JPH0513730B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to a learning control method for a strip crown using a hot rolling mill or the like having a shape control function.

<Conventional technology> Conventionally, as a method of controlling the strip crown using a rolling mill with a shape control function, the error from the target value is calculated based on the actual measurement value of the strip crown obtained in the previous coil, and the error is calculated based on this. It is known that by changing the target strip crown, the set value of the manipulated variable for controlling the shape of the next coil is corrected, thereby improving the accuracy of the strip crown. As shown in Fig. 2, which shows the control concept, first, if the band plate crown estimation formula and the band plate flatness estimation formula are expressed as i = 1, 2, ..., N (N is the number of constituent stands), the band plate crown The estimation formula is Ch _i =C _i +a _i C _hi-1 +b _i ε _i-1 +d _i Q _i +e _i x _i C _Ri ...(1), and the strip flatness estimation formula is ε _i = ξ _i (C _hi / h _i −C _hi / h _i-1 ) + f _i ε _i-1 + ε _ci ……(
2). However, C _hi is the strip crown on the exit side of the i-th stand, ε _i is the strip flatness on the exit side of the i-th stand, and C _i is the constant term in the equation for estimating the strip crown on the i-th stand (rolling function based on condition),
ε _ci is the constant term in the equation for estimating the strip flatness in the i-th stand, Q _i is the preset value for shape control in the i-th stand, h _i is the strip thickness at the exit side of the i-th stand, C _Ri is the work roll initial crown in the i-th stand, x _i is the work roll initial crown in the i-th stand
This is a coefficient for converting C _Ri into plate width equivalent.
In addition, a _i is the crown genetic coefficient, b _i is the influence coefficient of the flatness of the inlet side of the strip on the outlet side of the strip crown C _hi , and d _i is the influence coefficient of the shape control preset value Q _i on the strip crown. , ξ _i is the shape change coefficient, f _i is the influence coefficient of the flatness of the inlet side of the strip on the flatness of the outlet side ε _i , and e _i is the influence coefficient of the work roll initial crown C _Ri on the strip crown. .

The crown of the strip 2 rolled in the tandem rolling mill 1 is measured by a crown meter 3, and the result is given to a learning correction amount calculation section 4. In this learning correction amount calculation unit 4, a target value correction amount C _hLRN is determined by calculation such as the following equation (3) based on the error between the target value and the actual measurement value of the band crown.
In other words, C _hLRN (-1) is the previous target value correction amount, C _hf
is the target band crown, C _h is the measured band crown,
The target value correction amount C _hLRN is calculated by C _hLRN = kC _hLRN (-1) + (1-k) (C _hf - C _h ) (3) where k is the learning gain. and,
In the preset value determination unit 5, the target plate crown of the next coil is determined by taking into account the target value correction amount C _hLRN obtained as described above, and sets the upper and lower limits of the strip flatness at the intermediate stand as C _hf C _hf ⁺ C _hLRN . A combination of shape control preset values Q _i that achieves the value, strip flatness, and target strip crown and satisfies the upper and lower limits thereof is determined.

In this way, when rolling is performed using a hot rolling mill, etc., the error between the target value and the strip crown is obtained based on the actual measured value of the strip crown obtained in the previous coil, and the target strip crown is changed based on this. , the learning control of the strip crown is performed by modifying the control preset value Q _i of the next coil.

<Problems to be Solved by the Invention> In the conventional control method, poor accuracy occurs when the rolling conditions are changed, and furthermore, there is also a lack of improvement in the accuracy of the flatness of the strip. In other words, the genetic term of the entrance side strip crown is included in the error between the target strip crown value and the measured value, and in the conventional method, the cause of the error is due to a poor estimation of the final stand mathematical model. It is not possible to determine whether this is due to an error in the crown of the strip on the entry side, or if the rolling conditions are changed, the accuracy may deteriorate.
Furthermore, if there is an error in equation (2) itself, the error in strip flatness cannot be corrected.

The present invention was made in view of problems such as a decrease in precision when rolling conditions change and the difficulty in improving the precision of strip flatness. It is an object of the present invention to provide a rolling control method that can prevent the occurrence of poor precision even when the rolling stock changes, and can gradually improve the flatness of the strip.

<Means for solving the problem> The strip crown control method according to the present invention uses actual measured values of both strip flatness and strip crown using a rolling mill having a shape control function, and calculates _N-1 〓 〓 ⁱ⁼¹ w _i (C _hi −C _hi ^* ) ² +w _N (C〓 _h −C _hN ^* ) ² +W (ε ^~ −
ε _N ^* ) ² (where, N is the number of constituent stands, C _h is the actual value of the strip crown, ε ^~ is the measured value of the strip flatness, C^ _hi is the strip crown to be estimated, C _hi ^* , C _hN ^* is the strip crown calculation value according to a predetermined mathematical model, ε _N ^* is the strip flatness according to a predetermined mathematical model, w _i , w _N , W are weighting coefficients). The present invention is characterized in that the band plate crown is estimated at , the correction amount for learning control is determined as a roll crown from this, and the preset value for shape control of the next coil is determined based on this correction amount for learning control. It is something to do.

<Example> As shown in FIG. 1, which shows the control concept of an example of a rolling apparatus that implements the method of the present invention, in this example, the error between the actual measured value of the strip crown and the value calculated by the mathematical model is reduced. Similarly, the strip crown at the intermediate stand is estimated so that the difference from the value calculated by the mathematical model is small at the intermediate stand, and the learning control correction amount is determined from this. Furthermore, in order to estimate the strip crown at this intermediate stand, not only the actual measured value of the strip crown but also the actual measured value of strip flatness is used. It is calculated by converting it to the work roll initial crown.

Specifically, the crown meter 3 and the shape detector 6
Further, a learning control section 9 is provided which includes an intermediate stand strip crown estimating section 7 and a learning correction amount calculating section 8. The crown and flatness of the strip 2 rolled in the tandem rolling mill 1 are as follows:
The measurement is performed by a crown meter 3 and a shape detector 6, and the respective results are given to an intermediate stand strip crown estimation section 7 of a learning control section 9. The intermediate stand strip crown estimating unit 7 calculates the estimated strip crown value at the intermediate stand by, for example, the following calculation based on the actual strip crown measurement value C _h and the actual strip flatness measurement value ε. First, let the evaluation function be F expressed by the following equation (4), and estimate the strip crown of the intermediate stand that minimizes this value.

F= _N-1 〓 〓 ⁱ⁼¹ w _i (C _hi −C _hi ^* ) ² +w _N (C〓 _h −C _hN ^* ) ² +W(ε〓−
ε _N ^* ) ² ...(4) However, C〓 _h1 ~ _{C hN-1} is the band plate crown to be estimated from now on, and C _h1 ^* ~ C _RN ^* is the band plate crown according to equation (1) above. In the crown calculation section, ε _N ^* is the flatness of the strip at the final stand according to equation (2), w _i -w _N , and W is a weighting coefficient. Here, C _hi ^* =C _i +a _i C _hi-1 +b _i ε _i-1 ^* +d _i Q _i +e _i x _i C _Ri ……(5) ε _i ^* =ξ _i (C _hi /h _i − C _hi /h _i-1 ) + f _i ε _i-1 ^* +ε _ci …
...(6), and ε _i ^* is the exit flatness of the strip in the i-th stand according to equation (2). Note that C _i , a _{i ,} etc. in formulas (5) and (6) are the same as those described above.
The combination of C _hi ^...C^N-1 with the minimum value under the above objective function is set as the estimated value of the strip crown of the intermediate stand. That is, it is obtained by solving ∂Q _i /∂C _hi =0 (7) with i=1, 2, . . . , N-1.

Next, after estimating the strip crown of the intermediate stand in this way, the correction amount for learning control is determined as a roll crown. The error from the value is defined as i = 1, 2, ..., N-1 as C _OF ′ _Si =C _hi −C _hi ^* C _OF ′ _SN =C〓 _h −C _hN ^* ……(8), and this is Assuming that this is due to poor estimation of the roll mechanical crown, the roll crown estimation error C _OFSi is calculated using the following equation (9).

In other words, C _OFSi = 1/x _i e _i・C _OF ′ _Si (9) where i = 1, 2, ..., N, and the learning control correction amount for the next coil is i.
=1, 2,...,N, C _LRNi =k _i C _LRNi (-1) + (1-k _i )C _OFSi (10). However, in equation (10), C _LRNi (-1) is the previous learning control correction amount, and k _i is the learning control gain at the i-th stand.

Once the correction amount C _LRNi for learning control is determined by equation (10), the preset value Q _i for shape control of the next coil is determined based on this. That is, the preset value determination unit 5 converts the value of the work roll initial crown C _Ri in equation (1) into the correction amount for learning control based on equation (10).
Using C _LRNi , convert this roll crown C _Ri to C _Ri +
This is used to replace C _LRNi and determine the preset value Q _i for shape control of the next coil.

In this way, using the measured values ε and _Ch of both the strip flatness and the strip crown, the strip crown at the intermediate stand is adjusted so that the value of the function shown in equation (4) is minimized. The band plate crown estimating section 7 estimates the correction amount C _LRNi for learning control as a roll crown, and the learning correction amount calculation section 8 calculates the preset shape control preset for the next coil based on this correction amount C _LRNi obtained by the learning correction amount calculation section 8. Determine the value Q _i . For learning control correction amount
C _LARi is calculated by estimating the band crown at the intermediate stand so that the difference between the measured value C _h of the band plate crown and the calculated value is small, and the difference from the calculated value at the intermediate stand is also reduced. Furthermore, in order to estimate the strip crown of this intermediate stand, we use not only the actual strip crown value C _h as in the conventional method, but also the actual strip flatness value ε. Control correction amount C _LRNi
is converted to the work roll initial crown C _Ri , which is independent of the plate width. As in the past, when the rolling conditions change due to the inability to determine whether the cause of the error is due to poor estimation of the number of final stands in the model or an error in the strip crown on the entry side, the accuracy is reduced. It is possible to prevent deterioration, and even if the rolling conditions are changed, accuracy defects do not occur, and the flatness of the strip is gradually improved, and the accuracy of not only the strip crown but also the strip flatness can be improved. I can do it. Furthermore, since the work roll initial crown C _Ri is converted into the correction amount C _LRNi for learning control, it is possible to respond to changes in board width.

Note that the present invention can be applied not only to hot rolling mills but also to plate rolling mills and other rolling mills having shape control capabilities.

<Effects of the Invention> According to the strip crown control method according to the present invention, it is possible to prevent the occurrence of poor precision even when rolling conditions are changed, and it is possible to improve the precision of not only the strip crown but also the strip flatness. It has the advantage of being able to accommodate changes in board width.

[Brief explanation of the drawing]

Fig. 1 is a conceptual diagram of control according to an embodiment of the present invention;
The figure is a conceptual diagram of the conventional method. In the drawings, 1 is a tandem rolling machine, 2 is a strip, 3 is a crown meter, 5 is a preset value determining section, 6 is a shape detector, 7 is an intermediate stand strip crown estimation section, 8 is a learning correction amount calculation section, 9 is a learning control section.

Claims (1)

[Claims] 1 Using the actual measured values of both the strip flatness and the strip crown using a rolling mill with a shape control function, _N-1 〓 〓 ⁱ⁼¹ wi(C^ _hi h _i −C _hi ^* ) ² +w _N (C _h −C _hi ^* ) ² +W(ε−
ε _N ^* ) ² (However, N is the number of constituent stands, C〓 _h is the measured value of the strip crown, ε〓 is the measured value of the strip flatness, Ch _i is the strip crown to be estimated, Ch _i ^* , C _hN ^* is the strip crown calculation value according to a predetermined mathematical model, ε _N ^* is the strip flatness according to a predetermined mathematical model, and W _i , W _N , W are weighting coefficients). The present invention is characterized in that the band plate crown is estimated at , the correction amount for learning control is determined as a roll crown from this, and the preset value for shape control of the next coil is determined based on this correction amount for learning control. Band crown control method.