Method for slicing material to be slicing and method for manufacturing steel billets

Abstract

To provide a technique that can suppress bumping caused by scarfing of a material to be scarfed.SOLUTION: A scarfing method for a material to be scarfed includes steps of: inputting information on a material to be scarfed including a surface temperature of the material to be scarfed obtained before scarfing the material to be scarfed and a scarfing condition including arbitrarily set conveying speed and scarfing-oxygen pressure to a machine learning model; calculating a predicted scarfing amount and a predicted noise value of the material to be scarfed and calculating a difference between the predicted scarfing amount and a preset target scarfing amount; changing either or both of the conveying speed and the scarfing-oxygen pressure and repeatedly executing calculation of the predicted scarfing amount and the target scarfing amount, until the difference becomes an allowable value or less and the predicted noise value becomes a preset reference noise value or less; and scarfing the material to be scarfed while conveying the material at the preset conveying speed by the scarfing-oxygen pressure, when the difference between the predicted scarfing amount and the target scarfing amount becomes equal to the allowable value or less and the predicted noise value becomes equal to the reference noise value or less.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a method for melt cutting a material to be melt cut, and to a method for producing a steel slab using the method. [Background technology]

[0002] In steelworks, cast steel pieces such as slabs and billets are transported to the rolling process after passing through a process in which impurities and defects such as cracks, pinholes, and scabs on the surface of the steel piece are thermochemically removed using combustion gas and oxygen to maintain high quality. If the amount of slag removed per unit time is large, it may not be possible to completely blow it off with high-pressure jet water. In such cases, the slag remaining on the surface of the steel piece may come into contact with moisture and cause a steam explosion, or slag ingots may adhere to the neck of the lift roll or remain on the surface of the steel piece, causing quality defects in subsequent processes.

[0003] As a method for preventing the slag from remaining on the billet, for example, the method shown in Patent Document 1 is known. The bumping prevention method disclosed in Patent Document 1 automatically adjusts the height of a high-pressure jet that washes away the molten iron so that it can follow the shape of the billet, thereby washing away the slag and preventing bumping.

[0004] Patent Document 2 discloses a control method for performing appropriate cutting. In this method, a machine learning model is used to determine cutting conditions for performing cutting at a target cutting amount before cutting, thereby ensuring a stable cutting amount. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2008-212988 A [Patent Document 2] JP 2020-157307 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, the conventional technology has the following problems. The technology disclosed in Patent Document 1 did not understand or experiment the state of bumping caused by the reaction of the high-pressure water used during the slag cutting with the slag cutting, so the effectiveness of the measures to prevent steam explosions was unknown. The technology disclosed in Patent Document 2 aims to control the amount of slag cutting, so measures to prevent bumping were not considered.

[0007] The present invention has been made in consideration of the above circumstances, and has an object to provide a technique capable of suppressing bumping that accompanies the melt-cutting of a material to be melt-cut. [Means for solving the problem]

[0008] The method for scalding a material to be scalded according to the present invention, which advantageously solves the above problem, includes: a prediction value calculation step of inputting scalding material information, including the surface temperature of the material to be scalded, obtained before scalding for the material to be scalded, and scalding conditions, including an arbitrarily set conveying speed and scalding oxygen pressure, into a machine learning model to calculate a predicted scalding amount and a predicted noise value of the material to be scalded; a determination step of calculating a difference between the predicted scalding amount and a preset target scalding amount, and repeatedly executing the prediction value calculation step by changing either one or both of the conveying speed and the scalding oxygen pressure until the difference becomes equal to or less than an allowable value and the predicted noise value becomes equal to or less than a preset reference noise value; and a scalding step of scalcing the material to be scalded while conveying it at the conveying speed and scalding oxygen pressure that are set when the difference between the predicted scalding amount and the target scalding amount becomes equal to or less than an allowable value and the predicted noise value becomes equal to or less than the reference noise value.

[0009] The method for sintering a workpiece according to the present invention includes the steps of: (a) the information on the material to be scalded includes the surface temperature, component value, thickness, length and width of the material to be scalded, and the scalding conditions include the scalding oxygen pressure, the transport speed of the material to be scalded, the pressure and flow rate of high-pressure jet water for removing slag, and the scalding preheating time; (b) The machine learning model is a model that has been created in advance by adjusting the weighting coefficients of each explanatory variable, using the information on the material to be scalded and the scalding conditions contained in the past performance data as explanatory variables, and the amount of scalding and the noise value as objective variables, so that the error between the predicted value and the actual value of the amount of scalding is small, and the error between the predicted value and the actual value of the noise value is small; (c) further including a thickness measurement step of measuring the thickness of the workpiece before and after the laser cutting step using a thickness measurement device, a noise measurement step of measuring noise during the laser cutting step using a sound level meter, and a model update step of updating the machine learning model by sequentially learning the machine learning model using an actual value of the laser cutting amount of the workpiece calculated from the difference in thickness before and after the laser cutting and an actual noise value measured during the laser cutting as objective variables, and using information on the workpiece that has been laser cut and laser cutting conditions as explanatory variables; This may be a more preferable solution.

[0010] The method for producing steel pieces according to the present invention, which advantageously solves the above-mentioned problems, includes a carrying-in step of transporting steel pieces to a laser cutting device, a charging step of charging the steel pieces to a processing position within the laser cutting device, and a first laser cutting step of laser cutting the surface of the steel pieces charged in the charging step, and further optionally includes a photographing step of photographing the laser-cut surface of the steel pieces, and, if necessary, a grinding step of automatically grinding areas of the steel piece surface that need treatment, which are obtained from the surface photograph taken in the photographing step using a grinding cart and a reverse traverser, and is characterized in that, after the processing of the steel pieces is completed, when the laser cutting work is repeatedly performed from the carrying-in step on the next steel piece to be processed, when the surface of the steel piece is laser-cut in the first laser cutting step, laser cutting is performed under laser cutting conditions determined according to any of the laser cutting methods for the material to be laser-cut described above.

[0011] In addition, in the method for producing steel pieces according to the present invention, the laser cutting operation further includes a re-loading step of automatically inverting or rolling the steel pieces using a rolling machine and automatically pulling the steel pieces back to the processing position as necessary, and a second laser cutting step of laser cutting the uncut surface of the steel pieces, and further, optionally, applying the photographing step and, if necessary, the grinding step to the unphotographed laser-cut surface of the steel pieces, automatically flipping or rolling the steel pieces as necessary, and repeating the re-loading step through the grinding step until there are no more surfaces that need to be laser cut, a more preferable solution may be to perform laser cutting under laser cutting conditions determined according to any of the above-mentioned laser cutting methods for the material to be laser cut when laser cutting the surface of the steel pieces in the second laser cutting step. Effect of the Invention

[0012] According to the method for melt-cutting a material to be melt-cut and the method for manufacturing a steel billet of the present invention, it is possible to suppress the frequency and scale of bumping during melt-cutting work. It is also effective in preventing explosion defects in the material to be melt-cut due to large-scale steam explosions. Therefore, it is possible to stably manufacture safe, high-quality melt-cut steel billets, which can improve productivity and are therefore industrially useful. [Brief description of the drawings]

[0013] [Figure 1] 1 is an explanatory diagram showing a schematic configuration of a slag cutting facility suitable for use in a slag cutting method for a workpiece according to one embodiment of the present invention; [Diagram 2] FIG. 2 is a functional block diagram showing the functional configuration of a calculation device used in the embodiment. [Diagram 3] FIG. 2 is a flow diagram illustrating a method for melt-cutting a workpiece according to the embodiment. [Figure 4] 1 is a graph showing the relationship between the actual noise level during surface cutting work and the maximum scattering distance of molten iron during bumping. [Diagram 5] This is a photographic image showing the state of the surface of a steel piece after laser cutting when large bumping occurred with a noise level of 96 dB. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, the embodiments of the present invention will be described in detail. Note that the drawings are schematic and may differ from the actual ones. The following embodiments are merely examples of equipment and methods for embodying the technical idea of ​​the present invention, and are not intended to limit the configuration to the following. In other words, the technical idea of ​​the present invention can be modified in various ways within the technical scope described in the claims.

[0015] [1. Equipment configuration] An equipment configuration suitable for application to a method for melt-cutting a material to be melt-cut according to one embodiment of the present invention will be described with reference to Fig. 1. Fig. 1 is an explanatory diagram showing a schematic configuration of melt-cutting equipment 100 according to this embodiment.

[0016] The slag cutting equipment 100 according to this embodiment is equipment for removing defects on the surface of a material to be slag cut, such as a steel billet, by slag cutting. In the slag cutting equipment 100, as shown in Fig. 1, the rotation speed of a plurality of slag cutting material transport rolls 6 arranged on a line is controlled, and the surface of the material to be slag cut is slag cut by a slag cutting device 9 while the material to be slag cut 7 is transported at a predetermined transport speed.

[0017] In this embodiment, the process computer 1 instructs the calculation device 2 on the target amount of slag cutting. The calculation device 2 instructs the slag cutting conditions to the slag cutting device 9 and the conveying speed control device 6A, respectively. The conveying speed control device 6A controls the rotation speed of the slag cutting material conveying roll 6 so that the slag cutting material 7 is conveyed at the instructed conveying speed. The slag cutting device 9 injects fuel gas and oxygen onto the surface of the slag cutting material 7, ignites it, and first preheats the slag cutting material 7 with the heat. After preheating, high-pressure oxygen for slag cutting is injected and sprayed onto the surface of the slag cutting material 7, and slag cutting is performed on one or more surfaces of the slag cutting material 7 by utilizing the heat generated by the continuous oxidation reaction between the molten iron on the surface of the slag cutting material 7 and the oxygen.

[0018] The position of the slag-cutting unit of the slag-cutting device 9 is adjusted according to the thickness, length, and width of the material to be slag-cut, and in response to instructions from the calculation device 2, the fuel gas and oxygen are controlled to slag-cut the surface of the material to be slag-cut 7. The slag-cutting oxygen pressure of the slag-cutting device 9 and the rotation speed (conveying speed) of the material to be slag-cut transport roll 6 are determined by the calculation device 2.

[0019] Thickness gauges 3 and 4 are installed on the upstream and downstream sides of the slag cutting device 9 in the conveying direction FD, respectively. The thickness gauges 3 and 4 as thickness measuring devices are arranged to sandwich the material 7 to be slag cut from above and below, and can be configured as devices that measure the distance to the front and back surfaces of the material 9 to be slag cut. Examples of such devices include a laser distance meter, an ultrasonic distance meter, and an eddy-current distance meter. The thickness of the material 7 to be slag cut measured by the thickness gauges 3 and 4 is sent to the calculation device 2 as thickness data, and the actual amount of slag cutting is calculated from the difference in thickness before and after slag cutting.

[0020] A thermometer 5 is installed upstream of the cutting device 9 in the transport direction FD to measure the surface temperature of the material to be cut. A contact thermocouple or a non-contact radiation thermometer can be used as the thermometer 5. An image sensor may also be used. The thermometer 5 can be installed above the line so that it can measure the center of the material to be cut in the width and length directions, for example. The surface temperature measured by the thermometer 5 is sent to the calculation device 2 as surface temperature data of the material to be cut 7.

[0021] A sound level meter 8 is installed around the slag cutting device 9. The sound level during slag cutting is measured by the sound level meter 8 and transmitted to the calculation device 2. The sound level is preferably measured by the sound level meter 8 installed indoors at a predetermined distance from the slag cutting device 9. The sound level meter 8 converts the measured sound pressure into an A-weighted sound pressure level (dB) as specified in JIS Z 8731:2019 and outputs it as a sound level (dB). The A-weighted sound pressure (Pa) refers to the sound pressure (Pa) measured by multiplying it by the frequency weighting characteristic A (see Annex JA) as specified in JIS C 1509-1. The sound pressure level (dB) is calculated as 10 times the common logarithm of the value obtained by dividing the square of the effective value of the sound pressure by the square of the reference sound pressure. Although the Z-weighted sound pressure level can also be used, it is preferable to use the A-weighted sound pressure level, which is close to the hearing of the worker, as the evaluation index. The noise level is also called the noise value.

[0022] The amount of scalding of the material 7 to be scalded by the scalding device 9 is controlled by changing and adjusting either or both of the conveying speed of the material 7 to be scalded and the oxygen pressure for scalding. The conveying speed and the oxygen pressure for scalding during scalding are calculated by the calculation device 2. The calculation device 2 determines the conveying speed and the oxygen pressure for scalding of the material 7 to be scalded using a machine learning model based on information acquired from the process computer 1, which is a higher-level command device, and various measuring instruments. The determined conveying speed of the material 7 to be scalded is output to the conveying speed control device 6A, which controls the rotation speed of the material conveying roll 6. The determined oxygen pressure for scalding is output to the scalding device 9. The conveying speed of the material 7 to be scalded is controlled based on the value determined by the calculation device 2, and the material 7 to be scalded is scalded by the scalding device 9.

[0023] [2. Measurement data, etc.] The calculation device 2 receives input of thickness data of the material 7 to be scalculated before and after scalculation, measured by the thickness gauges 3 and 4. The calculation device 2 converts the thickness data into the amount of scalculation. The surface temperature measured by the thermometer 5 is input to the calculation device 2 as surface temperature data of the material 7 to be scalculated. The noise level measured by the sound level meter 8 is input to the calculation device 2 as noise level data during scalculation of the material 7 to be scalculated. The calculation device 2 receives input of the target amount of scalculation of the material 7 to be scalculated from the process computer 1, as well as information including the component composition of the material 7 to be scalculated, its size such as length and width, the preheating temperature and preheating time of the material 7 to be scalculated, the conveying speed and scalculation oxygen pressure of the material 7 to be scalculated, and the pressure and flow rate of the high-pressure jet water for removing scalculation slag, as performance data, for example, from a performance database 1A provided in the process computer 1. Here, the calculated amount of swarf cutting, the measured surface temperature of the material 7 to be swarfed, its component composition, and its size such as length and width are defined as material to be swarf cut information, and the conveying speed, swarf cutting oxygen pressure, measured noise level, high-pressure jet water pressure and flow rate, preheating temperature and preheating time are defined as swarf cutting conditions. These performance data are collected in the storage medium 10 of the calculation device 2 shown in FIG.

[0024] [3. Arithmetic device] The functional configuration of the arithmetic device 2 according to this embodiment will be described with reference to Fig. 2. Fig. 2 is a functional block diagram showing the functional configuration of the arithmetic device 2 according to this embodiment.

[0025] The calculation device 2 according to this embodiment is a device that has a machine learning model and outputs operation instructions to each device of the slag cutting equipment 100. It has a function of calculating a predicted slag cutting amount and a predicted noise value from the workpiece information and the set slag cutting conditions, and determining appropriate slag cutting conditions for the workpiece 7. The calculation device 2 can be configured as a computer equipped with a CPU. The calculation device 2 may have other functions such as an input unit, a display unit, and an output unit, which are not shown.

[0026] As shown in Figure 2, the calculation device 2 of this embodiment has a storage medium 10, a machine learning model creation unit (machine learning model update unit) 11, a predicted value calculation unit 12, a cutting condition determination unit 13, and a cutting condition change unit 14.

[0027] The machine learning model creation unit 11 creates a machine learning model for calculating a predicted amount of the workpiece 7 to be sculpted and a predicted noise value during sculpting, based on past performance data.

[0028] The machine learning model is created by adjusting the weighting coefficients of each explanatory variable, with the above-mentioned information on the workpiece material and the slag cutting conditions of the performance data as explanatory variables, and the amount of slag cutting and the noise value as objective variables, so that the error between the predicted value and the actual value of the amount of slag cutting is small, and the error between the predicted value and the actual value of the noise value is small. When creating the machine learning model, the prediction accuracy of the objective variable calculated by the model can be improved by using parameters that affect each objective variable as explanatory variables.

[0029] For example, the information on the material to be cut as explanatory variables that affect the amount of swarfing includes the surface temperature, composition, thickness, width, and length of the material to be cut 7. The surface temperature of the material to be cut 7 affects the amount of energy to be applied to the material to be cut 7. The composition is related to the melting point of the material to be cut 7, and the amount of energy to be applied to the material to be cut 7 changes depending on the melting point. For example, carbon equivalent can be used as an index that represents the melting point due to the composition. In addition, the width of the material to be cut 7 represents the effect on the amount of swarfing due to the shape of the material to be cut 7 that has a different thickness in the width direction, such as a dogbone shape. The length affects the ease of heat transfer during preheating. The shorter the material to be cut 7, the greater the effect on the ease of heat transfer during preheating. The thickness affects the temperature drop between the start and end of swarfing.

[0030] For example, the cutting conditions as explanatory variables that affect the amount of slag cutting include the slag cutting oxygen pressure, the transport speed of the material 7 to be slag cut, and the preheating time. The slag cutting oxygen pressure and the transport speed of the material 7 to be slag cut are indicators that control the energy applied per unit area of ​​the surface of the material 7 to be slag cut. The preheating time refers to the time during which preheating is performed at the start of slag cutting, in which oxygen gas is sprayed for a certain period of time against the cutting start position of the material 7 to be slag cut to create a pool of molten steel. The preheating time is a value empirically determined depending on the type of steel and the surface temperature of the material 7 to be slag cut, and is an indicator that controls the amount of molten steel that is created during preheating.

[0031] For example, explanatory variables that affect noise levels include the cutting oxygen pressure, high-pressure jet water pressure and flow rate, the amount of material 7 to be cut, the transport speed, and preheating time. The amount of material 7 to be cut and the transport speed represent the amount of molten iron generated per unit time and unit width. As mentioned above, the preheating time affects the amount of molten iron at the beginning of cutting. The cutting oxygen pressure and high-pressure jet water conditions affect the ability to remove molten iron per unit time and unit width. If the ability to remove the molten iron that has been generated is insufficient, the remaining molten iron may react with the jet water, causing bumping or steam explosions. These are the causes of noise.

[0032] In the machine learning model creation unit 11, for example, a regression model or deep learning may be used to generate a model. The regression model may be used to predict the amount of swarf cutting, and the deep learning may be used to predict the noise level. The model may be selected arbitrarily.

[0033] Moreover, the machine learning model creation unit 11 preferably has a function of updating the machine learning models. The updating of each machine learning model may be performed, for example, every time one piece of material 7 to be cut is cut. Each time performance data is accumulated in the storage medium 10, the machine learning model creation unit 11 sequentially updates each machine learning model using the performance data. This makes it possible to improve the prediction accuracy of the amount of cutting and the noise value, which are the objective variables of each machine learning model, from the latest information.

[0034] The predicted value calculation unit 12 has a predicted cutting amount calculation function and a predicted noise value calculation function. The predicted cutting amount calculation function of the predicted value calculation unit 12 calculates a predicted cutting amount of the material 7 to be cut using a machine learning model created in advance by the machine learning model creation unit 11. The predicted cutting amount calculation function inputs the surface temperature of the material 7 to be cut measured by the thermometer 5 into the machine learning model, and calculates a predicted cutting amount of the material 7 to be cut when cutting is performed with the cutting oxygen pressure and the transport speed of the material 7 to be cut that are to be set.

[0035] The predicted noise value calculation function of the prediction value calculation unit 12 calculates a predicted noise value during cutting of the material 7 to be cut, which is the target of cutting, using a machine learning model created in advance by the machine learning model creation unit 11. The predicted noise value calculation function inputs the predicted cutting amount calculated by the predicted cutting amount calculation function and the cutting conditions used for the calculation into the machine learning model, and calculates a predicted noise value when cutting is performed at the planned cutting oxygen pressure and transport speed of the material 7 to be cut.

[0036] The slag cutting condition determination unit 13 calculates the difference between the predicted slag cutting amount calculated by the predicted slag cutting amount calculation function of the predicted value calculation unit 12 and the target slag cutting amount, and determines whether the difference is within the allowable value. The target slag cutting amount is determined in advance for each workpiece 7 by the process computer 1. The slag cutting condition determination unit 13 determines whether the predicted noise value calculated by the predicted noise value calculation function of the predicted value calculation unit 12 is equal to or lower than the reference noise value. The reference noise value is determined in advance based on the positional relationship between the slag cutting device 9 and the sound level meter 8 and the relationship with bumping. When the slag cutting condition determination unit 13 determines that the difference between the predicted slag cutting amount and the target slag cutting amount is equal to or lower than the allowable value and that the predicted noise value is equal to or lower than the reference noise value, the slag cutting material 7 is slag cut under the slag cutting conditions at that time. The slag cutting condition determination unit 13 outputs the slag cutting oxygen pressure at that time to the slag cutting device 9. The slag cutting condition determination unit 13 outputs the conveying speed at that time to the conveying speed control device 6A. If the difference between the predicted cutting amount and the target cutting amount is greater than the allowable value, or if the predicted noise value exceeds the reference noise value, the cutting conditions are changed by the cutting condition changing unit 14 and the respective amounts are predicted again.

[0037] The cutting condition change unit 14 changes the cutting conditions when the cutting condition determination unit 13 determines that the difference between the predicted cutting amount and the target cutting amount is greater than the allowable value or that the predicted noise value exceeds the reference noise value. The cutting condition change unit 14 changes one or both of the conveying speed and the cutting oxygen pressure among the cutting conditions.

[0038] For example, in order to suppress the noise level with the same amount of slag cutting, it is necessary to lower the slag cutting oxygen pressure and slow down the conveying speed.

[0039] The cutting condition changing unit 14 outputs the changed values ​​of the cutting conditions to the predicted value calculation unit 12. Based on the cutting conditions changed by the cutting condition changing unit 14, the predicted cutting amount calculation function and the predicted noise value calculation function of the predicted value calculation unit 12 again calculate the predicted cutting amount and the predicted noise value, respectively.

[0040] [4. Method of cutting the material to be cut] The method for melt-cutting a material to be melt-cut according to this embodiment will be described with reference to Fig. 3. Fig. 3 is a flow diagram showing the method for melt-cutting a material to be melt-cut according to this embodiment.

[0041] <s1>Start calculation Before starting the cutting of the workpiece 7, as a pre-processing step, a machine learning model is generated by the machine learning model creation unit 11 of the computing device 2. As described above, the machine learning model creation unit 11 uses the past cutting performance data acquired in the storage medium 10 to create and update each machine learning model by adjusting the weight coefficients, etc., of the explanatory variables so as to reduce the error between the predicted value and the performance value of the objective variable.

[0042] At the start of the cutting operation, information on the material 7 to be cut is input from the process computer 1 to the predicted value calculation unit 12 of the arithmetic device 2. At this time, a target amount of the material 7 to be cut is given (S2).

[0043] <s3>Surface temperature measurement of material to be cut Thermometer 5 outputs the measured surface temperature of material 7 to be cut to calculation device 2. When calculation device 2 receives the surface temperature of material 7 to be cut measured by thermometer 5, it calculates the preheating time. The preheating time is determined by the surface temperature and component composition of material 7 to be cut, that is, the melting point. In addition, thickness gauge 3 outputs the measured thickness of the material to be cut before cutting to calculation device 2.

[0044] <s4>Initial setting of cutting conditions The calculation device 2 is provided with the initial value of the transport speed and the initial value of the oxygen pressure for cutting, in addition to the information on the material to be cut from the process computer 1 (S4). These may be the cutting conditions for the immediately preceding material to be cut 7 that has a similar component composition, thickness, length, width, etc., or may be the average cutting conditions for similar materials to be cut 7 in the past.

[0045] <s5>Calculation of predicted cutting volume and predicted noise value The calculation device 2 calculates a predicted cutting amount using a predetermined machine learning model based on the acquired surface temperature of the workpiece 7, the thickness of the workpiece, and the workpiece information and cutting conditions input from the process computer 1, using a predicted cutting amount calculation function of the prediction value calculation unit 12. Next, the calculation device 2 calculates a predicted noise value using a predetermined machine learning model based on the calculated predicted cutting amount, the workpiece information and cutting conditions used in the calculation, using a predicted noise value calculation function of the prediction value calculation unit 12 (S5).

[0046] <s6>Judging the predicted amount of swarf The calculation device 2 calculates the difference between the predicted amount of swarf cutting and the target amount of swarf cutting by the swarf cutting condition determination unit 13, and determines whether or not the difference is within a predetermined allowable value (threshold value) (S6). The allowable value as the threshold value can be set based on, for example, the measurement error of the thickness gauge used to measure the amount of swarf cutting, the error between the set value and the actual value of the conveying speed of the material to be swarfed 7, etc.

[0047] <s7>Change of cutting conditions In step S6, if the difference between the predicted cutting amount and the target cutting amount exceeds the allowable value, the cutting condition change unit 14 changes, for example, the conveying speed (S7), and then the calculation of the predicted cutting amount and predicted noise value in step S5 is performed again.

[0048] In order to bring the predicted amount of swarf cutting closer to the target amount of swarf cutting, either the conveying speed or the swarf cutting oxygen pressure or both are changed. For example, it is preferable to fix the swarf cutting oxygen pressure and change the conveying speed. In this case, in step S7, the conveying speed is increased or decreased by a predetermined change amount depending on the magnitude relationship between the predicted amount of swarf cutting and the target amount of swarf cutting. This change amount of the conveying speed can be set, for example, to an integer multiple of the minimum unit rotation speed of the material-to-be-swarfed conveying roll 6 that can be changed by the conveying speed control device 6A. For example, when the predicted amount of swarf cutting is larger than the target amount of swarf cutting by a tolerance, the conveying speed is increased by a predetermined change amount. Also, when the predicted amount of swarf cutting is smaller than the target amount of swarf cutting by a tolerance, the conveying speed is decreased by a predetermined change amount.

[0049] The processes of steps S5 to S7 are repeatedly executed until the difference between the predicted cutting amount and the target cutting amount falls within the allowable value in the determination process of step S6.

[0050] <s8>Judging predicted noise level When it is determined in the determination process of step S6 that the difference between the predicted cutting amount and the target cutting amount is within the allowable value, the calculation device 2 determines whether or not the predicted noise value is equal to or lower than the reference noise value by the cutting condition determination unit 13 (S8). The reference noise value can be set as a noise level at which the magnitude of bumping is equal to or lower than a predetermined value, based on the relationship between the magnitude of bumping during cutting and the actual noise level, as shown in Fig. 4, for example. In the example of Fig. 4, the noise level is set to 80 dB as the reference noise value. The reference noise value varies depending on the distance between the cutting device 9 and the sound level meter 8.

[0051] <s7>Change of cutting conditions In step S8, if the predicted noise value exceeds the reference noise value, the cutting condition change unit 14 changes either or both of the conveying speed and the cutting oxygen pressure. For example, after the cutting oxygen pressure and the conveying speed are changed (S7), the calculation of the predicted cutting amount and the predicted noise value in step S5 is executed again. For example, in order to reduce the predicted noise value while maintaining the predicted cutting amount, the cutting oxygen pressure and the conveying speed are each reduced by a predetermined change amount. For example, the change amount of the cutting oxygen pressure can be an integer multiple of the smallest measurement unit of the measuring device for the cutting oxygen pressure. The change amount of the conveying speed is set to a magnitude that allows the predicted cutting amount to be maintained in relation to the change amount of the cutting oxygen pressure.

[0052] The processes of steps S5 to S8 are repeatedly executed until the difference between the predicted cutting amount and the target cutting amount falls within the allowable value and the predicted noise value falls below the reference noise value in the determination processes of steps S6 and S8.

[0053] <s9>Setting the cutting conditions and carrying out the cutting If it is determined in the determination processes of steps S6 and S8 that the difference between the predicted amount of swarf cutting and the target amount of swarf cutting is within the allowable value and that the predicted noise value is equal to or less than the reference noise value, the swarf cutting condition determination unit 13 determines to swarf the material 7 to be swarf cut at the conveying speed of the material 7 to be swarf cut that is set at that time. The conveying speed control device 6A rotates the material to be swarf cut conveying roll 6 so that the material 7 to be swarf cut is conveyed at the determined conveying speed. The conveyed material to be swarf cut is swarf cut by the swarf cutting device 9 at the determined swarf cutting oxygen pressure (S9). The thickness gauge 4 outputs the measured thickness of the material 7 to be swarf cut after swarf cutting to the calculation device 2. When swarf cutting of the surface of the material 7 to be swarf cut is completed, the calculation ends (S10).

[0054] <Update of machine learning model> After the cutting operation in step S9 is completed, the machine learning model may be updated.

[0055] <Steps for updating the model for calculating the predicted amount of swarf> The predicted cutting amount calculation model update step of updating the predicted cutting amount calculation process of the machine learning model is executed as follows.

[0056] The thickness of the material 7 to be cut measured by the thickness gauges 3 and 4 before and after cutting the material 7 to be cut is obtained as actual thickness data, and the difference is calculated as an actual cutting amount value and stored in the storage medium 10. The machine learning model creation unit 11 sequentially learns the predicted cutting amount calculation process of the machine learning model based on the actual cutting amount value of the material 7 to be cut, the material to be cut information of the material 7 to be cut, and the set cutting conditions. That is, the machine learning model creation unit 11 corrects the weighting coefficients for each parameter in the predicted cutting amount calculation process of the machine learning model based on the material to be cut information and cutting conditions of the material 7 to be cut so that the difference between the predicted cutting amount value calculated in the predicted cutting amount calculation process of the machine learning model and the actual cutting amount value of the material 7 to be cut is reduced. When the material 7 to be cut is next cut, the predicted cutting amount is calculated using the predicted cutting amount calculation process of the machine learning model updated by sequential learning.

[0057] <Predicted noise value calculation model update step> The predicted noise value calculation model update step for updating the predicted noise value calculation process of the machine learning model is executed as follows.

[0058] In addition to the performance information of the predicted cutting amount calculation model update step, the performance noise value measured by the sound level meter 8 during cutting is stored in the storage medium 10. The machine learning model creation unit 11 sequentially learns the predicted noise value calculation process of the machine learning model based on the performance value of the cutting amount of the cut material 7, the performance noise value during cutting, the cut material information of the cut material 7, and the set cutting conditions. That is, the machine learning model creation unit 11 corrects the weighting coefficients for each parameter in the predicted noise value calculation process of the machine learning model based on the performance value of the cutting amount of the cut material, the cut material information of the cut material 7, and the cutting conditions so that the difference between the predicted noise value calculated in the predicted noise value calculation process of the machine learning model and the actual noise value becomes smaller. The next time the cut material 7 is cut, the predicted noise value is calculated using the predicted noise value calculation process of the machine learning model updated by sequential learning.

[0059] [5. Manufacturing method of steel billets] Next, a method for manufacturing a slab according to another embodiment of the present invention will be described. The method for manufacturing a slab according to this embodiment includes a slab cutting operation using a slab cutting device, and includes a carry-in process, a charging process, and a first slab cutting process, and further includes, optionally, a photographing process and a grinding process that is performed as necessary. After the slab cutting process of the previous slab is completed, the slab cutting process is repeated from the carry-in process on the next slab to be processed. Furthermore, the slab cutting process includes, as necessary, a recharging process and a second slab cutting process.

[0060] In the carrying-in process, the steel billet cast by continuous casting or the like is transported to a machine scarf as a swarf cutting device. For example, the steel billet is transported onto a turntable of the machine scarf.

[0061] In the charging step, for example, the steel pieces on the turntable are charged to a processing position in the machine scarf.

[0062] The laser cutting process includes an ignition step, a preheating step, and a laser cutting step. In the ignition step, for example, combustible gas is supplied from a scarfer unit, and the combustible gas is ignited by an ignition device. Next, in the preheating step, a part of the surface of the steel piece is melted by the heat of the burning combustible gas to form a molten pool, which serves as a heat source. Next, in the laser cutting step, oxygen gas is supplied toward the molten pool, and the surface of the steel piece is melted to a depth of about 1 to 4 mm by the heat of the oxidation reaction between the oxygen gas and iron, thereby removing surface defects of the steel piece.

[0063] In this scouring process, the scouring method for the material to be scouring is applied, and scouring is carried out so as to meet the target scouring amount and the reference noise value of the steel billet as the material to be scouring.

[0064] In the photographing process, the surface of the steel billet that has been machined is photographed. The presence or absence of surface defects is then judged, and further machining or the next grinding process is decided. This photographing process is preferably applied to steel types that require strict surface quality.

[0065] In the grinding process, based on the judgment in the photographing process, the grinding carriage and the inverted traverser are used to automatically grind the areas on the steel billet surface that need to be treated, which are obtained from the surface photographs taken in the photographing process, thereby removing surface defects.

[0066] In the laser cutting step, the entire circumference of the billet can be laser cut in one go. The billet can also be laser cut one side at a time or multiple sides at the same time. In the recharging process, a turnover machine is used to automatically turn or roll the billet in order to laser cut the sides of the billet that were not processed in the first laser cutting process, and the billet is automatically pulled back to the processing position again.

[0067] The second scouring process scouring the unscoated surface of the steel piece. Each processing step is the same as the first scouring process. Then, optionally, a photographing process and, if necessary, the grinding process are applied.

[0068] Furthermore, the recharging step to the grinding step are repeated until there is no more surface of the steel billet that needs to be ground.

[0069] As described above, the method of scouring the material to be scouring and the method of manufacturing a steel billet using this method as the scouring process make it possible to accurately control the amount of scouring, improve work safety, and achieve a synergistic effect of increasing productivity. EXAMPLES

[0070] Using the equipment configuration shown in Figure 1, steel slabs with thicknesses of 235 mm and 260 mm, widths of 1000 to 1625 mm, and lengths of 6.10 to 8.93 m were slag-cut. The target slag-cut amount was 2.3 mm. The slag-cutting oxygen pressure was set to 230 Pa in gauge pressure. The conveying speed was changed in the range of 9.6 to 16.5 m / min. The noise was measured using a sound level meter installed in a building about 20 m away from the slag-cutting equipment. Figure 4 shows the relationship between the maximum molten iron scattering distance (m) during bumping and the noise level (dB) at that time. The noise level exceeds 80 dB and the maximum molten iron scattering distance during bumping is 5 m or more, which is "bumping." Figure 5 shows an image of the surface of the slag-cut steel piece when "large bumping" with a noise level of 96 dB occurred. A defect D caused by slag-cutting slag is observed on the surface of the steel piece. When such abnormalities in the surface properties of steel billets occur, offline grinding processing may be required, which may hinder productivity and increase the risk of damage to surrounding equipment due to scattered molten iron.For example, by setting the grinding conditions to a noise level of 80 dB or less, it was possible to improve the quality of the steel billet surface, safety, and productivity. [Explanation of symbols]

[0071] 100 Cutting equipment 1. Process Computer 1A Performance Database 2 Arithmetic unit 3 (Before shaving) Thickness gauge 4 (After cutting) Thickness gauge 5 Thermometer (workpiece surface temperature measuring device) 6. Carrying roll for material to be cut 6A Conveyor speed control device 7 Material to be machined 8. Sound level meter 9. Scaling equipment (machine scarf) 10 Storage medium 11 Machine learning model creation department (machine learning model update department) 12 Prediction value calculation section 13 Cutting condition judgment section 14. Cutting condition change section FD Transport direction D (Steel surface) Defect

Claims

1. A predicted value calculation step of inputting information on the material to be cut, including the surface temperature of the material to be cut, acquired before cutting, and cutting conditions, including an arbitrarily set conveying speed and cutting oxygen pressure, into a machine learning model to calculate a predicted cutting amount and a predicted noise value of the material to be cut; a determination step of calculating a difference between the predicted amount of swarf cutting and a preset target amount of swarf cutting, and repeatedly executing the predicted value calculation step by changing either one or both of the conveying speed and the swarf cutting oxygen pressure until the difference becomes equal to or smaller than a tolerance value and the predicted noise value becomes equal to or smaller than a preset reference noise value; a cutting step of cutting the workpiece while conveying it at a conveying speed and a cutting oxygen pressure that are set when the difference between the predicted cutting amount and the target cutting amount becomes equal to or less than a tolerance value and the predicted noise value becomes equal to or less than the reference noise value; A method for fusion-cutting a material to be fused, comprising the steps of:

2. The information on the workpiece includes a surface temperature, a component value, a thickness, a length, and a width of the workpiece, 2. The method for thermally cutting a material to be thermally cut according to claim 1, wherein the thermal cutting conditions include a thermal cutting oxygen pressure, a transport speed of the material to be thermally cut, a pressure and flow rate of high-pressure jet water for removing thermal cutting slag, and a thermal cutting preheating time.

3. 2. The method for melt-cutting a material to be melt-cut according to claim 1, wherein the machine learning model is a model that has been created in advance by using information on the material to be melt-cut and the melt-cutting conditions contained in past performance data as explanatory variables, and the amount of melt-cutting and the noise value as objective variables, and adjusting the weighting coefficients, etc. of each explanatory variable so that the error between the predicted value and the actual value of the amount of melt-cutting is small, and the error between the predicted value and the actual value of the noise value is small.

4. a thickness measuring step of measuring the thickness of the workpiece before and after the cutting step by a thickness measuring device; a noise measuring step of measuring noise during the cutting step using a sound level meter; a model updating step of updating the machine learning model by sequentially learning the machine learning model using an actual value of the amount of the material to be scalded calculated from the difference in thickness before and after the scalding and an actual noise value measured during the scalding as objective variables, and information on the material to be scalded and scalding conditions of the scalded material as explanatory variables; The method for melt-cutting a material to be melt-cut according to claim 3, further comprising:

5. a loading process for transporting the steel billet to a sintering device; a charging step of charging the steel piece to a processing position in the swarf cutting device; a first spalling process for spalling a surface of the steel billet charged in the charging process; Including, Optionally, a photographing step of photographing the surface of the steel piece that has been machined; a grinding step of automatically grinding areas of the steel billet surface that need to be treated, the areas being obtained from the surface photograph taken in the photographing step, using a grinding carriage and a reversing traverser, as required; After the processing of the steel piece is completed, the next steel piece to be processed is repeatedly subjected to the processing from the carrying-in step, A method for manufacturing a steel slab, wherein, during the first laser cutting step, the surface of the steel slab is laser cut under laser cutting conditions determined according to the method for laser cutting a material to be laser cut according to any one of claims 1 to 4.

6. The above-mentioned cutting operation further comprises: a recharging step of automatically reversing or turning the steel piece over using a turnover machine as necessary, and automatically pulling the steel piece back to the treatment position again; a second laser cutting step of laser cutting the unlaser-cut surface of the steel piece; Including, and optionally applying said photographing step and, optionally, said grinding step to an unphotographed surface of said piece of steel, The steel piece is automatically inverted or turned over as necessary, and the steps from the recharging step to the grinding step are repeated until no surface is left that needs to be ground; The method for producing a steel billet according to claim 5, wherein, when the surface of the steel billet is laser-cut in the second laser-cutting step, the laser-cutting is performed under laser-cutting conditions determined in accordance with the method for laser-cutting a material to be laser-cut according to any one of claims 1 to 4.

Images