JPH04111910A - Method for controlling shape of rolled stock in multistage rolling mill - Google Patents

Abstract

PURPOSE:To execute shape control with extremely high shape convergence without the generation of hunting by adapting the estimated shape of rolling material as the detected value of sheet shape at present point of time, calculating the manipulated variable of each actuator to minimize the value of overall performance function and operating each actuator. CONSTITUTION:The manipulated variable of a means 12 for pushing back-up roll (i.e., the increment for pushing the a back-up roll 5), manipulated variable of a means 13 for moving tapered roll (i.e., movement of a pair of upper and lower tapered rolls 3, 3) and manipulated variable of a means 11 for moving the drawing down position (i.e., correction that is added to the control signal (e) from a thickness controller 9) are calculated based on the detected signal (f) from a sheet shape detector 6 and the target sheet shape signal (g) that is preliminarily set with a sheet shape controller 10 and they are respectively outputted as control signals (h), (i), (d). And, the positions of the back-up roll 5 and tapered rolls 3, 3 are respectively operated by manipulated variables that is instructed according to the control signals (h), (i) with the means 12 for pushing back-up roll and means 13 for moving the tapered rolls and the sheet shape of rolled stock 1 is controlled.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a method for controlling the shape of a rolled material in a multi-high rolling mill.

(Prior technology) In recent years, in the rolling of thin plates of copper alloys, etc., in order to meet the requirements for the thickness accuracy of products, not only automatic plate thickness control is performed in multi-high rolling mills, but also high precision is applied to the plate shape. As precision is required, automatic shape control methods are being developed.

Shape control of the rolled material in this type of multi-high rolling mill is generally based on shape data detected from a shape detector installed on the exit side of the rolling mill to control the control amount of the shape control actuator unique to each rolling mill. , is calculated by a unique control algorithm built into the shape control device, and is performed by operating each actuator by the calculated control amount (for example, Japanese Patent Laid-Open No. 55-45562,
JP-A-63-16804, JP-A-63-1680
(See Japanese Patent Application Laid-Open No. 62-21814).

Among these conventional techniques, the one disclosed in Japanese Patent Application Laid-Open No. 63-16806 takes into account the time delay caused by the transfer time of rolling 10A from the roll hide to the shape detector, and calculates the shape parameters. A method of correcting the estimated values has been adopted, and it is said that this allows for a wide range of rolling conditions, a fast response from low rolling speeds, and a better plate shape. ing.

(Problem to be Solved by the Invention) By the way, the shape detector installed in this type of multi-high rolling mill detects minute tension changes in the rolled material detected by the shape detection row that rotates in contact with the rolled material. Generally, the rolling mill is configured to convert the shape into a change, and therefore, the time required to detect the shape is sometimes longer than the control cycle of the rolling mill, that is, the data sampling time interval.

Also, if there are many errors in the data from the detection roller,
In order to prevent malfunctions due to this error, a time-weighted average value of the difference between the detected shape and the target shape is sometimes calculated and so-called PI or PID control is performed. It takes more time to obtain accurate shape data, and sometimes it takes more than twice the control cycle of the plate shape.

In this way, the time delay due to time for shape detection is
Normally, this is much longer than the time delay caused by the transfer time t of the rolled material, which is often a problem in practice, but the above-mentioned conventional technology does not take this point into account. .

In other words, in the conventional technology described above, no consideration is given to the case where a time delay longer than the control cycle occurs during the control algorithm, and the control amount is directly calculated for the detected plate shape. Therefore, the shape correction could only be adjusted by adjusting the control gain, and as a result, the shape correction took a very long time. Also, conversely, control gain
If an attempt is made to correct the shape all at once, hunting will occur if there is a time delay longer than the control period.

The present invention has been developed in view of the above circumstances, and even when a time delay longer than the control cycle occurs, it is possible to prevent hunting by appropriately predicting the plate shape due to the delay. It is an object of the present invention to provide a method for controlling the shape of a rolled material in a multi-high rolling mill, which can perform shape control with extremely high shape convergence.

(Means for Solving the Problems) In order to achieve the above object, the present invention has taken the following technical means.

That is, the invention described in claim (1) is based on the difference between the plate shape of the rolled material detected by the plate shape detector installed on the exit side of the multi-high rolling mill and a preset target plate shape, and the difference between the plate shape of the rolled material and the target plate shape set in advance, and the Evaluate the outlet side plate shape and plate thickness of the rolled material using the difference between the amount of change in thickness of the rolled material in response to a change in the amount of operation of the rolled material and the preset target amount of change in thickness when changing the amount of operation of each of the actuators. A comprehensive evaluation function is defined and set in advance,
During plate shape control of the rolled material, the plate shape of the rolled material is constantly detected by the plate shape detector, and the current detection result from the plate shape detector and the past control amount of each actuator are used to determine the current state of the rolled material. Calculate the predicted shape of the rolled material at , use the predicted shape of the rolled material as the detected value of the current plate shape, calculate the operation amount of each actuator that minimizes the value of the comprehensive evaluation function, and calculate the operation amount of each actuator that minimizes the value of the comprehensive evaluation function. The method is characterized in that each actuator is actuated based on the manipulated variable to sequentially control the shape of the rolled material.

In addition, the invention described in claim (2) is based on the difference between the plate shape of the rolled material detected by a plate shape detector installed on the exit side of the multi-high rolling mill and a preset target plate shape, and Evaluate the outlet side plate shape and plate thickness of the rolled material using the difference between the amount of change in thickness of the rolled material in response to a change in the amount of operation of the rolled material and the preset target amount of change in thickness when changing the amount of operation of each of the actuators. A comprehensive evaluation function is defined and set in advance,
During plate shape control of the rolled material, the plate shape of the rolled material is constantly detected by the plate shape detector, and the current detection result from the plate shape detector, the past control amount of each actuator, and the control cycle are The predicted shape of the rolled material at the present time is calculated from the amount of change in rolling speed that occurred during the period, the change in plate thickness that occurred during the control cycle, and the amount of load change due to the change in rolling speed, and the predicted shape of the rolled material is calculated at the current time. The operation amount of each actuator that minimizes the value of the comprehensive evaluation function is calculated by employing it as the detected value of the rolling plate shape, and each actuator is operated in accordance with the calculated operation amount to determine the shape of the rolled material. It is characterized by sequential control.

(Function) In the invention described in claim (1), the predicted shape fi'(k) of the rolled material at the current time k is calculated by combining the detection result fi'(k) from the plate shape detector at the current time k and each actuator. Based on this predicted shape fi'(k), each actuator is operated to minimize the value of the overall evaluation function J. A quantity Δxj (k) is calculated.

Therefore, even if, for example, the time to obtain the detection result fi0(k) is longer than the control period,
The actual shape of the rolled material at each time point k is the same as that obtained in real time by the predicted shape fi' (k), which makes it possible to prevent hunting from occurring, and even if the control gain is increased, the shape of the rolled material remains unchanged. The convergence is extremely good.

Further, in the invention according to claim (2), the predicted shape fi
'(k), not only the past control amount Δxj(k-1) of each actuator, but also the amount of change Δ■ in rolling speed that occurred during the control cycle, change in plate thickness, change in rolling speed By also considering the amount of load change ΔP due to the change in load, the influence on the plate shape when these changes occur is also predicted in advance.

(Example) Hereinafter, one embodiment of the present invention will be described in detail based on the drawings.

FIG. 1 is an overall configuration diagram showing an apparatus to which the method of the present invention is applied, and FIG. 2 is a front view of a multi-high rolling mill to which the method of the present invention is applied. This example shows a case where the method of the present invention is applied to a 20-high rolling mill.

In Fig. 1.2, 1 is a rolled material that is a thin plate, 2 is a pair of upper and lower work rolls that contact the rolled material 1, 3 is a first intermediate roll that is a taber roll installed behind the work roll 2, and 4 is a A second intermediate roll 5 is installed behind the first intermediate roll 3, and 5 is a backup roll installed further behind the second intermediate roll 4. These rolls 2 to 5 constitute a 20-high rolling mill. .

Further, numeral 6 denotes a plate shape detector which is arranged at a downstream position slightly away from the 20-high rolling mill and detects the elongation (plate shape) of the rolled material 1 in the rolling direction. In the example, it is configured by arranging n shape sensor elements.

7. 8 is a plate thickness gauge that detects the entrance plate thickness and exit plate thickness of the rolled material 1, which are placed at appropriate positions on the upstream and downstream sides of the 20-high rolling mill, respectively; 9 is based on plate thickness gauges 7 and 8; A plate thickness control device outputs the operation amount as a control signal e to an appropriate number of roll reduction position moving means (actuators for plate thickness control) 11 based on the detection results, and controls the plate thickness control device 10 as appropriate based on the detection results by the plate shape detector 6. This is a plate shape control device that outputs an operation amount to a number of backup roll pushing means 12 and Taber roll moving means 13 (both actuators for plate shape control) to control them.

In this embodiment, with the apparatus having such a configuration, control of the plate shape of the rolled material 1 by the method of the present invention is performed as follows.

First, the plate thickness control device 9 performs normal feedforward type plate thickness control and feedback type plate thickness control based on detection signals a and b from the plate thickness gauges 7 and 8 and a preset target exit side plate thickness signal C. The operation amount is calculated by thickness control and a control signal e is output. This control signal e changes the operation amount of the back unroll pushing means 12 and the Taber roll moving means 13 by adding the control signal d which is the operation amount calculated by the plate shape control device 10 described later. Corrections will be made taking into account changes in plate thickness caused by the above. After such correction, the control signal is output to the roll reduction position moving means 11, the roll reduction position in the 20-high rolling mill is operated by the instructed operation amount, and the thickness of the rolled material 1 is controlled. .

On the other hand, the plate shape control device 10 controls the amount of operation of the backup roll pushing means 12 (i.e., increases the amount of pushing of the backup roll 5) based on the detection signal f from the plate shape detector 6 and the preset target plate shape signal g. Quantity) ΔX
, ~ΔX4, and the operation amount of the tapered roll moving means 13 (that is, the amount of movement of the pair of upper and lower Taper rolls 3, 3) ΔXS+
ΔX6 and the operation amount of the roll reduction position moving means 11 (i.e., the correction bone added to the control signal e from the plate thickness control device 9) ΔX7+Δx8 are calculated, and the control signals i and i are calculated.
, d.

Then, the backup roll pushing means 12 and the tapered roll moving means 13 send control signals, respectively.
Buffer roll 5 by the amount of operation specified according to i
The positions of the tapered rolls 3 and 3 are operated, and the rolled material 1
The shape of the plate is controlled.

By the way, the characteristic part of the present invention is that the plate shape control device 1
The operation amount ΔX, which is performed at 0, is in the calculation means of Δ×8. The calculation means will be explained in detail below.

That is, the plate shape control device 10 has a comprehensive evaluation function J that evaluates the outlet side plate shape and plate thickness of the rolled material 1 using the following equation (4).
are defined and set in advance.

This comprehensive evaluation function J is calculated based on the difference between the outlet side plate shape from the plate shape detector 6 and a preset target plate shape, and the amount of change in plate thickness corresponding to the change in the operation amount of each actuator 11 to 13. It is defined using the difference from the target plate thickness change amount when changing the operation amount of each actuator 11 to 13.

Here, the difference between the exit side plate shape and the target plate shape is calculated using the following formula (1).
The error shape ei(k), ei(k) − fio(k) − fi”(k) (i =
1 to n) - (1), as shown in the following equation (2), the difference between the current error shape ei (k) and the error shape one point before the current time ei (k) - ei (k-1 ) is calculated as the weighted sum of

ε1(k)=to+・ei(k) +Kp・I:ei(
k)-ei(k-1))(i-1~n)...[2) In other words, this equation (2) indicates that so-called PI control is performed regarding the deviation signal εi (k), Although this lengthens the shape detection time, residual deviations are removed and malfunctions due to detection errors can be prevented.

In addition, the amount of plate thickness change corresponding to the change in the operation amount of each actuator 11 to 13'' and the preset value of each actuator 11
The difference from the target plate thickness change amount at the time of changing the operation amount of ~13 is determined by the following formula (3).

εn++(k)=L++'(k)-f+m"(k)...
・(3) In this case, the comprehensive evaluation function J is calculated by the following formula (4
) is as follows.

, J−Σ ei (k)”・匈12 ・・
-(4) However, fio(k) (i = 1 to n) is the plate elongation value measured at the time by the i-th shape sensor element constituting the plate shape detector 6, fi''(k) (i =
l to n) are the target plate elongation values at time k for the i-th shape sensor element, f7 and +. (k) is the amount of plate thickness change, f, with respect to the change in the operation amount of each actuator 11 to 13. 11k) is the target plate thickness change amount when changing the operation amount of each actuator 11 to 13, wi (i = 1 to n −1
-1) is the weighting coefficient for the deviation ε1(k), K, ,
are the difference between the current error shape ei (k) and the error shape one point before the current time ei (k), respectively.
-ei is a weighting factor for (k-1).

In addition to introducing such an evaluation function J, in this example, the influence coefficient formula of the plate shape and plate thickness of the rolled material 1 due to the change in the operation amount of each actuator 11 to 13 is expressed by the following formula (5).
Create it like this.

(i-1 to n+1) ... (5) However, Δxj
(k) (j -1 to m; m-8 in this embodiment) is the amount of change in the operation amount of each actuator 11 to 13 at the current moment k to be determined here, Δxj (k-/) (l-1 to M )
is the amount of change in the operation amount of each actuator 11 to 13 at the current point in time, K.

is the prediction gain for the above Δxj(k-1''), Δfi
(k”) (i-1 to n) are each actuator 11 to 13
The predicted shape change amount detected by the first shape sensor element when the operation amount of is changed by Δxj (k), Δf7
゜1(k゛) is the predicted amount of plate thickness change detected by the plate thickness gauge 8 when the operation amount of each actuator 11 to 13 is changed by Δxj (k), αji(j=1 to m, i= 1~n
+1) is the influence coefficient of Δxj (k) on Δfi(k').

In this way, in this embodiment, each actuator 11 to 1
By incorporating the past output data Δxj (k-1) in 3 as a weight into the conventional influence coefficient equation as a series sum, the shape change due to the detection time delay is predicted in advance, and the present invention is advantageous in this respect. It has the biggest feature.

Furthermore, how much past data should be considered, that is, (5
), if the detection time delay is 8T and the control period is Δt, then M - ΔT/Δ
It is considered that it is sufficient if M
= 3.

In addition, the prediction gain for Δxj(k-1>) is a parameter representing the delay of the detected shape, and may be selected as appropriate for the time delay in the shape detector. In this example, KI=xz =L=1.0.

Then, in equations (2) to (4), fi'(k) −fi”(k)−Δf i (k)(i
−1 to n+1) ... (6) is substituted, and the plate shape detector 6 detects every moment during the plate shape control so that the overall evaluation function J of the plate thickness and plate shape becomes the minimum every moment. Detected plate shape values fi'(k), fio(k-
1) Based on (i = 1 to n), each operation amount change amount Δxj (k) (j-1 to m) is calculated by the following algorithm, and each actuator 11 to 13 is operated.

Now, the deviation signal ε1(k) (i-1 to n+1) is expressed as ε1(
k) − to +・ei(k)+Kp・(ei(k)−ei(
k-1)) ei(k)-fi'(k) fi''(k)
ei(k-1)=fi0(k-1)-fi"(k-1)
−(7)(i −1~n) ε□I(k) = .

When each actuator 11 to 13 is moved by Δxj (k), the comprehensive evaluation function J is expressed as (8). In order to minimize this comprehensive evaluation function J, it must be as follows. That is, from equations (8) and (9),
...00) (However, s=1 to m). By solving this equation 00) for Δxj (k), the amount of change in the operation amount of each actuator 11 to 13 for minimizing the comprehensive evaluation function J regarding the plate thickness and plate shape can be obtained.

By the way, to express the above equation 00) from another perspective, when predictive control was not performed in the past, apart from the operation in equation (7) above, the measured value fi0(k) was used as it was for shape control. However, when performing predictive control in this embodiment, instead of using the measured value fi0(k) as it is, the following equation 00
) This means the same thing as using the value of the predicted shape fi'(k) defined by the equation.

fi' (k) −fio(k)+Σ αji°(ΣK
k'Δxj (k-ZN...00) In other words, in this embodiment, the past output data ΔxDk is added to the measured value fi'(k) at the current point k that is sequentially obtained.
−/), add the sum of the attached series to the weight of the predicted shape fi'(
k), and it can be said that shape control is performed based on this fi'(k).

According to the above-mentioned algorithm, when the shape is extremely defective, such as at the beginning of rolling, the control target signal level becomes excessive, and some of the actuators 11 to 13 are unable to follow the target signal due to constraints due to response characteristics. Therefore, in this embodiment, the plate shape control device 10 performs the following steps (1) to (2) to prevent the actuators 11 to 13 from being unable to follow the target signal.

(2) Calculate the movable speed of each actuator 11 to 13 based on a function defined in advance according to rolling conditions (rolling speed, rolling load).

■Calculate the movable limit value constrained by the position limit from each actuator's current position.

■Each actuator 11 to 13 determined from the movable speed
The movable limit value per control cycle is calculated.

■Set the smaller of the movable limit values obtained in steps (■) and (2) as the final movable limit value.

(2) Based on the error shape between the detected shape from the plate shape detector 6 and the target shape, the movement amount target value of each actuator 11 to 13 that minimizes the comprehensive evaluation function J is calculated as described above.

■If there is an actuator for which the target value calculated in the previous step ■ exceeds the movable limit value determined in step Calculate the amount of shape change when moving to the movable limit value, subtract it from the current error shape, exclude the relevant actuator from the usable actuators, and minimize the overall evaluation function J in step ① of the remaining actuators. Find the movement target value and change the movement limit. This process is repeated until there are no actuators subject to the movement limit restriction, or until all actuators 11 to 13 are available.

■Multiply the control gain to obtain the final value for each actuator 11
Calculate the moving target value of ~13.

In this example, the operation amount of each actuator 11 to 13 obtained in this way is expressed as Δxj(k) (j = 1 to 8)
Based on the above-mentioned roll rolling position moving means 11.backup roll pushing means 12. The plate shape of the rolled material 1 is controlled by the tapered roll moving means 13.

Next, FIGS. 3 and 4 show experimental results obtained by applying the method of the present invention to shape control of an actual rolled material, and experimental results obtained when shape control was performed using a conventional method, respectively.

Here, experiments were conducted using thin plates made of copper alloy (RFC) with a width of 630 mm and thicknesses of 0 and 5 mm.

In addition, in the figure, 1.1-unit indicates that the elongation in the rolling direction of a rolled material having a length of 1 m is longer than the reference value by 10-5 m. In addition, (Ds L) is the lateral direction on the drive side, (Ws L) is the lateral direction on the work side, (Cr#1~#4) is the crown position, respectively.
(Ws T) indicates each control amount in the tilt (tilt downward) direction on the work side, and the horizontal axis indicates time, and one square indicates one minute.

As shown in these figures, according to this example, so-called hunting does not occur and the time is very short (approximately 1 to 10 minutes).
It is clear that extremely high convergence performance is obtained in about 2 minutes).

Next, the theoretical validity of the shape prediction adopted in this example will be explained by actually analyzing the simplest modeled problem.

That is, here, the explanation will proceed assuming that i = 1 and j = 1, and the problem of controlling the shape of the rolling mill Q is assumed to be a one-dimensional problem of a virtual wire rod, as shown in FIG.

In the figure, 21 is a work roll, 22 is a virtual wire,
23 indicates a shape detector that detects the shape of the wire 22;
Measured value f at the current moment k detected by the detector 23
(k), the control amount Δx(k) is set within the shape control device 24.
is calculated, and the shape control of the wire rod 23 is performed according to the calculated value. Also, roll 21 at the current moment k
Let the actual shape immediately below be fR(k).

Under the above assumptions, the control system that progresses over time is shown in a table as shown in Table 1 below.

(Next leaf) Table 1 (Generally fR(k) -fR(k -1) -gf(k
-1)) where g is the control gain and α is the influence coefficient when the control amount is changed by Δx (k).

Then, in the shape control device 24, the measured value f (k)
Assuming that it takes ΔT to obtain ΔT sequentially, and assuming that the time delay 8T is about twice the control period Δt, then f (k)−ge(k−2)(k≧4) ・...(11)
The following relationship holds true, and the initial three time points k = 1.2.3
In this case, f(1)-f(2)-f(3)=C (constant) is detected.

Here, if the initial value fR(1) -f(1) of the actual shape is used, and fR(k) is sequentially calculated in the case where predictive control as in this embodiment is not performed, the following is obtained. Become. However, for k≧4, the above equation (11) is used.

r”(1) −fR(1) =C fR(2) =fR(1) gf(1)−(1−g)
CfR(3) −fR(2)−gf(2) = (1−
g) C-gc-(1-2g)CfR(4) -fR(3
) −gf (3) = (1, −2g) −gC=
(1-3g)CfR(5) -fR(4)-gf (4
) −(1-3g)C-gfR(2)=(1-4g+
g2) CrR(6) = fR(5) −gf (5)
-(1-4g+g2)C-gfR(3)-(1-5g+
3g2) C fR(7) -fR(6)-gf(6) = (1-5
g+3g”)C-gfR(4)=<1-6g+6g”
)C fR(8) =fR(7)-gf(7)-(1-6g
+6g”)C-gfR(5)-(1-7g+10g2-
g')C r"(9)=f"(8) gf(8)-(1-7g4
40g2-g')C-gf! (6)-(1-8g+15
g2-4g3)C rR(10) = fR(9) -gf (9) =
(1-8g+15g2-4g') C=gfR(7)=
(1-9g+21g''-10g3)C Then, calculate fR(k) as above, for example, g=0.5,C
If plotted with =1.0, the result will be as shown in FIG. 6, and it can be clearly seen that hunting occurs when predictive control is not performed.

Next, a description will be given of how the same problem as above occurs when the predictive control according to this embodiment is used.

Now, let f''(k) be the predicted shape at the current moment k, and express it by the actual measurement value f(k). From the above equation θ0)°, the following equation 02) is obtained. However, regarding the weight, are all set to 1.0.

f'(k) = f(k)+αΔx(k-1) + (X
Δx(k-2)-02) Also, conventionally, the actual measured value f (k)
Instead of using the shape as it is to control the shape, this predicted shape f'(k) is used, so αΔx(k) −−gf' (k) ...(13)
holds true, and from this equation 02) (13), the following equation is obtained.

f' (k) = f (k) - gf' (k-1)
-gf' (k-2) ...θ leakage Also, in this case,
Similarly to the above, if we tabulate the control system that progresses over time, it will look like the following (Table 2): Table 2 -fR(k-1)-gf'(k-1))fR(k) =
f'(1) -f(1)=C (If the constant is calculated sequentially, fR(k) will be obtained as shown below (generally fR(k) and the initial value number).

fR(1) -f' (1) =C fR(2) =fR(1) -gf' (1) -(1
fR(3) −fR(2) −gf' (2) = (
1fR(4) -fR(3) -gf' (3) =
(1g)C g)C-g(f(2)-gf' (1))g)C-g(
1-g)C g) 2G g) 2C-g (f (3) -gf' (2)g
)2C-g(1-g)2G g) 3C gf' (1)) fR(5) -fR(4) -gf' (4) -(1
-g)3C-g(f(4)-gf"(3)-gf'
(2)) - (1-g>3C-g(1-g)3C -(1-g) 'C Therefore, it is clear that f R(k) = (1-g)''・C below. If g = 0.5 is substituted here, fR(k) will converge in a hexaseries manner.

By the way, this means that the above-mentioned equation (II)04) and the control relational equation fR(k) -f R(k-1) -gf'
(k-1) ...05).

That is, from the above formula 05), gf' (k-1) = fR
(k-1)-f''(k), and by substituting this into formula 04, f'(k) = f(k)-gf'(k-1)-gf'(
k-2)-f(k)-(fR(k-1)-fR(k)
1-(fR(k-2)-fR(k-1))-f(k)
+(fR(k)-fR(k-2)).

Here, from the formula (11) f(k) −fR(k-2), f'(k) = fR(k-2)+ (fR(k
)-fR(k-2)1=fR(k), and in the end, the predicted shape f''(k) and the actual shape fR(k)
will match.

Therefore, this f'(k) −fR(k) is the original 05)
Substituting into the equation, f R(k) = fR(k-1) - gfR(k-1
)−(1−g)・fR(k−1), and fR(k) =(1−g) l′−′・f”(
1) is obtained.

As can be seen from the above explanation, as long as the formula 00102) for predictive control adopted in this embodiment is used, eventually,
This means that the current actual shape of the rolled material, which would have been obtained directly under the work roll, is obtained in real time, regardless of the time loss to obtain the measured value, and the control gain can be increased. This results in shape control with extremely excellent convergence performance without causing any hunting.

Although one embodiment of the present invention has been described above, the present invention is not limited to this.

For example, instead of the above equation 00), the predicted shape fi'(k
) may be formulated as the following equations 00 and 0'7).

+βi·ΔV...06) However, 8) is the amount of change in rolling speed that occurred during the control cycle, and βi is the coefficient of influence on the shape by the ΔV at the i-th shape sensor element.

+γi・ΔP ・・・07) However, Δ
P is the amount of load change due to changes in plate thickness and rolling speed that occur during the control cycle, and γi is the influence coefficient on the shape of the i-th shape sensor element due to ΔP.

By formulating the predicted shape fi'(k) in this way, it is possible to grasp the shape disturbance due to changes in rolling speed and plate thickness, and changes in load due to changes in rolling speed as a predicted shape before measurement. Disturbances in the plate shape due to changes in rolling speed and load are prevented.

Furthermore, in the above embodiment, as shown in equation (2), ε1(
k) was given as the weighted sum of the difference between the current error shape and the error shape one point before the current time, but the following equation 03
), it may be replaced with a weighted sum of the current and past error shapes.

e1(k) = Ko ・ei(k)+K, -ei
(k-1) + K2 ・ei(k-2) ten...
(i-1 to n) ...08) In this case, by appropriately setting the weighting coefficient, so-called PID11 control can be performed, and a clearer deviation signal εi (k
) can be obtained, however, the calculation time for shape detection becomes longer, so it is preferable to set the value of M in the above-mentioned formula (5)2 or 00) to a larger value accordingly.

In addition, although the said Example demonstrated the case where the method of this invention was applied to a 20-high rolling mill, the method of this invention is not limited to this.

(Effects of the Invention) As explained above, according to the invention set forth in claim (1), the predicted shape of the rolled material at the present time is based on the current detection result from the plate shape detector and the past control amount of each actuator. is calculated, and based on the predicted shape, the operation amount of each actuator that minimizes the value of the comprehensive evaluation function is calculated, and each actuator is operated based on this operation amount to sequentially control the shape of the rolled material. Even when the time required to obtain a detected shape value is so long as to exceed the control cycle, it is possible to control the shape of the rolled material with extremely high shape convergence without causing so-called hunting.

Further, according to the invention as set forth in claim (2), when forming the predicted shape of the rolled material described above, the amount of change in rolling speed or the change in plate thickness that occurs during the control cycle, and the change in load due to the change in rolling speed. Since the amount is also taken into account, disturbances in the plate shape due to changes in rolling speed or load can be prevented.

[Brief explanation of the drawing]

1 to 6 show a method for controlling the shape of a rolled material using a multi-high rolling mill as an embodiment of the present invention. Figure 2 is a front view of a multi-high rolling mill to which the method of the present invention is applied;
5 and 4 are graphs showing the experimental results when the predictive control of this embodiment was performed and when it was not performed, respectively.
The figure shows the configuration of the rolled material in the simplest model, and FIG. 6 is a graph showing the state of hunting. 1... Rolled material, 2... Work roll, 3... First
Intermediate roll, 4... Second intermediate roll, 5... Backup roll, 6... Plate shape detector, 7, 8... Plate thickness gauge, 9... Plate thickness control device, 10...・Plate shape control device, 11... Roll reduction position moving means (plate thickness control actuator), 12... Backup roll pushing means (plate shape control actuator), 13... Taber roll moving means (plate shape control actuator). Hmm (LS 0 me ψ 5

Claims (2)

[Claims] (1) The difference between the plate shape of the rolled material detected by the plate shape detector installed on the exit side of the multi-high rolling mill and the preset target plate shape, and the change in the operation amount of each actuator. A comprehensive evaluation function is defined and set in advance to evaluate the outlet side plate shape and plate thickness of the rolled material using the difference between the thickness change amount and the target plate thickness change amount when changing the operation amount of each actuator set in advance. , During plate shape control of the rolled material, the plate shape of the rolled material is constantly detected by the plate shape detector, and the current detection result from the plate shape detector and the past control amount of each actuator is used to determine the current value of the rolled material. The predicted shape of the rolled material is calculated, and the predicted shape of the rolled material is used as the detected value of the current plate shape to calculate the operation amount of each actuator that minimizes the value of the comprehensive evaluation function. 1. A method for controlling the shape of a rolled material in a multi-high rolling mill, comprising operating each actuator based on an operation amount to sequentially control the shape of the rolled material. (2) The difference between the plate shape of the rolled material detected by the plate shape detector installed on the exit side of the multi-high rolling mill and the preset target plate shape, and the change in the operation amount of each actuator. A comprehensive evaluation function is defined and set in advance to evaluate the outlet side plate shape and plate thickness of the rolled material using the difference between the thickness change amount and the target plate thickness change amount when changing the operation amount of each actuator set in advance. , During plate shape control of the rolled material, the plate shape of the rolled material is constantly detected by the plate shape detector, and the current detection result from the plate shape detector, the past control amount of each actuator, and the control The predicted shape of the rolled material at the present moment is calculated from the amount of change in rolling speed that occurred during the rolling period, the change in plate thickness that occurred during the control cycle period, and the amount of load change due to the change in rolling speed. The operation amount of each actuator that minimizes the value of the comprehensive evaluation function is calculated by employing it as the detected value of the plate shape in A method for controlling the shape of a rolled material in a multi-high rolling mill.