JPS61193716A - Control method of rolled plate thickness of strip - Google Patents

Abstract

PURPOSE:To correct the inlet side plate thickness influence coefficient and controlled valuable influence coefficient and to control the plate thickness with high accuracy by applying a Kalman filter for the difference in the control data of a control time and the control data before the fixed control time. CONSTITUTION:The inlet side plate thickness influence coefficient to minimize the outlet side plate thickness deviation and control quantity influence coefficient are calculated by applying Kalman filter to these differences DELTAH1(K), DELTAU(K) and DELTAH(K) by finding the difference in the deviation DELTAh1(K) in the inlet side measured plate thickness against the inlet side reference plate thickness and the deviation DELTAh1(K-1) in the inlet side measured plate thickness before the fixed control period, the controlled valuable DELTAU(K) of the strip part which obtains the inlet side measured value and the difference DELTAH2(K) between the outlet side plate thickness deviation DELTAh2(K) against the target plate thickness of the strip part which gives the controlled valuable and the outlet side plate thickness deviation DELTAh2(K-1) before the fixed control period, both influence coefficients are corrected by the calculated value and the controlled valuable is determined by the corrected influence coefficient.

Description

[Detailed Description of the Invention] “ [Industrial Application Field] The present invention relates to a method for controlling the rolled thickness of a strip, and in particular,
The present invention relates to a feed-forward plate thickness control method that controls the thickness of the plate on the exit side to a target plate thickness by controlling the reduction amount, tension, etc. according to the plate thickness on the input side of a rolling stand.

[Conventional technology]

Strip rolling, for example, Sendzimir rolling mill (hereinafter referred to as 2M
When rolling special steel such as stainless steel using a mill, the rolling reaction force cannot be accurately determined due to the structure of the mill. or by controlling the tension in the strip (forward tension control or rear tension control) at least either downstream or downstream.

In such plate thickness control, the plate thickness cannot be measured directly below the rolling rolls, so the plate thickness measurement must be performed at a point on the exit side (downstream side) of the rolling rolls. Therefore, there is a distance difference between the thickness measurement point (thickness measurement point on the exit side) and the control point (directly below the roll), and there is a delay time in control for the time the rolled material moves this distance, so the normal Feedback control cannot achieve a highly accurate control system. Therefore, in this type of plate thickness control, instead of normal feedback control,
Feedforward control, which measures the plate thickness at the entrance side (upstream side) of the rolling rolls and controls the roll reduction amount or tension in accordance with the measured value, has become mainstream.

The control gain of feedforward control is generally determined from the characteristics of the rolling mill and the rolled material. Those characteristics include;
This includes the characteristics of the rolling mill main body, the characteristics of the rolling device and tension control device, the characteristics of the control circuit that drives and energizes these devices, and the characteristics of the rolled material. Some of these characteristics cannot be known in advance with sufficient accuracy. For example, the deformation resistance of a rolled material differs from material to material even in the same type of material, and further changes depending on rolling conditions such as the amount of reduction during rolling and the rolling speed. Further, even the characteristics of the rolling mill main body, such as the spring constant of the rolling mill, may change due to replacement of rolling rolls. These fluctuation factors that cannot be grasped in advance cause control errors in feedforward control.

Therefore, the applicant of this application first applied for Japanese Patent Application No. 56-125036 (
In Japanese Patent Application Laid-Open No. 58-25807), the main purpose is to improve the accuracy of plate thickness control in feedforward control, and also to prevent plate thickness deviation due to fluctuation factors during rolling, and to prevent thickness deviation due to fluctuation factors during rolling. A feedforward type rolling plate thickness control method has been disclosed that is intended to correct excess or deficiency of Ifi on-line.

Since the present invention is based on the technology disclosed in this prior application, this prior art will be briefly explained.

In general, the characteristics of plate thickness rolling by a rolling mill are analyzed using many theories collectively called rolling theory, and the following approximate equation (1) holds true within one rolling bus of a reversible rolling mill such as a 2M mill. It has been known.

K-δP/δh2 where: Δh2(k) Thickness deviation on the two-person side Δtl2(k) : Thickness deviation on the exit side Δ5(k) : Rolling control amount ΔT(k) : Tension control amount Ii λ : Influence on the board thickness on the inlet side Coefficient ν : Roll reduction amount influence coefficient ξ : Tension influence coefficient P : Rolling pressure K : Mill spring constant From formula (1), if λ, ν and ξ can be accurately grasped,
Corresponding to the inlet plate pressure deviation Δh r (k), the control amount (change N) ΔS (k
) and ΔT (k) can be determined in advance. λ, ν and ξ
can be calculated in advance under the rolling conditions of each pass based on rolling theory, but as described above, these values include errors due to various factors.

Therefore, attention is paid to the errors in λ, ν, and ξ, and these are corrected by learning control to values that can be considered to have the smallest error at each control point. Simply, (Δh of equation 11
+ (k), Δh2(k) are obtained by measurement, Δ5(k), ΔT(k) are obtained by the control system, and λ, ν, and ξ are assumed to be variables, and three consecutive data are calculated. By substituting λ3 into equation (1) to obtain three equations and solving them simultaneously, the coefficient λ3. ν1. ξ
, and add the current entry side thickness deviation Δh + (k) to equation (1), which is composed of these coefficients, and add Δh 2 (k) =
0 is introduced and the current Δ5(k) and ΔT(k) are determined as the controlled variables, and these controlled variables and their rolling results Δh2(k) and Δh1(k) are substituted into the +11 formula again to create a new Create one equation, replace it with the one based on the oldest data of the three equations above, solve them simultaneously, and get the current coefficient λ2. ν2. Find ξ2, calculate the control amount as described above, measure and control Δh1(k), Δh2(k)
By repeating the measurements, it is possible to set a more or less reliable influence coefficient.

However, this alone may introduce a large amount of noise such as transient noise, and there is a risk that control stability may be impaired. Therefore, this prior invention has high stability,
This is an application of the Kalman filter, which is known as a state estimation method for linear discrete value systems. That is, the above-mentioned prior invention applies (applying a Kalman filter to Equation 11, measuring the plate thickness deviation of the material to be rolled, and finding a control gain that minimizes it,
The control gain in predictive control is corrected, and excesses and deficiencies in the control amount due to changes in the material being rolled during rolling, errors in estimating the rigidity of the rolling mill and roll system, etc. are corrected online using learning control.

[Problem that the invention seeks to solve]

In the conventional method as described above, the plate thickness measurement value and control amount at one control timing are taken into the Kalman filter, so there is no direct current caused by the thermal expansion of the rolls during rolling or the drift of the plate thickness detector that measures the plate thickness. Component disturbances are not removed, and as a result, there is a risk that the process of estimating each coefficient may be delayed or incorrect estimation results may be provided. In addition, since the feedforward control method is used, the roll reduction control amount during rolling has a nearly one-to-one correspondence with the entry side plate thickness variation, and therefore the entry side plate thickness deviation Δh + (+<) and the roll reduction control amount ΔS.
The correlation of (k) is very high. Therefore, in the method described above in which the Kalman filter is applied as is, assuming that each coefficient to be estimated is independent for such feedforward control, each influence coefficient λ.

There is a risk that an error will occur in the estimation of ξ and ξ.

The present invention improves the drawbacks of the prior art, and
The DC component of the disturbance is removed by taking the difference between the plate thickness measurement value and the control amount, and the reduction amount influence coefficient is calculated from the relationship between the entry side plate thickness influence coefficient λ and the reduction amount influence coefficient ν (λ + ν = 1). ν is expressed by the entrance side plate thickness influence coefficient λ, which is expressed as (1
) by substituting it into the equation and eliminating ν to make each coefficient estimated by the Kalman filter independent, increasing the accuracy of the plate thickness in feedforward control and further preventing plate thickness deviations due to fluctuation factors during rolling. It also aims to correct on-line any excess or deficiency in the control amount caused by fluctuation factors.

[Means for solving problems]

In order to achieve the above-mentioned objects, the present invention provides a strip rolled plate in which at least one of the roll reduction amount and the strip tension is controlled as a control amount in accordance with the inlet side plate thickness in a direction that matches the outlet side plate thickness. In the thickness control method, the deviation Δh of the measured board thickness on the entry side from the reference board thickness on the entry side
1(k) and the deviation Δh of the inlet side measured plate thickness before a certain control period
, (k-1), and the difference ΔU between the control amount ΔU (k) of the strip section from which the input side measurement value was obtained and the control amount ΔU (k-1) before a certain control period. (k), the outlet side plate thickness deviation Δh2(k) with respect to the target plate thickness of the strip section given the control amount, and the outlet side plate thickness deviation Δ before a certain control period.
Find the difference ΔH2(k) with h2(k-x) and calculate these differences ΔH1(k) and ΔU (k).

and ΔH, (k) to calculate the entry side thickness influence coefficient and control amount influence coefficient that minimize the exit side thickness deviation, and use these calculated values to change the input side thickness influence coefficient and control amount that were used until then. The quantity influence coefficient is corrected, and the controlled quantity is determined using the revised influence coefficient.

According to the second aspect of the present invention, which incorporates both the roll reduction amount and the strip tension as control variables, the deviation Δh1(k) of the measured board thickness on the entry side with respect to the reference board thickness on the entry side and the difference Δh1(k) before a certain control period The difference ΔH1(k) between the deviation of the measured board thickness on the entrance side Δh, (k-1), the reduction control amount ΔS (k) of the strip section where the entrance side measurement value was obtained, and the reduction control amount before a certain control period The difference ΔS (k) from ΔS (k-1), the tension control amount ΔT (k) of the strip section from which the input side measurement value was obtained, and the tension control amount ΔT (k-1) before a certain control period. Difference ΔT (
k) and the difference ΔHz (k) between the outlet side plate thickness deviation Δh2(k) with respect to the target plate thickness of the strip section given the control amount and the outlet side plate thickness deviation Δh2(k−1) before a certain control period. The difference between the input side plate thickness deviation ΔHI (k) and the reduction control amount difference ΔS (k), the difference between the exit side plate thickness deviation ΔH2 (k) and the reduction control amount ΔS (
k) and the difference between the tension control amount T (k)
A Kalman filter is applied to calculate the entrance plate thickness influence coefficient and tension influence coefficient that minimize the exit side plate thickness deviation, and the reduction amount influence coefficient is calculated from the input side plate thickness influence coefficient. The entrance plate thickness influence coefficient, tension influence coefficient, and reduction amount influence coefficient are corrected, and the control amount is determined using these revised influence coefficients.

Further, according to the third invention, the control amount is only the roll reduction amount, and the deviation Δh of the measured board thickness on the entry side with respect to the standard board thickness on the entry side.
+ (k) and the deviation Δ of the inlet side measured plate thickness before a certain control period
The difference ΔHI (k) between Difference Δ
S (k), the difference ΔH2(k) between the outlet side plate thickness deviation Δha(k) with respect to the target plate thickness of the strip portion given the reduction control amount and the outlet side plate thickness deviation Δh2(k−1) before a certain control period.
), and calculate the difference between the difference ΔH1(k) in the thickness deviation on the inlet side and the difference ΔS(k) in the reduction control amount, the difference ΔH2(k) in the thickness deviation on the outlet side and the reduction ? hll <1r
A Kalman filter is applied to the difference between the J amount and the difference ΔS (k) to calculate an inlet thickness influence coefficient that minimizes the outlet thickness deviation, and then a reduction amount influence coefficient is calculated from the inlet thickness influence coefficient. , the entrance plate thickness influence coefficient and the reduction amount influence coefficient that had been used up to that point are corrected, and the reduction control amount is determined using these revised influence coefficients.

FIG. 1 shows hardware for implementing the control method according to the present invention. (motor, cylinder, etc.) 19. The control of the first drive device D1 is controlled by the input/output device I10 of the reduction amount control device 31, which is controlled by the control device (computer) 40. Additionally, the thickness of the strip 10 is determined by controlling the tension in the strip. Tensioning is effected by means of a second drive 21, for example by controlling the speed between the stands of the rolling rolls. The second drive control device D2 is controlled by a tension control device 33 which is controlled by a control computer 40. Detectors 15 and 17 detect the thickness of the inlet and outlet sides of the strip 10, respectively, and their detection signals are input to the A/D converter (A/D) of the control device 40. The computer 40 includes a CPU that interprets and executes instructions and controls the entire system, and a read-only ROM that stores fixed data.
, and an RA that writes or reads data at any time.
It has M.

Hereinafter, the method for controlling plate thickness according to the first aspect of the present invention will be described as an example in which feedforward control is performed using both the roll reduction amount and the strip tension as control variables according to the plate thickness measurement value at the entrance side of the rolling machine. explain in detail.

Here, the equation (λ+C1) representing the relationship between the entrance side plate thickness influence coefficient λ and the reduction amount influence coefficient ν is defined as Equation (5).

λ+ν=1 (5) First, equation (1) is expressed as follows using the difference between each control data.

Here, the plate thickness control system is considered to be a discrete time control system, assuming that there may be some interference, and the equation (6) in the first stage is expressed as the following equation (7) by applying a Kalman filter.

Y(k) =h(k) X(k) +w(k)
...(7) Y(k)=ΔH21t=k・Δt
...(8) ΔHl (k) = Δh, (
k)-Δh, (k-1)...(11)ΔH
2(k)=Δh1(k)−Δh2(k−1)...
・(12)Δ5(k)=ΔS (k)−ΔS (k−1
) ... (13) ΔT (k) = ΔT (k) −
ΔT (k-1) ... (14) Xl Insect λ
...(15)XZ Mushiν
...(1-6)X, Δ
ξ...(17)w(k
) : whiteness noise with variance C1 h' : transposition of vector The state variables of variables λ, ν, ξ of the controlled system are expressed as follows.

X (k+1) = rb, K (k) + v
(k) - (18) Here, v(k): white normality noise vector U of dispersion matrix cv = unit matrix Equations (7) and (18) above indicate rolling characteristics according to the state space representation method. That is, the difference Δ) l2(
k) is the observed value Y (k), the difference Δ of the entrance side plate thickness deviation is
Difference Δ between H1(k) and the reduction control amount determined accordingly
5(k), the tension control amount difference ΔT (k) is one vector 1h(k), and the rolling characteristic coefficients λ, ν.

ξ is expressed as a state variable vector 1c(k), and a Kalman filter, which is known as a state estimation method for a linear discrete value system, is applied. That is, where C, (k) is the optimal estimate of X (k)
is the estimated error variance matrix of . From now on, at the start of control (k
If initial values (λ., ν., ξ.) of =o) are given, λ, ν become more optimal as control progresses.

Estimated values of ξ are obtained sequentially.

Next, learning control using equations (20) to (22) will be specifically explained (see FIG. 2).

As the initial value (0), use a calculated value based on rolling theory or an actual value up to that point. Optimum values for CX(0), Cv, and C,1 are determined experimentally for each rolling mill (step 101).

First, substitute the initial value (1), find ΔT(1) (step 107
). For example, in plate thickness control mainly based on the reduction control amount ΔS,
As T(1)=O, Δ5(1)=-Δh+(1)・λ/
It becomes ν. The usage of ΔS and ΔT is determined depending on the specifications of the rolling mill, the type of rolled material, etc.

Next, from Δ5(1), ΔT(1) and entrance side plate thickness deviation Δh + (1) obtained as described above, initial values Δ5(0), ΔT ( 0) and Δ
Difference from hl(0) ΔHl (1)−Δh+(1)−Δ′h+(0)mu5(1)=Δ5(1)−Δ5(0) 8T(1) −ΔT(1) − Step 109) to calculate ΔT(0), )1(1) = (ΔH+(1)mu5(1)ΔT(1)
) (step 111). In other words, the roll reduction control amount Δ5 corresponding to the entry side plate thickness deviation Δh, (1)
(1), setting the tension control amount ΔT(1). Next, the exit plate thickness deviation Δh obtained as a result of this control (1) is measured (step 113), constant control period (sampling period)
The difference from Δhz(0) obtained from the previous control result is calculated as Y(1
). That is, from equation (12), 8Hz (1) can be found (
Step 115), (Construct Y(1) from equation 81 (
Step 117) That is, Y(1) =Δhz(1)−Δ
h2(0). On the other hand, cX(0), cv, c.

A(1) is obtained from equation (20) using and th (1) (step 119). And A(1) and X(0), Y
(1).

1h(1) is put into equation (21) to obtain the first optimum estimated value S(1), that is, the characteristic influence coefficients λ, ν, and ξ (step 121). and c, (0), Ih (1)
, A (1) is used to find C, (1) from equation (22) (step 123).

Thereafter, in the same way, measure the entrance side plate thickness deviation Δh I (2), set the control amounts Δ5(2) and 8T(2) from equation (1) using step (1), and calculate the exit side plate thickness deviation. In the procedure of measuring Δhz(2), finding h(2) and Y(2), calculating A(2), and finding X(2), the characteristic control influence coefficient λ, The optimal estimated values of ν and ξ are determined, and based on this, the control it is determined and controlled (step 125).
．． 127).

Thus, according to the present invention, the control characteristics of a plate thickness control system mainly based on feedforward are estimated online and in real time from actual control data, and this is immediately applied to the next control. Since it is used to determine lI, highly accurate plate thickness control is possible even in rolling mills whose control characteristics cannot be accurately known in advance.

As an alternative method to the above, as described below, each coefficient is calculated using a Kalman filter by combining the method of differentiating each control data described above and the method of making the coefficients estimated from the relationship in equation (5) independent. It can also be estimated.

Hereinafter, the plate thickness control method according to the second and third aspects of the present invention will be explained using as an example a case where feedforward control is performed using both the rolling reduction amount and the tension as control variables according to the plate thickness measurement value at the entrance side of the rolling machine. (See Figure 3).

Expressing the relationship between the entrance side plate thickness influence coefficient λ and the reduction amount influence coefficient ν, which is known from rolling theory.
(24) is substituted into the equation (6) obtained by differentiating the respective control data described above, and the equation (6) is expressed as follows without including the reduction amount influence coefficient.

ΔH2(k)−Δ5(k)=λ(ΔH1(k)−ΔS(k))+ξ・muT(k)
(25) Then, by applying a Kalman filter to the above equation (25), the equation (25) is expressed as follows, similar to the above equations (7) to (17).

Y (k) = h (k) 7t: (k) +
w (k) ... (26) Y (k) =
ΔH2−ΔS lt=k・Δt...(27)
ΔH1(k)=Δh1(k)−Δh, (k−1)
...(30)ΔH2(k)=Δ)lx(k)−
Δh! (k-1) ・(31)Δ5(k)=
ΔS (k) −ΔS (k-1) ...
(32)ΔT(k)=ΔT(k) −ΔT(k−1
)...(33)Xl Δλ
...(34)X3 Δξ
...(35)w
(k): whiteness noise with variance Cw h': transpose of vector h and state variables are expressed as follows in the same way as equations (18) and (19).

S(k+1)=1! ', K(k) +V(k)
(36) where w(k): white normality noise vector I of dispersion matrix Cv two unit matrices Equations (26) and (36) above indicate the rolling characteristics according to the state space representation method. In other words, the exit side plate thickness The deviation is the difference ΔH2 (
k) and the reduction control amount that caused the exit side plate thickness deviation Δ
S (k) is the observed value Y (k), and the difference between the difference ΔH1 (k) in the entry side plate thickness deviation and the difference ΔS (k) in the reduction control amount determined according to this and the difference in the tension control amount Let the difference ΔT (k) be one vector h (k), and express each coefficient λ, ξ as a state variable vector X (k),
The Kalman filters of equations (20) to (22) are applied. From now on, the initial value (λ.,
ξ. ), as the control progresses, more optimal estimates of λ and ξ are successively selected. For the initial value, use a calculated value based on rolling theory or an experimental value up to that point. The optimum values of C, (0), Cv and Cw are determined experimentally for each rolling mill (step 201).

Next, to specifically explain the learning control using equations (20)2 to (22), first, from the initial value X (0), each control data is Calculate the difference ΔH1(1), h5(1), 8T(1) (steps 203 to 209), h(1) = (ΔH1(1) - m5(1), ΔT(1)
)) (step 211). That is, the control amount Δ5(1) corresponds to the entrance side plate thickness deviation Δh+(1).

Set ΔT (1). Next, step 213) measures Δhz(1) obtained as a result of this control, and calculates the difference from 8hg(0) obtained from the control result before a certain control period by Y(
1) (steps 215 and 217). On the other hand, C, (
0), Cv, C, and Ih (1) to find A(1) from equation (2o) (step 219). and A(1
) and X (0), Y (1), and Ih (1) into equation (21) to obtain the first optimal estimate 1(1), that is, the influence coefficients λ and ξ (step 221), and this λ ν is calculated from equation (24). So"i'c, (0
), Ih(1), A(1) t- to find C,(1) from equation (22) (step 223). Similarly,
Measure the entry side plate thickness deviation Δh + (2) and use X (1) to calculate the control amount Δ5 (2) from equation (1).

Set ΔT(2), measure the outlet thickness deviation Δhz(2), calculate h(2) and Y(2), calculate A(2), find λ(2), and calculate ν from the results. As the control progresses, the optimum estimated values of the successive coefficients λ, ν, and ξ are determined by the procedure
Based on this, the control amount is determined and control is performed (steps 225, 227).

Note that the above explanation was based on an example in which both the reduction amount and the tension are controlled according to the measured value of the entrance side plate thickness.
When performing feedforward control with the tension constant and only the reduction amount as the control variable, use the equations (25) to (38) above excluding the terms related to the tension to calculate the entrance plate thickness influence coefficient and the reduction amount influence coefficient. All you have to do is ask for it.

In addition, when performing feedforward control with only the tension as the controlled variable while keeping the reduction amount constant,
), the entry side plate thickness influence coefficient and tension influence coefficient may be determined using an expression that excludes the term related to the rolling reduction amount. These controls are the same as those described above except that there is no term regarding tension or reduction amount, and therefore the explanation will be omitted.

In feedforward plate thickness control using this learning control, ΔS is calculated from the time at which the plate thickness is measured on the entry side and the entry side plate thickness deviation Δh obtained thereby.

Since the time at which the control is performed using ΔT and the time at which the exit side plate thickness, which is the control result, is measured are different, the rolled material is tracked and the measurement data and control amount for the same point on the strip are used as the value at the same point. Therefore, the control amount .DELTA.U may be determined using the next optimal estimated value several times before, rather than one step before. Furthermore, in order to obtain the difference between the thickness deviation and the control amount, it is necessary to store the sheet thickness deviation and the control amount before a certain control period in a storage device.

In addition, in the present invention, the reduction control amount ΔS and the tension control amount Δ
T is collectively referred to as the control amount ΔU, and therefore, when simply referred to as the control amount, it includes the case where only the reduction amount, the case where only the tension is controlled, or the case where both are controlled. Also, "before a certain control cycle" means a control time (k-N) that is N control cycles before the control time. Also, in the above explanation, the time before a certain control period is set to (k-1), but (k-N
) of course.

〔Example〕

An example in which the present invention was implemented in cold rolling of stainless steel using a Sendzimir rolling mill (2M mill) is shown below. The specifications are as follows.

Below is a margin.As described above, the present invention had a remarkable effect of improving plate thickness accuracy.

Incidentally, in the present invention, a control method in which both the reduction amount and the tension are controlled variables is adopted. Further, as a conventional technique, a method disclosed in Japanese Patent Application Laid-open No. 58-25807 was adopted in which a plate thickness measurement value and a control amount at one control timing are used in a Kalman filter, and each influence coefficient is estimated independently. According to the conventional technology, errors occur in the estimation of each influence coefficient due to thermal expansion of the rolls during rolling, drift of the plate thickness detector, etc.
The control system sometimes diverged and became uncontrollable, resulting in a low application rate of about 20%.

〔Effect of the invention〕

As described above, the plate thickness control of the present invention estimates the control characteristics of the plate thickness control system, which is mainly based on feedforward, online and in real time from actual control data, and immediately applies this to the next control. Since it is used to determine the amount of il, it is possible to control the plate thickness with high precision even in rolling mills where control characteristics cannot be accurately known in advance. In particular, in order to remove DC component disturbances that delay the estimation process or give incorrect estimation results when applying the Kalman filter, control data at the control time and a certain period before the control time (k-1
) By applying a Kalman filter to the difference in control data, it becomes possible to control plate thickness with higher accuracy than before. In addition, by making the influence coefficient estimated by the Kalman filter independent, it is possible to control the plate thickness with high precision.Furthermore, when controlling the plate thickness using only either roll reduction or tension as the control variable, a Kalman filter can be formed. Since the dimensions of vectors and matrices can be reduced by one, calculation time can be significantly reduced.

[Brief explanation of drawings]

FIG. 1 is an illustrative diagram showing hardware for implementing the control method of the present invention, and FIGS. 2 and 3 are flowcharts showing two different control methods of the present invention. 10... Strip, 11.13... Rolling roll,
31... Reduction amount control device, 33... Tension control device,
40...control device. Figure 1 Figure 2

Claims (1)

[Claims] 1. In a strip rolling plate thickness control method in which at least one of the roll reduction amount and the strip tension is controlled as a control amount according to the input side plate thickness, feedforward control is performed in a direction that matches the output side plate thickness, comprising: Obtain the difference ΔH_1(k) between the deviation Δh_1(k) of the entrance-side measured plate thickness with respect to the standard plate thickness and the deviation Δh_1(k-1) of the entrance-side measured plate thickness before a certain control cycle, and the entry-side measured value. The difference ΔU(k) between the controlled amount ΔU(k) of the strip section and the controlled amount ΔU(k-1) before a certain control period, and the deviation of the exit side plate thickness from the target plate thickness of the strip section to which the controlled variable was applied. The difference ΔH_2(k) between Δh_2(k) and the exit side plate thickness deviation Δh_2(k-1) before a certain control period is calculated, and these differences ΔH_1(k), ΔU(k), and ΔH_2(k
) by applying a Kalman filter to calculate the inlet plate thickness influence coefficient and control amount influence coefficient that minimize the outlet plate thickness deviation,
A strip rolling plate thickness control method characterized by correcting the entrance plate thickness influence coefficient and control amount influence coefficient that have been used up to then using these calculated values, and determining the control amount using the revised influence coefficients. 2. The strip-rolled plate thickness control method according to claim 1, wherein the tension is constant, the control amount is only the reduction amount, and the control amount influence coefficient is the reduction amount influence coefficient. 3. The strip-rolled plate thickness control method according to claim 1, wherein the rolling blade is kept constant, the control amount is only tension, and the control amount influence coefficient is the tension influence coefficient. 4. The strip-rolled plate thickness control method according to claim 1, wherein the control amount is a tension and a reduction amount, and the control amount influence coefficient includes a tension influence coefficient and a reduction amount influence coefficient. 5. In a strip rolling plate thickness control method in which the roll reduction amount and strip tension are controlled as control variables according to the input side plate thickness, and the output side plate thickness is feedforward controlled in a direction that matches the target plate thickness, the input side with respect to the input side reference plate thickness The difference ΔH_1(k) between the deviation Δh_1(k) of the measured plate thickness and the deviation Δh_1(k-1) of the measured plate thickness on the entrance side before a certain control cycle, and the reduction control of the strip section where the measured value on the entrance side was obtained. The difference ΔS(k) between the amount ΔS(k) and the reduction control amount ΔS(k-1) before a certain control period
) and the tension control amount Δ of the strip section from which the entry side measurement value was obtained.
T(k) and tension control amount ΔT(k-1) before a certain control period
and the difference ΔH between the outlet side plate thickness deviation Δh_2(k) with respect to the target plate thickness of the strip section given the control amount and the outlet side plate thickness deviation Δh_2(k-1) before a certain control cycle.
_2(k) is calculated, and the difference ΔH_2 of the entrance side plate thickness deviation is calculated.
(k) and the difference ΔS(k) in the roll control amount, the difference between the exit side plate thickness deviation ΔH_2(k) and the difference ΔS(k) in the roll control amount, and the difference ΔS(k) in the roll control amount, and Difference Δ
A Kalman filter is applied to T(k) to calculate the entry side plate thickness influence coefficient and tension influence coefficient that minimize the exit side plate thickness deviation, and further, the reduction amount influence coefficient is calculated from the input side plate thickness influence coefficient. A strip rolling plate thickness control method characterized by correcting the previously used entry side plate thickness effect coefficient, tension effect coefficient, and reduction amount effect coefficient, and determining the control amount using these corrected effect coefficients. 6. In a strip rolling plate thickness control method in which the roll reduction amount is used as a control variable according to the input plate thickness and the output plate thickness is feedforward controlled in a direction that matches the target plate thickness, the input side measured plate thickness is compared to the input side reference plate thickness. The difference ΔH_1(k) between the deviation Δh_1(k) and the deviation Δh_1(k-1) of the plate thickness measured on the entrance side before a certain control period, and the reduction control amount ΔS( k) and the reduction control amount ΔS(k-1) before a certain control period ΔS(k
), and the difference ΔH_2(k) between the outlet side plate thickness deviation Δh_2(k) with respect to the target plate thickness of the strip section given the reduction control amount and the outlet side plate thickness deviation Δh_2(k-1) before a certain control period.
and the difference between the input side plate thickness deviation difference ΔH_1(k) and the said reduction control amount difference ΔS(k), and the difference between the said exit side plate thickness deviation difference ΔH_2(k) and the said reduction control amount ΔS
A Kalman filter is applied to the difference from A strip rolling plate thickness control method, characterized in that an entry side plate thickness influence coefficient and a reduction amount influence coefficient are corrected, and a reduction control amount is determined using these corrected influence coefficients.