by the crusher 50 is cooled by a cooler 52. The sintered ore cooled by the cooler 52 is sieved by a screening device 54 equipped with multiple sieves into finished sintered ore 56 (hereinafter, finished sintered ore 56 may be referred to as sintered ore 56) having a predetermined particle size or larger, and return ore 58 having a particle size smaller than the predetermined particle size. The resulting finished sintered ore 56 is called sintered ore, and the yield of sintered ore is the sintered ore yield. The sintered ore 56 is charged into the blast furnace as blast furnace raw material, while the return ore 58 is transported to a blending tank 10 and used as raw material for sintered ore 56.

The process computer 100 is, for example, a general-purpose computer such as a workstation or personal computer. The process computer 100 controls the operation of each device constituting the sinter ore production equipment 140 to produce sinter ore 56 of the desired product quality. The process computer 100 also collects and stores the operating parameters of the sinter ore production equipment 140, raw material information for the sinter raw materials, sensor information measured by sensors installed in each device of the sinter ore production equipment 140, and actual sinter ore yield values. The sinter ore yield is preferably measured at intervals of 5 to 180 minutes. Measuring the sinter ore yield at intervals of less than 5 minutes is not preferable because changes in operating conditions are small and the actual yield value is unlikely to change. Measuring the sinter ore yield at intervals longer than 180 minutes is also not preferable because it is difficult to obtain accurate actual yield values in response to changes in operating conditions. It is more preferable to measure the sinter ore yield at intervals of 30 to 90 minutes.

Operational parameters of the sinter ore production equipment 140 include, for example, the blending ratio of coke powder, the amount of moisture added to the sintering raw materials, the transport speed of the pallets 34, the thickness of the charging bed S, the fuel gas flow rate, air flow rate, and air-fuel ratio (air ratio) supplied to the ignition furnace 36. Sensor information includes, for example, the surface temperature of the charging bed S in the ignition furnace 36 (described below), the amount of exhaust gas sucked into the wind box 40, the oxygen concentration of the atmospheric gas above the charging bed S in the ignition furnace 36, and the oxygen concentration of the exhaust gas sucked into the wind box 40. Raw material information for the sintering raw materials includes the amount of quicklime and silica contained in the sintering raw materials, the brand of iron ore, and the average particle size of the sintering raw materials.

The sintered ore yield prediction device 110 inputs input data, including the surface temperature of the charging layer S in the ignition furnace 36, into a yield prediction model, outputs the sintered ore yield, and predicts the sintered ore yield. The sintered ore yield prediction device 110 also identifies the surface temperature at which the predicted sintered ore yield will be equal to or greater than the target yield value, and identifies the operating parameters of the ignition furnace 36 that can achieve the identified surface temperature. The sintered ore yield prediction device 110 outputs the identified operating parameters of the ignition furnace 36 to the process computer 100, thereby setting the identified operating parameters as the operating conditions of the ignition furnace 36.

Next, the ignition furnace 36 will be described. Figure 3 is a schematic cross-sectional view of the ignition furnace 36. Figure 3(a) is a cross-sectional view taken along a plane perpendicular to the direction of travel of the pallet 34, and Figure 3(b) is a cross-sectional view taken along a plane parallel to the direction of travel of the pallet 34.

The length of the ignition furnace 36 (distance in the pallet transport direction) is, for example, 2 to 4 m. The thickness of the charging bed S is, for example, 400 to 800 mm. The width of the charging bed S is, for example, 3000 to 6000 mm. Pallet side panels 88 are provided on the side portions of the charging bed S. The pallet side panels 88 are provided to support the charging bed S from the sides and maintain the layer thickness distribution of the charging bed S.

The ignition furnace cover 74 of the ignition furnace 36 has a top cover 80 covered with castable refractory and steel shell, and a side cover 82, and is provided to cover the charging bed S. The top cover 80 covers the top of the charging bed S, and the side cover 82 covers the side of the charging bed S. Although a certain gap exists between the side cover 82 and the pallet side plate 88, the size of this gap is configured to be kept relatively small. This limits air leakage from the sides, maintaining a certain degree of airtightness inside the ignition furnace 36.

A baffle 70 is provided at the entrance of the ignition furnace 36, and is configured to maintain a narrow gap between the baffle 70 and the upper surface of the charging bed S. Similarly, the exit of the ignition furnace 36 is configured to maintain a narrow gap between the baffle 70 and the upper surface of the charging bed S. In this way, the baffle 70 keeps the gap narrow at both the charging entrance and the unloading exit, thereby maintaining a certain degree of airtightness inside the ignition furnace 36.

The ignition furnace burner 72 has multiple burner nozzles in the width direction of the charging bed S. The ignition furnace burner 72 generates combustible gas by mixing fuel gas and combustion air, and sprays a flame from the tip of the burner nozzle. The number of burner nozzles provided in the width direction of the charging bed S may be between two and eight.

The ignition furnace burner 72 is attached to the top cover 80, and the burner nozzle of the ignition furnace burner 72 is located inside the ignition furnace 36. The ignition furnace burner 72 is connected to supply pipes for fuel gas and combustion air used for burner heating. The fuel gas and combustion air are supplied from a supply device installed outside the ignition furnace 36. The fuel gas supply pipe to the ignition furnace burner 72 is equipped with a fuel gas flow control valve 90 for adjusting the fuel gas flow rate and a fuel gas flow meter 92 for measuring the fuel gas flow rate. This controls the flow rate of fuel gas combusted in the ignition furnace burner 72. Furthermore, the combustion air supply pipe to the ignition furnace burner 72 is equipped with an air flow control valve 94 for adjusting the combustion air flow rate and an air flow meter 96 for measuring the combustion air flow rate. This controls the air ratio during combustion in the ignition furnace burner 72.

Figure 4 is a schematic diagram showing one of the multiple burner nozzles 76 that make up the ignition furnace burner 72. The ignition furnace burner 72 has multiple burner nozzles 76 that emit flames from the top cover 80 toward the surface of the charging bed S along the width direction of the pallet 34. The burner nozzle 76 shown in Figure 4 injects combustion air toward the gas flow into which the fuel gas is injected. This causes the fuel gas and combustion air to mix inside the ignition furnace 36, generating combustible gas and causing a combustion reaction. For this reason, the burner nozzle 76 injects combustion air at a certain angle relative to the gas flow into which the fuel gas is injected, so that the fuel gas and combustion air are mixed.

The fuel gas for the ignition furnace burner 72 may be any of coal gas, city gas, natural gas, methane gas, ethane gas, propane gas, and shale gas. A mixed gas of two or more fuel gases selected from these may also be used as the fuel gas. The coal gas may be coke oven gas, blast furnace gas, converter gas, or electric furnace gas. These gases are by-product gases generated during the manufacturing process at steel mills, and by using these gases, these by-product gases can be reused as fuel gas.

The combustion conditions of the ignition furnace burner 72 are, for example, that the fuel gas is supplied at a flow rate of about 1200 ^Nm3 /(m·hr) per meter of width of the sintering bed S, and combustion air is supplied so that the air-to-fuel gas ratio is about 0.9. This allows the surface temperature of the sintering bed S to be raised to about 1200°C.

Referring again to Figure 3, the ignition furnace cover 74 is provided with a measurement window 86 positioned at a position where the top surface of the charging bed S can be directly viewed. A position where the top surface of the charging bed S can be directly viewed is a position where the top surface of the charging bed S inside the ignition furnace cover 74 can be visually captured from outside the ignition furnace cover 74 through the measurement window 86.

Figure 5 is a schematic diagram showing an example of an ignition furnace 36 in which a measurement window 86 is provided in the side cover 82. Figure 5(a) is a schematic side view of the ignition furnace 36. As shown in Figure 5(a), the measurement window 86 is positioned higher than the upper surface of the charging bed S in the ignition furnace 36. One or more measurement windows 86 may be provided in the ignition furnace cover 74. As shown in Figure 5, by providing multiple measurement windows 86 in the ignition furnace cover 74, the surface temperature of the charging bed S can be measured at different positions in the traveling direction of the pallet 34.

Figure 5(b) is a schematic diagram showing the field of view observed when looking directly at the top surface of the charging bed S through the measurement window 86. As shown in Figure 5(b), the measurement window 86 is preferably located in a position that allows direct viewing of the area where the flame 98 sprayed from the burner nozzle 76 collides with the top surface of the charging bed S (hereinafter, this area will be referred to as the ignition area 99). By measuring the temperature of the ignition area 99, the maximum temperature on the surface of the charging bed S can be determined.

The measurement window 86 may be installed in any position that allows direct viewing of the top surface of the charging bed S, and is not limited to the side cover 82. The measurement window 86 may also be installed in the top cover 80 of the ignition furnace 36, in a position that allows visual observation of the top surface of the charging bed S. However, if the measurement window 86 is installed in the top cover 80, the field of view of the top surface of the charging bed S may be limited by the supply piping for the ignition furnace burner 72 and fuel gas, etc. Furthermore, if part of the top cover 80 is opened, the strength of the top cover 80 decreases, making it easier for firebricks, etc., to fall off. For this reason, it is preferable to install the measurement window 86 in the side cover 82.

The measurement window 86 is preferably made of heat-resistant glass such as quartz glass or borosilicate glass. It is also preferable that the measurement window 86 be made of a material that transmits infrared rays, such as barium fluoride, calcium fluoride, zinc sulfide, zinc selenide, or germanium.

Figure 6 is a graph showing the transmittance of light including infrared wavelength bands. The horizontal axis of Figure 6 represents wavelength (μm), and the vertical axis represents transmittance (%). As shown in Figure 6, although the transmittance of light varies depending on the material selected, by selecting a wavelength of light that provides high transmittance, it is possible to receive light (radiated light) generated inside the ignition furnace cover 74 on the outside of the ignition furnace cover 74 through the measurement window 86.

The ignition furnace 36 is provided on the outside of the ignition furnace cover 74 and has a radiation thermometer 84 that measures the surface temperature of the charging bed S through a measurement window 86. The radiation thermometer 84 uses a measurement wavelength selected from the range of 0.5 μm to 4.0 μm.

The installation position of the radiation thermometer 84 will be explained with reference again to Figure 3. The radiation thermometer 84 is installed in a position where it can receive radiant light from the surface of the charging bed S inside the ignition furnace 36 through the measurement window 86. Because the measurement window 86 is installed at a position higher than the surface of the charging bed S, the radiation thermometer 84 is installed with its light-receiving portion facing downward. Meanwhile, the direction in which the light-receiving portion of the radiation thermometer 84 faces can be freely changed left or right toward the measurement window 86. Therefore, by adjusting the left-right position of the light-receiving portion of the radiation thermometer 84, the surface temperature of the charging bed S can be measured at any position relative to the traveling direction of the pallet 34, such as the direction of the entrance or exit of the ignition furnace 36, or the direction of the ignition portion 99 where the flame 98 from the ignition furnace burner 72 collides with the charging bed S.

The radiation thermometer 84 may be a spot radiation thermometer capable of measuring localized temperatures, a scanning radiation thermometer capable of scanning the field of view in one axis direction using a rotating mirror or the like, or a thermograph capable of measuring in-plane temperature distribution. When using a scanning radiation thermometer, it may be scanned in the direction of travel of the charging bed S or in the width direction.

Figure 7 is a schematic diagram showing the temperature measurement positions using the radiation thermometer 84. Figure 7(a) is a diagram showing an example of measuring the temperature at a specific position on the surface of the charging bed S using a spot radiation thermometer. In the example shown in Figure 7(a), the surface temperature of the charging bed S is measured at the center in the width direction (black circle in the figure) within the range of the ignition section 99 where the flame 98 from the ignition furnace burner 72 collides with the charging bed S. This makes it possible to determine the maximum temperature on the top surface of the charging bed S.

Figure 7(b) shows an example of measuring the surface temperature by scanning measurement points across the width of the charging bed S using a scanning radiation thermometer. In the example shown in Figure 7(b), the average temperature across the width (black line in the figure) is measured within the ignition zone 99 where the flame 98 from the ignition furnace burner 72 collides with the charging bed S. This makes it possible to grasp the surface temperature distribution across the width of the charging bed S.

Figure 7(c) shows an example in which a thermograph is used to measure the surface temperature of a certain range in the longitudinal direction, with the measurement range including the entire width of the charging bed S. In the example shown in Figure 7(c), the temperature is measured in the range including the ignition point 99 where the flame 98 from the ignition furnace burner 72 collides with the charging bed S (the range surrounded by the black line in the figure). This makes it possible to measure the surface temperature distribution of the charging bed S in the width and longitudinal directions.

The radiation thermometer's measurement wavelength is selected from the range of 0.5 μm to 4.0 μm. Using a measurement wavelength within this range reduces temperature measurement error, allowing the surface temperature of the charging bed S to be measured with high accuracy. On the other hand, if the radiation thermometer's measurement wavelength is less than 0.5 μm, the change in radiant energy is small in the temperature range of 600 to 1400°C, resulting in large temperature measurement errors. Furthermore, if the radiation thermometer's measurement wavelength exceeds 4.0 μm, it will be affected by the flame 98 from the ignition furnace burner 72, resulting in large temperature measurement errors. Furthermore, if the radiation thermometer's measurement wavelength exceeds 4.0 μm, it will be affected by atmospheric gases such as carbon dioxide generated in the ignition furnace, resulting in even larger temperature measurement errors.

Figure 8 is a graph showing the intensity distribution of infrared radiation emitted from a flame. The horizontal axis of Figure 8 represents wavelength (μm), and the vertical axis represents relative radiation intensity (%). Relative radiation intensity is the relative value of radiation intensity normalized using the maximum value of the measured radiation intensity. The flame shown in Figure 8 is a flame resulting from the combustion of coke oven gas.

Because coke oven gas contains a relatively high amount of hydrogen, a flame fueled by coke oven gas appears transparent to visible light. However, as shown in Figure 8, the wavelength of the infrared light emitted from the flame overlaps with the wavelength measured by the radiation thermometer, meaning that the flame can interfere with temperature measurement. For this reason, in the ignition furnace 36 of this embodiment, the measurement wavelength of the radiation thermometer is set to 4.0 μm or less. This prevents the flame from affecting the surface temperature of the charging bed S even when measured through the flame. As a result, temperature measurement errors by the radiation thermometer 84 can be reduced. Note that if the combustion of the ignition furnace burner 72 is in an oxygen-deficient state, the flame color may turn yellow. To avoid this, it is preferable to perform combustion of the ignition furnace burner 72 under oxygen-rich conditions, where the air ratio of the fuel gas to the theoretical air volume is 1 or greater.

Figure 9 is a graph showing the transmittance of light passing through carbon dioxide gas. The horizontal axis of Figure 9 represents wavelength (μm), and the vertical axis represents transmittance (%). As shown in Figure 9, carbon dioxide (carbon dioxide) generated by burning fuel gas in the ignition furnace burner can also cause disturbances in temperature measurement. For this reason, in the ignition furnace 36 of this embodiment, the measurement wavelength of the radiation thermometer is set to 4.0 μm or less. This makes it less likely that light transmission will be obstructed by carbon dioxide generated by the combustion of fuel gas, thereby reducing errors in temperature measurement by the radiation thermometer 84.

The radiation thermometer 84's detection element is preferably a photoelectric type. Thermoelectric detection elements such as pyroelectric elements and thermopiles have a measurement wavelength of 8 to 13 μm, which is undesirable because the measurement wavelength is affected by flames and atmospheric gases. For example, it is preferable to use one of the following photoelectric detection elements: lead selenide (PbSe), lead sulfide (PbS), indium gallium arsenide (InGaAs), and silicon (Si). Lead selenide has a measurement wavelength of 4 μm, lead sulfide has a measurement wavelength of 2 μm, indium gallium arsenide has a measurement wavelength of 1.55 μm, and silicon has a measurement wavelength of 0.9 μm. Regardless of which type is used, the radiation thermometer 84 will have a measurement wavelength in the range of 0.5 μm to 4.0 μm.

Furthermore, it is preferable that the transmittance of the measurement window 86 at the measurement wavelength of the radiation thermometer 84 be 50% or more. Figure 10 is a graph showing the transmittance when lead selenide (PbSe) and lead sulfide (PbS) are selected as the radiation thermometer detection element and quartz glass is used as the measurement window 86. Since the measurement wavelength of lead sulfide is 2.0 μm, the transmittance of quartz glass at the measurement wavelength is nearly 100%. Therefore, the 2.0 μm wavelength radiation emitted from the surface of the charging bed S in the ignition furnace 36 passes through the measurement window 86 without being attenuated by the flame or atmospheric gas within the ignition furnace. This shows that the surface temperature of the charging bed S can be measured with high accuracy using the radiation thermometer 84 installed outside the measurement window 86.

On the other hand, when lead selenide is used as the detection element of the radiation thermometer 84, the measurement wavelength is 4.0 μm, so the transmittance of quartz glass for the measurement wavelength is approximately 70%. Therefore, although the 4.0 μm wavelength radiation emitted from the surface of the charging bed S in the ignition furnace 36 reaches the measurement window 86 without being attenuated by the flame or atmospheric gas in the ignition furnace, approximately 30% of the radiation is blocked when passing through the measurement window 86. Therefore, the radiation thermometer 84, which is installed outside the measurement window 86, receives 70% of the intensity of the radiation that passes through the measurement window 86 to measure the surface temperature of the charging bed S.

Therefore, in such cases, it is preferable to correct the temperature measured by the radiation thermometer 84 using the transmittance of the measurement window 86. This allows the surface temperature of the charging layer S to be measured with high accuracy, even when a measurement window 86 with low transmittance at the measurement wavelength is used. On the other hand, as the transmittance decreases, the amount of transmittance correction increases, resulting in greater variability in the measurement values. For this reason, it is preferable that the transmittance at the measurement wavelength through the measurement window 86 be 50% or higher.

The correction method when the transmittance of the measurement wavelength through the measurement window 86 is less than 100% is as follows. The radiation thermometer 84 receives radiation from the object to be measured and converts it into temperature. The self-luminous intensity E of the loading layer to be measured is expressed by the following equation (1).

In the above formula (1), E is the spontaneous emission intensity (W/m ³ ), ε is the emissivity (-) of the charging layer to be measured, λ is the measurement wavelength (m), T is the temperature of the charging layer to be measured (K), constant _C1 is 3.741×10 ⁻¹⁶ (W·m ² ), and constant _C2 is 1.439×10 ⁻² (m·K). Note that (-) means that it is dimensionless. The emissivity ε may be set to 0.9 as a representative value for iron oxide. Alternatively, the emissivity ε may be set in advance by offline measurement. Specifically, the emissivity ε can be set so that the measurement value obtained by the radiation thermometer and the measurement value obtained by the contact thermocouple match for the charging layer to be measured.

Furthermore, when the radiated light is blocked by the measurement window 86, the light receiving intensity Er (W/m ³ ) of the light received by the radiation thermometer is expressed by the following equation (2) using the self-emission intensity E and transmittance ξ (%) in the above equation (1).

Er=ξE...(2)

When making corrections using the above equations (1) and (2), the transmittance ξ of the window material of the measurement window is determined in advance by measuring the electromagnetic waves emitted from the blackbody furnace offline using a radiation thermometer through the window material used for measurement. Then, by using the transmittance ξ and the above equations (1) and (2), the surface temperature T of the charging layer S can be calculated with high accuracy even if the transmittance of the measurement window 86 is not 100% for the selected measurement wavelength.

In addition, in the ignition furnace 36 according to this embodiment, it is preferable to create a negative pressure above the charging layer S inside the ignition furnace cover 74 by suctioning air using a wind box 40 installed below the pallet 34. Here, negative pressure means that the pressure inside the ignition furnace 36 is lower than the pressure outside.

Referring again to Figure 1, the exhaust fan 46 is equipment that sucks in gas from the main duct 42, and is, for example, a blower or pump. The exhaust fan 46 sucks in gas from the main duct 42, creating a negative pressure inside the wind box 40 connected to the main duct 42. When the wind box 40 becomes negative pressure, the wind box 40 sucks in air from inside the charging bed S, and the pressure above the charging bed S inside the ignition furnace cover 74 becomes negative. The pressure above the charging bed S may be represented by the pressure measured inside the ignition furnace cover 74. In this case, it is preferable to measure the pressure downstream of the ignition furnace 36, and it is even more preferable to measure the pressure near the center of the ignition furnace 36 in the width direction.

In this way, the wind box 40 creates a negative pressure above the charging bed S inside the ignition furnace cover 74. The negative pressure above the charging bed S inside the ignition furnace cover 74 is preferably 2 to 10 kPa lower than the pressure outside the ignition furnace 36. By creating a negative pressure above the charging bed S, water vapor generated by the combustion of fuel gas is drawn into the charging bed S from above. Furthermore, dust generated by ignition of the charging bed S is prevented from floating above the charging bed S. This allows the radiation emitted from the surface of the charging bed S to pass through the measurement window 86 without being affected by the water vapor and dust generated inside the ignition furnace 36. As a result, the surface temperature of the charging bed S can be measured with even greater accuracy using the radiation thermometer 84.

Specifically, if the fuel gas used in the ignition furnace burner 72 contains hydrogen ( _H2 ) or methane ( _CH4 ), water vapor will be generated by the combustion of this fuel gas. Figure 11 is a graph showing the light transmittance of water vapor. The horizontal axis of Figure 11 represents wavelength (μm), and the vertical axis represents transmittance (%).

As shown in Figure 11, water vapor has low transmittance for part of the radiation light within the range of 0.5 μm to 4.0 μm (the 3-4 μm range), which is the wavelength measured by the radiation thermometer 84. Therefore, the water vapor within the ignition furnace 36 attenuates the radiation light emitted from the surface of the charging bed S, reducing the light intensity received by the radiation thermometer 84 installed outside the measurement window 86. Therefore, by drawing the water vapor within the ignition furnace 36 into the interior of the charging bed S, the attenuation of radiation light due to the water vapor is suppressed, and it can be seen that the surface temperature of the charging bed S can be measured with high accuracy using the radiation thermometer 84 installed outside the measurement window 86.

Next, the sintered ore yield prediction device 110 will be described. As described above, the sintered ore yield prediction device 110 is a device that inputs input data, including the surface temperature of the charging layer S in the ignition furnace 36, into a yield prediction model, outputs the sintered ore yield, and predicts the sintered ore yield. Furthermore, the sintered ore yield prediction device 110 is also a device that identifies the surface temperature at which the predicted sintered ore yield will be equal to or greater than the target yield value, and identifies the operating parameters of the ignition furnace 36 that can achieve the identified surface temperature.

Figure 12 is a schematic diagram showing an example configuration of a sintered ore yield prediction device 110. The sintered ore yield prediction device 110 is, for example, a general-purpose computer such as a workstation or personal computer. The sintered ore yield prediction device 110 has a control unit 112, an input unit 114, an output unit 116, and a storage unit 118. The control unit 112 is, for example, a CPU, and executes various programs stored in the storage unit 118, causing the control unit 112 to function as a data acquisition unit 120, a yield prediction unit 122, an ignition furnace operating parameter identification unit 124, and a yield prediction model generation unit 126.

The input unit 114 is, for example, a keyboard, a touch panel integrated with a display, or the like. The output unit 116 is, for example, an LCD or CRT display. The storage unit 118 is, for example, an updatable flash memory, a built-in hard disk or a memory card connected via a data communication terminal, or an information recording medium and its read/write device. The storage unit 118 stores programs and data for realizing the production of sintered ore 56 by the sintered ore production equipment 140. The storage unit 118 also stores a database 128 and a yield prediction model 130. The database 128 stores 30 or more, preferably 100 or more, and more preferably 500 or more data sets, each set consisting of actual values for the surface temperature of the charging bed S and actual values for the yield of sintered ore previously produced by the sintered ore production equipment 140. If necessary, the database 128 may acquire from the process computer 100 the actual values of the operating parameters of the ignition furnace 36 (described below), the actual values of the operating parameters of the sintering process, and the actual values of the attribute information of the sintering bed S, and store a data set that associates these with the actual values of the surface temperature of the sintered ore sintering bed S and the actual values of the yield.

Next, the sinter yield prediction process performed by the sinter yield prediction device 110 will be described. The data acquisition unit 120 acquires the surface temperature of the sintering bed S from the process computer 100 as input data. The data acquisition unit 120 acquires the same type of surface temperature of the sintering bed S as the input data for the yield prediction model 130, so that the data can be input into the yield prediction model 130 generated by the yield prediction model generation unit 126 (described below). Therefore, the data acquisition unit 120 may acquire the surface temperature of the sintering bed S at a specific position as the surface temperature of the sintering bed S, or may acquire the average surface temperature of the sintering bed S. For example, the surface temperature of the sintering bed S at the position (ignition point) where the flame emitted from the ignition furnace burner 43 in the ignition furnace 36 impinges on the sintering bed S, or at a representative position set downstream of the ignition point, or the average temperature in the width direction of the sintering bed S may be used. The data acquisition unit 120 outputs the acquired input data to the yield prediction unit 122.

When the yield prediction unit 122 acquires input data from the data acquisition unit 120, it reads the yield prediction model 130 stored in the storage unit 118. The yield prediction unit 122 predicts the sintered ore yield by inputting the input data into the yield prediction model and outputting the sintered ore yield. The yield prediction unit 122 may also display the predicted sintered ore yield on the output unit 116. This allows the operator to confirm the predicted sintered ore yield by visually checking the output unit 116.

In this way, the sintered ore yield prediction device 110 of this embodiment predicts the sintered ore yield by inputting the surface temperature of the charging layer S in the ignition furnace 36 into the yield prediction model 130 and outputting the sintered ore yield. The surface temperature of the charging layer S is measured from the outside of the ignition furnace cover 74 through the measurement window 86 using a radiation thermometer 84 with a measurement wavelength selected from the range of 0.5 μm to 4.0 μm. This surface temperature is measured with high accuracy, with the effects of the flame 98 and carbon dioxide suppressed. Therefore, by predicting the sintered ore yield using input data including this surface temperature, the sintered ore yield can also be predicted with high accuracy.

Next, the operational parameter setting process for the ignition furnace 36 performed by the sinter ore yield prediction device 110 will be described. The yield prediction unit 122 outputs the predicted sinter ore yield to the ignition furnace operational parameter specification unit 124. The ignition furnace operational parameter specification unit 124 determines whether the predicted sinter ore yield value obtained from the yield prediction unit 122 is equal to or greater than the target yield value. The target yield value may be input by the operator via the input unit 114, or may be input in advance by the operator and stored in the storage unit 118.

If the acquired predicted value of the sintered ore yield is equal to or greater than the target yield, the ignition furnace operational parameter identification unit 124 identifies the surface temperature acquired by the data acquisition unit 120 as the surface temperature that can make the yield of the sintered ore bed S equal to or greater than the target yield. In this case, the ignition furnace operational parameter identification unit 124 does not change the ignition furnace operational parameters, such as the fuel gas flow rate and air ratio, so that the current surface temperature is maintained.

On the other hand, if the acquired predicted value of the sintered ore yield is less than the target yield value, the ignition furnace operation parameter identification unit 124 changes the surface temperature acquired by the data acquisition unit 120 and outputs the changed surface temperature to the yield prediction unit 122. The yield prediction unit 122 inputs the changed surface temperature back into the yield prediction model 130, outputs the sintered ore yield, and predicts the sintered ore yield. The yield prediction unit 122 outputs the predicted value of the sintered ore yield to the ignition furnace operation parameter identification unit 124. The ignition furnace operation parameter identification unit 124 determines whether the acquired predicted value of the sintered ore yield is equal to or greater than the target yield value. The yield prediction unit 122 and the ignition furnace operation parameter identification unit 124 repeat this process until the predicted value of the sintered ore yield is equal to or greater than the target yield value.

If the sintered ore yield predicted at the changed surface temperature is equal to or greater than the target yield value, the ignition furnace operational parameter identification unit 124 identifies the changed surface temperature as the surface temperature that will make the sintered ore yield equal to or greater than the target yield value.

13 is a graph showing the relationship between the flow rate of the fuel gas (C gas) supplied to the ignition furnace burner 72 and the surface temperature of the sintering bed S. In the example shown in FIG. 13, coke oven gas (C gas) was used as the fuel gas for the ignition furnace burner 72. The fuel gas flow rate on the horizontal axis represents the flow rate of the fuel gas supplied per unit width of the sintering bed S (Nm ³ /(h·m)). The radiation thermometer 84 used to measure the surface temperature of the sintering bed S is a thermograph capable of measuring the surface temperature distribution of the sintering bed S. The measuring element is InGaAs, the measuring wavelength is 1.55 μm, and the measuring window 86 is made of quartz glass. Air was sucked in from the wind box 40 to control the pressure inside the sintering bed 36 to be 5 kPa lower than the outside pressure.

Figure 13(a) is a graph showing an example of measuring the spot temperature at the widthwise center of the charging bed S as the surface temperature of the charging bed S. The spot temperature at the widthwise center is a temperature measured using surface temperature data from a specific position 100 mm away from the ignition unit 99 in the ignition furnace 36 in the conveying direction of the pallet 34.

Figure 13(b) is a graph showing an example in which the average temperature in the width direction of the charging bed S is used as the surface temperature of the charging bed S. The average temperature in the width direction of the charging bed S was calculated as an average value from the surface temperature distribution data at a position 100 mm away from the ignition section 99 in the ignition furnace 36 in the conveying direction of the pallet 34.

As shown in Figures 13(a) and (b), increasing the fuel gas flow rate also increases the surface temperature of the sintering bed S. The storage unit 118 stores the correspondence between the fuel gas flow rate and the surface temperature of the sintering bed S, as shown in Figures 13(a) and (b). The ignition furnace operation parameter identification unit 124 reads this correspondence from the storage unit 108 and, by using this correspondence and the surface temperature that can achieve the target yield value or higher, can identify the fuel gas flow rate, which is an operation parameter of the ignition furnace 36. The ignition furnace operation parameter identification unit 124 outputs the identified operation parameters of the ignition furnace 36 to the process computer 100. The process computer 100 sets the operation parameters acquired from the ignition furnace operation parameter identification unit 124 as the operation conditions of the ignition furnace 36 and adjusts the fuel gas flow control valve 90 to satisfy these conditions. In this way, the sintered ore yield prediction device 110 according to this embodiment can identify the operational parameters of the ignition furnace that will enable the sintered ore yield to be equal to or greater than the target yield, and set the operational conditions of the ignition furnace 36 to the identified operational parameters. By changing the operational conditions of the ignition furnace 36 in this way, it becomes possible to produce sintered ore 56 at a yield equal to or greater than the target yield in the sintered ore production facility 140.

Next, we will explain how to generate the yield prediction model 130 used to predict the sintered ore yield. The data acquisition unit 120 acquires the surface temperature of the charging bed S and the actual values of the sintered ore yield from the process computer 100, and stores these as a set of data in the database 128 of the storage unit 118. The number of data sets stored in the database 128 is at least 30 or more, preferably 100 or more, and even more preferably 500 or more.

The yield prediction model generation unit 126 reads a pre-stored machine learning model from the storage unit 118, and trains the machine learning model using multiple data sets stored in the database 128 as training data to generate a trained machine learning model. This trained machine learning model becomes the yield prediction model 130. Note that the machine learning model used in the sinter ore yield prediction method and sinter ore yield prediction device 110 according to this embodiment may be, for example, a neural network, decision tree learning, random forest, or support vector regression. Furthermore, an ensemble model combining multiple machine learning models may also be used as the machine learning model. Furthermore, a classification model such as k-nearest neighbor or logistic regression may also be used as the machine learning model.

Figure 14 is a schematic diagram showing an example of a yield prediction model 130 generated using a neural network. The yield prediction model 130 shown in Figure 14 is a yield prediction model 130 that takes the surface temperature of the sintering bed S and the operational parameters of multiple ignition furnaces described below as input data, and the sintered ore yield as output data. L1, L2, and L3 represent the input layer, intermediate layer, and output layer, respectively. The yield prediction model 130 shown in Figure 13 has two intermediate layers, three nodes, and uses a sigmoid function as the activation function. However, if the operational parameters of the ignition furnace are not used as input data for the yield prediction model 130, a neural network is constructed that takes only the surface temperature of the sintering bed S as input data.

By using such a machine learning model, it is possible to freely select input data, such as ignition furnace operating parameters, that are correlated with sintered ore yield without considering the problem of multicollinearity. By including ignition furnace operating parameters that are correlated with sintered ore yield in the input data, it is possible to improve the accuracy of sintered ore yield predictions by the yield prediction model 130.

The yield prediction model generation unit 126 may also improve the accuracy of sinter ore yield predictions by dividing the data set stored in the database 128 into training data and test data and performing machine learning. For example, the yield prediction model generation unit 126 may perform machine learning of the weight coefficients of the neural network using the training data, and generate the yield prediction model 130 while appropriately changing the structure of the neural network (number of intermediate layers and number of nodes) so as to increase the accuracy rate of sinter ore yield for the test data. The error propagation method may be used to update the weight coefficients.

The yield prediction model 130 may be updated to a new model through re-learning, for example, every six months or every year. This is because the more data stored in the database 128, the more accurately a yield prediction model can be generated that can predict yield. By generating and updating the yield prediction model using the latest data, it is possible to generate a yield prediction model that reflects changes over time in the sinter ore production equipment 140.

The data acquisition unit 120 may acquire the surface temperature of the charging bed S at a specific position as the surface temperature of the charging bed S, or may acquire multiple surface temperatures of the charging bed S measured at multiple positions. When the data acquisition unit 120 acquires actual data measured at multiple positions as the surface temperature, the number of nodes can be adjusted so that the actual data measured at those positions is input as the surface temperature of the charging bed S to the input layer L1 of the neural network. Furthermore, when the data acquisition unit 120 acquires a temperature chart created by scanning the charging bed S in the width direction or length direction using a radiation thermometer 84 as the surface temperature, or when it acquires two-dimensional image data of the temperature distribution on the surface of the charging bed using a thermograph capable of measuring in-plane temperature distribution, a sinter yield prediction model may be generated using a convolutional neural network that uses these as inputs.

The data acquisition unit 120 may also acquire the heat retention index of the charging bed S as the surface temperature of the charging bed S. The heat retention index is the value obtained by integrating the temperature at or above a predetermined lower limit temperature over the retention time. This heat retention index is also an index representing the surface temperature of the charging bed S.

Figure 15 is a graph showing the change in surface temperature of the charging bed S over time from ignition. The heat retention index is the area Sk where the surface temperature of the charging bed S is maintained at 1000°C or above in the graph showing the change in surface temperature over time from ignition shown in Figure 15, when the lower limit temperature is 1000°C. The change in surface temperature of the charging bed S over time is determined by extracting longitudinal temperature data at the center of the width of the charging bed S from surface temperature distribution data measured by thermography downstream of the ignition unit 99 in the ignition furnace 36 in the conveying direction of the pallet 34. The change in surface temperature over time is determined from this temperature data and the conveying speed of the pallet 34, and the retention time for maintaining the temperature at 1000°C or above can be calculated from this change in surface temperature over time.

Figure 16 is a graph showing the relationship between the flow rate of fuel gas (C gas) supplied to the ignition furnace burner 72 and the heat retention index of the charging bed S. As shown in Figure 16, increasing the flow rate of fuel gas also increases the heat retention index of the charging bed S. Therefore, if the ignition furnace operation parameter identification unit 124 identifies a heat retention index that can achieve the target yield or higher using a method similar to that described above, the fuel gas flow rate, which is an operation parameter of the ignition furnace 36 that can achieve the target yield or higher, can be identified by using the correspondence shown in Figure 16.

In addition to the surface temperature of the charging bed S, the data acquisition unit 120 may also acquire, as input data, the operating parameters of the ignition furnace 36, the operating parameters of the sintering process, and attribute information of the charging bed S. This information is collected and stored by the process computer 100.

The operating parameters of the ignition furnace 36 are the set values of at least one of the fuel gas flow rate, air volume, and air ratio of the ignition furnace burner 72, which are among the sinter ore production conditions. These affect the flame temperature and injection speed injected from the ignition furnace burner 72, and therefore affect the ignition behavior and sintering reaction of the sintering bed S. Therefore, by including the operating parameters of the ignition furnace 36 in the input data, the yield prediction model 130 becomes a model that can predict the sinter ore yield with greater accuracy.

The operating parameters for the sintering process are at least one of the following: the conveying speed of the pallet 34, which is one of the sinter ore production conditions; the amount of exhaust gas drawn into the wind box 40, which is one of the sensor information; the oxygen concentration of the atmospheric gas above the charging layer S in the ignition furnace 36; and the oxygen concentration of the exhaust gas drawn into the wind box 40. If multiple wind boxes 40 are installed, an oxygen concentration meter may be installed in each wind box 40, and the oxygen concentration exhausted from each wind box 40 may be used.

When the conveying speed of the pallets 34 changes, the contact time between the flame and the surface of the sintering bed S changes, changing the surface temperature of the sintering bed S. Therefore, the conveying speed of the pallets 34 affects the sintering reaction. Furthermore, the amount of exhaust gas sucked into the wind box 40 can be considered the amount of gas that has passed through the interior of the sintering bed S. The temperature inside the sintering bed S changes depending on the amount of gas passing through the interior of the sintering bed S, so the amount of exhaust gas also affects the sintering reaction. Furthermore, the oxygen concentration of the atmospheric gas above the sintering bed S in the ignition furnace 36 affects the progress of the sintering reaction as the gas passes through the sintering bed S. The oxygen concentration of the exhaust gas sucked into the wind box 40 affects the progress of the sintering reaction within the sintering bed S. Therefore, by including these sintering process operational parameters in the input data, the yield prediction model 130 becomes a model that can predict sinter yield with even greater accuracy.

Attribute information of the sintering bed S refers to at least one of the following sinter ore production conditions: the thickness of the sintering bed S, the content of quicklime and silica contained in the sintering raw materials, the blending ratio of coke powder, the brand of iron ore, the moisture content of the sintering raw materials, and the average particle size of the sintering raw materials. A change in the thickness of the sintering bed S affects the sintering reaction because it changes the distance between the ignition furnace burner 72 and the surface of the sintering bed S and alters the ignition characteristics of the flame 98. Changes in the content of quicklime, silica, and coke powder contained in the sintering raw materials or the brand of iron ore change the rate and heat generation of the melting reaction in the sintering bed S, which also affect the sintering reaction in the sintering bed S. Furthermore, changes in the moisture content and average particle size of the sintering raw materials change the gas permeability passing through the sintering bed S, affecting the amount of oxygen passing through the sintering bed S, which also affect the sintering reaction in the sintering bed S. Therefore, by including attribute information of the sintering layer S in the input data, the yield prediction model 130 becomes a model that can predict the yield of sintered ore with even greater accuracy.

Furthermore, if these are included in the input data, in the process of identifying the surface temperature that will result in a sintered ore yield equal to or greater than the target yield value, instead of changing the surface temperature, it is possible to change the operational parameters of the sintering process or adjustable items in the attribute information of the sintering bed S. Then, the operational parameters used when the predicted sintered ore yield equals or exceeds the target yield value may be set as the production conditions for the sintered ore production equipment 140.

In the example shown in Figure 1, the process computer 100 and the sintered ore yield prediction device 110 are separate devices, but this is not limited to this. The sintered ore yield prediction device 110 may have the functions of the process computer 100, in which case the sintered ore production equipment 140 does not need to have the process computer 100.

Furthermore, while the sinter ore yield prediction device 110 has been described using an example in which it has an ignition furnace operation parameter identification unit 124 and a yield prediction model generation unit 126, this is not limited to this. If the yield of sinter ore is to be predicted, the sinter ore yield prediction device 110 does not need to have the ignition furnace operation parameter identification unit 124. Furthermore, if the yield prediction model 130 is generated externally, acquired via the data acquisition unit 120, and stored in the storage unit 118, the yield prediction model generation unit 126 does not need to be included. Furthermore, the yield prediction model 130 may be stored in another device that can communicate with the sinter ore yield prediction device 110, rather than in the storage unit 118 of the sinter ore yield prediction device 110.

Next, we will explain an example in which sintered ore was produced by predicting the sintered ore yield using the sintered ore yield prediction device 110 of the sintered ore production facility 140. In this example, the surface temperature of the charging bed S inside the ignition furnace 36 was measured from the outside of the ignition furnace cover 74 installed in the ignition furnace 36 shown in Figure 3 through the measurement window 86 using a radiation thermometer 84 with a lead selenide (PbSe) element measuring at a wavelength of 4.0 μm. The measurement window 86 was made of calcium fluoride, quartz glass, or germanium, with a radiation light transmittance of 92%, 70%, or 65%, respectively. The radiation thermometer 84 was a scanning radiation thermometer thermograph, and measured the in-plane temperature distribution of the charging bed S downstream in the pallet transport direction from the ignition section 99 inside the ignition furnace.

In the examples of the invention, the surface temperature of the charging bed S was calculated using the spot temperature at the widthwise center of the charging bed S, the average temperature in the widthwise direction of the charging bed S, and the heat retention index calculated from the surface temperature distribution of the charging bed S. The heat retention index was calculated by calculating the integral value of the surface temperature during the time the surface temperature of the charging bed S was maintained at or above 1000°C, with a lower limit of 1000°C.

In the examples of the invention, the yield of sintered ore was predicted not only using yield prediction models (Examples 1 to 5) that used the above-mentioned surface temperature as input data, but also using yield prediction models (Examples 6 to 9) that included as input data the surface temperature and operational parameters such as the fuel gas flow rate supplied to the ignition furnace burner 72, the conveying speed of the pallet 34, and/or the content of quicklime and coke powder in the sintering raw materials.

The yield prediction model 130 obtained multiple data sets from the process computer 100, each set consisting of the actual values of the above input data and the actual values of the sintered ore yield, and stored them in the database 128.

When the number of data sets stored in database 128 reached 3,000, 1,500 pieces of training data were extracted from database 128, and the remaining 1,500 were used as test data. The 1,500 pieces of training data were used to train a machine learning model. In the machine learning, the type of input data used in the yield prediction model was changed, and a yield prediction model corresponding to each input data was generated.

The generated yield prediction model was evaluated for its sinter yield prediction accuracy using 1,500 test data sets. The machine learning model used was a neural network with three hidden layers and five nodes in each hidden layer. The activation function used was a sigmoid function.

In Comparative Example 1, a yield prediction model was generated using 1,500 pieces of training data under the same conditions as in Example 2, except that the measurement wavelength of the radiation temperature system was set to 4.5 μm, and the sintered ore yield was evaluated using 1,500 pieces of test data. In Comparative Example 2, a yield prediction model was generated using 1,500 pieces of training data under the same conditions as in Example 2, except that the measurement window was removed and the surface temperature was measured, and the sintered ore yield was evaluated using 1,500 pieces of test data. The results of evaluating the sintered ore yield prediction error 3σ as a measure of the sintered ore yield prediction accuracy in the inventive and comparative examples are shown in Table 1 below.

As shown in Table 1, the prediction error 3σ of the sintered ore yield in Examples 1 to 9 was smaller than the prediction error 3σ of the sintered ore yield in Comparative Examples 1 and 2. These results confirmed that the yield prediction model for Examples 1 to 9 can be used to predict the sintered ore yield with high accuracy. Measuring the surface temperature of the sintered ore bed S from outside the measurement window 86 using a radiation thermometer 84 with a measurement wavelength of 4.0 μm suppresses the effects of flames and carbon dioxide, allowing the surface temperature of the sintered ore bed S to be measured with high accuracy. Therefore, using this surface temperature as input data is thought to have enabled the sintered ore yield to be predicted with high accuracy. Furthermore, in Examples 6 to 9, the input data also included operational parameters correlated with the sintered ore yield, so the prediction error 3σ of the sintered ore yield was even smaller than in Examples 1 to 5. These results confirmed that sintered ore yield can be predicted with even greater accuracy by using a yield prediction model that includes these operational parameters as input data.

On the other hand, in Comparative Example 1, although the conditions were the same as those of Inventive Example 2, the value of the yield prediction error 3σ was larger than that of Inventive Example 2. This result is thought to be due to the fact that the measurement wavelength of the radiation thermometer 84 was a long wavelength of 4.5 μm, and carbon dioxide (carbonic acid gas) generated by the combustion of the fuel gas acted as a disturbance to the temperature measurement, increasing the measurement error of the surface temperature of the sintered ore bed obtained by the radiation thermometer 84, which in turn reduced the accuracy of the sintered ore yield prediction.

In addition, although Comparative Example 2 was under the same conditions as Inventive Example 2, the value of the yield prediction error 3σ was larger than Inventive Example 2. In Comparative Example 2, the measurement window 86 was opened and the surface temperature of the sintering bed S was measured directly using the radiation thermometer 84. Therefore, when the measurement window 86 was opened, a local temperature drop occurred on part of the surface of the sintering bed S due to the inflow of outside air. This local temperature drop increased the variance in the surface temperature measured by the radiation thermometer 84, which is thought to have reduced the accuracy of the sintered ore yield prediction.

The yield prediction model generated in Example 9 was used to identify the operating parameters of the ignition furnace that would achieve a target yield or higher, and sintering was carried out. Here, the target yield was, for example, 80.0%. The average temperature in the width direction that would achieve the target yield or higher was identified, and the C gas flow rate that would achieve the identified average temperature in the width direction was identified using the correspondence relationship in Figure 12(b), and the C gas flow rate was set as the operating condition of the ignition furnace 36.

When sintered ore was produced using the sintered ore production equipment 140 and the operating conditions of the ignition furnace 36 for three days, the sintered ore yield was 81.6%. In contrast, when sintered ore was produced using the sintered ore production equipment 140 for three days without resetting the operating conditions of the ignition furnace 36, the sintered ore yield was 77.3%. These results confirm that sintered ore can be produced with a high yield by using the sintered ore production equipment control method and sintered ore production method according to this embodiment.

10 Blending tank 12 Surge hopper 14 Blended raw material transport conveyor 16 Drum mixer 18 Sintering raw material 20 Sintering raw material transport conveyor 30 Sintering machine 32 Sintering raw material charging device 34 Pallet 36 Ignition furnace 38 Gaseous fuel supply device 40 Wind box 42 Main duct 44 Dust collector 46 Exhaust fan 48 Chimney 60 Ore feeding hopper 62 Drum feeder 64 Charging gate 66 Chute 70 Baffle 72 Ignition furnace burner 74 Ignition furnace cover 76 Burner nozzle 80 Top cover 82 Side cover 84 Radiation thermometer 86 Measurement window 88 Pallet side plate 90 Fuel gas flow rate adjustment valve 92 Fuel gas flow meter 94 Air flow rate adjustment valve 96 Air flow meter 98 Flame 99 Ignition unit 100 Process computer 110 Sintered ore yield prediction device 112 Control unit 114 Input unit 116 Output unit 118 Storage unit 120 Data acquisition unit 122 Yield prediction unit 124 Ignition furnace operation parameter identification unit 126 Yield prediction model generation unit 128 Database 130 Yield prediction model

Claims (8)

A method for predicting the yield of sintered ore in a sintering ore production facility that charges sintering raw materials including carbonaceous material onto an endless moving pallet to form a charging layer, and then produces sintered ore using an ignition furnace that combusts the carbonaceous material in the charging layer, comprising:
inputting input data including the surface temperature of the charging layer inside the ignition furnace cover, which is measured from the outside of the ignition furnace cover through a measurement window of the ignition furnace cover with a radiation thermometer having a measurement wavelength selected from a range of 0.5 μm to 4.0 μm, into a yield prediction model, and outputting the yield of sintered ore to predict the yield of sintered ore;
A method for predicting a yield of sintered ore, wherein the transmittance of the measurement window at the measurement wavelength is 50% or more . The sintered ore yield prediction method according to claim 1, wherein the pressure above the sintering bed inside the ignition furnace cover is made negative. When the surface of the charging bed ignited by a flame sprayed from an ignition furnace burner provided in the ignition furnace is defined as an ignition part,
The method for predicting a yield of sintered ore according to claim 1 , wherein the surface temperature includes a surface temperature of the ignition part. When the surface of the charging bed ignited by a flame sprayed from an ignition furnace burner provided in the ignition furnace is defined as an ignition part,
The method for predicting a yield of sintered ore according to claim 2 , wherein the surface temperature includes a surface temperature of the ignition part. A surface temperature at which the yield of sintered ore predicted by the method for predicting the yield of sintered ore according to any one of claims 1 to 4 becomes equal to or greater than a target yield value is specified;
A method for controlling a sintered ore manufacturing facility, comprising: setting operating conditions of the ignition furnace so that the surface temperature of the sintering bed becomes a specified surface temperature. A method for producing sintered ore, which produces sintered ore using the control method for sintered ore production equipment described in claim 5. A method for generating a yield prediction model for predicting the yield of sintered ore in a sintering ore production facility that produces sintered ore using an ignition furnace that burns the carbonaceous material in the sintering bed after charging sintering raw materials including carbonaceous material into an endless moving pallet to form a sintering bed, comprising:
an actual value of input data including a surface temperature of the charging bed inside the ignition furnace cover measured from the outside of the ignition furnace cover of the ignition furnace through a measurement window of the ignition furnace cover with a radiation thermometer having a measurement wavelength selected from a range of 0.5 μm to 4.0 μm;
a plurality of data sets each consisting of a set of the input data and an actual value of the sintered ore yield, and by machine learning using the plurality of data sets obtained as training data, a yield prediction model is generated in which the input data is used as an input and the yield of the sintered ore is used as an output ;
A method for generating a yield prediction model, wherein the transmittance of the measurement window at the measurement wavelength is 50% or more . A sintered ore yield prediction device for predicting the yield of sintered ore in a sintered ore production facility that produces sintered ore by using an ignition furnace that burns the carbonaceous material in the sintered ore charging bed after charging sintered raw materials including carbonaceous material into an endless moving pallet to form a charging bed,
a data acquiring unit that acquires input data including a surface temperature of the charging bed inside the ignition furnace cover measured from the outside of the ignition furnace cover through a measurement window of the ignition furnace cover by a radiation thermometer having a measurement wavelength selected from a range of 0.5 μm to 4.0 μm;
a sintered ore yield prediction unit that inputs the input data into a yield prediction model, outputs the sintered ore yield, and predicts the sintered ore yield;
and
The sintered ore yield prediction device , wherein the transmittance of the measurement window at the measurement wavelength is 50% or more .