JP2719215B2 - Edge drop control method for sheet rolling - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a setup method in edge drop control of sheet rolling.

[Prior Art] In the cold rolling of continuous rolling, a multi-pass rolling mill having a plurality of passes on a rolling line is installed on a rolling line for high-speed rolling of a rolling mill and improvement of edge drop. Edge drop control is performed so that the drop is small and the shape is good. As one of the edge drop control methods,
There is a method of adjusting the shift amount in the sheet width direction of a work roll having a tapered end of the roll body (hereinafter, referred to as a taper). For example, in the technique disclosed in Japanese Patent Application Laid-Open No. 58-209402, the shift amount of the tapered work roll pair in the sheet width direction is adjusted according to the sheet width. Also, JP-A-60-12213
In the technique disclosed in the above publication, the measured value of the edge drop amount of the steel sheet from the thickness gauge on the exit side of the final pass rolling mill is compared with a target edge drop amount, and the work roll shift is performed based on the comparison calculated value. The amount operation and the pressure operation of the roll vendor are performed to control the edge drop.

In the edge drop control, usually, the optimum set value of the work roll shift amount is obtained in advance based on the rolling conditions, and the work roll shift amount is set to the optimum set value during rolling.

[Problems to be Solved by the Invention] In the edge drop control, the setup uses a mathematical model (hereinafter referred to as an edge drop prediction model) that predicts an edge drop created in advance so as to match a target edge drop. In the actual rolling process, the rolling conditions constantly change, and the accuracy is not always guaranteed by the coefficients of the model formula obtained off-line.

According to the present invention, in the edge drop control, in order to improve or maintain the prediction accuracy of the edge drop, the values of various model coefficients included in the model are sequentially updated off-line based on actual measured values of plant data and the like. It discloses the method of moving.

[Means for Solving the Problems] The present invention relates to a method for rolling a rolled material at the time of plate rolling by a multi-pass multi-rolling mill having a mechanism for shifting a work roll having a tapered end portion in a body axis direction. The relative position of the work roll taper start point and the end of the rolled material side is set to match the predetermined edge drop target value on the exit side of the final pass rolling mill using the edge drop prediction formula expressing the drop Edge drop control method,
The coefficient of the edge drop prediction formula is updated based on the value of the edge drop obtained from the above edge drop prediction formula and the measured value of the plant data obtained during rolling, and the updated coefficient is used for the work roll shift setting of the next material This is an edge drop control method in sheet rolling characterized by the following.

[Operation] FIG. 1 shows an embodiment of plate rolling to which the present invention is applied.
The start point of the line is the S point and the end point is the O point. At the O point, an edge drop target value of the material is given. On the line, there is an n-pass rolling mill 7 as a plate rolling device. According to the properties of the material and the target value of the edge drop at the point O, the number of use of the work roll shift incorporated in the rolling mill in the n-pass rolling mill is restricted from the line upstream side. The operation of the edge drop mainly uses a work roll shift. An edge drop detector 8 for actually measuring the edge drop on the exit side of the final pass rolling mill is installed at the point O, and the calculation of the edge drop control is performed by the control computer 4.

At point S, a detector 5 for detecting an edge drop of a plate on the entry side of the rolling mill and a detector 6 for detecting a plate width are provided. The plate 1 to be rolled is rolled from the right to the left in the drawing. Information on the material to be rolled at point S,
The rolling conditions and the edge drop amount after rolling at the point O are input to the control computer, and the work roll shift amount of the rolling mill to be used is output from the control computer.

FIG. 2 shows an example of a change in edge drop when the material is rolled in the line configuration as shown in FIG. In FIG. 2, the abscissa indicates the pass number from the upstream of the multi-pass multiple rolling mill, and 0 indicates the entry side of the rolling mill. The vertical axis indicates the edge drop amount, in which the downward direction is represented by drop and the upward direction is represented by up. D ₁ and D ₂ in the figure is a parameter showing the edge drop quantitatively, chain line indicates the target value of edge drop. Each time the number of passes increases from the first pass, the edge drop approaches the target value.

Embodiment Hereinafter, details of an edge drop control method and a method of updating coefficients of an edge drop prediction model will be described based on an embodiment of the present invention.

The edge drop is represented by Expression (1) proposed by the inventors in Japanese Patent Application Laid-Open No. 62-244506. That is, defined by two values of D ₁ and D _2.

Here, h ₅₀ is 50 mm from the width end to the center of the width.
Thickness at the position, h ₁₅ and h ₂₅ are toward the plate width center from the plate width end, represents the thickness of 15mm and 25mm positions. The edge drop defined in this way can express the work roll shift amount S _w , the work roll bender force F _w, and the edge drop の of the sheet before rolling as a function. Here, the work roll shift amount _Sw is the distance from the taper start point of the work roll to the plate side end. Edge drop is
It varies greatly depending on the material, and the tapered portion of the work roll is tapered on a straight line or a curve such as a sine curve or an arc, and the size thereof is greatly affected by the edge drop. Further, it has been clarified from a theoretical analysis or an experimental formula that the edge drop is also affected by the rolling load P, the sheet thickness H, the sheet width, and the like. Therefore, the edge drop D can be accurately expressed by using a mathematical model structure as shown in Expression (2).

Here, i is the rolling mill at the i-th pass from the upstream pass side, n is the final pass for performing edge drop control, k _fm is the material, _Tr is the radius of the tapered portion of the work roll, P is the rolling load, and Γ is the material The edge drop amount before rolling, H, is the sheet thickness before rolling. The A ₅ from A _1, B ₁ B ₅ from influence coefficients according to each variable, E is, takes into consideration the plate width with a constant term. Further, ε is a correction term, and α, β and γ are learning coefficients. Since the definition of the edge drop is defined as two values of D ₁ and D ₂ , in the embodiment, D _{1 is} also used for Expression (2).
And it is intended to represent in two formulas D _2. No. 3-1
FIG. 3 and FIG. 3-2 show the edge drop prediction model formula D _1.
And shows the relationship between the measured values D ₁ and D ₂ with respect to D _2, it can be seen that D ₁ and D ₂ by the model is a edge drop accurately represented.

Next, a method of setting the operation amount of the edge drop control in the sheet rolling by the multiple rolling mill having a plurality of passes will be described. In addition, the edge drop of the plate before rolling is not necessarily the same amount of edge drop at both side edges of the plate, so that on the drive side and the operation side of the rolling mill, the final pass rolling mill exit side edge drop is separately obtained, Also in the setting of the work roll shift, the upper and lower work roll shifts of each rolling mill are set independently. FIG. 4 shows a vertical work roll shift of a rolling mill. Obtains each difference between the target value for the D ₁ and D _2, adjust the deviation. Operation side,
In order to obtain the optimum value as the operation amount of the work roll shift on the driving side, as a function of the calculated edge drop value and its target value, the evaluation value is determined by a predetermined edge drop evaluation function so that the value is minimized. Ask. The evaluation function J is represented by Expression (3), and determines an optimal work roll shift amount from the upstream path.

_{J = W 1i (D 1 -} 1) 2 + W 2i (D 2 - 2) 2 ... (3) where i is the rolling mill number from the upstream side path, W ₁ and
W ₂ is a weighting factor according to the difference between D ₁ and D ₂ and the respective target values ₁ and _2, empirically is a numerical value determined for each target path. Larger W ₂ than W ₁ in the upstream path, downstream path
Let W _{1 be} greater than W ₂ . In the embodiment, the first pass
W ₁ and 1, W ₂ and 2, and the second pass, and the W ₁ to the last path that is targeted to the edge drop control 1, W ₂ and 1.

To find the optimal work roll shift amount for each pass,
The work roll shift amount given as an initial value is input to the edge drop prediction model formula, and the obtained edge drop amounts D ₁ and D ₂ are substituted into the formula (3), and the evaluation function J obtained is J ₁ . . Further, by adding the work roll shift of a minute amount to the initial value of the work roll shift amount, substituting a work roll shift amount is added to the edge drop prediction model equation, the edge drop amount D ₁ and D ₂ obtained formula (3)
Substituted for, those obtained an evaluation function J and J _2. Further, the initial work roll shift amount is slightly subtracted, the work roll shift amount is substituted into the edge drop prediction model formula, and the obtained edge drop amounts D ₁ and D ₂ are substituted into the formula (3) to obtain the evaluation function J. those seeking to J _3. The smallest J among J ₁ , J ₂ and J ₃ is selected and set as J ₀ .
A very small work roll shift amount is added to the work roll shift amount at that time, and the added work roll shift amount is substituted into an edge drop prediction model formula to obtain D ₁ and D _2. ) to by substituting seeking J, do it and J _2. Further, a very small work roll shift amount is subtracted from the work roll shift amount at J ₀ , and the subtracted work roll shift amount is substituted into an edge drop prediction model formula to obtain D ₁ and D _2, which are obtained by formulas ( substituted in 3) obtains the evaluation function J, is it with J _3. J ₀ , J ₂ and
Seeking the smallest J in J _3, compared to the J _0, if is smaller this time obtained J and the current J and J _0, as in the method obtains the J ₂ and J _3, J _0, J obtaining the minimum value of the _two and J _3. Wherein J ₀ is compared with the current J, if direction of this J is large, the work roll shift amount used for the calculation of the time of the J ₀ is the optimal value. Substituting the optimum work roll shift amount into the edge drop prediction model formula, obtaining D ₁ and D ₂ , comparing them with the edge drop target value, and if they match, the work roll shift amount is used as the work roll in the path. Set the shift value. If they do not match, the optimum value of the work roll shift amount in the edge drop controlled rolling mill is obtained based on the same method as the previous time, in order to obtain the optimum value of the work roll shift amount in the i + 1 pass. FIG. 5 shows a procedure for obtaining the optimum work roll shift amount.

By operating the work roll shift in each pass in this way, the shape at the plate side end may become unstable. For this reason, in the embodiment, the work roll shift amount is set from the first pass to the fourth pass, and the shape (flatness) of the plate side during this period is within a range where the plate can be passed or the plate does not break. Deterioration of the shape is guaranteed by changing the setting of the work roll vendor.However, the setting range of the work roll shift amount is limited by the degree of this shape defect. Settable upper and lower limits are defined.

By the above processing, the work roll shift amount of each pass is set at the start of rolling. The coefficient value of the above-described edge drop prediction formula is obtained by calculating the difference between the measured value of the edge drop at the point O on the line and the value calculated by the edge drop prediction model due to errors in the edge drop prediction model, changes in rolling conditions between lots, various disturbances, and the like. There is a difference between them. In order to cope with these, in the present embodiment, the coefficient value of the model is sequentially updated in order to make the edge drop at the point O coincide with the target edge drop in a feedback manner. By successively updating each preset control operation between lots, the coefficient value of the model having adaptability to factors that are difficult to measure (for example, the heat crown of a roll) in rolling conditions. Can be used for the control amount calculation, and accurate edge drop control can always be performed.
The update of the coefficient value of the above-described edge drop prediction model is calculated by Expression (5). In the embodiment, the edge drop parameter has two parameters (D ₁ and D ₂ ), and the sequential update gain for D ₁ is represented by α ₁ (t), and the sequential update gain for D ₂ is represented by α ₂ (t). Since the two updating methods are exactly the same, they are collectively referred to as shown in Expression (5).
Notated by α (t).

Here, t is a lot number and represents a temporal change.

Here, I is a unit matrix, which is a diagonal gain matrix of P _a (t) ∈R _a ^{6 × 6} . P at the start of control (that is, t = 0)
_a (0) is obtained in advance of the control by analyzing operation data or the like. According to this method, in order to perform self-learning of P _a (t) shown in equation (5) between lots in a normal rolling state by successively updating the edge drop prediction model coefficient values, a mathematical model is created by extensive data analysis, etc. Also becomes unnecessary. k _a (t) means calculation of a variable gain for always minimizing the estimation error at each time point. Due to the variable gain, a correction operation having a large influence coefficient acts at the start of learning. However, as the learning proceeds, the correction operation can be reduced, and the learning can be speeded up and stability can be promoted. FIG. 6 shows an example of the control result according to the embodiment, and it can be seen that the edge drop on the exit side of the final pass rolling mill sufficiently satisfies the target edge drop.

[Effects of the Invention] As described above, the present invention provides, as an operation end for adjusting an edge drop with respect to an edge portion in the plate width direction,
By using a work roll shift device in each pass rolling mill, setup control is possible, edge drop control ability can be exhibited, and a target edge drop amount can be easily obtained, so that a great effect is obtained. Bring. In addition, the present invention uses an evaluation function representing an index of a desired edge drop, and changes an operation end set value so as to approach the most desirable edge drop within an operation limit range of each operation end in lot units. By repeating the edge drop control, the control capability always ensures the edge drop as much as possible. In addition, by performing self-learning of the edge drop effect coefficient representing the degree of influence of the change in the operation amount on the edge drop, it is possible to always perform highly accurate edge drop control with adaptability to changes in rolling conditions, The control system can be reduced in size, and has a great economical effect.

[Brief description of the drawings]

Figure 1 is one embodiment of a plate rolling of interest of the present invention, FIG. 2 before learning edge drop pattern, 3-1 view and a 3-2 figures for edge drop prediction model equation D ₁ and D ₂ relationship Found D ₁ and D _2, embodiment of Figure 4 the upper and lower work roll shifting of the rolling mill, the procedure Figure 5 is obtaining an optimum work roll shift, Figure 6 is edge drop pattern after learning. 1 ... board, 4 ... control computer, 6 ... detector for detecting board width, 8 ... edge drop detector.

 ──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Mitsunobu Inaba 1-1-1, Edamitsu, Yawatahigashi-ku, Kitakyushu-shi, Fukuoka Prefecture Inside Nippon Steel Corporation Yawata Works (56) References JP-A-60-148610 (JP) , A) Tokiko 56-14362 (JP, B2)

Claims (1)

(57) [Claims] 1. An edge drop predicting expression for expressing an edge drop of a rolled material in a plate rolling by a multi-pass multiple rolling mill having a shift mechanism in a body axis direction of a work roll having a tapered end of a roll body. In the edge drop control method of setting the relative position of the work roll taper start point and the end of the rolled material plate side so as to match the predetermined edge drop target value on the exit side of the final pass rolling mill using the above, The coefficient of the edge drop prediction model formula is updated based on the value of the edge drop obtained from the edge drop prediction formula and the measured value of the plant data obtained during rolling, and the updated coefficient is used for the work roll shift setting of the next material An edge drop control method in strip rolling, characterized in that: