Shaped steel cold-rolled dimension prediction method, shaped steel manufacturing method, cold-rolled dimension prediction model generation method, and shaped steel cold-rolled dimension prediction device

Abstract

Provided are a shaped steel cold-rolled dimension prediction method and a shaped steel cold-rolled dimension prediction device, which can predict a cold-rolled dimension of a shaped steel with high accuracy using a hot-rolled dimension of the shaped steel.　The shaped steel cold-rolled dimension prediction method comprises: inputting data, which includes a hot-rolled dimension of a shaped steel manufactured by hot rolling a steel billet having been heated in a heating furnace in a plurality of passes and then cooling said steel billet, a hot rolling temperature of the shaped steel, and a cold rolling temperature of the shaped steel, to a cold-rolled dimension prediction model; and predicting the cold-rolled dimension of the shaped steel by outputting the cold-rolled dimension of the shaped steel. The cold-rolled dimension of the shaped steel outputted from the cold-rolled dimension prediction model is one or more of cross-sectional dimensions of the shaped steel and the hot-rolled dimension of the shaped steel included in the input data includes one or more of the cross-sectional dimensions.

Description

Method for predicting cold dimensions of shaped steel, method for manufacturing shaped steel, method for generating cold dimension prediction model, and device for predicting cold dimensions of shaped steel

[0001] The present invention relates to a method for predicting cold dimensions of shaped steel, a method for manufacturing shaped steel, a method for generating a cold dimension prediction model for shaped steel, and an apparatus for predicting cold dimensions of shaped steel.

[0002] H-beams, a type of shaped steel, are primarily manufactured by hot rolling. In the manufacturing process, billets of steel are preheated and then hot-rolled using multiple rolling mills, including rough rolling, intermediate rolling, and finish rolling, to form them into a desired cross-sectional dimension.

[0003] After finish rolling, the H-beams are cooled in a cooling bed. After being cooled in the cooling bed, the H-beams are straightened for any bends or warps using a roll straightener called a leveler, and then transported to an inspection table. The cross-sectional dimensions of the H-beams after cooling are measured by workers on the inspection table (hereafter, the cross-sectional dimensions after cooling will be referred to as "cold dimensions"). The cold dimensions of H-beams may also be measured automatically using a cold dimension gauge.

[0004] If the measured cold dimensions are outside the target dimension range, the roll spacing (roll gap) during rough rolling, intermediate rolling, and finish rolling is changed so that the cold dimensions are within the target dimension range. However, by the time the cold dimensions of the H-section steel are measured, several hours have passed since the finish rolling was completed due to cooling in a cooling bed and straightening with a roll straightener. This delay in adjusting the roll spacing causes the problem of producing a large number of defective products that are outside the target dimension range.

[0005] To address this issue, there is a method of measuring the cross-sectional dimensions of the H-section steel immediately after finish rolling (hereinafter, the hot cross-sectional dimensions after finish rolling will be referred to as "hot dimensions") and adjusting the roll gap during rough rolling, intermediate rolling, and finish rolling based on the hot dimensions, thereby quickly eliminating the dimensional defects. However, the temperature of the H-section steel immediately after finish rolling is much higher than room temperature. Therefore, the hot dimensions of the H-section steel are larger than the cold dimensions, so even if the hot dimensions are within the target dimension range, the cold dimensions do not necessarily fall within the target dimension range. For this reason, it is necessary to convert the hot dimensions to cold dimensions and then adjust the roll gap during rough rolling, intermediate rolling, and finish rolling.

[0006] As a technique for converting hot dimensions to cold dimensions, Patent Document 1 discloses a technique in which the cross-sectional dimensions after finish rolling are measured while the temperature of the measurement object is simultaneously measured, and the hot dimensions are corrected to cold dimensions using a linear equation for the temperature. According to Patent Document 1, by using a linear equation for the temperature at which the dimensions are measured, it is possible to correct the hot dimensions to cold dimensions while taking into account changes in the cross-sectional dimensions due to thermal expansion.

[0007] Patent Document 2 discloses a technology for converting hot dimensions into cold dimensions by measuring the temperature distribution in the width direction of the flange and the temperature distribution in the height direction of the web, correcting the amount of thermal expansion of the measured parts with a linear expansion coefficient according to the temperature, and calculating the sum of the amount of thermal expansion for each part.

[0008] Japanese Patent Laid-Open No. 7-120235 Japanese Patent Laid-Open No. 2008-249352

[0009] However, in Patent Document 1, the correction is made using a linear equation of the temperature at which the dimensions are measured, and therefore the technology does not take into consideration volume expansion due to phase transformation when the temperature drops. Therefore, the technology disclosed in Patent Document 1 has a problem in that the accuracy of correction to cold dimensions decreases when volume expansion due to phase transformation occurs.

[0010] In Patent Document 2, the temperature distribution is measured by moving the thermometer in the height direction or width direction, which results in a problem that it takes a long time to measure the temperature.This technology also has a problem that the volume expansion due to phase transformation when the temperature drops is not modeled in detail, so when volume expansion due to phase transformation occurs, the accuracy of conversion from hot dimensions to cold dimensions decreases.

[0011] The present invention has been made in view of the above-mentioned conventional techniques, and its object is to provide a method and an apparatus for predicting cold dimensions of shaped steel, which can predict the cold dimensions of shaped steel with high accuracy using the hot dimensions of the shaped steel. Furthermore, another object of the present invention is to provide a method for generating a cold dimension prediction model to be used in the method for predicting cold dimensions of shaped steel, and a method for manufacturing shaped steel.

[0012] The means for solving the above problems are as follows. [1] A method for predicting the cold dimensions of a shaped steel manufactured by hot rolling a slab in multiple passes and cooling it, comprising: inputting input data including the hot dimensions of the shaped steel, the hot temperature of the shaped steel, and the cold temperature of the shaped steel into a cold dimension prediction model; outputting the cold dimensions of the shaped steel to predict the cold dimensions of the shaped steel; the cold dimensions of the shaped steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the shaped steel; and the hot dimensions of the shaped steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the shaped steel output from the cold dimension prediction model. [2] The method for predicting the cold dimensions of a shaped steel according to [1], wherein the input data includes at least one of one or more rolling parameters of the hot rolling and one or more attribute parameters of the shaped steel. [3] The method for predicting the cold dimensions of a shaped steel according to [1] or [2], wherein the shaped steel is straightened by a roller straightener after cooling, and the input data includes one or more straightening parameters of the roller straightener. [4] A method for producing a shaped steel by hot rolling a billet in multiple passes and cooling it, the method comprising: specifying hot dimensions of the shaped steel such that the cold dimensions of the shaped steel predicted by the method for predicting the cold dimensions of the shaped steel according to any one of [1] to [3] fall within a target dimension range; specifying rolling parameters for hot rolling that can hot roll the shaped steel with the specified hot dimensions using a hot dimension prediction model that receives input data including one or more rolling parameters for the hot rolling and outputs the hot dimensions of the shaped steel; and hot rolling the billet in multiple passes under rolling conditions that include the specified rolling parameters for hot rolling.[5] A method for generating a cold dimension prediction model used to predict the cold dimensions of structural steel manufactured by hot rolling and cooling a steel slab in multiple passes, the method comprising: training a machine learning model using multiple data sets as training data, each set consisting of actual values ​​of input data including the hot dimensions of the structural steel manufactured in the past, the hot temperature of the structural steel, and the cold temperature of the structural steel, and actual values ​​of the cold dimensions of the structural steel; generating a cold dimension prediction model that uses the input data as input and outputs the cold dimensions of the structural steel; the cold dimensions of the structural steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the structural steel; and the hot dimensions of the structural steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the structural steel output from the cold dimension prediction model. [6] A device for predicting cold dimensions of a shape steel produced by hot rolling a steel slab in multiple passes and cooling it, the device comprising a cold dimension prediction unit that inputs input data including the hot dimensions of the shape steel, the hot temperature of the shape steel, and the cold temperature of the shape steel into a cold dimension prediction model and outputs the cold dimensions of the shape steel to predict the cold dimensions of the shape steel, the cold dimensions of the shape steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the shape steel, and the hot dimensions of the shape steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the shape steel output from the cold dimension prediction model. [7] The device for predicting cold dimensions of a shape steel according to [6], wherein the input data includes at least one of one or more rolling parameters of the hot rolling and one or more attribute parameters of the shape steel. [8] The cold dimension prediction device for structural steel according to [6] or [7], wherein the structural steel is straightened by a roller straightener after cooling, and the input data includes one or more straightening parameters of the roller straightener.[9] A cold dimension prediction device for structural steel as described in any of [6] to [8], which has a rolling parameter identification unit that identifies rolling parameters for hot rolling that can hot roll structural steel within a target cold dimension range, and the rolling parameter identification unit identifies hot dimensions of the structural steel that are within the target cold dimension range, and identifies rolling parameters for hot rolling that can hot roll structural steel of the identified hot dimensions using a hot dimension prediction model that receives input data including one or more of the rolling parameters for hot rolling and outputs the hot dimensions of the structural steel.

[0013] The cold dimension prediction model used in the method for predicting cold dimensions of sectional steel according to the present invention is generated using actual values ​​of the cold dimensions of sectional steel that have undergone volumetric expansion due to phase transformation, and therefore the prediction model is a cold dimension prediction model that takes into account volumetric expansion due to phase transformation. In this way, the cold dimension prediction method according to the present embodiment predicts cold dimensions using a prediction model that takes into account volumetric expansion due to phase transformation, making it possible to predict cold dimensions with high accuracy from hot dimensions.

[0014] The cold dimension prediction model used in the method for predicting cold dimensions of section steel according to the present invention uses the temperature measured in the hot state. Temperatures measured with a non-contact thermometer do not capture the temperature distribution in the thickness direction, and the amount of shrinkage varies depending on the thickness, even if the thickness shrinks from the same temperature to room temperature. In contrast, the cold dimension prediction model simultaneously measures the temperature and thickness during hot measurement, resulting in a prediction model that takes into account differences in the amount of thermal shrinkage due to differences in thickness. Thus, the method for predicting cold dimensions of section steel according to the present invention predicts cold dimensions using a cold dimension prediction model that takes into account differences in the amount of thermal shrinkage due to differences in thickness, making it possible to predict cold dimensions with high accuracy from the hot dimensions.

[0015] Fig. 1 is a cross-sectional schematic diagram showing the cross-sectional shape of an H-section steel. Fig. 2 is a schematic diagram showing an example of rolling equipment including a sectional steel cold dimension prediction device in which a sectional steel cold dimension prediction method according to an embodiment of the present invention can be implemented. Fig. 3 is a schematic diagram showing an example of the configuration of a sectional steel cold dimension prediction device. Fig. 4 is a flow chart showing the flow of a rolling parameter identification process.

[0016] First, the cross-sectional dimensions of an H-shaped steel 10, which is an example of structural steel, will be described using FIG. 1 . FIG. 1 is a cross-sectional schematic diagram showing the cross-sectional shape of the H-shaped steel 10. As shown in FIG. 1 , the relatively thick portions of the H-shaped steel 10 are the flanges 12, and the relatively thin portions that connect the pair of flanges 12 are the webs 14. The cross-sectional dimensions of the H-shaped steel 10 include the thickness of the web 14 (twb), the thicknesses of the flanges 12 in the top, bottom, left, and right directions (tf1b, tf2b, tf3b, tf4b), the height of the web 14 (hwb), and the width of the flanges 12 (wfb). In the following description, the thickness of the web 14 will be referred to as "twb." The thicknesses of the flanges 12 in the top, bottom, left, and right directions will be referred to as "tf1b, tf2b, tf3b, tf4b." The height of the web 14 will be referred to as "hwb," and the width of the flanges 12 will be referred to as "wfb."

[0017] There are many combinations of cross-sectional dimensions for the H-shaped steel 10, ranging from several tens for constant inside dimension H-shaped steel (JIS H) to several hundreds for constant outside dimension H-shaped steel. Here, the outside dimension is the web height hwb of the H-shaped steel 10, and the inside dimension is the dimension obtained by subtracting the flange thickness from the outside dimension.

[0018] The strength classification of the H-shaped steel 10 is roughly divided into tensile strength of 400N / mm ２ class, 490N / mm ２ class, 490N / mm ２ The H-beam steel 10 is divided into three types: 40k steel, 50k steel, and greater than 50k. The chemical composition of the H-beam steel 10 is adjusted for each steel type classification (40k steel, 50k steel, and greater than 50k) according to this strength classification. Such H-beam steel 10 is mainly manufactured by hot rolling. The H-beam steel 10 manufactured by hot rolling is called rolled H-beam steel.

[0019] In the method for predicting cold dimensions of shaped steel and the device 46 for predicting cold dimensions of shaped steel according to this embodiment, each cross-sectional dimension (hot dimension) of hot rolled shaped steel is used to predict each cross-sectional dimension (cold dimension) of the shaped steel after cooling. Hereinafter, an embodiment of the present invention will be described using an example in which it is applied to an H-shaped steel, which is an example of shaped steel.

[0020] 2 is a schematic diagram showing an example of a rolling facility 100 including a section steel cold dimension prediction device 46 capable of implementing a section steel cold dimension prediction method according to an embodiment of the present invention. The rolling facility 100 is installed in a rolling factory. The rolling facility 100 includes a heating furnace 20, a roughing mill 22, an intermediate rolling mill 24, a finishing mill 30, a hot shape gauge 32, and a hot thermometer 34.

[0021] The heating furnace 20 is a device that heats the billet to a predetermined temperature. In the heating furnace 20, the billet is heated to the austenite temperature range or higher (e.g., 1100 to 1300°C or higher). The billet extracted from the heating furnace 20 is rolled in a roughing mill 22. In the roughing mill 22, the billet with a rectangular cross section is subjected to about 5 to 30 passes of reverse rolling using grooved rolls, thereby roughly shaping it to approximate the cross-sectional dimensions of an H-beam. The example shown in FIG. 2 illustrates a case in which the billet is heated in the heating furnace 20, but as long as hot rolling is possible, the billet does not necessarily have to be heated in the heating furnace 20. The billet may be heated in a tunnel furnace or by induction heating. Alternatively, the billet after casting may be directly hot rolled without heating.

[0022] The intermediate rolling mills 24 are a combination of one or more intermediate universal rolling mills 26 and one or more edger rolling mills 28. The intermediate universal rolling mill 26, which is one of the intermediate rolling mills 24, is a rolling mill having a total of four rolls: upper and lower horizontal rolls that are driven to rotate about a horizontal axis, and left and right vertical rolls that rotate freely about a vertical axis.

[0023] The diameter of the upper and lower horizontal rolls of the intermediate universal rolling mill 26 is approximately 1000 to 1500 mm. The width of the upper and lower horizontal rolls is adjusted to the appropriate width depending on the hwb of the H-section steel 10 to be rolled. The diameter of the left and right vertical rolls is approximately 600 to 1000 mm. The upper and lower horizontal rolls and the left and right vertical rolls are each designed so that their positions can be adjusted using a reduction device, and the spacing between the upper and lower horizontal rolls and the spacing between the sides of the horizontal rolls and the vertical rolls can be set as desired. The intermediate universal rolling mill 26 can simultaneously reduce the web 14 and flange 12, reducing the flange 12 with a maximum outward tilt of approximately 10 degrees. The intermediate universal rolling mill 26 performs approximately 5 to 25 passes of reverse rolling.

[0024] The edger rolling mill 28, which is one of the intermediate rolling mills 24, has two horizontal rolls, one above the other, each equipped with a groove (hereinafter, the horizontal rolls will be referred to as "E1 rolls"). The diameter of the E1 roll is approximately 800 to 1200 mm, and both the upper and lower E1 rolls are driven. The E1 rolls press down the tip of the flange 12 from above and below, thereby adjusting the wfb. In the edger rolling mill 28, the E1 rolls perform approximately 5 to 25 passes of reverse rolling, forming the H-section steel into the cross-sectional dimensions.

[0025] A finish universal rolling mill is used as the finishing rolling mill 30. The finishing rolling mill 30 is a rolling mill that finishes the intermediate rolled rolled material into a product cross-sectional shape by rolling in one pass. The dimensions and configuration of the rolls of the finish universal rolling mill are the same as those of the intermediate universal rolling mill 26, but the finish universal rolling mill rolls the flange 12 so that it is perpendicular to the web 14. The finish rolled H-section steel 10 is cut to a predetermined product length by a hot saw (not shown).

[0026] The hot shape meter 32 is provided downstream of the finishing rolling mill 30 and measures hot dimensions, which are cross-sectional dimensions, of the H-shaped steel 10 rolled by the roughing mill 22, the intermediate rolling mill 24, and the finishing rolling mill 30. The hot shape meter 32 measures twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb of the H-shaped steel 10 in the hot state.

[0027] It is preferable that the hot shape meter 32 measures the above cross-sectional dimensions over the entire length and uses the average value of the measured cross-sectional dimensions. It is more preferable to use the average value of the cross-sectional dimensions excluding several meters from the leading and trailing ends of the H-section steel 10, which are the unsteady portions. The hot dimensions of the H-section steel 10 measured by the hot shape meter 32 are output to the process computer 44.

[0028] The hot thermometer 34 is provided near the hot shape gauge 32 and measures the temperatures of the flange 12 and web 14 of the H-shaped steel 10 whose hot dimensions have been measured by the hot shape gauge 32 (hereinafter, this temperature will be referred to as the "hot temperature"). The hot thermometer 34 measures the temperature of the flange 12, for example, at a position on the outer surface of the flange that is 1 / 4 of the height of the flange 12, and measures the temperature of the web 14, for example, at a position on the upper surface of the web that is the center of the web 14. The hot thermometer 34 outputs the hot temperatures to the process computer 44. If the hot shape gauge 32 has a temperature measurement function, the hot temperature of the H-shaped steel 10 may be measured by the hot shape gauge 32.

[0029] The rolling equipment 100 further includes a cooling bed 36, a roller straightener 38, a cold shape gauge 40, a cold thermometer 42, a process computer 44, and a steel cold dimension prediction device 46. The H-beam 10 whose hot dimensions and hot temperature have been measured is transported to the cooling bed 36. The cooling bed 36 is a device for cooling the hot-rolled H-beam 10. The H-beam 10 is left on the cooling bed 36 for several hours and cooled to room temperature by air cooling. In some cases, mist is sprayed onto the H-beam 10 on the cooling bed 36 to accelerate the cooling rate. Room temperature refers to the room temperature of a rolling factory where the rolling equipment 100 is installed and the H-beam 10 is manufactured.

[0030] The H-beam 10 cooled in the cooling bed 36 is straightened for bending and warping by a roller straightener 38. The roller straightener 38 is a device that straightens bending and warping of the flange 12 and web 14 of the H-beam 10. The roller straightener 38 is configured so that the roll pitch and reduction amount of the straightening rolls can be adjusted, and the roll pitch and reduction amount of the straightening rolls are adjusted so that the bending and warping of the H-beam 10 are appropriately straightened.

[0031] The H-beam 10 whose bending and warping have been corrected by the roller straightener 38 is transported to the cold shape meter 40. The cold shape meter 40 measures the cold dimensions, which are the cross-sectional dimensions, of the cooled H-beam 10. The cold shape meter 40 measures the twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb of the H-beam in the cold state. The cold dimensions may be measured by an operator on an inspection table (not shown).

[0032] For the cold dimensions, it is preferable to measure the above cross-sectional dimensions over the entire length and use the average value of the measured cross-sectional dimensions. It is even more preferable to use the average value of the cross-sectional dimensions excluding several meters from the leading and trailing ends of the H-section steel 10, which are the unsteady portions. The measured cold dimensions are output to the process computer 44.

[0033] The cold thermometer 42 is provided near the cold shape gauge 40 and measures the temperatures of the flanges 12 and web 14 of the H-beam 10 whose cold dimensions have been measured (hereinafter, this temperature will be referred to as the "cold temperature"). The cold thermometer 42 measures the temperature of the flange 12, for example, at a position that is ¼ of the height of the flange 12, and the temperature of the web 14, for example, at the center of the web 14. Since the temperature of the H-beam 10 cooled to room temperature is the same as the room temperature of the rolling mill, the cold thermometer 42 may measure the room temperature of the rolling mill instead of measuring the temperatures of the flanges 12 and web 14 of the H-beam 10, and use this temperature as the cold temperature. In this embodiment, an example will be described in which the room temperature of the rolling mill is used as the cold temperature. The cold thermometer 42 outputs the cold temperature to the process computer 44.

[0034] The process computer 44 may be a general-purpose computer such as a workstation or a personal computer. The process computer 44 is connected to the heating furnace 20, the roughing mill 22, the intermediate rolling mill 24, the finishing mill 30, the hot shape gauge 32, the hot thermometer 34, the cold shape gauge 40, and the cold thermometer 42 by wire or wirelessly, and controls the manufacturing process of the H-beam 10 by the rolling equipment 100. The process computer 44 acquires attribute parameters of the H-beam 10 to be manufactured from a higher-level computer. The attribute parameters of the H-beam 10 include information on the target dimensions of the H-beam 10, the steel type classification (40k steel, 50k steel, and three types above), the chemical composition (contents of C, Si, Mn, Cr, Mo, V, etc.), and target values ​​of mechanical properties (yield stress, tensile strength, elongation, toughness, hardness, etc.). The target dimensions of the H-shaped steel 10 are the target dimensions (representative values ​​and allowable values) of twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb.

[0035] The process computer 44 sets rolling parameters for hot rolling, such as the number of passes of each rolling mill and the roll spacing of each rolling roll, in accordance with the attribute parameters of the H-section steel 10. The set values ​​of the rolling parameters for normal hot rolling are set as table values ​​associated with the attribute parameters of the H-section steel 10 based on past rolling results.

[0036] The process computer 44 outputs the set hot rolling parameters to the roughing mill 22, the intermediate universal mill 26, the edger mill 28, and the finishing mill 30. Each rolling mill controls the roll gap, roll position, rolling load, etc. of each rolling mill in accordance with the acquired hot rolling parameters. The process computer 44 acquires attribute parameters, hot rolling parameters, measured hot dimensions / cold dimensions, measured hot temperatures / cold temperatures, identification numbers that identify the roll sets of each rolling mill, etc. The process computer 44 stores this data in its database in association with the production lot number of the H-beam 10.

[0037] The structural steel cold dimension prediction device 46 predicts the cold dimensions of the H-shaped steel 10 after the hot dimensions of the H-shaped steel 10 are measured by the hot shape meter 32 and the hot temperature is measured by the hot thermometer 34. The cold dimensions predicted by the structural steel cold dimension prediction device 46 may be one or more of twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb of the H-shaped steel 10. However, since all of the above cold dimensions are subject to management for the H-shaped steel 10, it is preferable that the structural steel cold dimension prediction device 46 predicts all of the cold dimensions twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb. In the following embodiment, an example will be described in which the structural steel cold dimension prediction device 46 predicts all cold dimensions of the H-shaped steel 10.

[0038] When the hot dimensions are measured by the hot shapemeter 32 and the hot temperature is measured by the hot thermometer 34, the section steel cold dimension prediction device 46 acquires input data including the hot dimensions, hot temperature, and cold temperature (room temperature of the rolling mill) of the H-section steel 10 from the process computer 44. The section steel cold dimension prediction device 46 inputs the acquired input data into a cold dimension prediction model and outputs the cold dimensions, thereby predicting the cold dimensions of the H-section steel 10 after it has been cooled to room temperature and has had any bending or warping corrected.

[0039] The cold dimension prediction device 46 for structural steel may identify hot dimensions that will result in the predicted cold dimensions of the H-shaped steel 10 falling within a target dimension range, and may identify rolling parameters for hot rolling that will produce the H-shaped steel 10 with the identified hot dimensions. The cold dimension prediction device 46 for structural steel may input input data including rolling parameters for hot rolling and perform inverse analysis of a hot dimension prediction model that outputs the hot dimensions of the H-shaped steel 10, thereby identifying rolling parameters for hot rolling that will produce the H-shaped steel 10 with hot dimensions that fall within the target dimension range. Here, inverse analysis of the hot dimension prediction model refers to an analytical method that uses the hot dimension prediction model to identify, through optimization calculation, a combination of input data that will output desired output data. That is, the cold dimension prediction device 46 for structural steel uses the hot dimension prediction model to identify, through optimization calculation, rolling parameters for hot rolling that will output the desired hot dimensions of the H-shaped steel 10 (hot dimensions of the H-shaped steel 10 that fall within the target dimension range). The section steel cold dimension prediction device 46 may output the identified rolling parameters for hot rolling to the process computer 44, thereby reflecting the rolling parameters in the rolling conditions for hot rolling.

[0040] Next, a structural steel cold dimension prediction device 46 that predicts the cold dimensions of the H-beam 10 will be described. FIG. 3 is a schematic diagram showing an example of the configuration of the structural steel cold dimension prediction device 46. The structural steel cold dimension prediction device 46 can be a general-purpose computer such as a workstation or a personal computer. The structural steel cold dimension prediction device 46 has a control unit 48, an input unit 50, an output unit 52, and a storage unit 54. The control unit 48 is, for example, a CPU, and functions as a data acquisition unit 56, a cold dimension prediction unit 58, a rolling parameter identification unit 60, and a prediction model generation unit 62 by executing programs stored in the storage unit 54.

[0041] The input unit 50 is, for example, a keyboard, a touch panel integrated with a display, or the like. The output unit 52 is, for example, an LCD or CRT display, or the like. The storage unit 54 is, for example, an updatable flash memory, a built-in hard disk or a hard disk connected via a data communication terminal, an information recording medium such as a memory card, and a read / write device for the same. The storage unit 54 stores programs and data for realizing each function of the section steel cold dimension prediction device 46.

[0042] The storage unit 54 further stores a database 64, a cold dimension prediction model 66, and a hot dimension prediction model 68. The database 64 stores 1,000 or more sets of data, each set containing the actual hot dimensions, actual hot temperatures, actual cold temperatures, and actual cold dimensions of the H-beam steel 10 previously manufactured by the rolling equipment 100. It is preferable that the number of stored data sets, each set containing the actual hot dimensions, actual hot temperatures, actual cold temperatures, and actual cold dimensions of the H-beam steel 10, be 3,000 or more. This data set is used as training data for generating the cold dimension prediction model 66.

[0043] The database 64 further stores 1,000 or more sets of data, each set consisting of rolling parameters for hot rolling of H-section steel 10 previously manufactured by the rolling equipment 100 and actual values ​​of the hot dimensions of the H-section steel 10. It is preferable that the number of stored data sets, each set consisting of rolling parameters for hot rolling and actual values ​​of the hot dimensions of the H-section steel 10, is also 3,000 or more. This data set is used as training data for generating the hot dimension prediction model 68.

[0044] It is preferable that the above-mentioned data sets for the H-section steel 10 manufactured by the rolling equipment 100 are sequentially acquired from the process computer 44 and stored in the database 64. In this case, an upper limit may be set on the number of data sets stored in the database 64, and the oldest data sets may be deleted so as not to exceed the upper limit.

[0045] The cold dimension prediction model 66 is a trained machine learning model that has been trained using multiple data sets stored in the database 64 as training data. The cold dimension prediction model 66 is a trained machine learning model that receives the hot dimensions (twb, tf1b, tf2b, tf3b, tf4b, hwb, wfb) of the H-shaped steel 10, the hot temperature, and the cold temperature as inputs, and outputs the cold dimensions (twb, tf1b, tf2b, tf3b, tf4b, hwb, wfb) of the H-shaped steel 10. The cold dimension prediction model 66 is used to predict the cold dimensions of the H-shaped steel 10.

[0046] The hot dimension prediction model 68 is also a trained machine learning model that has been trained using multiple data sets stored in the database 64 as training data. The hot dimension prediction model 68 is a trained machine learning model that takes hot rolling parameters as input and outputs the hot dimensions of the H-beam steel 10. The input hot rolling parameters include at least the roll spacing in each pass of the roughing mill 22, the intermediate mill 24, and the finishing mill 30. The hot rolling parameters may also include the rolling load in each pass of the roughing mill 22, the intermediate mill 24, and the finishing mill 30. The hot dimension prediction model 68 is used to identify hot rolling parameters that can hot roll an H-beam steel 10 of a specific hot dimension by back-analyzing the prediction model.

[0047] Next, we will explain the processing performed by the data acquisition unit 56 and the cold dimension prediction unit 58 when predicting the cold dimensions of the H-beam 10. The data acquisition unit 56 acquires the hot dimensions, hot temperature, and cold temperature of the H-beam 10 as input data from the process computer 44. The data acquisition unit 56 outputs the acquired input data to the cold dimension prediction unit 58. The cold temperature is the room temperature in the rolling mill at the time the cold temperature is predicted.

[0048] When the cold dimension prediction unit 58 acquires the input data, it reads out the cold dimension prediction model 66 from the storage unit 54. The cold dimension prediction unit 58 inputs the input data into the cold dimension prediction model 66 and outputs the cold dimensions of the H-shaped steel 10. In this way, the cold dimension prediction unit 58 predicts the cold dimensions of the H-shaped steel 10 using the hot dimensions of the finish-rolled H-shaped steel 10. The cold dimension prediction unit 58 may display the predicted cold dimensions of the H-shaped steel 10 on the output unit 52. This allows the worker to confirm the predicted values ​​of the cold dimensions of the H-shaped steel 10 by visually checking the output unit 52.

[0049] The cold dimension prediction model 66 used to predict the cold dimensions of the H-shaped steel 10 is generated using actual values ​​of the cold dimensions of the H-shaped steel 10 that have expanded in volume due to phase transformation, and therefore this model is a cold dimension prediction model that takes into account the volume expansion due to phase transformation. Therefore, it can be seen that by predicting the cold dimensions of the H-shaped steel 10 using this prediction model, the cold dimensions can be predicted with higher accuracy than a prediction model that does not take phase transformation into account.

[0050] The input data for the cold dimension prediction model 66 includes the hot dimensions, hot temperature, and cold temperature of the H-shaped steel 10. The hot dimensions, hot temperature, and cold temperature of the H-shaped steel 10 all affect the output cold dimensions of the H-shaped steel 10. Therefore, by including the hot dimensions, hot temperature, and cold temperature of the H-shaped steel 10 in the input data for the cold dimension prediction model, the prediction accuracy of the cold dimensions by the cold dimension prediction model 66 is improved.

[0051] Preferably, the input data for the cold dimension prediction model 66 further includes one or more rolling parameters for hot rolling. The rolling parameters for hot rolling are, for example, the parameters described below.

[0052] <Rolling parameters of hot rolling> Roll spacing between rolling rolls in each pass in the roughing mill 22, intermediate mill 24, and finishing mill 30 Time from extraction from the heating furnace 20 to finish rolling in the finishing mill 30 Rolling load in each pass in the roughing mill 22, intermediate mill 24, and finishing mill 30 Billet thickness, billet width, billet length, billet weight

[0053] These rolling parameters of the hot rolling affect the hot dimensions of the H-beam steel 10. As described above, since the hot dimensions affect the cold dimensions, it can be said that the rolling parameters of the hot rolling also affect the cold dimensions. Therefore, by including these rolling parameters of the hot rolling in the input data of the cold dimension prediction model, the prediction accuracy of the cold dimension prediction model 66 is improved.

[0054] The input data of the cold dimension prediction model 66 preferably further includes one or more straightening parameters of the roller straightener 38. The straightening parameters of the roller straightener 38 are, for example, the parameters described below.

[0055] <Straightening parameters> ・Straightening roll roll pitch ・Straightening roll reduction amount ・Roll axial position

[0056] Since the roller straightener 38 is a device that straightens out bending and warping of the cooled H-shaped steel 10, the straightening parameters of the roller straightener 38 have a significant effect on the cold dimensions of the H-shaped steel 10. For this reason, by including the straightening parameters of the roller straightener 38 in the input data of the cold dimension prediction model, the prediction accuracy of the cold dimension prediction model 66 is improved.

[0057] It is preferable that the input data of the cold dimension prediction model 66 used to predict the cold dimensions of the H-shaped steel 10 further includes one or more attribute parameters of the H-shaped steel 10. The attribute parameters of the H-shaped steel 10 are, for example, the parameters described below.

[0058] <Attribute parameters> ・Representative value of target dimensions ・Steel type classification (3 types: 40k steel, 50k steel, and over 50k) ・Chemical composition (content of C, Si, Mn, Cr, Mo, V, etc.) ・Target values ​​of mechanical properties (yield stress, tensile strength, elongation, toughness, hardness, etc.)

[0059] Of these attribute parameters, the representative value of the target dimension affects the hot and cold dimensions of the H-section steel 10. The steel type classification, chemical composition, and target values ​​of the mechanical properties affect the composition and structure of the oxide scale in the heating furnace 20, and affect the deformation resistance during hot rolling. For this reason, it can be said that these attribute parameters also affect the hot and cold dimensions of the H-section steel 10. Therefore, by including the above attribute parameters in the input data of the cold dimension prediction model 66, the prediction accuracy of the cold dimension prediction model 66 is improved.

[0060] Next, a description will be given of the rolling parameter specification process performed by the rolling parameter specification unit 60. Fig. 4 is a flow diagram showing the flow of the rolling parameter specification process. The flow shown in Fig. 4 is started, for example, on the condition that the process computer 44 has acquired the hot dimensions and hot temperature of the H-beam steel 10 that has been finish-rolled.

[0061] First, the data acquisition unit 56 acquires the hot dimensions, hot temperature, and cold temperature of the H-beam 10 as input data from the process computer 44 (step S101). The data acquisition unit 56 outputs the acquired input data to the cold dimension prediction unit 58.

[0062] When the cold dimension prediction unit 58 acquires the input data, it reads out the cold dimension prediction model 66 from the storage unit 54, inputs the input data into the cold dimension prediction model 66, and outputs the cold dimensions of the H-beam 10. As a result, the cold dimension prediction unit 58 predicts the cold dimensions of the H-beam 10 cooled to room temperature (step S102). The cold dimension prediction unit 58 outputs the predicted cold dimensions to the rolling parameter identification unit 60.

[0063] The rolling parameter specifying unit 60 determines whether all the predicted cold dimensions are within the range of the target dimensions of the H-beam (step S103). The range of the target dimensions of the H-beam may be stored in advance in the storage unit 54 or may be input by an operator via the input unit 50.

[0064] The rolling parameter specifying unit 60 compares all the predicted cold dimensions with the target dimension range for each dimension. If any one of the predicted cold dimensions is not included in the target dimension range, the rolling parameter specifying unit 60 determines that the predicted cold dimension is not within the target dimension range (step S103: No).

[0065] If the rolling parameter specifying unit 60 determines that the hot dimensions are not within the range of the target dimensions, it changes the hot dimensions used as input data (step S104). The rolling parameter specifying unit 60 changes the hot dimensions at the same positions as the cold dimensions determined not to satisfy the target dimensions, and executes the processing from step S102. In step S103, the rolling parameter specifying unit 60 repeatedly executes the processing of steps S104, S102, and S103 until all predicted cold dimensions are included in the range of the target dimensions. One unit of the hot dimensions changed by the rolling parameter specifying unit 60 may be stored in advance in the storage unit 54 for each hot dimension position, or may be input by an operator via the input unit 50.

[0066] On the other hand, if all predicted cold dimensions are included in the range of the target dimensions, the rolling parameter specifying unit 60 determines that the predicted cold dimensions are within the range of the target dimensions (step S103: Yes). The rolling parameter specifying unit 60 specifies the hot dimensions used in the input data as hot dimensions that will bring the cold dimensions into the range of the target dimensions (step S105). If it is determined in the first processing of step S103 that the cold dimensions are within the range of the target dimensions, there is no need to change the rolling parameters of the hot rolling, so the processing of steps S105 to S107 may be skipped.

[0067] The rolling parameter specifying unit 60 specifies rolling parameters for hot rolling that can hot roll the H-section steel 10 of the specified hot dimensions (step S106). Here, the rolling parameters for hot rolling are, for example, the roll spacing in each pass of the roughing mill 22, the intermediate mill 24, and the finishing mill 30. The rolling parameter specifying unit 60 reads out the hot dimension prediction model 68 from the storage unit 54, and uses the hot dimension prediction model 68 to specify the roll spacing in each pass that can hot roll the H-section steel 10 of the specified hot dimensions.

[0068] The rolling parameter specification unit 60 sets the roll spacing of the rolling rolls in each pass of each rolling mill to, for example, 10 levels in 0.1 mm increments, centered on the reference rolling conditions, for both the widening and narrowing sides of the roll spacing uniformly for each pass of each rolling mill. Then, the combined input data is input to the hot dimension prediction model 68, which outputs a predicted value of the hot dimension. Of the output predicted values ​​of the hot dimension, the roll spacing included in the input data used to output the predicted value of the hot dimension that is the same as or closest to the specified hot dimension is specified as the roll spacing that can roll the H-beam steel 10 with the specified hot dimension. In this way, the roll spacing for each pass that can roll the H-beam steel 10 with the specified hot dimension can be specified.

[0069] The rolling parameter specifying unit 60 outputs the specified hot rolling parameters to the process computer, and reflects the specified hot rolling parameters in the rolling conditions of each rolling mill (step S107), and this process ends. Thereafter, in the rolling facility 100, the H-beam steel 10 is manufactured by hot rolling under rolling conditions including the specified hot rolling parameters. This makes it possible to manufacture H-beam steel 10 that falls within the target cold dimension range.

[0070] Next, a method for generating the cold dimension prediction model 66 used to predict the cold dimensions of the H-section steel 10 will be described. The data acquisition unit 56 acquires, from the process computer, the actual hot dimensions, the actual cold dimensions, the actual hot temperature, and the actual cold temperature of the H-section steel 10 previously manufactured by the rolling equipment 100. The data acquisition unit 56 stores the acquired actual values ​​as a set of data in the database 64 of the storage unit 54. The number of data sets stored in the database 64 is preferably 1000 or more, and more preferably 3000 or more.

[0071] The prediction model generation unit 62 reads out a machine learning model stored in advance in the storage unit 54, and performs machine learning on the machine learning model using multiple data sets stored in the database 64 as training data to generate a trained machine learning model. This trained machine learning model becomes a cold dimension prediction model 66. The cold dimension prediction model 66 uses, as training data, a data set including actual values ​​of cold dimensions of shaped steel that has expanded in volume due to phase transformation. Therefore, the generated cold dimension prediction model 66 is a prediction model that takes into account volume expansion due to phase transformation. The prediction model generation unit 62 stores the generated cold dimension prediction model 66 in the storage unit 54. The machine learning model used in this embodiment may be any of commonly used neural networks, decision tree learning, random forests, support vector regression, Gaussian processes, and nearest neighbor methods.

[0072] The prediction model generation unit 62 may update the cold dimension prediction model 66 to a new cold dimension prediction model by, for example, retraining the machine learning model every month or every year. The data acquisition unit 56 acquires a data set every time an H-beam 10 is manufactured in the rolling equipment 100 and stores the data set in the database 64. Therefore, the database 64 stores measurement data of newly manufactured H-beam 10, and the number of data sets stored in the database 64 increases.

[0073] As the number of data sets and the number of training data increase, the prediction accuracy of the cold dimension prediction model 66 improves, making it possible to predict cold dimensions with higher accuracy. Furthermore, by using new data sets, the latest state of the rolling equipment 100 is reflected in the cold dimension prediction model 66, so by updating the cold dimension prediction model 66, it becomes possible to predict cold dimensions from hot dimensions with even higher accuracy. The prediction model generation unit 62 can also generate and update a hot dimension prediction model 68 using data sets stored in the database 64, just like the cold dimension prediction model 66.

[0074] As described above, in the cold dimension prediction method used in the structural steel cold dimension prediction device 46 according to this embodiment, cold dimensions are predicted from hot dimensions using the cold dimension prediction model 66 that reflects the actual values ​​of the cold dimensions of structural steel that have expanded in volume due to phase transformation. This makes it possible to predict cold dimensions from hot dimensions with higher accuracy than a method that predicts cold dimensions without considering volume expansion due to phase transformation.

[0075] The present invention is not limited to the above embodiment and various modifications can be made. In the present embodiment, the cold dimension prediction method is described as an example in which all cross-sectional dimensions of the H-shaped steel 10 are predicted in the cold state. However, the present invention is not limited to this. The cold dimension prediction method according to the present embodiment is only required to predict one or more of twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb in the cold state. In this case, the input data of the cold dimension prediction model 66 used to predict the cold dimensions only needs to include a cross-sectional dimension corresponding to the cold dimension of the steel beam to be predicted. For example, when the cold dimension prediction method predicts the cold dimension twb, the input data of the cold dimension prediction model 66 only needs to include at least the hot dimension twb, and other cross-sectional dimensions may also be included.

[0076] The hot and cold temperatures included in the input data preferably include the hot and cold temperatures at positions corresponding to the positions of the cross-sectional dimensions to be predicted. For example, when predicting the cold twb using the cold dimension prediction method, the input data preferably includes the hot and cold temperatures of the web 14.

[0077] In the present embodiment, an example has been shown in which the rolling facility 100 shown in Fig. 2 has the process computer 44 and the section steel cold dimension prediction device 46, but this is not limiting. For example, the process computer 44 may have the function of the section steel cold dimension prediction device 46, and these may be configured as a single device.

[0078] In the present embodiment, an example has been shown in which the control unit 48 in the section steel cold dimension prediction device 46 shown in Fig. 3 has the data acquisition unit 56, the cold dimension prediction unit 58, the rolling parameter identification unit 60, and the prediction model generation unit 62, but this is not limited to this. If the cold dimension prediction method is implemented in the section steel cold dimension prediction device 46, the control unit 48 does not need to have the rolling parameter identification unit 60. If the cold dimension prediction model 66 and the hot dimension prediction model 68 are generated using separate calculation devices and the generated prediction models are stored in the storage unit 54, the control unit 48 does not need to have the prediction model generation unit 62.

[0079] In the present embodiment, an example has been described in which the cold dimensions of the H-shaped steel 10 are predicted using the shaped steel cold dimension prediction device 46, but the shaped steel for which the cold dimensions are predicted is not limited to the H-shaped steel 10. The shaped steel cold dimension prediction device 46 can similarly predict the cold dimensions of other shaped steel manufactured by hot rolling in the roughing mill 22, the intermediate rolling mill 24, and the finishing mill 30, such as channel steel, I-shaped steel, angle steel, steel sheet pile, or rail.

[0080] Next, an example will be described in which a steel billet was hot-rolled into an H-section steel using the rolling equipment 100 shown in Fig. 2, the cold dimensions of the H-section steel were predicted from the hot dimensions of the H-section steel using a cold dimension prediction model, and 100 shaped steel pieces were manufactured while adjusting the rolling conditions using the predicted cold dimensions. In this example, of the 100 H-section steel pieces manufactured, the cold dimensions of 99 H-section steel pieces were predicted from the hot dimensions, excluding the first H-section steel piece immediately after the rolls were changed.

[0081] In Example 1, a cold dimension prediction model was used that takes the following input data as input and outputs cold dimensions. The cold dimensions to be output are twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb. In Example 1, 900 H-shaped steel beams with hwb of 400 to 900 mm and wfb of 200 to 400 mm were manufactured in advance, and the data set obtained from this manufacturing example was used as training data to train the machine learning model and generate a cold dimension prediction model. A neural network with three intermediate layers and five nodes each was used for the machine learning model. A sigmoid function was used as the activation function.

[0082] <Input data for Example 1> Hot dimensions (twb, tf1b, tf2b, tf3b, tf4b, hwb, wfb) Hot temperatures (web temperature, flange temperature) Cold temperatures (room temperature in the rolling mill at the time of prediction)

[0083] In Example 2, a cold dimension prediction model was used that takes the following input data as input and outputs cold dimensions. The cold dimensions to be output are twb, tf1b, tf2b, tf3b, tf4b, hwb, and wfb. In Example 2, 900 H-section steel beams with hwb of 400 to 900 mm and wfb of 200 to 400 mm were also manufactured in advance, and the data set obtained from this manufacturing example was used as training data to train the machine learning model and generate a cold dimension prediction model. A neural network with three intermediate layers and five nodes each was used for the machine learning model. A sigmoid function was used as the activation function.

[0084] <Input data for Example 2> Hot dimensions (twb, tf1b, tf2b, tf3b, tf4b, hwb, wfb) Hot temperatures (web temperature, flange temperature) Cold temperatures (room temperature in the rolling mill at the time of prediction) Steel type classification (3 types: 40k steel, 50k steel, and over 50k) Billet thickness, width, length, and weight Time from extraction from the heating furnace to finish rolling Rolling load for each pass in the roughing mill, intermediate mill, and finish mill

[0085] In Examples 1 and 2, the rolling conditions for hot rolling were adjusted based on the cold dimensions predicted using the hot dimensions. The adjustment of the rolling conditions for hot rolling was carried out using a hot dimension prediction model in which the roll gaps in each pass of the roughing mill, intermediate mill, and finishing mill are input and the hot dimensions are output. Specifically, the roll gaps for each pass in the rolling conditions for hot rolling were adjusted by the following steps 1 to 3.

[0086] Step 1: Using the cold dimension prediction model, the hot dimensions of the H-section steel were identified to fall within the range of the target cold dimensions. Step 2: The roll spacing for each pass was adjusted to 10 levels in 0.1 mm increments on both the widening and narrowing sides of each pass of each rolling mill, centered on the standard rolling conditions. These combinations were then input as input data into the hot dimension prediction model to output the hot dimensions. Step 3: Of the output hot dimensions, the roll spacing for each pass was adjusted so that it was the roll spacing for each pass included in the input data used to output the predicted hot dimensions that were the same as the identified hot dimensions.

[0087] On the other hand, in Comparative Example 1, H-section steel was manufactured by adjusting the rolling conditions based on the measured cold dimensions without using a prediction model that predicts cold dimensions from hot dimensions. In Comparative Example 2, H-section steel was manufactured by adjusting the rolling conditions based on the measured hot dimensions. The manufacturing results of Invention Examples 1 and 2 and Comparative Examples 1 and 2 are shown in Table 1 below.

[0088]

[0089] In Example 1, of the 99 shaped steel sections evaluated, 85 had all of their cold dimensions (twb, tf1b, tf2b, tf3b, tf4b, hwb, wfb) within the target range. In Example 2, of the 99 shaped steel sections evaluated, 96 had all of their cold dimensions within the target range.

[0090] Such results could not be obtained unless the cold dimensions of the H-section steel could be predicted with high accuracy. Therefore, from the above results, it was confirmed that in Inventive Examples 1 and 2, the cold dimensions of the H-section steel could be predicted with high accuracy, and that by adjusting the rolling conditions using the cold dimensions, it was possible to manufacture H-section steel with the target cold dimensions.

[0091] In Example 2, the input data included the billet thickness, width, length, and weight, the time from removal from the heating furnace to finish rolling, and the rolling load, resulting in a cold dimension prediction model that took these fluctuations into account. In fact, during the production of 100 H-beams, the billet weight, the time from removal from the heating furnace to finish rolling, and the rolling load for each pass fluctuated. Therefore, in Example 2, the cold dimensions could be predicted taking into account the effects of these fluctuations, improving the accuracy of cold dimension prediction. As a result, more billets were accepted than in Example 1.

[0092] On the other hand, in Comparative Example 1, the rolling of the first section steel was completed, and the steel was cooled on a cooling bed and straightened with a roller straightener, and then the cold dimensions were measured with a cold gauge. At this point, four hours had passed since the second H-section steel was rolled, and there was an operational concern that a large amount of H-section steel that did not meet the target cold dimensions would be produced, so the production of the H-section steel was stopped midway.

[0093] In Comparative Example 2, the rolling conditions were adjusted so that the hot dimensions were within the target range. The adjustment of the rolling conditions was the same as Procedure 3 in Invention Examples 1 and 2. However, because the hot dimensions and the cold dimensions are different, adjusting the rolling conditions based on the hot dimensions did not allow for the production of H-section steel with the target cold dimensions. Of the 99 sections evaluated, only 60 had cold dimensions that were all within the target range. For sections whose cold dimensions did not fall within the target range, additional dimensional correction processes or remanufacturing were required, resulting in a significant increase in the manufacturing cost of the sections.

[0094] DESCRIPTION OF SYMBOLS 10 H-section steel 12 Flange 14 Web 20 Heating furnace 22 Roughing rolling mill 24 Intermediate rolling mill 26 Intermediate universal rolling mill 28 Edger rolling mill 30 Finishing rolling mill 32 Hot shape gauge 34 Hot thermometer 36 Cooling bed 38 Roller straightener 40 Cold shape gauge 42 Cold thermometer 44 Process computer 46 Cold dimension prediction device for section steel 48 Control unit 50 Input unit 52 Output unit 54 Storage unit 56 Data acquisition unit 58 Cold dimension prediction unit 60 Rolling parameter identification unit 62 Prediction model generation unit 64 Database 66 Cold dimension prediction model 68 Hot dimension prediction model 100 Rolling equipment

Claims

1. A method for predicting the cold dimensions of structural steel produced by hot rolling a slab in multiple passes and cooling it, comprising: inputting input data including the hot dimensions of the structural steel, the hot temperature of the structural steel, and the cold temperature of the structural steel into a cold dimension prediction model; outputting the cold dimensions of the structural steel to predict the cold dimensions of the structural steel; the cold dimensions of the structural steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the structural steel; and the hot dimensions of the structural steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the structural steel output from the cold dimension prediction model.

2. A method for predicting cold dimensions of a section steel as set forth in claim 1, wherein the input data includes at least one of one or more rolling parameters of the hot rolling and one or more attribute parameters of the section steel.

3. A method for predicting cold dimensions of structural steel as described in claim 1 or claim 2, wherein the structural steel is straightened with a roller straightener after cooling, and the input data includes one or more straightening parameters of the roller straightener.

4. A method for manufacturing shaped steel by hot rolling a slab in multiple passes and cooling it, comprising: specifying hot dimensions of the shaped steel such that the cold dimensions of the shaped steel predicted by the method for predicting cold dimensions of shaped steel described in any one of claims 1 to 3 are within a target dimension range; specifying hot rolling parameters that can hot roll the shaped steel with the specified hot dimensions using a hot dimension prediction model that receives input data including one or more of the rolling parameters of the hot rolling and outputs the hot dimensions of the shaped steel; and hot rolling the slab in multiple passes under rolling conditions that include the specified hot rolling parameters.

5. A method for generating a cold dimension prediction model used to predict the cold dimensions of structural steel manufactured by hot rolling and cooling steel billets in multiple passes, comprising: training a machine learning model using multiple data sets as training data, each set consisting of actual values ​​of input data including the hot dimensions of the structural steel manufactured in the past, the hot temperature of the structural steel, and the cold temperature of the structural steel, and actual values ​​of the cold dimensions of the structural steel; generating a cold dimension prediction model that takes the input data as input and outputs the cold dimensions of the structural steel; the cold dimensions of the structural steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the structural steel; and the hot dimensions of the structural steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the structural steel output from the cold dimension prediction model.

6. A cold dimension prediction device for structural steel manufactured by hot rolling and cooling a slab in multiple passes, comprising a cold dimension prediction unit that inputs input data including the hot dimensions of the structural steel, the hot temperature of the structural steel, and the cold temperature of the structural steel into a cold dimension prediction model and outputs the cold dimensions of the structural steel to predict the cold dimensions of the structural steel, wherein the cold dimensions of the structural steel output from the cold dimension prediction model are one or more of the cross-sectional dimensions of the structural steel, and the hot dimensions of the structural steel included in the input data include cross-sectional dimensions corresponding to the cold dimensions of the structural steel output from the cold dimension prediction model.

7. The apparatus for predicting cold dimensions of structural steel as set forth in claim 6, wherein the input data includes at least one of one or more rolling parameters of the hot rolling and one or more attribute parameters of the structural steel.

8. The cold dimension prediction device for structural steel as described in claim 6 or claim 7, wherein the structural steel is straightened by a roller straightener after cooling, and the input data includes one or more straightening parameters of the roller straightener.

9. A cold dimension prediction device for structural steel as described in any one of claims 6 to 8, comprising a rolling parameter identification unit that identifies rolling parameters for hot rolling that can hot roll structural steel within a target cold dimension range, wherein the rolling parameter identification unit identifies hot dimensions of the structural steel that fall within the target cold dimension range, and identifies rolling parameters for hot rolling that can hot roll structural steel to the identified hot dimensions using a hot dimension prediction model that receives input data including one or more of the rolling parameters for hot rolling and outputs the hot dimensions of the structural steel.

Images