JP2867885B2 - Roll gap setting method in rolling mill - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for setting a roll gap of a rolling mill, which can set the roll gap of the rolling mill with high accuracy by learning a two-dimensional deformation resistance model formula.

[0002]

2. Description of the Related Art Conventionally, there are the following techniques for setting a roll gap in consideration of deformation resistance of a material to be rolled. The rolling load p is expressed by the equation (1) using the function f ₁ and the advanced rate f
x is given by equation (2) using the function f ₂ . p = f ₁ (H, h, σf, σb, Km, R, μ) (1) fx = f ₂ (H, h, σf, σb, Km, R, μ) (2) where H: stand Entry side plate thickness h: Stand exit side plate thickness σf: Stand exit side tension stress σb: Stand entry side tension stress R: Roll radius Km: Average deformation resistance of rolled material μ: Friction coefficient

Note that p, H, h, and p in the equations (1) and (2)
For σf, σb, R and fx, actual values collected from online are used, and the average deformation resistance Km,
The friction coefficient μ is calculated backward by simultaneously combining the following equations (3) and (4) that embody the equations (1) and (2).

[0004]

(Equation 6)

[0005]

(Equation 7)

In the technique disclosed in Japanese Patent Application Laid-Open No. 61-222618, the actual average deformation resistance Kma calculated back by the above method, the predicted deformation resistance Kmc and the normalized correction coefficient Zka (= Kma / Kmc) And a correction coefficient Zk for each stand by a known exponential smoothing method.
Is calculated. This correction coefficient Zk is used as a correction coefficient of the deformation resistance of each stand in the calculation of the setup after the second steel change, and a new predicted deformation resistance is determined from Zk · Kmc. Make the settings for Similar processing is performed for the friction coefficient.

In the technique disclosed in Japanese Patent Application Laid-Open No. 58-65506, the base material deformation resistance Kst is obtained from the rolling performance data of the first stand, and the average deformation resistance Km of each sundado is calculated based on the value. 5) Estimate according to equation. Km = Kst · f (ε _m ) · g (εv) (5) where Kst: base material deformation resistance ε _m : logarithmic strain ε _V : strain rate

[0008] The rolling load Pc is predicted using the average deformation resistance Km, and the error from the actual rolling load Pa is calculated as a friction coefficient error Δμ.
The roll gap of the next rolled material is set by learning the error for each stand.

[0009]

By the way, in the former method, the deformation resistance is corrected for each stand by determining the correction coefficient value unique to the stand although the deformation resistance originally has no value specific to the stand. The following inconveniences occur.

FIGS. 6A and 6B show true deformation resistance curves,
Taking a deformation resistance model equation, the actual average deformation resistance is a graph showing the relationship between the predicted deformation resistance of the following rolled material, the deformation resistance of the logarithmic strain ε _ai ~ε _di of i stand each horizontal axis and the vertical axis Shown. 6 (a) and 6 (b), solid lines are deformation resistance model formulas, broken lines are true deformation resistance curves, crosses indicate actual average deformation resistances during learning, and circles indicate predicted deformation resistances for the next rolled material, respectively. Is shown.

[0011] Now, as shown in FIG. 6 (a), the logarithmic strain ε _ai, each actual average deformation resistance corresponding to the ε _bi ... ε _di is a value indicated by each × mark at the time of learning, for example, a logarithmic strain ε
There is an error of DerutaKm _i between the actual average deformation resistance corresponding to _ai and deformation resistance model equation, and obtain the correction coefficients Zi to correct this. However, after learning, the logarithmic distortion of the material to be rolled becomes ε _ai ′, as shown in FIG.
In the case of changing to ε _bi ′, ε _ci ′, ε _di ′, for example, by multiplying the deformation resistance Km at the logarithmic strain ε _ai ′ by the correction coefficient Zi to obtain the predicted deformation resistance indicated by ○, the following material to be rolled is obtained. An error Δi occurs between the predicted deformation resistance and the deformation resistance given by the true deformation resistance curve.

The latter method has the following problem. That is, generally, there is a relationship as shown in FIG. 7 between the deformation resistance and the logarithmic distortion during cold rolling. Figure 7 is a graph showing the relationship between the deformation resistance Km and logarithmic strain epsilon _m,
The horizontal axis shows logarithmic strain ε _m and the vertical axis shows deformation resistance Km. In the graph, the solid line shows the relationship between high carbon steel, and the broken line shows the relationship between low carbon steel. It is found that there is a close relationship between the apparent deformation resistance Km and logarithmic strain epsilon _m As from this graph. However, in the latter conventional technique, the learning of f (ε _m ) in equation (5), that is, the relationship between logarithmic distortion and deformation resistance is not performed.
It is inevitable that an error occurs in the estimated deformation resistance after the stand.

[0013] Figure 8 is a graph showing the relationship between the deformation resistance Km and logarithmic strain epsilon _m, a number strain epsilon _m in the horizontal axis, also is shown taking deformation resistance Km on the vertical axis. In the graph, the solid line is the predicted deformation resistance, the x mark is the true deformation resistance of the first stand,
The circles indicate the true deformation resistance of the second to fourth stands. As is apparent from this graph, based on the true deformation resistance of the first stand, even if the deformation resistance of each stand at the subsequent stage is predicted as shown by a solid line, if there is a deviation in the prediction, the deviation goes to the subsequent stage. The error between the predicted deformation resistance and the true deformation resistance is increased, and the error becomes ΔKm at the fourth stand, so that accurate prediction of the rolling load cannot be performed.

The present invention has been made in view of the above circumstances, and its purpose is to grasp the deformation resistance model formula as a function of logarithmic distortion or total draft, and to convert the parameters into rolling performance data of a plurality of stands. It is another object of the present invention to provide a method for setting a roll gap in a rolling mill in which learning is performed based on this and a roll gap is set for the next rolled material based on the learning.

[0015]

According to a first aspect of the present invention, there is provided a method for setting a roll gap in a rolling mill, comprising the steps of:
The thickness, advance rate, and rolling load of at least two or more stands are measured, and the coefficient of friction between the rolling roll and the material to be rolled and the deformation of the material to be rolled are measured using the measured values and the theoretical rolling equation. Calculate the resistance, express the two-dimensional deformation resistance model formula as a function of logarithmic distortion or total draft based on the calculation result, learn this function, and set the roll gap for the next rolled material based on the learning result. It is characterized by.

According to a second aspect of the present invention, there is provided a method for setting a roll gap in a rolling mill, comprising the steps of:

[0017]

(Equation 8)

As a feature, learning for obtaining parameters a ₁ and a ₂ is performed.

A method for setting a roll gap in a rolling mill according to a third aspect of the present invention is a method of setting a two-dimensional deformation resistance model formula before learning.

[0020]

(Equation 9)

The deformation resistance model formula after learning is

[0022]

(Equation 10)

In this case, learning is performed to obtain a _1n and a _2n according to the following equations.

[0024]

[Equation 11]

[0025]

(Equation 12)

[0026]

According to the first invention, the two-dimensional deformation resistance model formula is expressed as a function of logarithmic distortion or total draft, and by learning this, there is an error in the actual rolling data. Even when the deformation resistance is to be obtained, the erroneous learning is suppressed because it is governed by the previous learning.

According to the second aspect of the present invention, the initial value of the roll gap can be accurately set by learning the deformation resistance model equation as a function equation of logarithmic distortion having the highest dependence on the equation.

According to the third aspect of the invention, it is possible to learn a two-dimensional deformation resistance model formula in real time by a simple method of solving a determinant.

(Principle) The principle of the present invention will be described. When performing the initial setting of the rolling rolls, first, for a tandem mill of N stands (using i as a generalized stand number), the rolling load of the i-th stand #i is Pi, the roll peripheral speed vri, the i-th stand exit side plate. The actual average deformation resistance Km _i is calculated from the rolling performance data such as the rolling load formula and the advanced ratio formula (fi = (vi-vri) / vri) using the rolling performance data such as the speed vi and the i-th stand exit side plate thickness hi. I do.

[0030] and the actual average deformation resistance of each stand _{# 1~ # N Km 1, Km} 2 ... Km i ... Km N. From each of the actual average deformation resistances, a two-dimensional deformation resistance model formula is expressed as a function of average logarithmic distortion or total draft by the following method,
Learn this. Average logarithmic distortion εm ₁ , εm ₂ ... εm _i
... .epsilon.m _N is given by matrix (material to be rolled before rolling), the plate thickness h ₀ and exits each stand from the side thickness _{h 1, h 2 ... h N} (6) formula.

[0031]

(Equation 13)

On the other hand, according to the study by the inventors, for materials having substantially the same hot rolling finish temperature history, the deformation resistance Km of cold rolling is calculated by using the logarithmic strain ε _m as shown in equation (7) (two-dimensional deformation resistance model). ).

[0033]

[Equation 14]

[0034] Thus, using the actual average deformation resistance Km _i,
The parameters of the two-dimensional deformation resistance model equation of equation (7) are learned using the evaluation function J of equations (8) and (9) so that the evaluation function J is minimized, and the parameters a ₁ and a ₂ are determined. Determine.

[0035]

(Equation 15)

Substituting equation (8) into equation (9), a _1n , a _2n
Is moved to the left side, the equation (10) is obtained. That is,
Solving equation (9) is equivalent to finding a simple solution as shown in equation (10).

[0037]

(Equation 16)

[0038]

[Equation 17]

In the equation (3), the case where the two-dimensional deformation resistance model equation is expressed as a logarithmic distortion function has been described. However, the present invention is not limited to this, and the total reduction ratio (1−stand exit side plate thickness / base metal plate) (Thickness), and this may be learned.

FIG. 1 is an explanatory diagram of the learning contents described above.
The horizontal axis shows logarithmic strain ε _m and the vertical axis shows deformation resistance Km. The broken line in the figure is the deformation resistance model formula before learning,
The solid line indicates the deformation resistance model formula after learning, and the crosses indicate the actual average deformation resistance. Δε is a constant that determines how much the actual average deformation resistance is considered.

In the learning, an error area which is an area of a region U indicated by a point between a pre-learning deformation resistance model formula shown by a broken line and a post-learning deformation resistance model formula shown by a solid line in FIG. , The sum of the error area of the region W shown by hatching between the deformed resistance model formula after learning shown by the solid line and the actual average deformation resistance shown by the crosses (shown by the formula (8)) is minimum. Means that the parameters a ₁ and a ₂ are determined so that

In this manner, the following equation (11), which is a deformation resistance model equation after learning in which the parameters a ₁ and a ₂ are determined.
The rolling load for each stand is determined based on the formula and the exit side plate thickness of each stand, the roll gap for each stand is determined from the gauge meter formula, the roll gap for each stand is set, and the next rolled material is set. May be prepared for rolling.

[0043]

(Equation 18)

[0044]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below with reference to the drawings showing the embodiments. FIG. 2 is a schematic view of a tandem rolling mill composed of five stands to which the method for setting a roll gap in a rolling mill according to the present invention is applied.
Reference numerals # 5 to # 5 denote first to fifth stands having a work roll and a backup roll, respectively, reference numeral 2 denotes a reduction device provided for each of the stands # 1 to # 5, reference numeral 3 denotes a load cell, reference numeral 4 denotes a thickness gauge, and reference numeral 5 denotes a sheet speed. Reference numeral 6 denotes a tension meter for detecting the tension between the stands # 1 to # 5 and the entrance / exit side, and reference numeral 10 denotes an arithmetic control unit. The material to be rolled 1 is supplied to the rolling mill in the direction of the arrow, and the roll gap is set under the arithmetic and control unit 10 at each stand.
The plate thickness is controlled at 1 to $ 5.

Next, the learning process and the roll gap setting process in the arithmetic and control unit 10 will be described with reference to the flowchart shown in FIG. FIG. 3A is a flowchart showing the process of the learning process. First, the arithmetic and control unit 10 determines the rolling loads detected by the load cells 3, the thickness gauge 4, the sheet speed meter 5 and the tensiometer 6, The thicknesses of the inlet and outlet sides of # 1 to # 5, the sheet speeds of the inlet and outlet sides of each stand # 1 to # 5, and the tension between each of the stand # 1 to # 5 are taken into account. (Step S ₁ ), the actual average deformation resistance Km _i and the actual friction coefficient μ are calculated (Step S ₂ ), and the two-dimensional deformation resistance model formula is logarithmically distorted (or the total pressure reduction) as shown in Expression (7) according to these. Rate), learning to obtain the parameters a ₁ and a ₂ of the deformation resistance model equation based on the actual rolling data, and updating them to a _1n and a _2n (step S ₃ ).

FIG. 3B shows the roll gap of each stand.
5 is a flowchart showing a calculation setting process of FIG.
Determine the target thickness for each exit side of the stand (step
S ₁₁), Parameter a₁, A_TwoA_1n, A_2nUpdated to
Based on the deformation resistance model equation after learning as shown in equation (11),
Deformation resistance Km for each stand_i(Step
S₁₂), According to the rolling theory formulas such as the rolling load formula and advanced rate formula
Calculate the rolling load for each stand (Step S₁₃),following
Each stun according to the gauge meter formula as shown in formula (12)
Calculate the roll gap for each node (Step S₁₄), FIG.
The setting is performed by the pressure reduction device 2 shown in FIG.

S = h−P / M (12) where S: roll gap h: thickness of each stand exit side P: rolling load M: mill rigidity

The following effects are obtained in such an embodiment. FIG. 4 is an explanatory diagram corresponding to FIG. 1, in which the horizontal axis represents logarithmic distortion ε _m and the vertical axis represents deformation resistance Km. In the figure, the broken line indicates the deformation resistance model formula before learning, the solid line indicates the deformation resistance model formula after learning, and the crosses indicate the actual average deformation resistance.

In the method of the present invention, the error area which is the area of the region U indicated by a point between the deformed resistance model formula after learning indicated by the solid line and the deformed resistance model formula before learning indicated by the broken line is added. And the error area of the hatched region W between the learned deformation resistance model formula shown by the solid line and the actual average deformation resistance marked with x is given by the formula (8).
Is determined so as to minimize the following equation. Generally, the actual average deformation resistance to be monotonically increased with the increase of the logarithmic strain ε _m is shown in FIG.
Even if an abnormal actual average deformation resistance that does not increase monotonically as shown by A occurs, erroneous learning is suppressed by considering the error area between the deformation resistance model formula before learning and a high learning resistance. Reliability is obtained.

FIG. 5 is a graph showing test results obtained by performing rolling by applying the method of the present invention and the conventional method to a cold tandem rolling mill. The horizontal axis represents the number of rolled materials, and the vertical axis represents rolling. The load prediction accuracy (actual value / predicted value) is shown.
In the graph, the solid line shows the results according to the method of the present invention, and the broken line shows the results according to the conventional method. As is apparent from this graph, when the method of the present invention is used, the prediction accuracy of the rolling load is greatly improved.

In such an embodiment of the present invention, FIG.
As shown in Fig. 1, the deformation resistance model formula after learning and the
The area of error from the model resistance model equation and the deformation resistance model after learning
The sum of the error area between the model and the actual average deformation resistance is minimized.
Parameter a₁, A _TwoTo decide
Therefore, considering the previous deformation resistance model formula, it is shown in FIG.
As a result, the actual rolling data is not normal,
Even if uniform deformation resistance is obtained,
Erroneous learning is sharply governed by the
Learning of deformation resistance model formula is performed stably without performing
It is.

Further, by learning the deformation resistance model equation as a function equation of logarithmic distortion having the highest dependence, the initial value of the roll gap can be accurately set. Furthermore, (10)
Since it is a simple method to solve the equation, it is possible to learn the deformation resistance model equation in real time. In the examples, the tandem rolling mill was described, but in the case of the reversible rolling mill, the same as that of rolling at the subsequent stand of the continuous rolling mill as the number of times of rolling of the same material to be rolled is advanced. Is also applicable.

[0053]

According to the first aspect of the invention, the deformation resistance model formula is expressed as a function of logarithmic strain or the total draft, and learning is performed to obtain its parameters. The gap can be determined stably and accurately, and the thickness accuracy and the yield can be improved.

According to the second aspect of the present invention, the roll gap can be accurately set by using the deformation resistance model formula as a function formula of logarithmic distortion having the highest dependency.

According to the third aspect of the present invention, since the learning of the parameters is performed by finding the solution of a predetermined equation, the deformation resistance model equation can be learned in real time, and the roll gap can be quickly set.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram showing the contents of learning of a deformation resistance model equation.

FIG. 2 is a schematic view of a tandem rolling mill to which the method of the present invention is applied.

FIG. 3 is a flowchart showing a processing procedure of the arithmetic and control unit.

FIG. 4 is an explanatory diagram showing the effect of learning of a deformation resistance model equation according to the method of the present invention.

FIG. 5 is a graph showing test results of the method of the present invention and the conventional method.

FIG. 6 is an explanatory view showing a drawback of the conventional method.

FIG. 7 is a graph showing a relationship between deformation resistance and logarithmic distortion.

FIG. 8 is an explanatory diagram showing a drawback of the conventional method.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Rolled material 2 Roll-down device 3 Load cell 4 Thickness gauge 5 Sheet speed meter 6 Tensiometer 10 Operation control device

Continuation of the front page (72) Inventor Toshio Ishibashi 4-5-33 Kitahama, Chuo-ku, Osaka-shi, Osaka Sumitomo Metal Industries, Ltd. (56) References JP-A-64-75111 (JP, A) JP-A-64 -62205 (JP, A) JP-A-60-250816 (JP, A) JP-A-60-15010 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) B21B 37/16- 37/20

Claims (3)

(57) [Claims] In a continuous rolling mill, at least two or more stands are measured for the thickness at the entrance and the exit, an advance rate, and a rolling load, and a rolling roll and a material to be rolled are calculated using these measured values and a rolling theoretical formula. And the deformation resistance of the material to be rolled are calculated. Based on the calculation result, a two-dimensional deformation resistance model formula is expressed as a logarithmic strain or a function of the total draft, and this function is learned. A roll gap setting method comprising setting a roll gap for a next rolled material. 2. The two-dimensional deformation resistance model formula is given by _2. The roll gap setting method according to claim 1, wherein learning for obtaining parameters a ₁ and a ₂ is performed. 3. A two-dimensional deformation resistance model formula before learning is expressed by the following equation. The transformed resistance model formula after learning is given by 2. The roll gap setting method according to claim 1, wherein learning is performed to obtain a _1n and a _2n according to the following equation. (Equation 4) (Equation 5)