JP2668845B2 - Method of controlling shape of rolled material of Sendzimir 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 of controlling a shape of a material to be rolled when a material to be rolled is rolled by a Sendzimir rolling machine, and particularly to a steepness obtained by classifying an output signal from a shape detector by a neural network. The present invention relates to a method for controlling the first intermediate roll when the second degree steepness degree pattern is included in the degree pattern.

[0002]

2. Description of the Related Art There is a strong demand for cost reduction by improving the quality of rolled sheet materials and improving production efficiency. In response to such requirements, rolling mills have been multi-staged and various rolling control methods have been developed. . The 20-high Sendzimir rolling mill is best known as one of the above-mentioned multi-high rolling mills, and the outline of this rolling mill is shown in FIG.

As is apparent from FIG. 6, a 20-high Sendzimir rolling mill 1 comprises a pair of opposing work rolls 2, a set of four first intermediate rolls 3 sandwiching the work rolls, and a first intermediate roll 3. A set of six second intermediate rolls 4 sandwiching the roll 3
And the upper and lower 6 arranged so as to sandwich the second intermediate roll 4
It is composed of two backup rolls 5 and one set of two As-U rolls 6, and FIG. 7 shows the action of the intermediate first roll 3 of the 20-high Sendzimir mill 1.

The first intermediate roll 3 having a taper cut on one side edge portion of the roll is a plate width direction of the material 7 to be rolled (see FIG. 7).
It is possible to control the shape of the end portion (ear portion) in the plate width direction by shifting in the direction indicated by arrow A).

Conventionally, the shape control of the material to be rolled in the 20-high Sendzimir rolling mill is performed by the above-mentioned As-U roll and the first roll.
Although it was performed by manually operating the intermediate roll, in recent years, the shape detector 10 which detects the shape in the width direction of the rolled plate material is used to quantitatively measure the shape, and the shape detector 10
A method has been developed in which the above two types of actuators are operated in response to output signals from the automatic control to automatically control the shape of the material to be rolled. Among them, automatic shape control by neural network and phage-reasoning is a pattern of plate flatness (type of elongation such as ear extension, belly extension, second extension).
, And each actuator is operated in accordance with the pattern to control the flatness satisfactorily. This is described, for example, in "Hitachi Review" Vol. 73, No.
8 (August 1991), pp. 21-28.

According to this, for example, as shown in FIG.
(H) The eight characteristic patterns are stored. When the shape detection value is measured, pattern detection of the shape detection value is performed, and a feedback signal is output in order to obtain a degree of conformity to the characteristic pattern. Input to the neural net. The neural network outputs the degree of classification into the characteristic pattern of the current plate shape. When it is recognized that the degree of (a) is the highest and the degree of (e) is medium, the As-U roll is operated as a result of fuzzy inference, and then the first intermediate roll 3 is output. Such as shifting
Appropriate shape control is performed by prioritizing the operation parts and simultaneously performing calculation and control for determining the operation amount.

The shape detection value of the material to be rolled is obtained by calculating the steepness of each point in the width direction of the material to be rolled and patterning the steepness in FIG. 8, where the steepness is As shown in FIG. 9, when the material to be rolled is placed on a plane, and the height at which the material to be rolled floats above the plane is h, the ratio to the distance 1 in the longitudinal direction of the material to be rolled: h / It is defined as 1 × 100 (%). Therefore, at a place where the steepness is high, extra elongation is performed due to extra rolling, and conversely, at a place where the steepness is low, the elongation is small, indicating that a high tension acts during rolling.

[0008] The steepness is obtained and patterned.
One of the patterns is a “second extension” pattern. This "No. 2
The "stretch" pattern refers to a number of shapes arranged on the exit side of the rolling mill.
When the shape of the rolled material is detected by the shape detector, the center of the strip width
Consider only one of the plate width sides (1/2 of the plate width)
The steepness at the edge of the plate is λE, and the steepness at the center of the plate width is λC.
The width is 25 mm or more closer to the center of the width than the width, and the width is
Largest steepness in the area 100mm or more from the center and closer to the plate edge
If the degree is λ N, it means a pattern in which the conditions of λ N> λ C and λ N> λ E are satisfied.
The elongation of the steepness λ N part when standing is called “second elongation”.
U. A specific example of the “second elongation” pattern is shown in FIG.
(F), the above “No. 2 elongation” pattern
Apart from that, usually the first intermediate roll 3 is moved outward.
By controlling the shape of the material to be rolled (specifically,
7 Upper right and lower first intermediate rolls respectively to the right
And either left or right or left
Control the shape of the material to be rolled by moving it to one side.
).

[0009]

SUMMARY OF THE INVENTION
Recognizing the "second elongation" pattern, the conventional first intermediate roll
Is satisfactory even if it is controlled to move
Shape control may not be possible. It is the first intermediate roll
Is moved excessively.

If the first intermediate roll is moved excessively, the ear extension may be larger than the second elongation before control. The first intermediate roll 3 is moved inward in order to improve the stretched ear length again. Thus the first
When the intermediate roll 3 is moved inward, the state returns to the initial state, and the second extension occurs again. In order to make this good again, the first intermediate roll is moved to the outside, and thereafter,
The same phenomenon control is repeated, and the shape of the material to be rolled becomes a state in which the second elongation and the edge elongation are alternately repeated, and the conventional technology has a problem that stable control and the shape of the material to be rolled cannot be obtained. .

This problem will be described in more detail with reference to a steepness graph showing the shape of the material to be rolled rolled by the conventional shape control method.

The steepness shown in FIG. 10 is 1.60% 2.
When the pattern in which the extension has occurred is detected, the first intermediate roll that influences the shape of the right side plate edge in FIG. 7 is moved outward by the conventional neural network and fuzzy inference (that is, in FIG. 7). The shape of the material to be rolled is controlled by moving the first intermediate roll 3 in the upper stage to the right), and the resulting shape is shown in FIG.

As is clear from FIG. 11, the steepness at the position where the maximum steepness of the second elongation is shown is reduced to 1.50% close to the target shape, and the elongation pattern is also the second elongation. No longer a pattern, ear steepness is 1.31%
From 1.90% to 1.90%, the pattern changes to ear extension.

Since such a change is not desirable, the pattern shown in FIG. 12 is obtained as a result of moving the first intermediate roll 3 on the side that influences the plate edge shape on the right side against the edge extension in FIG. The steepness is improved from 1.90% to 1.46% in the ears, but the sharpness of the No. 2 stretch is deteriorated from 1.50% to 1.58%, exceeding the target shape. It has returned to the second growth pattern.

Further, as a result of moving the first intermediate roll on the side that influences the plate edge shape on the right side again to the No. 2 elongation in FIG. 12 to the outside, the pattern changes to that shown in FIG. The stretch is steep at 1.58% to 1.47%
It has been improved to 1.84% from 1.46% to 1.84.
% Again, and again changed to a pattern of ear extension, and after that, it was repeated as it was, and the shape and control became very unstable.

As described above, when the first intermediate roll is moved by the conventional shape control method, the ear may be larger than the second elongation before the control, and the generated ear may be generated. When the first intermediate roll is moved inward to improve the elongation again, the first intermediate roll returns to the initial state again, and returns to the initial state. The same phenomenon and control are repeated thereafter, and there is a problem that stable control and the shape of the material to be rolled cannot be obtained.

[0017]

Means for Solving the Problems Accordingly, the present inventors have
Upon investigating the cause of the repetition of the second elongation and the edge elongation, it was found that the excessive movement of the first intermediate roll during the control of the first intermediate roll caused the above-mentioned first In order to prevent excessive movement of the intermediate roll, when the pattern of the second elongation is separated by the neural network, attention is paid to the steepness of the end in the width direction at that time, and this value is constant with respect to the target shape. When the ratio becomes equal to or more than the ratio of 2nd elongation control, a condition to set the output amount of movement control to the outside of the first intermediate roll, which is a rule for the second elongation control and obtained by fuzzy inference, is added, We obtained the knowledge that problems such as elongation and instability of the obtained shape can be solved.

The present invention has been made on the basis of such knowledge, and the shape of the material to be rolled is detected by a large number of shape detectors arranged on the exit side of the Sendzimir rolling mill, and the output from the shape detector is detected. The signal is separated into several steepness patterns by a neural network, and the
The degree to which each steepness pattern fits is obtained as a fuzzy amount, and the shift control amount of the As-U roll and the shift of the first intermediate roll due to the saddle eccentricity of the Sendzimir rolling mill are calculated by fuzzy reasoning. In a rolled material control method for a Sendzimir rolling mill, which determines a control amount to control the shape of the rolled material, the No. 2 elongation steepness pattern is classified from a large number of steepness patterns, and the strip width is determined. If the steepness value at the end of the direction is equal to or more than a certain ratio with respect to the steepness value of the target shape, the condition that the amount of movement control output to the outside of the first intermediate roll becomes zero is added. The present invention is characterized by a method for controlling a material to be rolled of a mill.

The method for controlling the rolled material of the Sendzimir rolling mill of the present invention will be described in more detail.

When the steepness value of the edge portion in the strip width direction of the separated No. 2 stretch pattern is λE and the steepness value of the edge portion in the strip width direction of the target shape is λ, (i) λE ≧ λα (however, α Is a predetermined constant, the value of which is determined by empirical rules and differs depending on the type of material, for example, 0.9 for 18-8 stainless steel, 0.8 for 13 chrome stainless steel, and generally 0. .9 or less), the first intermediate roll is not moved,
(ii) When the condition of λE <λα is satisfied, the first intermediate roll is moved according to the result of fuzzy inference.

When the condition is added so that the first intermediate roll operates as described above, the state in which the No. 2 elongation and the ear elongation are frequently repeated unlike the conventional case is eliminated, and stable control is performed. .

FIG. 1 shows an example of a program when the shape control method for a material to be rolled according to the present invention described above is carried out by a microcomputer. This program is started at a constant cycle, for example, 3 seconds.

[0023]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to FIGS. 2, 3, 4 and 5.

FIG. 2 is a graph showing the shape of the material to be rolled before control, showing that the second elongation is 1.55% in the steepness, and the steepness at the end in the sheet width direction at this time is λE. = 1.41%
It is. The target steepness at the end in the sheet width direction: λ is 1. in FIG.
50%, and in the case of 18-8 stainless steel, α = 0.
90, λE = 1.41, λα = 1.50 × 0.
90 = 1.35, so in the case of FIG. 2, λE ≧ λα
Since the condition of is satisfied, the outward movement control amount of the first intermediate roll obtained by fuzzy inference becomes 0, and the movement of the first intermediate roll is not performed. After this,
The shape after about 1 minute is shown in FIG. 3, and according to FIG. 3, the second elongation is not desirable because it hardly changes from 1.55% to 1.58% with a steepness exceeding the target shape. However, the steepness λE at the end in the sheet width direction is also 1.41.
% From 1.45%, which is far better than the conventional control examples shown in FIGS.
It is also understood that no significant change is seen in the shape even after about 1 minute has passed, and both the control and the shape of the rolled material are stable.

Further, when the shape of the rolled material before control is in the state shown in FIG. 4, the steepness λE of the edge portion in the plate width direction is 0.98%.
At this time, λα is 1.50 × 0.90% = 1.35
Therefore, λE <λα, and thus the output amount of movement control to the outside of the first intermediate roll obtained by fuzzy inference is output as it is, and the movement of the first intermediate roll is performed according to the result of fuzzy inference. After this, about 40
A graph of the shape of the rolled material after a lapse of seconds is shown in Fig. 5. According to this graph, the second elongation is 1.71% in terms of steepness.
To 1.50%, and the steepness of the plate width direction end portion deteriorates from 0.98% to 1.04%, but it can be seen that both are below the target shape and are good.

[0026]

According to the present invention, output signals from a large number of shape detectors arranged on the outlet side of a rolling mill are classified into a steepness pattern by a neural network, which is obtained as a fuzzy amount, and shape control is performed by fuzzy inference. In addition to the conventional shape control for performing the above, when the No. 2 stretch pattern is detected by the neural network when trying to control the No. 2 stretch, the steepness of the plate width direction end portion at that time is focused,
If this value exceeds a certain ratio with respect to the target shape, the condition is set to 0 for the outward movement control amount of the first intermediate roll obtained by the above fuzzy inference, so that it becomes higher than the conventional one. It enables precise shape control,
It has excellent effects such that the shape and control of the material to be rolled are more stable than before.

[Brief description of the drawings]

FIG. 1 shows a program of a method for controlling a shape of a material to be rolled according to the present invention.

FIG. 2 is a graph showing the shape of a material to be rolled before being subjected to rolling control.

FIG. 3 is a graph showing the shape of the material to be rolled after being rolled by the method for controlling the shape of the material to be rolled of the present invention.

FIG. 4 is a graph showing a shape of a material to be rolled before being subjected to rolling control.

FIG. 5 is a graph showing the shape of the material to be rolled after being rolled by the method for controlling the shape of the material to be rolled of the present invention.

FIG. 6 is an explanatory diagram showing a roll arrangement of a 20-high Sendzimir rolling machine.

FIG. 7 is an explanatory diagram for explaining the action of the intermediate first roll of the 20-high Sendzimir rolling machine.

FIG. 8 is an explanatory diagram for explaining a conventional rolled material shape control method using a neural network and fuzzy inference.

FIG. 9 is an explanatory diagram for explaining the definition of steepness.

FIG. 10 is a graph showing the shape of the material to be rolled before being subjected to rolling control.

FIG. 11 is a graph showing the shape of a material to be rolled rolled by a conventional method for controlling the shape of a material to be rolled.

FIG. 12 is a graph showing the shape of a material to be rolled before being subjected to rolling control.

FIG. 13 is a graph showing the shape of a material to be rolled rolled by a conventional method for controlling the shape of a material to be rolled.

[Explanation of symbols]

 1 20-high Sendzimir rolling mill 2 Work roll 3 1st intermediate roll 4 2nd intermediate roll 5 Backup roll 6 As-U roll 7 Rolled material 8 Dividing roll 9 Roll axis 10 Shape detector

Claims (1)

(57) [Claims] 1. A shape of a material to be rolled is detected by a number of shape detectors arranged on the exit side of a Sendzimir rolling mill, and an output signal from the shape detector is classified into several types of steepness patterns by a neural network. That was sorted
Seeking steepness pattern matching degree of les as fuzzy amounts, the difference in the Zenjimia mill by fuzzy inference
A rolled material control method for a Sendzimir rolling mill, wherein the As-U roll shift control amount and the first intermediate roll shift control amount due to dollar eccentricity are determined to control the shape of the rolled material. When the second elongation steepness pattern is present among the several types of steepness patterns sorted out, and the steepness value at the end in the plate width direction is a certain ratio or more to the steepness value of the target shape, A method for controlling the shape of a material to be rolled in a Sendzimir rolling mill, characterized in that a condition that an output amount of movement control to the outside of the first intermediate roll is 0 is added.