Method for generating a nozzle spacing estimation model, method for estimating nozzle spacing, method for controlling the amount of hot-dip galvanized coating, method for manufacturing hot-dip galvanized steel strip, nozzle spacing estimation model generation device, nozzle spacing estimation device, and hot-dip galvanized coating amount control device.
A machine learning-based nozzle spacing estimation model accurately and efficiently determines the spacing between the gas wiping nozzle and steel strip surface, addressing calculation errors and time issues in conventional methods, thereby controlling the molten zinc plating on steel strips effectively.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- JFE STEEL CORP
- Filing Date
- 2024-02-26
- Publication Date
- 2026-05-15
AI Technical Summary
Conventional methods for controlling nozzle spacing in hot-dip galvanizing processes involve iterative calculations with errors and long computation times, leading to inaccuracies in estimating and adjusting the amount of molten zinc plating on steel strips.
A method using machine learning to generate a nozzle spacing estimation model that calculates the spacing between the gas wiping nozzle and the steel strip surface based on input data such as steel type, thickness, line speed, temperature, and gas pressure, enabling accurate and rapid determination of nozzle positioning.
The method allows for precise control of molten zinc plating on steel strips without calculation errors and in a shorter time, ensuring the amount of plating adheres to a target value during actual operation.
Smart Images
Figure 0007859454000002 
Figure 0007859454000003 
Figure 0007859454000004
Abstract
Description
Technical Field
[0001] The present invention relates to a method for generating a nozzle interval estimation model, a method for estimating a nozzle interval, a method for controlling the amount of molten zinc plating adhesion, a method for manufacturing a molten zinc plated steel strip, a nozzle interval estimation model generation device, a nozzle interval estimation device, and a molten zinc plating adhesion amount control device.
Background Art
[0002] In a continuous molten zinc plating line, generally, a steel strip annealed in a continuous annealing furnace in a reducing atmosphere passes through a snout and is introduced into a molten zinc bath in a plating bath. Then, the steel strip is pulled up above the molten zinc bath through a sink roll and a collector roll in the molten zinc bath. Thereafter, wiping gas is blown onto the surface of the molten zinc plated steel strip from gas wiping nozzles arranged on both sides of the molten zinc plated steel strip, and the excess molten zinc adhering to and pulled up on the surface of the molten zinc plated steel strip is scraped off, thereby adjusting the plating adhesion amount.
[0003] Here, the plating adhesion amount of the molten zinc plated steel strip is controlled by adjusting the nozzle interval between the surface of the molten zinc plated steel strip continuously taken out from the molten zinc bath and the tip of the gas wiping nozzle. As a method for controlling the plating adhesion amount by adjusting this nozzle interval, conventionally, for example, a plating adhesion amount control method shown in Patent Document 1, a plating adhesion amount control method shown in Patent Document 2, a plating adhesion amount control method shown in Patent Document 3, and a plating adhesion amount control method shown in Patent Document 4 have been proposed.
[0004] In the plating adhesion amount control method shown in Patent Document 1, the plating adhesion amount control device includes a plating adhesion amount prediction model, a control unit, and a steel plate path movement amount estimation unit. The plating adhesion amount prediction model describes the relationship between the plate speed, the nozzle pressure, the distance between the nozzle and the steel plate (nozzle interval), and the plating adhesion amount adhering to the steel plate, and the control unit controls at least one of the nozzle pressure and the nozzle position so that the plating adhesion amount adhering to the steel plate becomes a desired value by referring to the plating adhesion amount prediction model.
[0005] Furthermore, the plating adhesion amount control method described in Patent Document 2 uses a statistical method in which the operating conditions are used as explanatory variables and the corresponding plating adhesion amount is used as the objective variable to classify the plating adhesion amount into multiple groups according to the operating conditions, and the plating adhesion amount during actual operation is estimated based on this classification information, thereby controlling the plating adhesion amount of the steel strip.
[0006] Furthermore, in the plating adhesion amount control method shown in Patent Document 3, before performing the adhesion adjustment process on the subsequent material, a predicted value of the plating adhesion amount on the subsequent material is calculated using the setting operation information set for the subsequent material and the adhesion amount prediction model generated by the adhesion amount prediction model generation method. The adhesion amount prediction model is generated by machine learning using multiple sets of training data, which include input operation information data containing one or more data selected from steel sheet information relating to the steel sheet to be plated and one or more data selected from nozzle operation information relating to the wiping nozzle, and actual plating adhesion amount data using the input operation information data.
[0007] Furthermore, the plating adhesion amount control method shown in Patent Document 4 includes an adhesion amount prediction step that predicts the plating adhesion amount during actual operation by using a combination of a decision tree model that predicts the plating adhesion amount with operating conditions as explanatory variables and the corresponding plating adhesion amount as the objective variable, and a physical model based on physical phenomena. [Prior art documents] [Patent Documents]
[0008] [Patent Document 1] Japanese Patent Publication No. 2020-105623 [Patent Document 2] Japanese Patent Publication No. 2019-210535 [Patent Document 3] Japanese Patent Publication No. 2022-535 [Patent Document 4] Japanese Patent Publication No. 2020-139175 [Overview of the Initiative] [Problems that the invention aims to solve]
[0009] By the way, in conventional plating adhesion amount control methods such as those shown in Patent Documents 1 to 4, the plating adhesion amount is estimated using a machine learning model. Furthermore, regarding the nozzle spacing during actual operation, which acts as the actuator, conventionally, the movement of the gas wiping nozzle is calculated from the deviation between the estimated plating amount and the target plating amount, which is estimated by a machine learning model. Then, the nozzle spacing after calculating the movement is input into the machine learning model to estimate the plating amount, and the deviation between that estimated value and the target plating amount is recalculated. This process is repeated until the nozzle spacing that minimizes this deviation is derived, thereby deriving the nozzle spacing for actual operation. The initial position of the gas wiping nozzle before moving it is calculated using a physical formula that shows the relationship between the nozzle spacing and the plating amount.
[0010] However, conventional methods for deriving nozzle spacing involve iterative calculations of the deviation between estimated values of gas wiping nozzle movement and plating adhesion and target values of plating adhesion, as well as calculations using some physical formulas. This results in calculation errors and a large amount of computation time.
[0011] Accordingly, the present invention has been made to solve this conventional problem, and its objective is to provide a method for generating a nozzle spacing estimation model, a method for estimating the nozzle spacing, a method for controlling the amount of hot-dip galvanized coating, a method for manufacturing a hot-dip galvanized steel strip, a nozzle spacing estimation model generating device, a nozzle spacing estimation device, and a hot-dip galvanized coating amount control device, which can calculate the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel sheet in actual operation without calculation errors and in a short time. [Means for solving the problem]
[0012] To solve the above problems, a method for generating a nozzle spacing estimation model according to one aspect of the present invention involves using machine learning to generate a nozzle spacing estimation model that estimates the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip by taking the following as input data: the type of steel of the hot-dip galvanized steel strip continuously taken from the molten zinc bath, the thickness of the hot-dip galvanized steel strip, the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and actual values of the amount of plating deposited on the hot-dip galvanized steel strip. The output data is the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip, relative to the input data.
[0013] Furthermore, another aspect of the present invention relates to a method for estimating nozzle spacing, which involves inputting the following into a nozzle spacing estimation model generated by the method described above: the type of steel of the hot-dip galvanized steel strip continuously removed from the molten zinc bath during actual operation; the thickness of the hot-dip galvanized steel strip; the line speed of the hot-dip galvanized steel strip; the temperature of the hot-dip galvanized steel strip; the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip; and a target value for the amount of plating deposited on the hot-dip galvanized steel strip. The method is essentially based on estimating the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip during actual operation.
[0014] Furthermore, another aspect of the present invention relates to a method for controlling the amount of hot-dip galvanized coating, which involves controlling the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip so that the nozzle spacing is as estimated by the nozzle spacing estimation method described above, and then blowing a wiping gas from the gas wiping nozzle onto the hot-dip galvanized steel strip as it is continuously removed from the molten zinc bath to control the amount of plating that adheres to the hot-dip galvanized steel strip. Furthermore, another aspect of the present invention relates to a method for manufacturing a hot-dip galvanized steel strip, which includes a hot-dip galvanizing amount control step that controls the amount of galvanizing applied to the hot-dip galvanized steel strip using the aforementioned method for controlling the amount of galvanizing applied.
[0015] Furthermore, another aspect of the present invention relates to a nozzle spacing estimation model generation apparatus that takes the following as input data: the type of steel of the hot-dip galvanized steel strip continuously removed from the molten zinc bath, the thickness of the hot-dip galvanized steel strip, the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and the actual values of the amount of plating deposited on the hot-dip galvanized steel strip. The apparatus then uses machine learning to generate a nozzle spacing estimation model that estimates the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip, by using multiple training data sets, in which the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip is used as output data for this input data.
[0016] Furthermore, another aspect of the present invention provides a nozzle spacing estimation device that estimates the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip during actual operation by inputting the following values into the nozzle spacing estimation model generated by the nozzle spacing estimation model generation device: the type of steel of the hot-dip galvanized steel strip continuously taken out from the molten zinc bath during actual operation, the thickness of the hot-dip galvanized steel strip, the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and a target value for the amount of plating deposited on the hot-dip galvanized steel strip.
[0017] Furthermore, another embodiment of the present invention provides a hot-dip galvanizing deposition amount control device that controls the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip so that the nozzle spacing is the same as that estimated by the nozzle spacing estimation device described above, thereby controlling the amount of plating deposited on the hot-dip galvanized steel strip by blowing wiping gas from the gas wiping nozzle onto the hot-dip galvanized steel strip as it is continuously removed from the hot-dip zinc bath. [Effects of the Invention]
[0018] According to the method for generating a nozzle interval estimation model, the method for estimating a nozzle interval, the method for controlling the amount of hot-dip galvanizing deposition, the method for manufacturing a hot-dip galvanized steel strip, the nozzle interval estimation model generation device, the nozzle interval estimation device, and the hot-dip galvanizing deposition amount control device according to the present invention, the nozzle interval between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip during actual operation can be calculated without calculation error and in a short time.
Brief Description of the Drawings
[0019] [Figure 1] It is a schematic configuration diagram of a continuous hot-dip galvanizing line including a nozzle interval estimation model generation device, a nozzle interval estimation device, and a hot-dip galvanizing deposition amount control device according to an embodiment of the present invention. [Figure 2] It is a flowchart for explaining the processing flow in the nozzle interval estimation model generation device, the nozzle interval estimation device, and the hot-dip galvanizing deposition amount control device shown in FIG. 1. [Figure 3] It is a diagram for explaining the nozzle interval estimation model generated by the nozzle interval estimation model generation device shown in FIG. 1. [Figure 4] It is a flowchart for explaining the details of step S2 (nozzle interval estimation step) in the flowchart shown in FIG. 2. [Figure 5] It is a diagram for explaining the logic of inputting the steel type of the hot-dip galvanized steel sheet during actual operation, the plate thickness of the hot-dip galvanized steel sheet, the line speed of the hot-dip galvanized steel sheet, the temperature of the hot-dip galvanized steel sheet, the gas pressure of the gas wiping nozzle, and the target value of the deposition amount into the nozzle interval estimation model to estimate the nozzle interval in step S2 (nozzle interval estimation step). [Figure 6] It is a diagram for explaining the nozzle interval derivation method according to the reference example.
Embodiments for Carrying Out the Invention
[0020] Embodiments of the present invention will be described below with reference to the drawings. The embodiments shown below are illustrative examples of devices and methods for realizing the technical concept of the present invention, and the technical concept of the present invention is not limited to the following embodiments in terms of the material, shape, structure, arrangement, etc., of the components. Furthermore, the drawings are schematic. Therefore, it should be noted that the relationship and ratio of thickness and planar dimensions may differ from those in reality, and there may be parts where the dimensional relationships and ratios differ between drawings.
[0021] Figure 1 shows a schematic configuration of a continuous hot-dip galvanizing line equipped with a nozzle spacing estimation model generation device, a nozzle spacing estimation device, and a hot-dip galvanizing deposition amount control device according to one embodiment of the present invention. In the continuous hot-dip galvanizing line 1 shown in Figure 1, a steel strip S annealed in a continuous annealing furnace (not shown) in a reducing atmosphere passes through a snout 3 and is introduced into the molten zinc bath 4 in the plating tank 2. The steel strip S is then pulled up to the top of the molten zinc bath 4 via sink rolls 5 and collect rolls 6. Subsequently, wiping gas is blown onto the surface of the hot-dip galvanized steel strip SS from gas wiping nozzles 7 located on both sides of the hot-dip galvanized steel strip SS to scrape off excess molten zinc that has adhered to the surface of the hot-dip galvanized steel strip SS and been pulled up, thereby adjusting the amount of plating deposited. In Figure 1, reference numeral 8 denotes a guide roll.
[0022] Here, the amount of plating deposited on the hot-dip galvanized steel strip SS is controlled by adjusting the nozzle distance D between the surface of the hot-dip galvanized steel strip SS, which is continuously removed from the hot-dip zinc bath 4, and the tip of the gas wiping nozzle 7. In order to control this nozzle spacing D, in this embodiment, the continuous hot-dip galvanizing line 1 is equipped with a nozzle spacing estimation model generation device 10, a nozzle spacing estimation device 20, and a hot-dip galvanizing deposition amount control device 30.
[0023] The nozzle spacing estimation model generation device 10 uses machine learning on multiple training data to generate a nozzle spacing estimation model M1 (see Figure 3) that estimates the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS. The nozzle spacing estimation model generation device 10 is a computer system that has a computational processing function to realize the function of generating the nozzle spacing estimation model M1 by executing a program on computer software. This computer system is configured with ROM, RAM, CPU, etc., and realizes the above-mentioned function on software by executing various dedicated programs that are pre-stored in the ROM, etc.
[0024] Here, as shown in Figure 3, the multiple training data sets are input data (explanatory variables) that are actual values of the steel type of the hot-dip galvanized steel strip SS continuously taken from the hot-dip galvanized bath 4, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7 that blows wiping gas onto the hot-dip galvanized steel strip SS, and the amount of plating deposited on the hot-dip galvanized steel strip SS. The output data (dependent variable) is the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS relative to this input data. The steel type of the hot-dip galvanized steel strip SS is the steel type of the hot-dip galvanized steel strip SS when it becomes a finished product. The thickness of the hot-dip galvanized steel strip SS is the thickness of the hot-dip galvanized steel strip SS when it becomes a finished product. The line speed of the hot-dip galvanized steel strip SS is the speed at which the hot-dip galvanized steel strip SS moves when it passes through the continuous hot-dip galvanizing line 1. Furthermore, the temperature of the hot-dip galvanized steel strip SS is the temperature of the steel strip S before it is introduced into the molten zinc bath 4.
[0025] These multiple training data are input by an operator from an input device 11 connected to the nozzle spacing estimation model generation device 10 before the actual hot-dip galvanizing operation. The nozzle spacing estimation model generation device 10 then uses this input training data to perform machine learning and generates a nozzle spacing estimation model M1 that estimates the nozzle spacing D. The machine learning method used in the nozzle spacing estimation model generation device 10 is a neural network, and the nozzle spacing estimation model M1 is an estimation model constructed using a neural network.
[0026] Furthermore, as shown in Figure 5, the nozzle spacing estimation device 20 takes the target values of the steel type of the hot-dip galvanized steel strip SS, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7, and the amount of plating adhesion to the hot-dip galvanized steel strip SS as input to the nozzle spacing estimation model M1 generated by the nozzle spacing estimation model generation device 10, and estimates the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS during actual operation. The nozzle spacing estimation device 20 is a computer system that has a calculation processing function to realize the nozzle spacing D estimation function by executing a program on computer software. This computer system is configured with ROM, RAM, CPU, etc., and realizes the above-mentioned function on software by executing various dedicated programs that are pre-stored in ROM, etc.
[0027] Here, information regarding the steel type of the hot-dip galvanized steel strip SS during actual operation, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7, and the target value of the plating adhesion amount of the hot-dip galvanized steel strip SS is input to the nozzle spacing estimation device 20 from the higher-level computer 40 connected to the nozzle spacing estimation device 20. The nozzle spacing estimation device 20 then inputs the information input from the higher-level computer 40 into the nozzle spacing estimation model M1 obtained from the nozzle spacing estimation model generation device 10 to estimate the nozzle spacing D during actual operation.
[0028] Furthermore, the hot-dip galvanizing deposition amount control device 30 controls the nozzle spacing between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS so that the nozzle spacing D is estimated by the nozzle spacing estimation device 20, and blows wiping gas from the gas wiping nozzle 7 onto the hot-dip galvanized steel strip SS as it is continuously taken out from the hot-dip zinc bath 4 to control the amount of plating deposited on the hot-dip galvanized steel strip SS. When the hot-dip galvanizing deposition amount control device 30 blows wiping gas from the gas wiping nozzle 7 onto the hot-dip galvanized steel strip SS, it blows the wiping gas at the gas pressure of the gas wiping nozzle 7 during actual operation, which is input from the host computer 40. As a result, the amount of plating on the hot-dip galvanized steel strip SS during actual operation is controlled to a target value for the amount of plating on the hot-dip galvanized steel strip SS input from the higher-level computer 40.
[0029] Next, the processing flow in the nozzle spacing estimation model generation device 10, the nozzle spacing estimation device 20, and the hot-dip galvanizing deposition amount control device 30 will be explained with reference to Figures 2 to 5. Figure 2 is a flowchart illustrating the processing flow in the nozzle spacing estimation model generation device 10, the nozzle spacing estimation device 20, and the hot-dip galvanizing deposition amount control device 30. First, before applying hot-dip galvanizing to the steel strip S in actual operation, in step S1, the nozzle spacing estimation device 20 generates a nozzle spacing estimation model M1 (see Figure 3). In generating this nozzle spacing estimation model M1, the nozzle spacing estimation model generation device 10 generates the nozzle spacing estimation model M1 by machine learning using multiple training data (nozzle spacing estimation model generation step).
[0030] Here, the multiple training data sets are input data (explanatory variables) that include the steel type of the hot-dip galvanized steel strip SS continuously extracted from the molten zinc bath 4, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7 that blows wiping gas onto the hot-dip galvanized steel strip SS, and the actual values of the amount of plating deposited on the hot-dip galvanized steel strip SS. The output data (dependent variable) is the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the steel strip SS relative to this input data.
[0031] The steel grade of the hot-dip galvanized steel strip SS is, as mentioned above, the steel grade of the hot-dip galvanized steel strip SS when it becomes a finished product. The sheet thickness of the hot-dip galvanized steel strip SS is the sheet thickness of the hot-dip galvanized steel strip SS when it becomes a finished product. The line speed of the hot-dip galvanized steel strip SS is the speed at which the hot-dip galvanized steel strip SS moves as it passes through the continuous hot-dip galvanizing line 1. The temperature of the hot-dip galvanized steel strip SS is the temperature of the steel strip S before it is introduced into the molten zinc bath 4.
[0032] As mentioned above, these multiple training data are input by the operator from the input device 11 to the nozzle spacing estimation model generation device 10 before the actual operation of hot-dip galvanizing. Next, in step S2, the nozzle spacing estimation device 20 estimates the nozzle spacing D during actual operation. In estimating the nozzle spacing D, as shown in Figure 5, the nozzle spacing estimation device 20 inputs the steel type of the hot-dip galvanized steel strip SS, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7, and target values of the amount of plating deposited on the hot-dip galvanized steel strip SS during actual operation into the nozzle spacing estimation model M1 generated in step S1, and estimates the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the steel strip SS during actual operation (nozzle spacing estimation step).
[0033] Step S2 (nozzle spacing estimation step) will be explained in detail with reference to Figure 4. First, in step S21, the nozzle spacing estimation device 20 reads the nozzle spacing estimation model M1 generated in step S1. Next, in step S22, the nozzle spacing estimation device 20 reads information from the host computer 40 regarding the steel type of the hot-dip galvanized steel strip SS during actual operation, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7, and the target value of the amount of plating deposited on the hot-dip galvanized steel strip SS.
[0034] Next, in step S23, the nozzle spacing estimation device 20 inputs the steel type of the hot-dip galvanized steel strip SS, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7, and the target value of the amount of plating deposited on the hot-dip galvanized steel strip SS during actual operation, which were read in step S22, into the nozzle spacing estimation model M1 read in step S21, and estimates the nozzle spacing D during actual operation. Next, in step S24, the nozzle spacing estimation device 20 outputs the result from step S23 (information on the estimated nozzle spacing D during actual operation) to the hot-dip galvanizing deposition amount control device 30. This completes step S2 (nozzle spacing estimation step).
[0035] Once step S2 is completed, in step S3, the hot-dip galvanizing amount control device 30 controls the nozzle distance between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS so that the nozzle distance D is the same as the nozzle distance D estimated in step S2, and blows wiping gas from the gas wiping nozzle 7 onto the hot-dip galvanized steel strip SS to control the amount of plating deposited on the hot-dip galvanized steel strip SS (plating amount control step). At this time, the hot-dip galvanizing amount control device 30 blows the wiping gas at the gas pressure of the gas wiping nozzle 7 during actual operation, which is input from the host computer 40.
[0036] As a result, the amount of plating on the hot-dip galvanized steel strip SS during actual operation is controlled to a target value for the amount of plating on the hot-dip galvanized steel strip SS input from the higher-level computer 40. Then, in actual operation, the hot-dip galvanized steel strip SS is manufactured through a hot-dip galvanized coating amount control process, which controls the amount of hot-dip galvanized coating using the aforementioned hot-dip galvanized coating amount control method (step S3, coating amount control step), followed by various other processes.
[0037] As described above, according to the nozzle spacing estimation model generation apparatus 10 and nozzle spacing estimation model generation method (step S1, nozzle spacing estimation model generation step) of the present embodiment, a nozzle spacing estimation model M1 is generated that estimates the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS by machine learning a plurality of training data. The plurality of training data are input data that includes the steel type of the hot-dip galvanized steel strip SS continuously taken from the hot-dip zinc bath 4, the plate thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7 that blows wiping gas onto the hot-dip galvanized steel strip SS, and actual values of the amount of plating deposited on the hot-dip galvanized steel strip SS, and the nozzle spacing between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS relative to this input data is used as output data.
[0038] Furthermore, according to the nozzle spacing estimation device 20 and nozzle spacing estimation method (step S2, nozzle spacing estimation step) of this embodiment, the generated nozzle spacing estimation model M1 is input with the steel type of the hot-dip galvanized steel strip SS continuously taken out from the molten zinc bath 4 during actual operation, the thickness of the hot-dip galvanized steel strip SS, the line speed of the hot-dip galvanized steel strip SS, the temperature of the hot-dip galvanized steel strip SS, the gas pressure of the gas wiping nozzle 7 that blows wiping gas onto the hot-dip galvanized steel strip SS, and a target value of the amount of plating deposited on the hot-dip galvanized steel strip SS, in order to estimate the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS during actual operation. This allows the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS during actual operation to be calculated quickly and without calculation error. This effect will be explained in comparison with the nozzle spacing derivation method related to the previous reference example shown in Figure 6.
[0039] In the nozzle spacing derivation method shown in the reference example in Figure 6, first, the nozzle spacing (initial nozzle position) is calculated using a physical formula that shows the relationship between the nozzle spacing (initial nozzle position) and the target value of the plating adhesion amount. Then, the calculated slack spacing (initial nozzle position), steel type during actual operation, and plate thickness are input into the plating adhesion amount estimation model M2 to estimate the predicted value of the plating adhesion amount. Next, the controller 101 calculates the nozzle movement amount of the gas wiping nozzle 7 from the deviation between the predicted value of the plating adhesion amount estimated by the plating adhesion amount estimation model M2 and the target value of the plating adhesion amount. Then, the nozzle spacing (nozzle position) after calculating the nozzle movement amount, as well as information such as the steel type and plate thickness during actual operation, are input again into the plating adhesion amount estimation model M2 to estimate the predicted value of the plating adhesion amount.
[0040] Then, the deviation between the predicted amount of plating and the target amount of plating is recalculated, and this process is repeated until the nozzle spacing (nozzle position) that minimizes the deviation is finally derived, thereby determining the nozzle spacing for actual operation. In the nozzle spacing derivation method of this previous reference example, the calculation involves iterative calculation of the deviation between the predicted values of the nozzle movement amount and plating adhesion amount of the gas wiping nozzle 7 and the target value of the plating adhesion amount, as well as calculations using some physical formulas. As a result, errors occur in the calculation of the nozzle spacing, and the calculation time is excessively long.
[0041] In contrast, as described above, in the nozzle spacing estimation model generation device 10 and nozzle spacing estimation model generation method (step S1, nozzle spacing estimation model generation step) according to this embodiment, first, the steel type of the hot-dip galvanized steel strip SS and actual values of the amount of plating deposited on the hot-dip galvanized steel strip SS are used as input data, and multiple training data sets, with the nozzle spacing for this input data as output data, are subjected to machine learning to generate a nozzle spacing estimation model M1 that estimates the nozzle spacing D.
[0042] Then, in the nozzle spacing estimation device 20 and nozzle spacing estimation method (step S2, nozzle spacing estimation step), the nozzle spacing D during actual operation is estimated by inputting the operating conditions such as the steel type of the hot-dip galvanized steel strip SS and the target value of the plating adhesion amount of the hot-dip galvanized steel strip SS during actual operation into the generated nozzle spacing estimation model M1. Therefore, when deriving the nozzle spacing D during actual operation, it is unnecessary to perform iterative calculations of the deviation between the predicted values of the nozzle movement and plating adhesion amount of the gas wiping nozzle 7 and the target value of the plating adhesion amount, as well as calculations using some physical formulas. Instead, it is only necessary to input the operating conditions during actual operation and the target value of the plating adhesion amount of the hot-dip galvanized steel strip SS into the nozzle spacing estimation model M1. This makes it possible to calculate the nozzle spacing D between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS during actual operation in a short time without calculation errors. Furthermore, the nozzle spacing D is generally calculated using the following equation (1).
number
[0043] Furthermore, according to the hot-dip galvanizing adhesion amount control device 30 and hot-dip galvanizing adhesion amount control method (step S3, plating adhesion amount control step) of this embodiment, the nozzle distance between the tip of the gas wiping nozzle 7 and the surface of the hot-dip galvanized steel strip SS is controlled so that the estimated nozzle distance D is achieved. Then, the amount of plating adhesion to the hot-dip galvanized steel strip SS is controlled by blowing wiping gas from the gas wiping nozzle 7 onto the hot-dip galvanized steel strip SS as it is continuously removed from the hot-dip galvanized bath 4. As mentioned above, the nozzle spacing D is greatly influenced by the gas pressure P of the gas wiping nozzle, as can be seen from equation (1). Therefore, by using the gas pressure P of the gas wiping nozzle as input data for training data when generating the nozzle spacing estimation model M1, and as an input parameter to the nozzle spacing estimation model M1, it is possible to reduce calculation errors and computation time when calculating the nozzle spacing D.
[0044] This makes it possible to control the amount of plating adhering to the hot-dip galvanized steel strip (SS) during actual operation to the target value for the amount of plating adhering to the hot-dip galvanized steel strip (SS). Furthermore, the method for manufacturing a hot-dip galvanized steel strip according to this embodiment includes a hot-dip galvanizing amount control step in which the amount of galvanizing applied to the hot-dip galvanized steel strip SS is controlled using a method for controlling the amount of galvanizing applied (step S3, galvanizing amount control step). This makes it possible to manufacture hot-dip galvanized steel strips (SS) in which the amount of plating adhesion is controlled to a target value.
[0045] Although embodiments of the present invention have been described above, the present invention is not limited thereto and can be modified and improved in various ways. For example, the machine learning method used in the nozzle spacing estimation model generation device 10 is not limited to neural networks; it can be any of the following: gradient boosting, decision trees, random forests, neural networks, and multiple regression. By using such machine learning techniques, a highly accurate nozzle spacing estimation model M1 can be generated. [Explanation of Symbols]
[0051] 1. Continuous hot-dip galvanizing line 2 Plating tank 3 Snout 4. Molten zinc bath 5 Syncroll 6. Collectol 7. Gas wiping nozzle 8 Guide Roll 10. Nozzle Spacing Estimation Model Generation Device 11 Input devices 20. Nozzle Spacing Estimation Device 30. Control device for controlling the amount of hot-dip galvanizing coating. 40 High-level calculator M1 Nozzle Spacing Estimation Model S steel strip SS hot-dip galvanized steel strip
Claims
1. A method for generating a nozzle spacing estimation model, characterized by using machine learning to generate a nozzle spacing estimation model that estimates the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip, by taking all of the following as input data: the type of steel strip of hot-dip galvanized steel strip continuously taken out from a molten zinc bath, the thickness of the hot-dip galvanized steel strip (the thickness of the hot-dip galvanized steel strip (galvanized steel strip) when it becomes a product), the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of a gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and the actual value of the amount of plating deposited on the hot-dip galvanized steel strip (determined by measuring the amount of plating deposited on the hot-dip galvanized steel strip when it becomes a product), and using multiple training data sets, in which the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip is used as output data for this input data.
2. The method for generating a nozzle spacing estimation model according to claim 1, characterized in that the machine learning method uses one of the following: gradient boosting, decision trees, random forests, neural networks, and multiple regression.
3. A method for estimating nozzle spacing, characterized in that the nozzle spacing estimation model generated by the method for generating a nozzle spacing estimation model according to claim 1 or 2 is used to input all of the following into the nozzle spacing estimation model generated by the method for generating a nozzle spacing estimation model according to claim 1 or 2: the type of steel of the hot-dip galvanized steel strip continuously taken out from the molten zinc bath during actual operation, the thickness of the hot-dip galvanized steel strip (the thickness of the hot-dip galvanized steel strip (galvanized steel strip) when it becomes a product), the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and the target value of the amount of plating deposited on the hot-dip galvanized steel strip, thereby estimating the nozzle spacing estimation value between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip during actual operation.
4. A method for controlling the amount of hot-dip galvanized coating to adhere to a hot-dip galvanized steel strip, characterized by controlling the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip so that the nozzle spacing is the estimated value of the nozzle spacing estimated by the nozzle spacing estimation method described in claim 3, and by blowing wiping gas from the gas wiping nozzle onto the hot-dip galvanized steel strip as it is continuously removed from the hot-dip galvanized bath to control the amount of galvanized coating to adhere to the hot-dip galvanized steel strip.
5. A method for manufacturing a hot-dip galvanized steel strip, characterized by including a hot-dip galvanizing adhesion amount control step for controlling the amount of galvanizing adhesion of a hot-dip galvanized steel strip using the hot-dip galvanizing adhesion amount control method described in claim 4.
6. A nozzle spacing estimation model generation device is characterized by generating a nozzle spacing estimation model that estimates the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip by using machine learning on multiple training data sets, which are output data representing the nozzle spacing between the tip of the gas wiping nozzle and the surface of the hot-dip galvanized steel strip, with the input data being the steel type of the hot-dip galvanized steel strip continuously removed from the molten zinc bath, the thickness of the hot-dip galvanized steel strip (the thickness of the hot-dip galvanized steel strip (galvanized steel strip) when it becomes a product), the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and the actual value of the amount of plating deposited on the hot-dip galvanized steel strip (determined by measuring the amount of plating deposited on the hot-dip galvanized steel strip when it becomes a product).
7. The nozzle spacing estimation model generation device according to claim 6, characterized in that the machine learning method uses one of the following: gradient boosting, decision trees, random forests, neural networks, and multiple regression.
8. A nozzle spacing estimation device characterized in that it estimates the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip during actual operation by inputting all of the following into the nozzle spacing estimation model generated by the nozzle spacing estimation device according to claim 6 or 7: the type of steel of the hot-dip galvanized steel strip continuously taken out from the molten zinc bath during actual operation, the thickness of the hot-dip galvanized steel strip (the thickness of the hot-dip galvanized steel strip (galvanized steel strip) when it becomes a product), the line speed of the hot-dip galvanized steel strip, the temperature of the hot-dip galvanized steel strip, the gas pressure of the gas wiping nozzle that blows wiping gas onto the hot-dip galvanized steel strip, and a target value for the amount of plating deposited on the hot-dip galvanized steel strip.
9. A hot-dip galvanizing adhesion amount control device characterized by controlling the nozzle spacing between the tip of a gas wiping nozzle and the surface of a hot-dip galvanized steel strip so that the nozzle spacing is the estimated value of the nozzle spacing estimated by the nozzle spacing estimation device described in claim 8, and by blowing wiping gas from the gas wiping nozzle onto the hot-dip galvanized steel strip as it is continuously removed from the hot-dip zinc bath to control the amount of plating adhesion to the hot-dip galvanized steel strip.