Cold rolling method for steel sheet, method for producing cold-rolled steel sheet, and production equipment for cold-rolled steel sheet

Abstract

To provide a cold rolling method for a steel sheet, a method for producing a cold-rolled steel sheet, and production equipment for a cold-rolled steel sheet that can suppress sheet thickness variations over the entire coil length, including rolling acceleration / deceleration regions, and satisfy strict sheet thickness accuracy requirements based on a friction coefficient prediction method capable of predicting the friction coefficient between a work roll and a steel sheet at high accuracy and the friction coefficient obtained. Provided is a cold rolling method for a steel sheet that controls the work roll gap of a cold rolling mill by using a prediction model for predicting the friction coefficient between the work roll of the cold rolling mill and the steel sheet, wherein the cold rolling method for a steel sheet includes a step in which a prediction model is generated using rolling performance data as an explanatory variable and an estimated value of the friction coefficient as an objective variable; the rolling performance data include the deformation resistance of the steel sheet and work roll diameter; the friction coefficient between the work roll of the cold rolling mill and the steel sheet is predicted by inputting the rolling conditions of the steel sheet to the prediction model; and the roll gap of the rolling mill is controlled based on the changes in sheet thickness on the outlet side of the cold rolling mill.

Description

Cold rolling method for steel sheet, cold rolled steel sheet manufacturing method, and cold rolled steel sheet manufacturing equipment

[0001] The present invention relates to a method for cold rolling a steel sheet, a method for manufacturing a cold-rolled steel sheet, and a manufacturing facility for a cold-rolled steel sheet.

[0002] In the cold rolling process, the technical difficulty of rolling is increasing due to the increase in thin, hard materials, and strict thickness accuracy is increasingly required. As a result, conventional thickness control technology can sometimes result in thickness tolerance deviations. In particular, during the acceleration and deceleration regions of rolling, including low speeds, the rolling load fluctuates significantly, making thickness fluctuations more likely. This is due to fluctuations in the friction coefficient caused by changes in the amount of rolling oil introduced between the work roll and the steel plate. Therefore, accurate prediction of the friction coefficient is important to suppress thickness fluctuations during the acceleration and deceleration regions of rolling.

[0003] Therefore, techniques for predicting the friction coefficient during rolling have been developed. For example, Patent Document 1 proposes the following method for predicting the friction coefficient that occurs between the work rolls of a rolling mill and a steel plate during rolling of the steel plate. In this method, the friction coefficient is predicted using a friction coefficient model formula created in advance, and the roll gap setting value of the rolling mill is corrected according to changes in the predicted friction coefficient. In addition, at this time, the friction coefficient μ calculated back using the rolling load formula is Ａ and the friction coefficient μ predicted by the friction coefficient model formula Ｃ The difference between the friction coefficient change ε Ａ The change in the friction coefficient ε to correct Ｃ This is predicted using machine learning. Specifically, a neural network is used for machine learning.

[0004] Japanese Patent Application Publication No. 7-246411

[0005] However, it is difficult to accurately predict the sudden change in the friction coefficient in the low-speed rolling region simply by optimizing the coefficients of the friction coefficient model formula proposed in Patent Document 1. In addition, the variables used as explanatory variables in the machine learning model do not sufficiently cover the variables related to the fluctuation of the friction coefficient, which causes a problem of low prediction accuracy of the friction coefficient.

[0006] Therefore, the present invention has been made in consideration of the above, and an object of the present invention is to provide a friction coefficient prediction method that can predict the friction coefficient between a work roll and a steel sheet with high accuracy, and a cold rolling method for a steel sheet that can suppress thickness fluctuations over the entire length of the coil including the acceleration and deceleration regions of rolling based on the obtained friction coefficient, and can satisfy strict thickness accuracy requirements, a manufacturing method for a cold-rolled steel sheet, and manufacturing equipment for a cold-rolled steel sheet.

[0007] The method for cold rolling a steel sheet according to the present invention, which advantageously solves the above-mentioned problems, is configured as follows: [1] A method for cold rolling a steel sheet, which cold rolls a steel sheet by controlling a work roll gap in a stand of the cold rolling mill when cold rolling the steel sheet using a prediction model for predicting a friction coefficient between the work rolls of the cold rolling mill and the steel sheet, wherein the prediction model is generated using rolling history data from past cold rolling of the steel sheet as explanatory variables and an estimated value of the friction coefficient as a response variable, the rolling history data including the deformation resistance of the steel sheet and the work roll diameter, and the method includes the steps of continuously predicting the friction coefficient between the work rolls of the cold rolling mill and the steel sheet by inputting rolling conditions of the steel sheet to be rolled into the prediction model, and controlling the roll gap of the cold rolling mill based on changes in the predicted friction coefficient.

[0008] The method for producing a cold-rolled steel sheet according to the present invention, which advantageously solves the above-mentioned problems, is configured as follows: [2] A method for producing a cold-rolled steel sheet, comprising a step of cold-rolling a steel sheet using the method for cold-rolling a steel sheet according to the above-mentioned [1].

[0009] The present invention provides a cold-rolled steel sheet manufacturing facility that advantageously solves the above-mentioned problems and is configured as follows: [3] A cold-rolled steel sheet manufacturing facility comprising: a cold rolling mill that cold-rolls a steel sheet; and a control device that controls a control variable of a work roll gap of the cold rolling mill, wherein the control device controls the work roll gap in a stand of the cold rolling mill when cold-rolling the steel sheet using a prediction model that predicts the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet, the prediction model being generated using rolling history data from past cold-rolling of the steel sheet as explanatory variables and an estimated value of the coefficient of friction as a response variable, the rolling history data including the deformation resistance of the steel sheet and the work roll diameter, and wherein the rolling conditions of the steel sheet to be rolled are input into the prediction model to continuously predict the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet, and the roll gap of the cold rolling mill is controlled based on changes in the predicted coefficient of friction.

[0010] According to the present invention, by using a prediction model that has been subjected to machine learning using the deformation resistance of a steel plate and the work roll diameter as additional input data, it is possible to predict with high accuracy the amount of change in friction coefficient that is used when correcting the roll gap setting value of a rolling mill.

[0011] 1 is a schematic diagram showing the configuration of a cold-rolled steel sheet manufacturing facility according to one embodiment of the present invention. FIG. 2 is a block diagram showing the configuration of a calculation unit according to the embodiment. FIG. 3 is a diagram for explaining the function of a prediction model according to the embodiment. FIG. 4 is a graph comparing a friction coefficient calculated back from a rolling load formula with a friction coefficient predicted from a simplified formula that has been used in conventional plate thickness control. FIG. 5 is a graph showing the relationship between plate thickness deviation and rolling speed when a roll gap set value is corrected by a roll gap correction method proposed in the prior art. FIG. 6 is a graph comparing the prediction accuracy of the prediction model of the prior art and the present invention. FIG. 7 is a graph showing the relationship between plate thickness deviation and rolling speed when the steel sheet cold rolling method according to the embodiment is applied. FIG. 8 is a flowchart showing the flow of roll gap condition control processing according to the embodiment. FIG. 9 is a graph showing rolling speed pattern I and pattern II used in the examples.

[0012] Hereinafter, a method for cold rolling a steel sheet, a method for manufacturing a cold-rolled steel sheet, and a manufacturing facility for a cold-rolled steel sheet according to an embodiment of the present invention will be described in detail with reference to the drawings. Note that the components in the embodiments shown below include those that are easily replaceable by a person skilled in the art, or those that are substantially the same.

[0013] <Cold-rolled steel sheet manufacturing equipment> The cold-rolled steel sheet manufacturing equipment according to this embodiment includes a cold rolling mill that cold-rolls a steel sheet and a control device that controls a control variable for the work roll gap of the cold rolling mill. The control device controls the work roll gap in a stand of the cold rolling mill when cold-rolling the steel sheet using a prediction model that predicts the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet. The prediction model is generated using rolling performance data from past cold rolling of the steel sheet as explanatory variables and an estimated value of the coefficient of friction as a response variable. The rolling performance data includes the deformation resistance of the steel sheet and the work roll diameter. The control device continuously predicts the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet by inputting the rolling conditions of the steel sheet to be rolled into the prediction model. The control device is configured to control the roll gap of the cold rolling mill based on changes in the predicted coefficient of friction.

[0014] Fig. 1 is a schematic diagram showing the configuration of a cold-rolled steel sheet manufacturing facility according to this embodiment. As shown in Fig. 1, the cold-rolled steel sheet manufacturing facility according to this embodiment (hereinafter abbreviated as "manufacturing facility") is a continuous tandem rolling line having multiple stands. The manufacturing facility includes a payoff reel 1, a joining device 2, a looper 3, a cold tandem rolling mill 4, a cutting machine 5, a tension reel 6, a tension meter 7, and a thickness meter 8.

[0015] The payoff reel 1 is a device that pays out steel sheets S. The manufacturing facility may be equipped with a plurality of payoff reels 1. In this case, each of the plurality of payoff reels pays out a different steel sheet S.

[0016] The joining device 2 is a device that joins the steel plate that was first paid out from the pay-off reel 1, i.e., the tail end of the preceding material, to the steel plate that was later paid out from the pay-off reel 1, i.e., the front end of the following material, to form a joined steel plate. A laser welding machine is preferably used as the joining device 2.

[0017] The looper 3 is a device for storing the steel sheet S so that cold rolling by the cold tandem rolling mill 4 can continue until the steel sheets are joined by the joining device 2, that is, until joining is completed.

[0018] The cold tandem rolling mill 4 is a device that cold rolls the steel sheet S to a target thickness. In this embodiment, the cold tandem rolling mill 4 includes five rolling stands, numbered a first rolling stand (#1std) to a fifth rolling stand (#5std), arranged in order from the entrance side (the right side of the paper in FIG. 2 ) of the steel sheet S to the exit side (the left side of the paper in FIG. 2 ). The configuration of the rolling stands and the conveying device for the steel sheet S are not particularly limited. In this embodiment, the cold tandem rolling mill 4 is of a type called a 4Hi type having four rolls per stand, but is not limited to this. For example, other types, such as a 6Hi type, can also be applied.

[0019] The cutting machine 5 is a device that cuts the steel sheet S into coils after cold rolling.

[0020] The tension reel 6 is a device that winds up the steel sheet S cut by the cutting machine 5. The type of tension reel 6 is not limited, and it may be, for example, a carousel tension reel. The manufacturing facility may also be equipped with multiple tension reels 6. In this case, the multiple tension reels 6 continuously wind up multiple steel sheets S.

[0021] The tension meter 7 is installed in the rolling mill, for example, on the entry side and delivery side of each rolling stand. The tension meter 7 measures the tension of the steel sheet S in the rolling mill, for example, on the entry side and delivery side of each rolling stand.

[0022] The thickness gauges 8 are installed in the rolling mill, for example, at the entry side and delivery side of each rolling stand. The thickness gauges 8 measure the thickness of the steel sheet S at the entry side and delivery side of the rolling mill, for example, at the entry side and delivery side of each rolling stand.

[0023] The arithmetic unit 9 serving as a control device for the manufacturing equipment according to this embodiment predicts the coefficient of friction between the work rolls of the rolling mill and the steel sheet S using a prediction model while the steel sheet S is being rolled. The arithmetic unit 9 controls the roll gap of the rolling mill so as to suppress changes in the sheet thickness at the delivery side of the rolling mill due to changes in the predicted friction coefficient. In this embodiment, values ​​for each stand of the tandem cold rolling mill 4 are input as explanatory variables and objective variables of the prediction model, and the friction coefficient of each stand can be predicted using the same prediction model. Furthermore, roll gap control is performed for each stand for which the friction coefficient has been predicted. The roll gap control of this embodiment is not limited to this; a prediction model may be created for each stand of the tandem cold rolling mill 4, or roll gap control may be applied to only some of the stands rather than all of the stands.

[0024] 2 is a block diagram showing the configuration of the arithmetic unit 9. As shown in FIG. 2, the arithmetic unit 9 includes an arithmetic device 91, an input device 92, a storage device 93, and an output device 94.

[0025] The arithmetic unit 91 is connected to the input unit 92, the storage unit 93, and the output unit 94 by wire via a bus wiring 95. However, the connection form of the arithmetic unit 91, the input unit 92, the storage unit 93, and the output unit 94 is not limited to a wired connection, and they may be connected wirelessly. Furthermore, they may be connected in a form that combines wired and wireless connections.

[0026] The input device 92 functions as an input port into which control information from the roll gap control device and information from the operation information device 10 are input. The information from the operation information device 10 may include information about the steel sheet S to be rolled, such as the steel type and dimensions. The information from the operation information device 10 may also include cold rolling condition information set by a process computer or an operator before cold rolling, such as numerical information, text information, and image information. Furthermore, the information from the operation information device 10 may also include rolling condition information during cold rolling, such as numerical information, text information, and image information.

[0027] The storage device 93 is configured by a known storage device such as a hard disk drive, a semiconductor drive, an optical drive, etc. The storage device 93 is a device that stores information required by the arithmetic unit 9 according to this embodiment, for example, information required to realize the functions of the arithmetic processing unit 98 described later.

[0028] The output device 94 functions as an output port that outputs a control signal from the arithmetic device 91 to the roll gap control device of the tandem cold rolling mill 4 .

[0029] The arithmetic unit 91 includes a RAM 96, a ROM 97, and an arithmetic processing unit 98. The RAM 96, the ROM 97, and the arithmetic processing unit 98 are connected to an input device 92, a storage device 93, and an output device 94 via a bus wiring 95.

[0030] The RAM 96 is a main working memory used when the arithmetic unit 91 performs some processing.

[0031] The ROM 97 stores a prediction model 97A and a prediction model execution program 97B. As shown in FIG. 3 , when creating the prediction model 97A, rolling performance data of the steel sheet S to be rolled is used as an explanatory variable. Furthermore, the difference between the friction coefficient calculated back from Equation (1) shown in Equation 1 below and the friction coefficient calculated from Equation (2) shown in Equation 2 below is used as the objective variable. Then, machine learning is performed. Here, when predicting the friction coefficient between the steel sheet S to be rolled and the work rolls, it is common to estimate the friction coefficient by back-calculating from a known rolling load formula such as the following formula.

[0032] The rolling load formula shown in the above formula (1) is expressed as follows: the rolling load P is the plate width B, the entry plate thickness H, the delivery plate thickness h, and the front tension t ｆ , rear tension t ｂ , friction coefficient μ Ａ , deformation resistance K of steel plate S ｍ and is expressed as a function of the roll radius R. In the thickness control performed during rolling of the steel sheet S to be rolled, due to the problem of response of the thickness control system, the friction coefficient μ may be calculated using a simple function of the rolling speed v and the reduction rate r as shown in Equation 2 (2) below.

[0033]

[0034] Figure 4 shows the relationship between the friction coefficient calculated from Equation (1) and Equation (2) and the rolling speed. A significant difference can be seen when comparing the friction coefficient estimated using Equation (1) based on the actual values ​​obtained when rolling the steel sheet S to be rolled with the friction coefficient estimated using Equation (2) prepared in advance based on the relationship between the rolling speed v and the reduction ratio r. Furthermore, the plot of the friction coefficient estimated from Equation (1) shows that the friction coefficient decreases as the rolling speed increases. This is thought to be because the amount of rolling oil entrained in the roll bite increases as the rolling speed increases, resulting in a decrease in the friction coefficient. Furthermore, the rate of change of the friction coefficient with respect to the rolling speed is greater in the lower speed region. At the start of rolling each coil, i.e., in the region where the rolling speed at the front end of the coil changes from low to high, the rate of change of the friction coefficient and the rate of change of the rolling load are also high. Meanwhile, near the end of rolling each coil, i.e., in the region where the rolling speed at the tail end of the coil changes from high to low, the rate of change of the friction coefficient and the rate of change of the rolling load are also high. In these regions, highly accurate thickness control is required to suppress fluctuations in delivery thickness.

[0035] Therefore, in this embodiment, instead of the conventionally used formula (2), a prediction model capable of predicting with high accuracy the coefficient of friction between the steel sheet S and the work rolls during rolling is used to suppress delivery thickness fluctuations. When creating the prediction model 97A, it is preferable to perform machine learning using as much rolling performance data as possible for the steel sheet S to be subjected to thickness control. However, the method is not particularly limited. A neural network model such as deep learning, a regression tree model such as random forest, gradient boosting such as XGBOOST, etc. can be used. Note that other known machine learning methods may also be employed.

[0036] In this embodiment, a neural network is used as the machine learning method, and therefore the prediction model 97A is a neural network model. The neural network model is expressed, for example, by a functional formula. Specifically, hyperparameters used in the machine learning model are set, and learning is performed by the neural network model using those hyperparameters.

[0037] In the hyperparameter optimization calculation, a neural network model is created by gradually changing some of the hyperparameters for the training data, and the hyperparameters that maximize the prediction accuracy for the validation data are selected. The hyperparameters typically include, but are not limited to, the number of hidden layers, the number of neurons in each hidden layer, the dropout rate in each hidden layer, and the activation function and output number in each hidden layer. Here, the dropout rate refers to the blocking of neuronal transmission with a certain probability. The hyperparameter optimization method is not particularly limited, but may include grid search, which gradually changes parameters, random search, which randomly selects parameters, or Bayesian optimization. In this embodiment, the prediction model 97A is stored in the ROM 97 in advance. However, the form is not limited thereto. For example, the prediction model 97A may be created online by providing the calculation processing unit 98 with a function for creating the prediction model 97A.

[0038] Here, the parameters used as explanatory variables for the prediction model 97A were examined. Specifically, conventionally used parameters such as rolling oil concentration, rolling oil temperature, rolling oil flow rate, rolling speed, coil length, reduction, entry thickness, and exit thickness were used. These parameters are used in the control described in Patent Document 1. In this embodiment, in addition to these, the influence of the work roll diameter, which is a variable related to the deformation resistance of the steel sheet S and the fluctuation of the contact area between the steel sheet S and the work roll, was evaluated. In general, the friction force between the steel sheet S and the work roll is related to the actual contact area. Therefore, if an assumption such as the Hitchcock equation is made and a known theoretical formula such as the Bland & Ford approximation formula is used, the smaller the work roll diameter and the lower the hardness of the steel sheet, i.e., the lower the deformation resistance, the higher the friction coefficient.

[0039] Therefore, for example, when rolling a steel plate having high hardness, if the friction coefficient is predicted without taking into account the hardness of the steel plate and the work roll diameter, a large difference may occur between the actual friction coefficient and the predicted friction coefficient. As a result, the roll gap setting value of the rolling mill calculated based on the change in the friction coefficient may not be corrected sufficiently. For example, as shown in Figure 5, depending on the steel plate, thickness deviation occurs when the rolling speed is accelerated or decelerated. In Figure 5, (a) is a graph showing the transition of the rolling speed over time, and (b) is a graph showing the transition of the thickness deviation over time.

[0040] As an explanatory variable of the prediction model, it is considered necessary to newly consider the deformation resistance of the steel sheet S because it affects the rolling load. Therefore, instead of using constants, actual values ​​are obtained in advance and are sorted for each steel type, for example, and then applied.

[0041] Furthermore, work roll diameters shrink by several tens of microns due to wear over time during use. Thermal expansion during rolling can also increase the work roll diameter by several tens of microns. These wear and thermal expansion factors can be estimated based on the rolling distance of the steel plate. Work rolls that have been used to some extent, e.g., those with wear of several tens of microns, are removed from the rolling mill, their surfaces are polished with a polishing machine, and reused within the operating range. If the difference between the upper and lower limits of this operating range diameter is 50 mm, the diameter of the work roll used for the #1 standard rolling mill shown in Figure 1 may be, for example, 580 mm or 550 mm after polishing, initially 600 mm. This value, i.e., the work roll diameter after roll polishing, is recorded in the operation information device 10.

[0042] FIG. 6 shows the relationship between the friction coefficient back-calculated from Equation (1), the friction coefficient predicted from conventional technology, for example, the method described in Patent Document 1, and the friction coefficient predicted from the prediction model used in this embodiment and the rolling speed. In the prediction model used in this embodiment, the deformation resistance of the steel sheet S and the work roll diameter were added as explanatory variables in addition to the conventional technology. The friction coefficient predicted from the prediction model used in this embodiment is nearly consistent with the friction coefficient back-calculated from Equation (1) across the entire speed range, demonstrating improved prediction accuracy of the friction coefficient compared to the conventional technology. As a result, as shown in FIG. 7, for example, it is possible to suppress thickness deviation when the rolling speed is accelerated or decelerated. In FIG. 7, (a) is a graph showing the transition of the rolling speed over time, and (b) is a graph showing the transition of the thickness deviation over time.

[0043] 2 , the arithmetic processing unit 98 has an arithmetic processing function and is wired to a RAM 96 and a ROM 97 via a bus wiring 95. The arithmetic processing unit 98 determines the rolling conditions for the steel sheet S to be rolled, i.e., the roll gap conditions. In order to perform the above processing, the arithmetic processing unit 98 executes a prediction model execution program 97B stored in the ROM 97 when it receives a signal from the roll gap control device via the input device 92 notifying that cold rolling is being performed. Thereby, the arithmetic processing unit 98 functions as an information reading unit 98A, a data preprocessing unit 98B, a roll gap condition determination unit 98C, and a result output unit 98D.

[0044] The information reading unit 98A reads control information from the roll gap control device obtained from the input device 92 and information from the operation information device 10.

[0045] The data pre-processing unit 98B executes a process for creating data to be input to the roll gap condition determining unit 98C. Specifically, the data pre-processing unit 98B executes processes such as unit conversion, deletion of coil data containing abnormal data or unacquired data, or data compensation, in order to read the rolling performance data and the rolling conditions of the steel sheet S to be rolled into the prediction model 97A.

[0046] The roll gap condition determination unit 98C reads the information obtained from the data pre-processing unit 98B into the prediction model 97A, and controls each unit to determine the roll gap conditions for the steel sheet S to be rolled. The result output unit 98D outputs the determined roll gap conditions to the roll gap control device of the tandem cold rolling mill 4 via the output device 94.

[0047] [Cold Rolling Mill Applicable to This Embodiment] In the above description, a cold tandem rolling mill has been used as an example of a cold rolling mill to which this embodiment is applied. However, the applicable cold rolling mill is not limited to a cold tandem rolling mill, and the present embodiment can also be applied to a reversing rolling mill or a Sendzimir rolling mill.

[0048] [Modifications] Although the above describes an embodiment of the present invention, the present invention is not limited to this embodiment and various modifications and improvements can be made. For example, the devices included in the manufacturing facility are not limited to those described above. Therefore, it is possible to continuously perform the cold rolling process and the preceding pickling process. A pickling device for pickling the steel sheet S may be located between the looper 3 and the cold tandem rolling mill 4. Furthermore, the cold tandem rolling mill 4 is not limited to a 4-high rolling mill but may be a multiple rolling mill such as a 5-high or 6-high rolling mill, and the number of rolling stands is not particularly limited. When creating the prediction model 97A, the friction coefficient may be directly estimated as the objective variable, rather than the difference between the friction coefficient predicted from the theoretical formula and the model formula. Furthermore, this embodiment does not need to be applied to all steel types rolled in the rolling line, and may be applied only to specific steel types, such as materials with strict thickness requirements. Furthermore, in the embodiment, roll gap control of the rolling mill was described as the target of control to suppress changes in the thickness at the delivery side of the rolling mill based on changes in the predicted friction coefficient. Roll peripheral speed control may also be performed simultaneously with this.

[0049] <Method for cold rolling steel sheet> The method for cold rolling steel sheet according to this embodiment is a method for cold rolling steel sheet, in which a prediction model for predicting the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet is used to control the work roll gap in a stand of the cold rolling mill when cold rolling the steel sheet. The prediction model is generated using rolling performance data from past cold rolling of the steel sheet as explanatory variables and an estimated value of the coefficient of friction as a response variable. The rolling performance data includes the deformation resistance of the steel sheet and the work roll diameter. The rolling conditions of the steel sheet to be rolled are input into the prediction model to continuously predict the coefficient of friction between the work rolls of the cold rolling mill and the steel sheet. The method for cold rolling steel sheet includes a step of controlling the roll gap of the rolling mill based on changes in the predicted coefficient of friction.

[0050] (Roll Gap Condition Control) The flow of the roll gap condition control process according to this embodiment will be described with reference to Fig. 8. Fig. 8 is a flowchart showing the flow of the roll gap condition control process according to this embodiment. The flowchart shown in Fig. 8 starts at the timing when a signal informing that cold rolling is being performed is input from the roll gap control device via the input device 92 (step S0). The roll gap condition control process proceeds to the processing of step S1.

[0051] In the process of step S1, the information reading unit 98A reads the prediction model 97A stored in the ROM 97. This completes the process of step S1, and the roll gap condition control process proceeds to the process of step S2.

[0052] In the processing of step S2, the information reading unit 98A reads the current rolling conditions of the steel sheet S to be rolled via the input device 92. This completes the processing of step S2, and the roll gap condition control processing proceeds to the processing of step S3.

[0053] In the processing of step S3, the roll gap condition determination unit 98C inputs the rolling conditions read in the processing of step S2 into the prediction model 97A read in the processing of step S1, and predicts the change amount Δμ in the friction coefficient. This completes the processing of step S3, and the roll gap condition control processing proceeds to the processing of step S4.

[0054] In the processing of step S4, the roll gap condition determination unit 98C predicts the amount of change ΔP in the rolling load from the amount of change Δμ in the friction coefficient continuously predicted in the processing of step S3. The amount of change ΔP in the rolling load is predicted from the amount of change Δμ in the friction coefficient using the following formula 3. This completes the processing of step S4, and the roll gap condition control processing proceeds to the processing of step S5.

[0055] Here, ΔP represents the amount of change in rolling load, a represents the influence coefficient of the amount of change in friction coefficient relative to the amount of change in rolling load, and Δμ represents the amount of change in friction coefficient.

[0056] In the process of step S5, the roll gap condition determination unit 98C predicts the amount of change in the roll gap from the amount of change in the rolling load predicted in the process of step S4. The amount of change in the roll gap is predicted from the amount of change in the rolling load using the following equation.

[0057] Here, ΔS represents the change amount of the roll gap, b represents the influence coefficient of the change amount of the rolling load on the change amount of the roll gap, and ΔP represents the change amount of the rolling load.

[0058] With the above processing, the roll gap condition determination unit 98C ends the series of roll gap condition control processing (step SE). In this way, the calculation processing unit 98 determines the roll gap conditions using the prediction model 97A from the rolling conditions of the steel sheet S to be rolled.

[0059] As is clear from the above explanation, in this embodiment, first, a prediction model 97A is created in which rolling performance data from past cold rolling of the steel sheet S is used as an explanatory variable, and the difference between equations (1) and (2) that predict the friction coefficient is used as a target variable. Then, a calculation processing unit 98 determines the roll gap conditions for the steel sheet S to be rolled from the rolling conditions of the steel sheet S to be rolled using the prediction model 97A. This makes it possible to suppress thickness fluctuations over the entire length of the coil, including the acceleration and deceleration regions of rolling, and to achieve improved thickness accuracy.

[0060] The present embodiment will be specifically described below using examples, although the present invention is not limited to these examples.

[0061] A cold rolling experiment was conducted using a cold tandem rolling mill consisting of five rolling stands according to the embodiment shown in Figure 1, using steel types A and B with a base thickness of 2.2 mm, a finished thickness of 0.35 mm, and a plate width of 1,300 mm. Steel types A and B have different deformation resistances. Table 1 shows the L value of the Swift equation. The effectiveness of the present invention was verified on multiple occasions with different work roll diameters and acceleration / deceleration patterns during rolling. Table 1 shows the steel type, work roll diameter, acceleration / deceleration pattern, explanatory variables of the prediction model used, and maximum plate thickness deviation. As shown in Figure 9, the acceleration / deceleration rate for pattern I is lower than that for pattern II. The required plate thickness deviation for both steel types A and B was within ±5 μm.

[0062] First, as a preliminary step, a prediction model for steel types A and B is created. Approximately 3,000 pieces of rolling performance data for each steel type are used as learning data, and deep learning is performed as a machine learning model. This prediction model is used in the following examples of the invention.

[0063] In No. 1, the steel plate was a high-deformation-resistance steel type A, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire coil length was 4.0 μm, achieving the required thickness accuracy. In No. 2, the steel plate was a high-deformation-resistance steel type A, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was II. The acceleration / deceleration rate was higher than in No. 1, and conditions were more likely to cause sudden fluctuations in rolling load, but the maximum thickness deviation over the entire coil length was 4.5 μm, achieving the required thickness accuracy. In No. 3, the steel plate was a high-deformation-resistance steel type A, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire coil length was 3.5 μm, achieving the required thickness accuracy. In No. 4, the steel plate was a high-deformation-resistance steel type A, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire length of the coil was 4.0 μm, achieving the required thickness accuracy.

[0064] In No. 5, the steel plate was made of low deformation resistance steel type B, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire coil length was 3.5 μm, achieving the required thickness accuracy. In No. 6, the steel plate was made of low deformation resistance steel type B, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire coil length was 4.0 μm, achieving the required thickness accuracy. In No. 7, the steel plate was made of low deformation resistance steel type B, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire coil length was 3.0 μm, achieving the required thickness accuracy. In No. 8, the steel plate was made of low deformation resistance steel type B, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire coil length was 3.5 μm, achieving the required thickness accuracy.

[0065] No. 9 uses the control system described in Patent Document 1. The steel plate is type A, which has high deformation resistance, the work roll diameter is 530 mm, and the acceleration / deceleration pattern is I. The maximum thickness deviation over the entire coil length is 9.0 μm, failing to meet the required thickness accuracy. This is believed to be due to the fact that the hardness of the steel plate and the fluctuations in the contact area between the steel plate and the work roll are not included in the explanatory variables of the prediction model, resulting in low prediction accuracy for the change in friction coefficient. No. 10 uses the same thickness control as No. 9. The steel plate is type A, which has high deformation resistance, the work roll diameter is 530 mm, and the acceleration / deceleration pattern is II. The maximum thickness deviation over the entire coil length is 10.0 μm, failing to meet the required thickness accuracy. Compared to No. 9, the thickness deviation is larger, which is believed to be due to the high acceleration / deceleration rate, which causes a sudden change in the rolling load and friction coefficient at the acceleration / deceleration section. No. 11 uses the same thickness control as No. 9. The steel plate used was steel type A, which has high deformation resistance, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire length of the coil was 7.5 μm, which did not meet the required thickness precision. The cause is believed to be the same as that of No. 9. No. 12 had the same thickness control as No. 9. The steel plate used was steel type A, which has high deformation resistance, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire length of the coil was 8.5 μm, which did not meet the required thickness precision. The cause is believed to be the same as that of No. 10.

[0066] No. 13 was subjected to the same thickness control as No. 9. The steel plate was a low-deformation-resistance steel type B, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire length of the coil was 7.5 μm, failing to meet the required thickness accuracy. The cause is believed to be the same as No. 9. No. 14 was subjected to the same thickness control as No. 9. The steel plate was a low-deformation-resistance steel type B, the work roll diameter was 530 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire length of the coil was 8.5 μm, failing to meet the required thickness accuracy. The cause is believed to be the same as No. 10. No. 15 was subjected to the same thickness control as No. 9. The steel plate was a low-deformation-resistance steel type B, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was I. The maximum thickness deviation over the entire length of the coil was 7.0 μm, failing to meet the required thickness accuracy. The cause is believed to be the same as in No. 9. No. 16 had the same thickness control as No. 9. The steel plate was type B, which has low deformation resistance, the work roll diameter was 560 mm, and the acceleration / deceleration pattern was II. The maximum thickness deviation over the entire coil length was 8.0 μm, which did not meet the required thickness accuracy. The cause is believed to be the same as in No. 10.

[0067] For Nos. 17 to 20, when the predictive model was created without including the work roll diameter but including the deformation resistance as explanatory variables and applied to cold rolling, the maximum thickness deviation was 5.5 to 7.0 μm, and the required thickness accuracy could not be met. The reason for this is thought to be that the work roll diameter was not included in the explanatory variables of the predictive model, resulting in low prediction accuracy of the amount of change in the friction coefficient.

[0068] For Nos. 21 to 24, when the predictive model was created without including deformation resistance but including work roll diameter as explanatory variables and applied to cold rolling, the maximum thickness deviation was 5.5 to 7.0 μm, and the required thickness accuracy could not be met. The reason for this is thought to be that deformation resistance was not included in the explanatory variables of the predictive model, resulting in low prediction accuracy of the amount of change in friction coefficient.

[0069]

[0070] From the above, it has been confirmed that the use of the cold rolling method for steel sheet and the manufacturing method for cold-rolled steel sheet according to the present invention to predict the change in friction coefficient during rolling with high accuracy is useful for achieving high plate thickness accuracy. Furthermore, this not only contributes to improving quality and yield, but also contributes to reducing the amount of energy used in the manufacturing process by eliminating the need to cut off parts with poor plate thickness on a separate line.

[0071] Although the present invention has been described above as an embodiment, the present invention is not limited to the description and drawings that form part of the disclosure of the present invention. In other words, other embodiments, examples, and operational techniques that can be made by those skilled in the art based on the present invention are all included in the technical scope of the present invention.

[0072] S Steel plate 1 Payoff reel 2 Joining device 3 Looper 4 Tandem cold rolling mill 5 Cutting machine 6 Tension reel 7 Tension meter 8 Plate thickness meter 9 Arithmetic unit 10 Operation information device 91 Arithmetic device 92 Input device 93 Storage device 94 Output device 95 Bus wiring 96 RAM 97 ROM 97A Prediction model 97B Prediction model execution program 98 Arithmetic processing unit 98A Information reading unit 98B Data preprocessing unit 98C Roll gap condition determination unit 98D Result output unit

Claims

1. A method for cold rolling a steel plate, comprising: using a prediction model for predicting a friction coefficient between a work roll of a cold rolling mill and a steel plate, and controlling a work roll gap in a stand of the cold rolling mill when the steel plate is cold rolled, to cold roll the steel plate; wherein the prediction model is generated using rolling performance data from past cold rolling of the steel plate as explanatory variables and an estimated value of the friction coefficient as a response variable; the rolling performance data includes the deformation resistance and work roll diameter of the steel plate; and the method for cold rolling a steel plate, comprising the steps of continuously predicting the friction coefficient between the work roll of the cold rolling mill and the steel plate by inputting rolling conditions of the steel plate to be rolled into the prediction model, and controlling the roll gap of the cold rolling mill based on changes in the predicted friction coefficient.

2. A method for producing cold-rolled steel sheet, comprising the step of cold-rolling a steel sheet using the method for cold-rolling a steel sheet according to claim 1.

3. A manufacturing facility for cold-rolled steel sheet, comprising: a cold rolling mill for cold-rolling a steel sheet; and a control device for controlling a control amount of a work roll gap of the cold rolling mill, wherein the control device controls a work roll gap in a stand of the cold rolling mill when the steel sheet is cold-rolled using a prediction model for predicting a friction coefficient between the work rolls of the cold rolling mill and the steel sheet, the prediction model being generated using rolling performance data from a past cold rolling of the steel sheet as explanatory variables and an estimated value of the friction coefficient as a target variable, the rolling performance data including the deformation resistance and the work roll diameter of the steel sheet, and the rolling conditions of the steel sheet to be rolled are inputted into the prediction model to continuously predict the friction coefficient between the work rolls of the cold rolling mill and the steel sheet, and the roll gap of the cold rolling mill is controlled based on changes in the predicted friction coefficient.

Images