Method for predicting plating defect of steel sheet, method for reducing defect of steel sheet, method for manufacturing hot-dip galvanized steel sheet, and method for generating model for predicting plating defect of steel sheet

Abstract

The steel sheet non-plating defect prediction method according to the present application is a steel sheet non-plating defect prediction method for a hot-dip galvanized steel sheet manufacturing apparatus provided with an annealing furnace and a plating device arranged on the downstream side of the annealing furnace, the steel sheet non-plating defect prediction method including a step of predicting non-plating defect information of a steel sheet on the exit side of the manufacturing apparatus using a non-plating defect prediction model learned through machine learning, the non-plating defect prediction model being learned using, as input data, one or two or more parameters selected from attribute information of a steel sheet charged into the manufacturing apparatus, one or two or more operation parameters selected from operation parameters of the annealing furnace, and one or two or more operation parameters selected from operation parameters of the plating device, and using non-plating defect information of a steel sheet on the exit side of the manufacturing apparatus as output data.

Description

Technical Field

[0001] This invention relates to a method for predicting plating defects in steel plates, a method for reducing defects in steel plates, a method for manufacturing hot-dip galvanized steel plates, and a method for generating a model for predicting plating defects in steel plates. Background Technology

[0002] In recent years, the demand for high-tensile steel sheets (high-strength steel sheets) that contribute to the lightweighting of structures has been steadily increasing in the automotive, home appliance, and building materials industries. As high-tensile steel sheets, it is known that, for example, by including Si in the steel, steel sheets with good porosity can be manufactured, and by including Si and Al, steel sheets with good ductility that easily form residual γ can be manufactured. However, when using high-tensile steel sheets containing a large amount of Si (especially 0.2% by mass or more) as the base material to manufacture hot-dip galvanized steel sheets, the following problems exist.

[0003] In other words, hot-dip galvanized steel sheet is manufactured by annealing the base steel sheet at a temperature of approximately 600–900°C in a reducing or non-oxidizing atmosphere, followed by hot-dip galvanizing of the steel sheet surface. It should be noted that alloyed hot-dip galvanized steel sheet can also be manufactured by heating and alloying the galvanized layer formed on the steel sheet surface. Here, silicon (Si) in the steel is an easily oxidized element, and even in commonly used reducing or non-oxidizing atmospheres, it is selectively oxidized, accumulating on the surface of the steel sheet to form oxides. These oxides reduce the wettability between the steel sheet and the hot-dip galvanized layer during the hot-dip galvanizing process, resulting in incomplete galvanizing defects.

[0004] Against this backdrop, technologies for suppressing plating defects have been proposed. Specifically, Patent Document 1 describes a method that increases the dew point inside an annealing furnace by introducing humidifying gas, thereby suppressing surface enrichment of Si through internal oxidation. Patent Document 2 describes a method that sets a target temperature for a steel plate at a specific location before immersion in the plating bath, and controls the temperature of the steel plate entering the plating bath by controlling the cooling section of the annealing furnace, thus controlling the temperature of the plating bath. Patent Document 3 describes a method that suppresses the vaporization of the hot-dip galvanized layer and the formation of an oxide film on the plating bath surface by controlling the dew point inside the furnace nose, thereby reducing the frequency of plating defects.

[0005] Existing technical documents

[0006]

[0007] Patent Document 1: Japanese Patent No. 6607339

[0008] Patent Document 2: Japanese Patent No. 4507681

[0009] Patent Document 3: Japanese Patent No. 5834775 Summary of the Invention

[0010] However, even in the methods described in Patent Documents 1-3, plating defects can still occur, especially with differences frequently appearing between the surface and back sides of the steel sheet. Eliminating these differences between the surface and back sides of the steel sheet can help improve the yield, but sometimes the side with the most plating defects can change depending on the production line, time, and timing, and the reasons for this are not yet clear.

[0011] This invention was made in view of the above-mentioned problems, and its object is to provide a method for predicting incomplete plating defects in steel sheets that can accurately predict the occurrence of incomplete plating defects. Another object of this invention is to provide a method for reducing the frequency of incomplete plating defects in steel sheets. Furthermore, another object of this invention is to provide a method for manufacturing hot-dip galvanized steel sheets that can improve the yield of hot-dip galvanized steel sheets. Moreover, yet another object of this invention is to provide a method for generating a prediction model for incomplete plating defects in steel sheets, which can generate a high-precision prediction model for incomplete plating defects in steel sheets.

[0012] The present invention relates to a method for predicting plating defects in steel plates. This method is used in a manufacturing apparatus for hot-dip galvanized steel plates equipped with an annealing furnace and a plating device disposed downstream of the annealing furnace. The method includes the following steps: taking one or more parameters selected from attribute information of the steel plate loaded into the manufacturing apparatus, one or more operating parameters selected from operating parameters of the annealing furnace, and one or more operating parameters selected from operating parameters of the plating device as input data; taking plating defect information of the steel plate at the outlet side of the manufacturing apparatus as output data; and using a plating defect prediction model learned by machine learning to predict the plating defect information of the steel plate at the outlet side of the manufacturing apparatus.

[0013] The aforementioned annealing furnace is a vertical annealing furnace that is sequentially configured with a heating section, a soaking section, and a cooling section from the upstream side. The operating parameters of the aforementioned annealing furnace, which constitute the aforementioned input data, may include the dew point information of the aforementioned heating section and the aforementioned soaking section.

[0014] The aforementioned plating apparatus is a plating apparatus in which a furnace nose, a plating bath, and a wiping device are arranged sequentially from the upstream side. The operating parameters of the aforementioned plating apparatus, which constitute the aforementioned input data, may include the temperature of the plating bath and the temperature of the steel plate immersed in the plating bath.

[0015] The operating parameters of the plating apparatus that constitute the input data may include one or more operating parameters selected from the operating parameters of the wiping apparatus.

[0016] The method for reducing defects in steel plates according to the present invention includes the following steps: using the method for predicting incomplete plating defects in steel plates according to the present invention, before the plating device is installed at the front end of the steel plate, the incomplete plating defect information of the steel plate is predicted using the attribute information of the steel plate, the actual value of the operating parameters of the annealing furnace, and the set value of the operating parameters of the plating device, so as to make the incomplete plating defect generation rate based on the predicted incomplete plating defect information of the steel plate fall within a preset allowable range, and then the operating parameters of the plating device are set again.

[0017] The method for manufacturing hot-dip galvanized steel sheet of the present invention includes the step of manufacturing hot-dip galvanized steel sheet using the defect reduction method of the present invention.

[0018] The method for generating a prediction model for incomplete plating defects in steel plates according to the present invention includes the following steps: taking at least one or two real data selected from the attribute information of the steel plate loaded into the manufacturing equipment, one or two or more real operation data selected from the operating parameters of the annealing furnace, and one or two or more real operation data selected from the operating parameters of the plating device as input real data; acquiring multiple learning data that use the incomplete plating defect information of the steel plate at the outlet side of the manufacturing equipment using the above-mentioned input real data as output real data; and generating an incomplete plating defect prediction model for predicting the incomplete plating defect information of the steel plate at the outlet side of the manufacturing equipment by using machine learning of the acquired multiple learning data.

[0019] As a form of machine learning, one can use machine learning methods selected from neural networks, decision tree learning, random forests, and support vector regression.

[0020] The method for predicting plating defects in steel sheets according to the present invention can predict the occurrence of plating defects in steel sheets with high accuracy. Furthermore, the defect reduction method for steel sheets according to the present invention can reduce the frequency of plating defects in steel sheets. Additionally, the manufacturing method for hot-dip galvanized steel sheets according to the present invention can improve the yield of hot-dip galvanized steel sheets. Moreover, the method for generating a plating defect prediction model for steel sheets according to the present invention can generate a plating defect prediction model with high accuracy that predicts the occurrence of plating defects in steel sheets. Attached Figure Description

[0021] Figure 1 This is a schematic diagram illustrating the configuration of a manufacturing apparatus for hot-dip galvanized steel sheets according to one embodiment of the present invention.

[0022] Figure 2 It means Figure 1 The diagram shows the structure of the annealing furnace.

[0023] Figure 3 It means Figure 1 A schematic diagram of the plating apparatus shown.

[0024] Figure 4 This is a block diagram illustrating the configuration of a steel plate plating defect prediction model generation device according to one embodiment of the present invention.

[0025] Figure 5 This is a graph showing the effect of the temperature difference between the steel sheet immersed in the hot-dip galvanizing bath and the temperature of the hot-dip galvanizing bath (immersion temperature of the sheet - bath temperature) and the dew point in the annealing furnace on the coating properties.

[0026] Figure 6 This is a diagram showing the effect of the dew point inside the furnace nose on the coating properties.

[0027] Figure 7 This is a graph showing the effect of the temperature difference between the steel sheet immersed in the hot-dip galvanizing bath and the temperature of the hot-dip galvanizing bath (immersion plate temperature - bath temperature) and the wiping gas pressure on the coating properties.

[0028] Figure 8 This is an example image representing a plating defect.

[0029] Figure 9 This is a block diagram illustrating the configuration of a steel plate plating defect prediction device according to one embodiment of the present invention. Detailed Implementation

[0030] Hereinafter, one embodiment of the present invention will be described with reference to the accompanying drawings.

[0031] [Composition of equipment for manufacturing hot-dip galvanized steel sheets]

[0032] First, refer to Figures 1-3 The configuration of a manufacturing apparatus for hot-dip galvanized steel sheet according to one embodiment of the present invention will be described.

[0033] Figure 1 This is a schematic diagram illustrating the configuration of a manufacturing apparatus for hot-dip galvanized steel sheets according to one embodiment of the present invention. Figure 2 It means Figure 1 A schematic diagram of the structure of the annealing furnace 2 shown. Figure 3 It means Figure 1 A schematic diagram of the structure of the plating device 3 shown.

[0034] like Figure 1 As shown, a hot-dip galvanized steel sheet manufacturing apparatus 1 according to one embodiment of the present invention includes a vertical annealing furnace 2 and a plating device 3 disposed on the downstream side of the annealing furnace 2.

[0035] The annealing furnace 2 includes a heating section 21, a soaking section 22, and a cooling section 23. The heating section 21 and soaking section 22 anneal the steel plate S by heating it to a predetermined annealing temperature. The cooling section 23 cools the annealed steel plate S and then conveys it to the plating apparatus 3. Furthermore, in this annealing furnace 2, appropriate amounts of humidifying gas and drying gas (introduced gas) are introduced into the annealing furnace 2 to make the dew point within the annealing furnace 2 uniform at the target dew point. It should be noted that, for example... Figure 2 As shown, in this embodiment, there are 12 and 10 locations for introducing humidifying gas and drying gas, respectively. However, as long as the dew point in the annealing furnace 2 becomes uniform at the target dew point, the number of locations for introducing humidifying gas and drying gas is not particularly limited.

[0036] In addition, measuring devices 24a and 24b for measuring the flow rate, temperature, and dew point of the introduced gas are provided at the inlet side of the gas introduction location. While it is preferable to measure the flow rate of the introduced gas at each introduction location, the total amount of introduced gas in the annealing furnace 2 as a whole can also be measured. It should be noted that in this embodiment, four dew point meters are provided in the heating section 21 and two dew point meters are provided in the soaking section 22 to measure the dew point inside the annealing furnace 2, but providing multiple dew point meters makes it easier to manage the dew point, so more dew point meters can be provided. In addition, in the heating section 21 and the soaking section 22, the steel plate S is heated from about 300°C to about 700-900°C, so at least three plate thermometers for measuring the temperature of the steel plate S are provided at the inlet and outlet sides of the heating section 21 and the outlet side of the soaking section 22. Similar to the dew point meters, providing multiple plate thermometers makes it easier to manage, so more can be provided.

[0037] Back Figure 1 The plating apparatus 3 includes a furnace nose 31, a pot body 32, and a wiping nozzle 33. The furnace nose 31 is a rectangular component with a cross-sectional shape perpendicular to the traveling direction of the steel plate S, which divides the space through which the steel plate S passes. Its front end is immersed in a hot-dip galvanizing bath (hereinafter referred to as the plating bath) P within the pot body 32. In this embodiment, the steel plate S annealed in the annealing furnace 2 is continuously immersed in the plating bath P within the pot body 32 through the furnace nose 31. An ingot I (referencing the hot-dip galvanizing bath composition) is added to the pot body 32. Figure 3 This replenishes the consumed hot-dip galvanized layer. Then, the steel sheet S is pulled up above the galvanizing bath P via the submerged rollers and support rollers in the galvanizing bath P.

[0038] The wiping nozzle 33 is positioned above the pot body 32, separated by the steel plate S. The wiping nozzle 33 blows wiping gas from an opening (slit) extending in the width direction of the steel plate S towards the steel plate S, which is being pulled up from the plating bath P. By blowing the wiping gas, excess hot-dip galvanized layer present on both sides of the steel plate S is extruded, adjusting the amount of hot-dip galvanized layer adhering to both sides of the steel plate S and homogenizing it in both the width and length directions of the steel plate S.

[0039] In addition, in this plating apparatus 3, appropriate amounts of humidifying gas and drying gas (introduced gas) are introduced into the nose 31 to make the dew point within the nose 31 uniform at the target dew point. Figure 3 As shown, in this embodiment, there are two locations for introducing humidifying gas and drying gas. However, as long as the dew point within the furnace nose 31 becomes uniform at the target dew point, the number of locations for introducing humidifying gas and drying gas is not particularly limited. Furthermore, the nozzle piping for the humidifying gas is configured to supply humidifying gas to both the surface and back sides of the steel plate S, respectively. Additionally, since the introduction of humidifying gas aims to suppress the formation of an oxide film on the plating bath surface and the vaporization of zinc, it is preferable that the humidifying gas is introduced close to the plating bath surface.

[0040] In addition, a dew point meter is installed inside the nose 31 to measure the dew point within the nose 31. Similar to the location of the nozzle piping for the humidifying gas, the dew point meter is preferably located near the plating bath surface. Furthermore, in this embodiment, the dew point is located on the back side 1 of the steel plate S, but a dew point can also be located on the surface side of the steel plate S. Additionally, to suppress the formation of zinc oxides from the solidification of molten zinc vapor and the adhesion of zinc oxides to the steel plate S, a nose heater 34 is installed on the nose 31 to heat the wall surface of the nose 31 by means of electricity. The wall temperature of the nose 31 is measured by a thermometer 35. It should be noted that in this embodiment, the thermometer 35 is located upstream of the nose heater 34, but it can also be located downstream of the nose heater 34.

[0041] Furthermore, the pot body 32 is equipped with an online plating bath analyzer 36 for continuously monitoring the Al concentration and other components of the plating bath P, an induction coil 37 for heating the plating bath P by electricity, and a thermometer 38 for measuring the temperature of the plating bath P. Furthermore, a measuring device 39 for measuring the flow rate, temperature, and pressure of the wiping gas is connected to the wiping nozzle 33. The flow rate, temperature, and pressure of the wiping gas are measured before the wiping gas outlet. Additionally, the ingot I used to replenish the consumed hot-dip galvanized layer is an ingot with a pre-prepared composition. Furthermore, the concentration of the plating bath components in the plating bath P can be managed using a concentration meter capable of online measurement, or it can be managed by periodically collecting samples from the plating bath P and analyzing them using methods such as ICP (Inductivity Coupled Plasma) analysis.

[0042] [Composition of the device for generating a prediction model for plating defects in steel plates]

[0043] Next, refer to Figures 4-8 The configuration of a steel plate plating defect prediction model generation device according to one embodiment of the present invention will be described.

[0044] Figure 4 This is a block diagram illustrating the configuration of a steel plate incomplete plating defect prediction model generation device according to one embodiment of the present invention. Figure 4 As shown, the steel plate plating defect prediction model generation device 100 of one embodiment of the present invention is composed of an information processing device such as a workstation, and includes a database 101 and a machine learning unit 102.

[0045] Database 101 consists of a non-volatile storage device that stores real data of the operating parameters of annealing furnace 2, real data of the operating parameters of plating device 3, real data of the property parameters of steel plate S, and real data of plating defects in steel plate S.

[0046] Examples of operating parameters for the annealing furnace 2 include the pass-through speed of the steel plate S, information on the introduced gas (temperature, flow rate, composition, dew point, etc.), dew point information within the annealing furnace 2, and temperature information of the steel plate S. This is because the pass-through speed of the steel plate S affects the residence time of the steel plate S within the annealing furnace, which in turn affects the surface enrichment of Si and Mn. Especially when the pass-through speed is slow (approximately below 60 mpm), due to the longer annealing time, even with a constant dew point within the annealing furnace, a surface enrichment layer of Si and Mn is easily formed, which is detrimental to plating performance. This is because the information on the introduced gas, the dew point information within the annealing furnace 2, and the temperature information of the steel plate S affect the surface enrichment and material properties of Si and Mn. In steels containing Si and Mn, the surface enrichment of Si and Mn is most promoted when the dew point in the annealing furnace 2 is in the range of -40 to -30°C. However, within the range of -30 to 0°C, the higher the dew point, the more suppressed the surface enrichment of Si and Mn, showing a tendency to promote internal oxidation of Si and Mn. Furthermore, even when the dew point is set below -45°C, there is still a tendency to suppress the surface enrichment of Si and Mn. Additionally, when the annealing temperature is above 750°C, there is a tendency to promote the surface enrichment of Si and Mn.

[0047] Si and Mn in steel are easily oxidized elements, and are selectively oxidized even in commonly used reducing or non-oxidizing atmospheres, accumulating on the surface of steel plate S to form oxides. When steel plate S is immersed in a plating bath, the Si and Mn oxides formed on its surface easily reduce the wettability of molten zinc, resulting in reduced plating adhesion. In this embodiment, the incomplete plating defect of the steel plate refers to the localized area on the surface of steel plate S where no zinc adheres due to the reduced wettability of molten zinc. Therefore, the incomplete plating defect of the steel plate can be understood as a concave defect where the base material is partially exposed on the surface of steel plate S.

[0048] The temperature information of the steel plate S, which can be used as an operating parameter of the annealing furnace 2, can be obtained by using plate temperature data measured by at least one plate thermometer selected from those installed in the annealing furnace 2. Furthermore, the dew point information within the annealing furnace 2 can be obtained by using dew point data measured by at least one dew point meter selected from those installed in the annealing furnace 2. Moreover, the information of the introduced gas (temperature, flow rate, composition, dew point, etc.) can be obtained by using measurement data from measuring devices 24a and 24b used to measure the flow rate, temperature, and dew point of the introduced gas within the furnace 2. This is because this information affects the amount of Si and Mn oxides that can be formed on the surface of the steel plate S.

[0049] The operating parameters of the annealing furnace 2 are not limited to those described above, and may include information related to the output of the combustion device, such as the combustion rate of the radiant tube burner installed in the annealing furnace 2 and the amount of fuel (fuel gas) supplied. This is because it affects the temperature history of the steel plate S within the annealing furnace 2 and the enrichment behavior of Si and Mn on the surface of the steel plate S. Furthermore, the hydrogen concentration and carbon monoxide concentration within the annealing furnace 2 are measured and included as atmospheric information within the annealing furnace 2 in the operating parameters of the annealing furnace 2. This is because, depending on the hydrogen concentration within the annealing furnace, even in an atmosphere with the same dew point, the oxygen potential within the furnace will change, thereby affecting the surface enrichment behavior of Si and Mn. For example, if the hydrogen concentration within the annealing furnace is high, even in an atmosphere with the same dew point, the oxygen potential within the furnace will decrease, thus showing a trend of suppressed surface enrichment of Si and Mn. In addition, this is because if the carbon monoxide concentration within the annealing furnace 2 is high, decarburization may occur on the surface of the steel plate S due to the influence of trace amounts of moisture within the annealing furnace 2, thereby affecting the wettability of the molten zinc.

[0050] Here, the effects of the temperature difference between the steel sheet S immersed in plating bath P and the temperature of plating bath P (immersion sheet temperature - bath temperature) and the dew point in the annealing furnace on the plating properties were investigated, and the results are presented in... Figure 5 .like Figure 5As shown, the difference between the temperature of the steel plate S immersed in plating bath P and the temperature of plating bath P, as well as the dew point in the annealing furnace, can be used to classify the areas of good and bad plating performance. Since the dew point in the annealing furnace is related to the surface enrichment of Si and Mn, and the temperature difference between the steel plate S immersed in plating bath P and the temperature of plating bath P is related to plating performance, it suggests that the combination of these two factors is important. Therefore, it can be inferred that, in order to accurately predict plating defects, it is crucial to use data analysis of both parameters related to the surface enrichment of Si and Mn and parameters related to plating performance.

[0051] The operating parameters of the plating apparatus 3 include information related to the nose conditions, the pot conditions, and the wiping conditions. Examples of nose-related information include information about the nose heater 34 (input power, etc.), the temperature of the steel plate S inside the nose 31, the input gas information (temperature, flow rate, composition), the dew point information inside the nose 31, and the temperature of the nose 31 wall. Examples of pot-related information include information about the induction coil 37 (input power, etc.), information about the ingot I (composition, etc.), the temperature of the steel plate S immersed in the plating bath P, and the temperature of the plating bath P. Examples of wiping-related information include information about the wiping gas (temperature, flow rate), the wiping gas pressure setting, the nozzle height, the nozzle spray angle, the distance between the nozzle and the steel plate, and the unit area setting.

[0052] Information from the furnace nose heater 34, the temperature of the steel plate S inside the furnace nose 31, and the temperature of the wall surface of the furnace nose 31 all affect the temperature of the steel plate S immersed in the plating bath P and the temperature of the plating bath P. Specifically, if the temperature of the steel plate S immersed in the plating bath P decreases, the reactivity between the steel plate S and the plating bath P decreases, reducing the adhesion of the coating and making it prone to incomplete plating defects. On the other hand, if the temperature of the steel plate S immersed in the plating bath P is too high, the temperature of the plating bath P increases, making it difficult to maintain a stable temperature for the plating bath P. Alternatively, the amount of iron dissolving from the steel plate S increases, leading to an increase in the size of the Fe-Zn-Al intermetallic compounds in the so-called slag, which adhere to the steel plate S, thereby increasing slag-related surface defects. Furthermore, if the temperature of the plating bath P increases, the temperature difference between the steel plate S and the plating bath P (immersion plate temperature - bath temperature) decreases, reducing the plating performance.

[0053] The information regarding the introduced gas and dew point within the furnace nose 31 affects the oxide film on the plating bath surface. The oxide film formed on the plating bath surface within the furnace nose 31 adheres to the surface of the steel sheet S when it is immersed in the plating bath P, reducing the coating properties. Specifically, as... Figure 6As shown, when the dew point inside the furnace nose is below -33°C, zinc fume generation becomes significant. The zinc fume adheres to, solidifies, and deposits on the inner wall of the furnace nose 31, randomly falling and adhering to the steel plate S, thus becoming ash defects. On the other hand, when the dew point inside the furnace nose is above -27°C, an oxide film forms on the plating bath surface inside the furnace nose 31, hindering the plating adhesion along with the steel plate S. Therefore, the dew point inside the furnace nose can be controlled within the range of -33°C to -27°C. However, the preferred dew point range varies depending on the temperature of the steel plate S immersed in the plating bath P, the conveying speed of the steel plate S, and the steel grade of the steel plate S.

[0054] Information related to the pot conditions, which are operating parameters of the plating apparatus 3, can include information related to the induction coil 37, the temperature of the steel plate S immersed in the plating bath P, and the temperature of the plating bath P. The temperature information of the steel plate S immersed in the plating bath P and the temperature information of the plating bath P affect the temperature management and plating properties of the plating bath P, while information related to the ingot I affects the temperature of the plating bath P and the concentration of the plating bath components. The target temperature of the plating bath P can be set within the range of 450–460°C, and the output of the induction coil 37 can be adjusted to keep it constant. However, the temperature of the plating bath P constantly changes due to important factors such as the temperature decrease caused by the addition of the ingot I, the temperature change caused by the temperature change of the steel plate S immersed in the plating bath P, and the temperature decrease caused by the wiping gas. Therefore, the temperature of the plating bath P can be controlled, in particular, by taking into account the temperature of the steel plate S immersed in the plating bath P. By controlling the factors affecting the temperature change of the plating bath P as operating parameters, the plating properties of molten zinc on the steel plate S can be adjusted. Furthermore, since the Al concentration in the plating bath P affects the adhesion between the steel sheet S and zinc, the alloying reaction of the zinc coating, and the formation of oxides on the plating bath surface, it can be controlled using preferred values ​​within the following ranges: 0.125–0.14% for alloyed hot-dip galvanized steel sheet (GA) and 0.19–0.23% for hot-dip galvanized steel sheet (GI). Therefore, the Al concentration in the plating bath P can be included in the pot conditions, which are operating parameters of the plating apparatus 3.

[0055] Information related to the wiping gas, including the wiping gas pressure setting, nozzle height, nozzle spray angle, distance between the nozzle and the steel plate, and unit area setting, affects the surface temperature of the plating bath and the wall temperature of the furnace nose 31. Specifically, the wiping gas pressure is adjusted to control the zinc plating adhesion amount; the faster the plate passage speed, the higher the pressure, and the lower the target plating adhesion amount, the higher the pressure. The wiping gas is sprayed from the wiping nozzle at a high speed of 100-300 m / s at a temperature of 30-150°C, colliding with the steel plate S, flowing along the steel plate S, and circulating around the plating bath surface. Therefore, the plating bath surface and the wall of the furnace nose 31 are cooled by the wiping gas, and the temperature of the hot-dip galvanizing bath near the bath surface also decreases. Due to the decrease in the temperature of the hot-dip galvanizing bath, the reactivity of the plating bath P and the steel plate S changes; therefore, the wiping gas pressure (YP gas pressure) can be considered as an operating parameter of the plating device 3 (refer to...). Figure 7 ).

[0056] As attribute parameters of steel plate S, information related to the composition of steel plate S and its dimensions (thickness, width) can be provided. The composition information of steel plate S can be determined based on the content of the constituent elements. For example, the C content, Si content, and Mn content of steel plate S. In addition to the contents of C, Si, and Mn, the composition information of steel plate S may also include the contents of Cr, Mo, Nb, Ni, Al, Ti, Cu, Ni, and B. Preferably, the composition information of steel plate S includes the Si content and Mn content. This is because the enrichment behavior of the steel plate S surface changes in the annealing furnace 2 depending on the Si and Mn contents, affecting the plating properties. However, the enrichment behavior of Si and Mn on the surface of steel plate S is affected not only by the Si and Mn contents but also by the combination with other elements; therefore, in addition to the Si and Mn contents, the aforementioned composition information can be added. Furthermore, the C content is preferably used as the composition information of steel plate S. This is because, in the annealing furnace 2, if the moisture inside the furnace combines with the carbon in the steel plate S, causing carbon to detach from the surface of the steel plate S, the moisture consumption inside the furnace 2 will change according to the carbon content, thus affecting the dew point inside the furnace. On the other hand, the thickness and width of the steel plate S affect various temperature parameters.

[0057] In this embodiment, the information regarding the incomplete galvanizing defect of steel plate S refers to information related to the occurrence state of incomplete galvanizing defects of steel plate S observed at the exit side of the hot-dip galvanized steel plate manufacturing equipment 1. If incomplete galvanizing occurs on the surface of steel plate S, the required rust-proof effect as a hot-dip galvanized steel plate cannot be achieved because there is no zinc adhesion in that area. Therefore, incomplete galvanizing occurring on the surface of steel plate S is determined as an incomplete galvanizing defect of steel plate S. In the information regarding the incomplete galvanizing defect of steel plate S, any index representing the occurrence state of incomplete galvanizing defects of steel plate S, such as the presence or absence of incomplete galvanizing defects, the number of incomplete galvanizing defects, and the area ratio, can be used. Alternatively, steel plate S can be divided along its long side to determine the presence, number, and area ratio of incomplete galvanizing defects within the divided area.

[0058] Information on incomplete galvanizing defects in steel sheet S can be obtained from images and videos of steel sheet S captured by a camera or similar device installed on the exit side of the hot-dip galvanizing steel sheet manufacturing equipment 1. Specifically, this can be obtained through image processing-based evaluation of the coating appearance using surface inspection equipment, micro-defect gauges, etc., or through visual evaluation of the coating appearance by the operator. As evaluation methods, there are methods that utilize differences in the polarized light reflection characteristics of surface patterns to identify incomplete galvanizing defects. Actual photographs of the incomplete galvanizing appearance are shown... Figure 8 (a), (b). For example, Figure 8 Images of steel plate S shown in (a) and (b) are used to determine the quality of the coating through image processing, and the data is then digitized as information related to the evaluation of the coating appearance. This is achieved through... Figure 8 Image processing was performed on the images of steel plate S shown in (a) and (b), and the images with highlighted light and dark areas are shown in... Figure 8 (c) and (d). Thus, by performing image processing on the image of the steel plate S, it is possible to identify the unplated defects as dot-like or stripe-like defects. As a judgment criterion, for example, when taking an image of a 1000mm wide steel plate S during the through-plate process (width direction 1000mm × length direction 3000mm), if the following defects exist... Figure 8 When the plating defects shown in (c) and (d) are detected, the area is deemed to have "plating defects". The captured images can be video or still images. When capturing still images, it is preferable to capture images at a high frequency, capturing images of at least three locations: the front end, the stable part, and the tail end of the coil.

[0059] Furthermore, when detecting plating defects using image processing with a surface inspection device, the brightness, size, and morphology (dot clusters, stripes, irregular shapes, etc.) of the defect area are determined, and the number and area ratio per unit area of ​​the steel plate S are calculated. It should be noted that, in addition to plating defects, slag defects are also detected on the surface of the steel plate S at the exit side of the hot-dip galvanized steel sheet manufacturing equipment 1. Plating defects are observed as concave defects on the surface of the steel plate S, while slag defects are observed as convex defects with granular intermetallic compounds adhering to them. Therefore, in the surface inspection device, both can be identified by utilizing the brightness and shadow of the reflected image of light illuminating the surface of the steel plate S.

[0060] In the machine learning unit 102, real data on the operating parameters of the annealing furnace 2, the operating parameters of the plating device 3, and the attribute parameters of the steel plate S stored in the database 101 are used as input real data. Real data on the missing plating defect information of the steel plate S corresponding to the input real data is used as output real data. By using machine learning with multiple training data, a missing plating defect prediction model M is generated to predict the occurrence of missing plating defects in the steel plate S at the outlet side of the hot-dip galvanized steel sheet manufacturing equipment 1. The missing plating defect prediction model M is a computer program that takes the operating parameters of the annealing furnace 2, the operating parameters of the plating device 3, and the attribute parameters of the steel plate S as input data, and the missing plating defect information of the steel plate S at the outlet side of the hot-dip galvanized steel sheet manufacturing equipment 1 as output data.

[0061] The machine learning model used to generate the prediction model M for plating defects can be any machine learning model, as long as it achieves sufficient prediction accuracy for practical use. For example, common neural networks (including deep learning, convolutional neural networks, etc.), decision tree learning, random forests, support vector regression, etc., can be used. Alternatively, ensemble models combining multiple models can be used. Furthermore, classification models such as the k-nearest neighbors algorithm and logistic regression can also be employed.

[0062] [Composition of the device for predicting plating defects in steel plates]

[0063] Next, refer to Figure 9 The configuration of a steel plate plating defect prediction device according to one embodiment of the present invention will be described.

[0064] Figure 9 This is a block diagram illustrating the configuration of a steel plate plating defect prediction device according to one embodiment of the present invention. Figure 9As shown, the steel plate incomplete plating defect prediction device 110 of one embodiment of the present invention is composed of an information processing device such as a personal computer. Before the plating device 3 is installed at the front end of the steel plate S, the steel plate incomplete plating defect prediction device 110 inputs the actual values ​​of the operating parameters of the annealing furnace 2, the set values ​​of the operating parameters of the plating device 3, and the attribute parameters of the steel plate S into the incomplete plating defect prediction model M, thereby predicting whether there will be incomplete plating defects on the steel plate S at the outlet side of the hot-dip galvanized steel plate manufacturing equipment 1. Furthermore, the steel plate incomplete plating defect prediction device 110 outputs information related to the prediction results to the manufacturing equipment operating condition setting device 120.

[0065] The manufacturing equipment operating condition setting device 120 determines the quality of the plating defects in the steel plate S by comparing the target range of the plating defect generation rate obtained from the main computer 130 with multiple prediction results output from the plating defect prediction device 110 of the steel plate. Even with the same operating parameters, there may be cases where plating defects occur and cases where plating defects do not occur, so it is preferable to use the probability of plating defects occurring in the index. Therefore, in this embodiment, the manufacturing equipment operating condition setting device 120 determines the quality of the plating defects in the steel plate S based on the plating defect generation rate. The plating defect generation rate can be calculated, for example, using an image of the steel plate S with the same operating parameters, by the formula (1) shown below. In addition, the same definition can be used for the predicted probability of plating defects occurring. At this time, "the number of times or the time of occurrence of plating defects" can be replaced with "the number of times or the time of occurrence of plating defects predicted".

[0066] The rate of missing plating defects = (number of times or time of missing plating defects) / (number of times or shooting time of steel plates with the same operating parameters) ... (1)

[0067] Furthermore, when it is determined that there is no problem with incomplete plating (the prediction result for incomplete plating is good), the manufacturing equipment operating condition setting device 120 continues to operate the hot-dip galvanized steel sheet manufacturing equipment 1. On the other hand, when it is determined that there is a problem with incomplete plating (the prediction result for incomplete plating is bad), the operating condition resetting unit 121 of the manufacturing equipment operating condition setting device 120 re-sets the operating conditions of the hot-dip galvanized steel sheet manufacturing equipment 1 before the plating device 3 is inserted into the front end of the steel sheet S. As a result, the frequency of incomplete plating defects is reduced, and hot-dip galvanized steel sheets with improved manufacturing yield can be produced.

[0068] In conventional methods, the occurrence of incomplete plating defects cannot be predicted with high accuracy because the influencing factors on both the steel plate side (annealing dew point, annealing temperature, etc.) and the plating bath side (furnace nose dew point, bath temperature, Al concentration, relative flow rate with the surrounding plating bath, etc.) are not considered simultaneously. Furthermore, while the state of the plating bath itself and the temperature of the steel plate immersed in the bath are considered as influencing factors on the plating bath side, wiping conditions are not taken into account. Wiping conditions are originally a factor used to control the amount of plating adhesion, but since the temperature of the wiping gas is mostly between room temperature and 100°C, the plating bath and furnace nose wall are cooled when excess zinc plating is scraped off. In this embodiment, wiping conditions and furnace nose wall temperature are added as parameters to optimize the plating bath conditions inside the pot. Therefore, in this embodiment, by simultaneously considering data from both the plating bath side and the steel plate side, including wiping conditions and furnace nose conditions, the occurrence of incomplete plating defects can be predicted with higher accuracy compared to conventional methods.

[0069] As described above, the plating defect prediction model M of this embodiment uses the property information of the steel sheet S loaded into the manufacturing equipment 1 for hot-dip galvanized steel sheet as input data. Therefore, a plating defect prediction model M can be generated that reflects the influence of Si and Mn oxides on the surface of the steel sheet S generated in the annealing furnace 2, the influence of the dew point in the annealing furnace 2, and the influence of the temperature of the steel sheet S when immersed in the plating bath. Furthermore, the plating defect prediction model M of this embodiment uses the operating parameters of the annealing furnace 2 as input data. Therefore, a plating defect prediction model M can be generated that reflects the formation state of Si and Mn oxides on the surface of the steel sheet S when immersed in the plating bath, and the influence of the temperature of the steel sheet S when immersed in the plating bath on the wettability of molten zinc. Furthermore, the plating defect prediction model M of this embodiment uses the operating parameters of the plating apparatus 3 as input data. Therefore, a defect prediction model M for incomplete plating can be generated, reflecting the temperature of the steel plate S when it is immersed in the plating bath, the state of oxides formed on the surface of the steel plate S or the plating bath surface, and the influence on the temperature of the plating bath and the reactivity of the steel plate S with molten zinc. In particular, even if the formation state of Si and Mn oxides formed on the surface of the steel plate S is the same, the state of incomplete plating of the steel plate S will change depending on conditions such as the immersion temperature of the steel plate S in the plating bath (immersion plate temperature) and the temperature of the plating bath (bath temperature). Therefore, by combining these parameters as input data, incomplete plating defects of the steel plate S can be predicted with high accuracy.

[0070] Example

[0071] The following describes the results of applying the method for predicting incomplete galvanizing defects of steel plates according to this embodiment to the following situation: In a continuous hot-dip galvanizing equipment, a galvanized layer is formed on the surface of a thin steel plate to obtain a hot-dip galvanized steel plate, and then the hot-dip galvanized steel plate is further alloyed to obtain an alloyed hot-dip galvanized steel plate.

[0072] In this embodiment, Figure 1 The hot-dip galvanized steel sheet manufacturing equipment 1 shown includes a reheating device for alloying treatment located downstream of the wiping nozzle 33 positioned on the outlet side of the pot body 32. The reheating device comprises an alloying section, a holding section, and a final cooling section, with an induction heating device installed in the alloying section. The alloying section forms a zinc coating with an alloy layer formed by a Zn-Fe alloying reaction on the surface of the thin steel sheet and is used to manufacture alloyed hot-dip galvanized steel sheets. The hot-dip galvanizing bath P is a Zn bath containing Al. Furthermore, a surface inspection device is installed downstream of the reheating device and upstream of the product winding device of the hot-dip galvanized steel sheet manufacturing equipment 1. The surface inspection device is capable of detecting incomplete coating defects in the alloyed steel sheet S. The surface inspection device determines whether incomplete coating defects have occurred on the surface of each steel sheet S along a predetermined length in the long side direction.

[0073] The alloyed hot-dip galvanized steel sheet manufactured using equipment 1 is produced from cold-rolled steel coils with a thickness of 1.0–2.3 mm and a width of 690–1550 mm. For the steel coils used as raw materials, steel coils with different thicknesses and widths are prepared for 10 different steel grades, and a total of 100 steel coils are used to manufacture the alloyed hot-dip galvanized steel sheet. During the manufacturing of these steel sheets, the operating parameters of the annealing furnace 2 and the plating apparatus 3 are changed along the long side of the steel sheet S, according to the manufacturing conditions of equipment 1. Then, a surface inspection device is used to obtain accurate data on the incomplete plating defects of the steel sheet S corresponding to the set operating parameters of the annealing furnace 2 and the plating apparatus 3.

[0074] It should be noted that in the surface inspection device, the long side of the steel plate S is divided into units of 100m, and the output shows whether any plating defects occur within the divided area. Then, along the long side of the steel plate S, using the divided area as the unit, the actual data of the operating parameters of the annealing furnace 2 and the plating device 3 are obtained from the control computer of the hot-dip galvanized steel plate manufacturing equipment 1, and stored in the database 101 of the steel plate plating defect prediction model generation device 100. At this time, the actual data of the steel plate's attribute parameters are also obtained from the control computer, and stored in the database 101 in correspondence with the actual data of the operating parameters of the annealing furnace 2 and the plating device 3.

[0075] As described above, when 20,000 datasets are stored in database 101, 1,600 training datasets are extracted from database 101, and the remaining 400 are used as test datasets. The extracted training datasets are used to generate a plating defect prediction model M through the machine learning unit 102. At this time, in the machine learning unit 102, various changes are made and selected to the input variables used in the plating defect prediction model M, generating multiple plating defect prediction models M. Then, using the test datasets, the prediction accuracy (correctness) of plating defects in the steel plate S is evaluated based on the generated multiple plating defect prediction models M. A neural network is used as the machine learning method, with 3 intermediate layers and 5 nodes in each intermediate layer. A sigmoid function is used as the activation function.

[0076] The accuracy of the plating defect prediction model M is evaluated by the following ratio: in each division along the long side of the steel plate S, the ratio of the total number of cases predicted by the plating defect prediction model M as having a plating defect when the plating defect is detected by the surface inspection device (true positive) to the total number of cases predicted by the plating defect prediction model M as not having a plating defect when the plating defect is not detected by the surface inspection device (true negative) relative to all test data.

[0077] The evaluation results of the accuracy of the incomplete plating defect prediction model M are shown in Table 1 below. Table 1 lists the parameters selected as input variables for the incomplete plating defect prediction model M. As property parameters for the steel sheet, parameters selected from the composition and dimensional information of the steel sheet S are used. As operating parameters for the annealing furnace 2, information on the furnace atmosphere is used, including the dew point of the soaking zone, the dew point of the heating zone, hydrogen concentration, and carbon monoxide concentration. The dew point of the soaking zone and the heating zone is the average value of the dew point in each region. The hydrogen concentration and carbon monoxide concentration are the values ​​measured by hydrogen and carbon monoxide concentration meters installed at the outlet of the soaking zone. The plate passage speed is the average speed of the region along the long side of the steel sheet S being evaluated passing through the annealing furnace 2.

[0078] The humidifying gas flow rate and drying gas flow rate, used as information on the input gases, are the average values ​​of the gases input into the soaking zone, and the fuel gas quantity is the total flow rate of the fuel gas input into the radiant tubes arranged in the heating zone. The steel plate temperature, used as an operating parameter of the annealing furnace, is measured using the temperature reading from a plate thermometer located at the outlet of the annealing furnace 2. On the other hand, the dew point inside the furnace nose, the steel plate temperature inside the furnace nose, the hydrogen concentration inside the furnace nose, the wall temperature of the furnace nose, the flow rate of the humidifying gas supplied to the furnace nose, and the flow rate of the drying gas supplied to the furnace nose are used as information on the furnace nose conditions, serving as operating parameters for the plating apparatus 3.

[0079] It should be noted that the temperature of the steel plate inside the furnace nose is considered to be the same as the temperature of the steel plate immersed in the plating bath P. Information regarding the bath conditions includes the bath temperature, the Al concentration in the bath, the power of the input induction coil, and the amount of ingot I added. Information regarding the wiping conditions includes the target plating adhesion amount of the steel plate S, the jet gas pressure (air pressure) of the wiping nozzle, the distance between the wiping nozzle and the steel plate S (nozzle-steel plate distance), the height of the wiping nozzle from the bath surface (nozzle height), and the jet angle of the wiping gas relative to the surface of the steel plate S. Furthermore, the operating parameters selected from these are used as input variables to generate a prediction model M for plating defects, and the accuracy relative to the test data is calculated.

[0080] As a result, in the incomplete galvanizing defect prediction model M, which uses one or more parameters selected from the attribute parameters of the steel sheet S loaded into the hot-dip galvanized steel sheet manufacturing equipment 1, one or more operating parameters selected from the operating parameters of the annealing furnace 2, and one or more operating parameters selected from the operating parameters of the plating apparatus 3 as input data, it was confirmed that it can accurately predict the incomplete galvanizing defect information of the steel sheet S at the outlet side of the hot-dip galvanized steel sheet manufacturing equipment 1. In contrast, when no attribute information of the steel sheet S, the operating parameters of the annealing furnace 2, or the operating parameters of the plating apparatus 3 are included as input variables, it was confirmed that the accuracy of predicting the incomplete galvanizing defect of the steel sheet S decreases.

[0081]

[0082] The embodiments of the invention made by the inventors have been described above, but the present invention is not limited to the description and drawings that constitute a part of the disclosure of the present invention based on these embodiments. That is, all other embodiments, examples, and applications made by those skilled in the art based on these embodiments are included within the scope of the present invention.

[0083] Industrial availability

[0084] According to the present invention, a method for predicting plating defects in steel sheets that can accurately predict the occurrence of plating defects in steel sheets can be provided. Furthermore, according to the present invention, a method for reducing defects in steel sheets that can reduce the occurrence frequency of plating defects in steel sheets can be provided. Additionally, according to the present invention, a method for manufacturing hot-dip galvanized steel sheets that can improve the manufacturing yield of hot-dip galvanized steel sheets can be provided. Furthermore, according to the present invention, a method for generating a plating defect prediction model for steel sheets that can generate a plating defect prediction model with high accuracy in predicting the occurrence of plating defects in steel sheets can be provided.

[0085] Symbol Explanation

[0086] 1. Manufacturing equipment for hot-dip galvanized steel sheets

[0087] 2 Annealing furnace

[0088] 3. Coating device

[0089] 21 heating sections

[0090] 22 heat exchange sections

[0091] 23 Cooling Section

[0092] 24a, 24b, 39 Measuring devices

[0093] 31 furnace noses

[0094] 32 pot body

[0095] 33 Wiping Nozzle

[0096] 34 Furnace Nose Heater

[0097] 35 and 38 thermometers

[0098] 36 Online Plating Bath Analyzer

[0099] 37 Induction Coil

[0100] Device for generating a prediction model of plating defects in 100 steel plates

[0101] 101 Database

[0102] 102 Machine Learning Department

[0103] Device for predicting plating defects in 110 steel plates

[0104] 120 Manufacturing Equipment Operating Condition Setting Device

[0105] 121 Operating Condition Resetting Section

[0106] I casting ingot

[0107] P hot-dip galvanizing bath

[0108] S steel plate

Claims

1. A method for predicting plating defects in steel sheets, which is a method for predicting plating defects in steel sheets produced by hot-dip galvanizing steel sheet manufacturing equipment, wherein the manufacturing equipment includes an annealing furnace and a plating device disposed downstream of the annealing furnace. The method for predicting plating defects includes using a plating defect prediction model to predict plating defect information of the steel plate at the outlet side of the manufacturing equipment. The plating defect prediction model takes one or more parameters selected from the attribute information of the steel plate loaded into the manufacturing equipment, one or more operating parameters selected from the operating parameters of the annealing furnace, and one or more operating parameters selected from the operating parameters of the plating device as input data, and outputs the plating defect information of the steel plate at the outlet side of the manufacturing equipment as output data. This model is learned through machine learning. in, The plating apparatus is a plating apparatus in which a furnace nose, a plating bath, and a wiping device are arranged sequentially from the upstream side. The operating parameters of the plating apparatus that constitute the input data include the temperature of the plating bath and the temperature of the steel plate immersed in the plating bath, as well as one or more operating parameters selected from the operating parameters of the wiping device. The operating parameters of the wiping device include the wiping gas pressure setting value, the nozzle height, the nozzle spray angle, the distance between the nozzle and the steel plate, and the unit area setting value.

2. The method for predicting plating defects in steel plates according to claim 1, wherein, The annealing furnace is a vertical annealing furnace with a heating section, a soaking section and a cooling section arranged sequentially from the upstream side. The operating parameters of the annealing furnace that constitute the input data include the dew point information of the heating section and the soaking section.

3. A method for reducing defects in steel plates, comprising the following steps: Using the method for predicting plating defects in steel plates according to claim 1 or 2, before the plating device is installed at the front end of the steel plate, the plating defect information of the steel plate is predicted using the attribute information of the steel plate, the actual values ​​of the operating parameters of the annealing furnace, and the set values ​​of the operating parameters of the plating device. The operating parameters of the plating device are then set again so that the plating defect generation rate based on the predicted plating defect information of the steel plate falls within a preset allowable range.

4. A method for manufacturing a hot-dip galvanized steel sheet, comprising the step of manufacturing the hot-dip galvanized steel sheet using the defect reduction method for steel sheet as described in claim 3.

5. A method for generating a prediction model of plating defects in steel plates, which is a method for generating a prediction model of plating defects in steel plates of hot-dip galvanized steel plates manufactured by equipment, wherein the manufacturing equipment includes an annealing furnace and a plating device disposed downstream of the annealing furnace. The generation method includes generating a plating defect prediction model by using machine learning with multiple training data to predict plating defect information of the steel plate at the outlet side of the manufacturing equipment. The multiple training data are obtained by using at least one or more real data selected from the attribute information of the steel plate loaded into the manufacturing equipment, one or more real operational data selected from the operating parameters of the annealing furnace, and one or more real operational data selected from the operating parameters of the plating device as input real data, and using the plating defect information of the steel plate at the outlet side of the manufacturing equipment using the input real data as output real data. in, The plating apparatus is a plating apparatus in which a furnace nose, a plating bath, and a wiping device are arranged sequentially from the upstream side. The operating parameters of the plating apparatus that constitute the input real data include the temperature of the plating bath and the temperature of the steel plate immersed in the plating bath, as well as one or more operating parameters selected from the operating parameters of the wiping device. The operating parameters of the wiping device include the wiping gas pressure setting value, the nozzle height, the nozzle spray angle, the distance between the nozzle and the steel plate, and the unit area setting value.

6. The method for generating a prediction model for plating defects in steel plates according to claim 5, wherein, As the machine learning method, machine learning selected from neural networks, decision tree learning, random forests, and support vector regression is used.

Images