JP3288546B2 - Method and apparatus for setting pass schedule of continuous 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 and an apparatus for setting a pass schedule for a continuous rolling mill, and more particularly to a method and an apparatus for setting a pass schedule for a continuous rolling mill that can accurately reflect the operational knowledge of an operator. It is.

[0002]

2. Description of the Related Art Generally, when a steel strip is processed by a continuous rolling mill, a reasonable operation such as a roll gap, a rolling speed, etc. is performed so as to satisfy a given specification such as a size, a shape and a strength of each steel strip. The setting value must be determined in advance. Conventionally, these setups have been automatically set by a computer based on a rolling model based on a known rolling theory. In order to bring such a setup model closer to actual rolling, the parameters necessary for the model were estimated and adjusted, and the model formula was improved. In addition, learning coefficients were set based on the operation results of the previous material. Some attempts have been made to improve the setting accuracy by reflecting the results in the rolling of the next material. Examples of approaches based on such a rolling model include the following. JP-A-1-233003
In the method disclosed in Japanese Patent Application Laid-Open Publication No. H10-209, the rolling power of each stand is predicted on the basis of a rolling model, and if the limit value is exceeded, the load distribution is distributed to other stands. However, there is a characteristic of operation that cannot be expressed by such a rolling model, and an attempt was made to systematize the correction performed by the operator on the setting result to further stabilize the operation. That is, in the method disclosed in Japanese Patent Application Laid-Open No. 4-162905, whether or not the learning coefficient can be updated is determined based on the rolling knowledge of the operator, and the fitting of the model to the operation results is controlled. Further, in the method disclosed in Japanese Patent Publication No. 63-27087, the rolling reduction pattern of the next material is determined so that the reduction of the rolling reduction is small by using the result appropriately corrected by the operator.

[0003]

In the conventional pass schedule setting method for a continuous rolling mill as described above, there are characteristics of the machine and phenomena that cannot be modeled in order for the rolling model to match the actual operation. Has certain limitations. For this reason, in the past, there were cases where it was not possible to operate using the setting result of the computer as it was, and at that time, the operator corrected the setting result based on the knowledge obtained from the operation experience and started operation. For example, in Japanese Patent Application Laid-Open No. 4-162905, the degree to which the results are reflected in the rolling model is determined according to the situation, and based on the knowledge under what conditions the accuracy of the model is likely to deviate. Thus, the amount of change of the learning coefficient is limited. This method cannot capture operating conditions that cannot be reflected in the rolling model. In Japanese Patent Publication No. 63-27087, when rolling the same material continuously, the operation after the intervention can be stabilized because the intervention result for the operator's setting is reflected as a reduction rate pattern. On the other hand, when successively rolling significantly different materials, it was necessary to change the rolling reduction pattern of the preceding material, and the operation could not be sufficiently stabilized.
Furthermore, with some materials, the operator cannot deal with phenomena that cannot be modeled, such as roll surface roughness, plate breakage, scale flaws, and plate drawing, which make it difficult to model rolling settings. Operational knowledge could not be systematized. However, mere determination of the setting by simply incorporating the operational knowledge has a problem in the setting accuracy, and the accuracy of the conventionally used rolling model must be sufficiently ensured. The present invention has been made in view of the above circumstances, and aims to ensure the accuracy of a rolling model by making full use of the knowledge of an operator and thereby constantly setting a pass schedule to thereby stabilize the rolling model. It is an object of the present invention to provide a method and an apparatus for setting a pass schedule of a continuous rolling mill capable of obtaining the above-mentioned operation state.

[0004]

According to a first aspect of the present invention, a roll gap of each stand of a continuous rolling mill is calculated from a thickness at an outlet side of each stand using a rolling model. In the pass schedule setting method for a continuous rolling mill that sets a pass schedule based on the results,
For the calculated roll gap of each stand, the roll gap difference between successive stands is calculated, and the relationship between the roll gap difference and the outlet plate thickness is determined based on the relationship between the roll gap of each stand and the outlet plate thickness. Estimating a sensitivity matrix to be represented, and repeatedly correcting the outlet plate thickness using the estimated sensitivity matrix so that the calculated roll gap difference pattern approaches a target pattern stored in a memory in advance. And a pass schedule setting method for a continuous rolling mill. According to a second aspect of the present invention, there is provided a roll gap calculating means for calculating a roll gap of each stand of a continuous rolling mill from a sheet thickness at an outlet side of each stand using a rolling model, and setting a pass schedule based on the calculation result. A roll gap difference calculating means for calculating a roll gap difference between successive stands with respect to the calculated roll gap of each stand; A sensitivity matrix estimating means for estimating a sensitivity matrix representing the relationship between the roll gap difference and the outlet plate thickness based on the relationship between the roll gap and the outlet plate thickness, and storing the calculated roll gap difference pattern in a memory in advance. Using the estimated sensitivity matrix above, the output of each of the above stands is brought closer to the estimated target pattern. A continuous rolling mill pass schedule setting device, characterized by comprising; and a delivery side thickness correction means for repeatedly modifying the thickness.

[0005]

According to the first and second aspects of the present invention, the roll gap of each stand of the continuous rolling mill is calculated from the exit side plate thickness of each stand using a rolling model, and a pass schedule is set based on the calculation result. At this time, the roll gap difference between the successive stands is calculated for the calculated roll gap of each stand. A sensitivity coefficient representing the relationship between the roll gap difference and the outlet plate thickness is estimated based on the relationship between the roll gap and the outlet plate thickness of each stand.
The outlet plate thickness is repeatedly corrected using the estimated sensitivity matrix so that the calculated roll gap difference pattern approaches the target pattern stored in the memory in advance. The target pattern is determined based on the experience of the operator and is stored in a memory in advance. As described above, by fully utilizing the knowledge of the operator, the accuracy of the model can be ensured, and a path schedule that can always perform a stable operation can be set.

[0006]

Embodiments of the present invention will be described below with reference to the accompanying drawings to provide an understanding of the present invention. The following embodiment is an example embodying the present invention, and does not limit the technical scope of the present invention. here,
FIG. 1 is a diagram showing a schematic flow of a pass schedule setting method for a continuous rolling mill according to an embodiment of the present invention, FIG. 2 is a schematic configuration diagram showing the periphery of a rolling mill stand, and FIG. It is a figure showing transition of sum. We have:
As a result of hearing the operator's knowledge of the operation,
It was found that the quality of the setting was empirically determined by the pattern of the difference between the roll gap of the entrance stand and the roll gap of the exit stand. Regarding the evaluation of the setting based on this roll gap difference pattern,
It has the following features.

First, the roll gap is a set value actually used when the material is operated, and is an output of a set calculation by a computer. Therefore, it is possible to directly correlate the result of the operation with the roll gap. This makes it easy to directly reflect the knowledge that the operator has. On the other hand, the thickness at the outlet of each stand, rolling reduction, load distribution rate, etc., which are conventionally used, are so-called prediction values that are neither set values of operation nor directly measured values, and include prediction errors of the model. Have been. Therefore, when troubles such as plate drawing, breakage, and flaws due to roll surface roughness occur during operation, the cause is estimated from the roll gap setting, and the setting accuracy is not improved. Secondly, regarding the use of the difference without using the absolute value of the roll gap, the target final thickness is specified in rolling, so the degree of freedom of the pass setting (thickness of each stand exit side) is output. (Each roll gap)
Only one less than the degree of freedom. This means that the absolute value of the roll gap cannot be adjusted to a desired value at each stand exit thickness except for the exit thickness of the final stand. On the other hand, in the relative pattern of each roll gap difference, the degree of freedom is exactly equal to the setting, so that adjustment can be performed by changing the thickness of each stand exit side excluding the exit side thickness of the final stand.

The pass schedule setting method for the continuous rolling mill according to the present embodiment is based on the above-described examination results. As shown in FIG. 1, the roll gap of each stand of the continuous rolling mill is changed. The roll thickness is calculated from the delivery side sheet thickness using a rolling model (S1 to S3), and based on the calculation result, a pass schedule is set (S11). The roll gap difference is calculated (S4),
A sensitivity matrix representing the relationship between the roll gap difference and the outlet plate thickness is estimated based on the relationship between the roll gap and the outlet plate thickness of each stand (S5, S6), and the calculated roll gap difference pattern is stored in a memory in advance. (S7-S9), the exit side plate thickness is repeatedly corrected using the estimated sensitivity matrix (S1).
0). FIG. 2 shows a schematic configuration of an apparatus to which the above method can be applied. The roll gap calculating means 2 shown in FIG.
Through S3, the path schedule setting means 3 executes step S1.
1, the roll gap difference calculating means 4 executes step S4,
The sensitivity matrix estimating means 5 executes steps S5 and S6, and the exit side sheet thickness correcting means 7 executes steps S7 to S10. Each of these means is prepared as a program of each execution form built in the computer connected to the continuous rolling mill, and a part or all of the means is supported by hardware separate from the above-mentioned computer. You can also.

The basic operation of this device will be described below. The knowledge that the operator has empirically considered to be more preferable rolling, which is less likely to cause trouble, is based on the strength of the material,
First, a roll gap difference table corresponding to material conditions such as a target sheet thickness and a sheet width is expressed. Then, this is stored in the memory 6 as a target pattern. Then, the roll gap is calculated by the roll gap calculating means 2 using the rolling model 1 and the roll gap difference is calculated by the roll gap difference calculating means 4. Next, the roll gap difference (the calculation result of the system setting) is made to approach the target pattern of the roll gap difference stored in the memory 6. More specifically, a sensitivity matrix indicating how much the roll gap difference changes when the exit plate thickness of each stand except the final stand is changed from the rolling model 1 is estimated by the sensitivity matrix estimating means 5, and the roll gap difference is estimated. By multiplying the error between the target pattern and the setting calculation result by the inverse matrix of the sensitivity matrix, the correction amount of the thickness at the outlet side of each stand excluding the last stand is obtained by the outlet thickness correcting means 7, and the setting calculation is rolled again. This is performed by the gap calculating means 2.

Here, a series of operation contents in the computer will be described in more detail. Now, let the roll gap difference vector, which is the operator knowledge corresponding to the next material, be d
s ＾, and the roll gap difference vector calculated from the rolling model 1 is ds. The roll gap difference, which is the calculation result, is calculated by giving the sheet thickness on the exit side of each stand. In this case, assuming that other setting conditions do not change, the stand-side exit thickness vector h ₀ , the sensitivity matrix J (= ∂ds / ∂h), and the thickness variation Δh
Is used to perform the Taylor expansion as follows, and is approximated up to the first-order term. Here, the sensitivity matrix J is a regular matrix due to the characteristics of hot rolling. ds = ds (h ₀ ) + JΔh The policy of bringing the roll gap difference vector ds as close as possible to ds ＾ is based on the following equation.
Formulate as a minimization for h. Here, T represents transposition (the same applies hereinafter). E = (ds ＾ −ds) ^T · (ds ＾ −ds) The necessary condition ∂E / ∂Δh = 0 and the sensitivity matrix J are regular, and the following equation is obtained. Δh = J ⁻¹ (ds ＾ −ds (h ₀ )) = J ⁻¹ Δds Here, Δds is an error vector between the roll gap difference of the operator and the roll gap difference of the pass schedule calculation result.

From the above calculation, the thickness at the exit side of each stand is Δ
h, and repeat the setting calculation several times.
Approaching the roll gap difference target pattern,
Therefore, update the stand-out side thickness of each stand as follows. h^{n + 1}= Hⁿ+ J^-1Δdsⁿ Where hⁱIs the thickness vector of each stand on the exit side
Yes, Δds^jIs the error vector of the j-th roll gap difference
It is torr. Next, the method of estimating the sensitivity matrix J is explained.
I do. Now, the number of stands of the continuous rolling mill is 7 for convenience of explanation.
Suppose there is. Discharge side thickness vector (final stand
H = [h₁ h_Two …
h ₆]^T, The element of the roll gap difference vector is ds =
[Ds₁ ds_Two … Ds ₆]^Tfar. h_iIs i
Outer side thickness of the stand, ds_i= S_{i + 1}−s_iIn
You. Also s_iIs the roll gap of the i-stand. This
At this time, the sensitivity matrix J is as follows.

[0012]

(Equation 1) Here, as shown in FIG. 2, among the variable, h ₁ to h ₆ is thickness at delivery side 6 stand, h ₇ is target thickness, s ₁ ~s ₇
Is 1 to 7 stand roll gap, P _{1 to} P ₇ is 1 to 7
The stand load, M _{1 to} M _7, are 1 to 7 stand mill constants. Next, if the roll gap difference is divided into roll gaps, the following equation is established.

(Equation 2) Here, in order to calculate the contents of the matrix, a well-known gauge meter formula used when calculating the roll gap in the pass schedule is used. Then, the following equation is established at each stand.

H _i = s _i + (P _i / M _i ) (i = 1, 2,..., 7) where M _i is a mill constant, that is, a coefficient when a rolling mill is considered as a spring. _Pi is the load. This load P _i
Is greatly affected by changes in the thickness of the entrance side to the stand (outside thickness of the previous stand) and the thickness of the exit side from the stand, but the effects of other thicknesses and other factors can be neglected. It is of the order of magnitude. Partial differentiation of the gauge meter roll gap with the plate thickness is as follows.

(Equation 3) Here, considering the relationship between the load of the rolling mill and the sheet thickness, a _i
> 0, b _i > 0. At this time, since the determinant of the sensitivity matrix J is as follows, it is understood that the sensitivity matrix J is a regular matrix.

[0014]

(Equation 4) That is, numerical instability does not pose a problem in performing the above method. However, mill modulus M _i, it is necessary to pre-determine the value. In addition, the influence coefficient ∂P _i / ∂h
_i-1, the value of ∂P _i / ∂h _i can be determined using a load prediction model setup calculations. For example, the load prediction data generally uses a function of the following format. P _i = F (h _i−1 , h _i , σ _i−1 , σ _i , R _i , w _i , θ _i ,...) Where i is the stand number, P _i is the load of the i stand,
F () is a load prediction model, h _i-1 is thickness at entrance side of the i stand (delivery side thickness of the i-1 stand), h _i is delivery side thickness of the i stands, sigma _i-1 is to i stand Entry tension (i-
The output tension of one stand), σ _i is the output tension of the i stand, R _i is the roll diameter of the i stand, w _i is the plate width of the i stand, and θ _i is the material temperature of the i stand.

The above-mentioned influence coefficient is calculated for the load prediction model as follows.

(Equation 5) Here, Δh _i is a minute change in the thickness of the sheet on the exit side of the i-stand, Δh _i
i-1 is a minute change in the thickness of the outlet side plate of the i-1 stand. Since the second term on the right side of the above two equations is calculated at the time of model calculation, the load of each first term on the right side may be calculated by slightly changing the plate thickness.

As described above, the respective calculations are performed, and the final path schedule can be set by the path schedule setting means 3 based on the results. Hereinafter, each data will be embodied and further explained for understanding the present invention. 1 and 2, the specifications input in steps S1 and S2 for each calculation by the roll gap calculation means 2 are the specifications of the material to be processed and the operating conditions.
It is assumed that the stand-side sheet thickness at the start of the calculation has been determined according to the operating conditions in step S1. In step S3, the roll gap calculating means 2 executes a setup calculation portion based on the rolling theory, and in step S3A
Then, the temperature and load are predicted, and the roll gap is calculated in step S3B. Roll gap difference calculation means 4
From the roll gap calculated in step S3 by
The roll gap difference between the front stand and the rear stand is calculated in step S4.

Next, in step S5, the sensitivity coefficient estimating means 5 obtains an influence coefficient from the sheet thickness and the load related to the settings used in the roll gap calculation, and in step S6, obtains a sensitivity matrix from the influence coefficient and the mill constant. . Step S7
Gives knowledge about the roll gap difference,
Data extracted from the operator in the following form, for example,
That is, the target pattern of the roll gap difference is stored in the memory 6. IF strength is 20K and target thickness ≤ 1.49mm THEN roll gap difference is (6,3,1.45,1.25,0.2, -0.5)

At step S7, the roll gap difference corresponding to the operating conditions of steps S1 and S2 is taken out by the exit side sheet thickness correcting means 7, and at step S8, the roll gap difference is obtained.
The difference from the calculated roll gap difference calculated by the above is calculated by the sum of squares of the error in the roll gap difference pattern. In step S9, it is determined whether the error is equal to or smaller than a predetermined value ε (eg, ε = 0.5). If the error is equal to or smaller than the predetermined value ε, a calculation result is output.
At 11, the actual operation is set. However, if the error is larger than the predetermined value ε, the output side thickness correction amount for bringing the roll gap difference in step S4 closer to the target pattern of the roll gap difference in step S7 is set in step S10.
Calculates the product of the inverse matrix of the sensitivity matrix and the product of the roll gap difference error, adds the correction amount to the exit side plate thickness, changes the thickness, and repeats the model calculation to obtain the roll gap setting.

According to the present embodiment, it is possible to approximate the roll gap difference which the operator has as an image of the stable operation by repeating the calculation about two or three times, thereby achieving the stable operation and reducing the load on the operator. It turns out that it can contribute. One example of the result is shown in FIG. This figure 3
Then, the number of calculations, the roll gap difference that is the calculation result, and the target roll gap difference that is set based on the operator's knowledge are two.
The manner in which the sum of squares (evaluation function) decreases is shown. In the figure, the first time is the calculation result based on the rolling model, and the deviation from the target roll gap difference is large. However, it can be seen that the modification according to the present embodiment quickly matches the target roll gap difference. Also,
According to the present embodiment, since the result obtained from the rolling model 1 is adapted to the operator's knowledge,
Sufficient model accuracy can be ensured.

As a result, by making full use of the knowledge of the operator, the accuracy of the model can be ensured, and a path schedule that can always perform a stable operation can be set. In the above embodiment, in order to make full use of the operator's knowledge, the target pattern of the roll gap difference based on the operator's experience is used. However, in actual use, when a good rolling result was obtained in the past, There is no problem even if the measured pattern of the roll gap difference is used.
In the above embodiment, after the roll gap difference calculation by the roll gap difference calculation means 4 (S4), the influence coefficient estimation (S5) and the sensitivity matrix calculation and the inverse matrix calculation (S6) by the sensitivity matrix estimation means 5 are executed. However, in actual use, there is no problem even if step S4 and steps S5 and S6 are executed in reverse or simultaneously.

[0021]

As described above, the pass schedule setting method and apparatus for a continuous rolling mill according to the present invention are constructed as described above. Therefore, the model accuracy can be ensured by making full use of the knowledge of the operator. It can set a pass schedule that enables stable operation at all times.

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic flow of a pass schedule setting method for a continuous rolling mill according to an embodiment of the present invention.

FIG. 2 is a schematic configuration diagram showing a stand around a rolling mill and a pass schedule setting device.

FIG. 3 is a diagram showing a transition of a sum of squared errors of a roll gap difference.

[Explanation of symbols]

 S1 to S3: Roll gap calculating step S4: Roll gap difference calculating step S5, S6: Sensitivity matrix estimation step S7 to S10: Outgoing side sheet thickness correcting step S11: Roll gap setting step

Continued on the front page (72) Inventor Yoshiyuki Takahashi 1 Kanazawacho, Kakogawa City, Hyogo Prefecture Inside Kobe Steel Works Kakogawa Works (72) Inventor Akira Kitamura 1-5-5 Takatsukadai, Nishi-ku, Kobe City, Hyogo Prefecture Kobe Corporation Steel Works Kobe Research Institute (72) Inventor Kazunori Haraguchi 1 Kanazawa-cho, Kakogawa City, Hyogo Prefecture Kobe Steel Works Kakogawa Works (56) References JP-A-59-7412 (JP, A) JP-A Sho 52-50956 (JP, A) JP-A-5-38511 (JP, A) JP-A-8-90020 (JP, A) JP-A-3-99710 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) B21B 37/00-37/78

Claims (2)

(57) [Claims] 1. A pass schedule setting method for a continuous rolling mill, wherein a roll gap of each stand of a continuous rolling mill is calculated from a thickness of an outlet side of each stand using a rolling model, and a pass schedule is set based on the calculation result. In the calculated roll gap of each stand, the roll gap difference between the successive stands is calculated, and based on the relationship between the roll gap of each stand and the outlet plate thickness, the difference between the roll gap difference and the outlet plate thickness is calculated. Estimating a sensitivity matrix representing the relationship, and repeatedly correcting the delivery plate thickness using the estimated sensitivity matrix so that the calculated roll gap difference pattern approaches the target pattern stored in the memory in advance. A pass schedule setting method for a continuous rolling mill. 2. A roll gap calculating means for calculating a roll gap of each stand of a continuous rolling mill from a thickness of an outlet side of each stand using a rolling model, and a pass schedule setting for setting a pass schedule based on the calculation result. Means for calculating a roll gap difference between successive stands with respect to the calculated roll gap of each stand, and a roll gap of each stand. Matrix estimating means for estimating a sensitivity matrix representing a relationship between the roll gap difference and the outlet plate thickness based on a relationship between the roll gap difference and the outlet plate thickness, and a target which stores the calculated roll gap difference pattern in a memory in advance. Using the estimated sensitivity matrix, the outlet plate thickness of each stand is approximated so as to approximate the pattern. A pass schedule setting device for a continuous rolling mill, comprising a delivery side thickness correcting means for repeatedly correcting.