JPH10328723A - Method and device for estimating profile of hot rolled sheet and rolling controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate the sheet profile of each stand of a tandem rolling mill without using a profile meter, and to enable the fine profile control of each stand. SOLUTION: The profile meters are installed on the inlet side of a first stand and the outlet side of the final stand of the rolling mill 1. The mechanical profile change of a work roll, caused by an actuator operation, is calculated by a mechanical change profile model calculating mechanism 6 as the function of the change of each operation quantity by a sensor 3. A thermal crown quantity by the thermal expansion of the work roll, is found from the measured values of plate thickness, rolling load and roll gap by the sensor 3 by a thermal crown calculating mechanism 7. From the fluctuations of these mechanical profile change and the thermal crown quantity, the work roll profile of each stand, that is, the sheet profile on the outlet side of the stand is estimated. Further, an influence coefficient of the sheet profile at the outlet side of each stand on the final stand, is learned by a learning mechanism 9, and it is used as a control parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rolling mill control system for a steel plant, and more particularly to a method for estimating a profile of a rolled sheet using a mathematical model and a profile control method based on the estimated value.

[0002]

2. Description of the Related Art Good control of the shape (elongation) and profile (thickness distribution in the width direction) of a rolled sheet makes it possible to improve productivity and product quality. The poor shape is caused by poor flatness due to uneven elongation in the longitudinal direction. In addition, the profile failure is caused by a plate crown (body crown) due to elastic deformation of the roll, a thermal crown, a wedge resulting from non-parallelity of upper and lower rolls, an edge drop at a plate width end, a high spot due to local wear of the roll, and the like. In particular, edge drop, which occurs when many factors are intertwined in a complicated manner, causes uneven rolling force in the width direction of the roll, which causes the rolled plate to meander, and requires trimming of the part less than the product thickness. The above problem is great.

[0003] In a conventional multi-high rolling mill, shape control, which is control of elongation, is performed by a post stand. In addition, since the instantaneous value of the outlet profile of each stand is unknown, the preset value is used to determine the amount of movement of the intermediate roll bender, work roll bender, and work roll shift that are actuators of each stand. doing. In this profile control, especially control of the profile shape of the plate edge is called edge drop control,
Control using an edge drop meter that measures only the profile of the plate edge is performed.

For example, JP-A-60-12213 discloses that
A work roll having a tapered portion is provided at least in the first stand of the multi-stage stand, and a plain work roll is provided at least in the final stand. The edge drop amount of the rolled plate on the exit side of the final stand is measured. It is disclosed that a shift amount of a work roll having a tapered portion in a plate width direction is controlled based on a deviation of the work roll. In Japanese Patent Application Laid-Open No. 4-200814, the edge drop amount of the last stand exit side of a work roll having a tapered portion is measured, and the shift amount of the work roll is controlled based on a deviation between the measured value and an intermediate target value. Further, it is disclosed that the edge drop amount of the final stand is measured, and the intermediate target value is changed based on a deviation between the measured value and the final target value.

[0005]

In the conventional profile control and edge drop control, since the profile meter and the edge drop meter use an X-ray apparatus, they are expensive, and their use is greatly restricted. In addition, there is a problem that a movable portion for scanning or the like is large, and if the movable portion is arranged on an intermediate stand, the distance between the stands becomes large, making it difficult to control the thickness in the longitudinal direction (AGC). Therefore, at present, only the entrance side of the first stand and the exit side of the last stand are arranged,
There is a difficulty in arranging it on an intermediate stand as in JP-A-4-200814.

As described above, at present, it is often difficult to measure the profile on the exit side of the intermediate stand, and the control amount is determined by collectively using the measurement information from the entrance side of the first stand and the exit side of the last stand. The distribution is performed at a predetermined ratio. For this reason, fine profile control in each stand cannot be performed, and high-precision control for crown and edge drop cannot be realized. This is also a problem that the profile control ability of the tandem rolling mill cannot be fully utilized.

[0007] The object of the present invention, in view of the state of the prior art,
The present invention provides a method and an observer for estimating a plate profile of each stand exit side of a rolling mill system composed of a plurality of stands by a numerical model, thereby applying optimal feedback control to an actuator of each stand to achieve good profile control. Is to make it happen.

The profile estimation method of the present invention described below includes the shape of the edge drop at the plate edge, and the profile control is used in a broad sense including the edge drop control.

[0009]

SUMMARY OF THE INVENTION It is an object of the present invention to provide a rolling control system comprising a plurality of stands, in which a variation in a rolled plate profile to be controlled is controlled by a mechanical deformation profile caused by a change in elasticity of a work roll caused by an operation amount of an actuator. And, capture and capture the synthetic change with the thermal crown caused by the thermal expansion of the work roll, and construct each mathematical model of the mechanical deformation profile and the thermal crown,
This is achieved by making it possible to estimate the plate profile on the exit side of each stand.

Another object of the present invention is to estimate the profile of each outlet side of a plurality of stands in real time, apply independent feedback control to the actuator of each stand, and provide a fine and highly responsive profile by the combined operation of the actuators. This is achieved by realizing a rolling control system that enables control.

The mechanical deformation profile model considers the mechanical deformation profile as a function of the operation amount deviation of the actuator at each stage, and calculates the deviation between the roll bender of the work roll and the intermediate roll, the roll shift and the leveling of the work roll. Calculate the profile based on it.

The thermal crown model regards the thermal crown as the deviation between the measured value of the roll gap and the set value, and uses a representative point (thickness measurement) for each stand based on the measured values of the roll gap, rolling load and exit side thickness. Calculate the thermal crown amount of point (1). Further, the thermal crown amount is developed in the width direction by the influence coefficient having the distribution in the width direction. The influence coefficient is updated by learning.

According to the structure of the present invention, the work roll profile of the final stand can be obtained from the sum of the calculated value of the mechanical deformation profile of each stage and the calculated value of the thermal crown multiplied by the influence coefficient. Further, the error between the work roll profile by this calculation and the plate profile on the exit side of the final stand by the measurement is taken as the influence of the thermal crown, and the influence coefficient is updated by learning.
Used to calculate a new work roll profile.

FIG. 20 is an explanatory diagram showing the basic concept of thermal crown and profile estimation in the present invention.
As shown in FIG. 3A, the distance between the centers of the pair of rolls 2001 that has not undergone thermal expansion is S ′, and the exit side thickness of the rolled plate is h. The pair of rolls 2002 that have undergone thermal expansion by rolling are deformed so that the plate edge direction becomes thinner with the center in the width direction as the maximum. Just spread out. This expanded amount Sα is defined as a thermal crown amount (STC) in the present invention.

The work roll profile calculation mechanism 2003 calculates a work roll profile 2004 from the deviation of the operation amount of the actuator and the thermal crown amount. This is compared with the measured shape 2005 of the final stand, and the error 2006 is learned by the learning mechanism 2007 as the influence of the thermal crown. The resulting effect coefficient 2008 is fed back to the calculation mechanism 2003. The influence coefficient updated by learning is used for calculation of a new work roll profile to improve the accuracy of profile control.

[0016]

Embodiments of the present invention will be described below in detail with reference to the drawings.

FIG. 1 shows the overall configuration of a rolling control system to which the present invention is applied. A rolling mill system 1 including a plurality of stands to be controlled, an actuator 2 for driving each stand, a sensor 3 for detecting a state quantity of each stand including a state of the actuator 2, a sensor 3 based on the movement of a plate material to be rolled A tracking processing mechanism 4 for collecting information; a tracking data table 5 for storing the collected information; a parameter table 10 to be described later; and a setup mechanism 11 for obtaining an operating point of the rolling mill system 1 based on values of the parameter data table 10. Control system design tool 1 that determines various operation setting parameters of feedback control based on the values of parameter data table 10
2, a feedback control mechanism 13 that outputs an operation command of the rolling mill system 1 to the actuator 2 based on information of the setup mechanism 11 and the control system design tool 12.

Further, as a unique configuration of this embodiment,
A mechanical change profile model calculation mechanism 6 for calculating a mechanical change of a work roll using the actuator state amount of the tracking data table 5, and a thermal crown calculation for calculating a thermal crown amount using a measured plate thickness of the tracking data table 5. Using the information of the plate profile data table 8, the tracking data table 5, and the plate profile data table 8 for storing information of the mechanism 7, the mechanical change profile model calculation mechanism 6, and the thermal crown calculation mechanism 7, control parameters to be described later are used. And a parameter data table 10 for storing the learning results.

FIG. 2 shows the structure of each stand of the tandem rolling mill system. Here, description will be made using a six-stage rolling mill. The six-high rolling mill includes a pair of work rolls 1101, a pair of intermediate rolls 1102, and a pair of backup rolls 11
03. The rolling plate 1104 is sandwiched between a pair of work rolls 1101 to perform rolling by applying a rolling load to the left and right of the pair of backup rolls 1103.

The operations of the actuator of the rolling mill, particularly those related to the profile control are roll bending, roll shift, and leveling. FIG. 3 shows the movement of the roll by the actuator. A work roll 11 having a tapered portion 1201 as shown in FIG.
01 will be described.

FIG. 2B shows a work roll bender. When a bending force FW is applied to the axis of the work roll 1101, the work roll 1101 undergoes vertical bending deformation, and the profile of the rolled plate 1104 changes accordingly. The same can be said for intermediate roll vendors not shown.

(C) shows a work roll shift. When the work roll 1101 having the tapered portion 1201 is moved by SW in the horizontal direction, the portion in contact with the tapered portion 1201 is deformed according to the taper, and the profile of the rolled plate 1104 changes. I do. The same can be said for the intermediate roll shift. The edge drop control is mainly performed by a roll shift.

(D) shows the leveling. When the left and right load values P1 and P2 are changed while keeping the total of the left and right rolling loads applied to the backup roll 1103 constant, the distribution of the load applied to the work roll 1101 in the plate width direction is obtained. Changes,
The work roll 1101 is vertically inclined by LEV, and the profile of the rolled plate 1104 changes along this inclination LEV.

When a fine operation by such an actuator is performed at each stand, the thickness control in the width direction, that is, the profile control with high accuracy including the edge drop can be performed. For this purpose, a plate profile on the exit side of each stand is required, and based on this, the actuator of each stand is operated by the feedback control mechanism 13. In the present embodiment, the change of the plate profile to be controlled is modeled by the following mathematical expression, and is estimated in real time without using a profile meter.

FIG. 4 is an explanatory diagram showing the coordinates of the rolled plate model. When treating the rolled plate 1104 as a mathematical model,
The coordinate axis x is set in the plate width direction. The origin of coordinate axis x is rolled plate 1
104 is the center (x = x ₀ = 0) in the plate width direction. Assuming that the width of the rolled plate 1104 is bw, the coordinate x of the plate end is (b
w / 2) and (−bw / 2).

For example, the delivery plate profile Cro (x) used in the mathematical model is a delivery-side plate thickness distribution 1301 having a distribution in the plate width direction. For all quantities that have a distribution in the width direction, such as the entry side plate profile Cre (x), the work roll profile CrWRi (x), and the load distribution,
Similar coordinate axes are set.

FIG. 5 is a conceptual diagram showing a mathematical model of rolling mill control. In the deformation of the plate material to be controlled, considering that the exit profile of the final stand is due to the influence of the entrance profile and the exit profiles of each stand,
The controlled object model is represented by Expression 1 including the influence coefficient. The work roll profile C of each stand
rWRi (x) can be considered as each outlet plate profile.

[0028]

(Equation 1)

That is, the exit side plate profile Cro (x) of the final stand (i = N) is the first stand entrance side plate profile Cre (x) and its influence coefficient B, and the work roll profile CrWRi (x) of each stand. Its influence coefficient A
It can be obtained by the sum of products of i. Influence coefficient A of each term on the right side on the exit side plate profile Cro (x) on the left side
i, B, and D are control parameters. The third term is the amount of influence of the rolling load due to the material-dependent component Pf (x), which is a value that can be calculated in advance when the material is determined, and is treated as a known value in this embodiment.

FIG. 6 shows a flow of processing for obtaining plate profile data based on sensor information. Information output from the sensor 3 is stored in the tracking data table 5 by the tracking processing mechanism 4. The tracking processing mechanism 4 divides the sheet material to be rolled by the rolling mill system 1 into m elements based on a constant mass flow (the product of the sheet thickness and the speed per unit width) of the sheet material, and passes through each element. The current position is tracked in real time, and sensor information when passing through each stand is collected and stored in the tracking data table 5.

The tracking data table 5 stores the following data associated with each element of the plate material. Incoming sheet speed vei and outboard sheet speed voi measured by a sheet speed meter, and each state quantity deviation of actuator 2 measured by an attached sensor, ie, ΔFWi (work roll bender), ΔFIi (intermediate roll bender), ΔSWi (work roll) Shift), ΔSIi (intermediate roll shift) and ΔLEVi
(Leveling) and the output side thickness hi measured by the thickness gauge
And the load Pi measured by the load meter, the roll gap measurement value Smi measured by the rolling position sensor (roll gap meter), and the entrance side plate profile Cre (x) and the exit side plate profile Cro (x) measured by the profile meter. is there.

Of these data, the inboard speed vei and the outboard speed voi are used in the tracking processing in the tracking processing mechanism 4. Each state variable deviation of the actuator 2 is input to a mechanical change profile model calculation mechanism 6, which calculates a mechanical change profile CrMi (x) generated by the mechanical action of the work roll of the rolling mill, and obtains a plate profile data table 8 Is stored in Delivery side thickness hi,
The load Pi and the roll gap measurement value Smi are input to the thermal crown calculation mechanism 7, and the thermal crown amount STCi
Is calculated and stored in the plate profile data table 8. The incoming side plate profile Cri (x) and the outgoing side plate profile Cro (x) are input to the parameter learning mechanism 9 and used for parameter learning processing.

Next, a method of estimating the work roll profile CrWRi (x) which cannot be directly detected in the mathematical model of Formula 1 will be described. Work roll profile CrWRi
(x) can be estimated from the mechanical change profile CrMi (x) due to the actuator operation and the thermal crown STCi due to the thermal expansion of the roll, and is expressed by Equation 2.

[0034]

(Equation 2)

The mechanical change profile CrMi (x) is a thickness distribution changed by the mechanical deformation of the work roll. As described above, the shape of the work roll at the portion in contact with the rolled plate changes due to the operation of the actuator such as a roll bender or a roll shift, and the shape of the rolled plate changes accordingly. Therefore, CrMi (x) is expressed as a function J using the state quantity deviation of the actuator as a variable, as shown in Expression 3.

[0036]

(Equation 3)

FIG. 7 shows a processing flow of the mechanical change profile model calculation mechanism. Data reading processing 601
Then, the deviation of the actuator state quantity required for calculating the roll profile model is read from the tracking data table 5.

In the calculation process 602, the equation (3) is calculated, and the profile C generated by the mechanical action of the work roll is calculated.
Find rMi (x). In the data writing process 603, CrMi (x) that is the result of the calculation process 602 is stored in the plate profile data table 8. In the drawing, solid arrows indicate the flow of processing, and dotted arrows indicate the flow of data.

On the other hand, the thermal crown amount STCi of Equation 2 is generated by thermal expansion of the roll. This thermal expansion is caused by the heat generated by rolling deformation of the rolled sheet being transmitted to the rolls and increasing the roll temperature, and has a distribution in the width direction.
Therefore, the thermal crown also has a distribution in the width direction.
However, the thermal crown amount STCi obtained by Equation 2 is a value of a representative point (plate thickness measurement point) having no distribution in the width direction.

The thermal crown amount STCi is a variation of the thermal crown, which is obtained by subtracting the roll gap set value Si from the roll gap Smi measured by the rolling position sensor, and can be expressed by Equation 4.

[0041]

(Equation 4)

Here, the roll gap set value Si can be calculated from Expression 6 obtained by transforming Expression 5 from the outlet plate thickness hi measured by a gauge meter and the load Pi measured by a load meter. Thus, the roll gap measured value Smi and the gauge meter measured value h
When the roll gap set value Si and the thermal crown STCi are separated by using i, the thermal crown amount STCi can be obtained from Expression 7.

[0043]

(Equation 5)

[0044]

(Equation 6)

[0045]

(Equation 7)

Here, Ki is a mill stiffness value, which is treated as a known value here.

FIG. 8 shows a processing flow of the thermal crown calculation mechanism. In the data reading process 701, the thermal crown amount STCi is obtained from the tracking data table 5.
Read out sheet thickness hi, load Pi, and roll gap measurement value Smi required for the calculation of.

In the calculation process 702, the roll gap set value Si is obtained by calculating the formula (6). In the calculation processing 703,
Equation 7 is calculated to determine the thermal crown amount STCi. In the data writing process 704, STCi, which is the result of the calculation process 703, is stored in the plate profile data table 8.

Next, the influence coefficient A, which is a parameter of the equation 1,
The learning process of i and B will be described. Here, a mechanical change profile CrWRi (x) is obtained, and a reference point x =
xo (for example, the center point in the plate width direction) is determined, and under these conditions, parameters Ai and B are obtained by learning process (1) based on Equation 1. This reference point is the thermal crown ST
It is the same position as the representative point of Ci.

Next, the final stand exit side profile Cro
In (x), the influence amount Cro ', which is affected only by the thermal crown STCi, is obtained by the following equation (8). This influence amount Cr
In order to obtain the influence amount Cro '(x) having the width distribution o', the equation 8 and the equation 9 obtained by substituting the equation 2 into the equation 1 are transformed into the equation 10.

[0051]

(Equation 8)

[0052]

(Equation 9)

[0053]

(Equation 10)

By the way, if the parameters Ai and B are obtained without error in the model of the equation (1), the equation (1) is established as an equation, and the equation (9) which is a modification of the equation (1) based on the equation (8) is represented by the equation (10). , The influence amount Cro '(x) does not have a distribution in the width direction, and it is impossible to expand the width direction. However, since the thermal crown actually has a distribution in the width direction, A which does not have a distribution in the width direction obtained in the learning process (1).
When i is used, Equation 1 does not hold as an equation when viewed in the entire width direction. That is, due to the error caused by the thermal crown, the influence amount Cro '(x) in Expression 10 can be developed with a distribution in the width direction.

Therefore, the learning process (2) is performed so as to eliminate the error of the influence coefficient Ai caused by the thermal crown by the development in the width direction. That is, the parameter Ai is obtained by the equation 11 as the sum of the reference value Aio obtained for the reference point xo and its deviation Ai (x).

[0056]

[Equation 11]

A series of equations (8 to 11) for expanding the estimated thermal crown amount in the sheet width direction is called a thermal crown model.

FIG. 9 shows a processing flow of the parameter learning mechanism. As described above, the parameter learning mechanism 9 obtains the parameters Ai, B and the deviation Ai (x) in the width direction with respect to the thermal crown influence of Ai by learning.

First, in the work roll profile calculation processing 901, a work roll profile CrWRi (x) is obtained from the result of the mechanical change profile model calculation mechanism 6 and the thermal crown calculation mechanism 7 according to equation (2).

Next, parameter learning processing (1) 902
Then, the parameters Ai and B for the reference point xo are obtained by regression estimation as described later. In the calculation processing 903, as a preparation for obtaining the parameter Ai (x), the thermal crown influence component Cro '(x) is obtained as described later. Finally, in the parameter learning process (2) 904, the parameter Ai (x) is a deviation of the parameter Ai with respect to the thermal crown influence, which reflects the thermal crown influence Cro '(x), and has a distribution in the width direction. Is obtained by regression estimation as described later. Hereinafter, each process in the parameter learning mechanism 9 will be described in detail.

FIG. 10 shows a processing flow of work roll profile calculation. The work roll profile calculation processing 901 reads CrMi (x) and STCi from the plate profile data table 8 in a data reading processing 9011.

Work roll profile calculation processing 901
2 calculates Equation 2. The data write processing 9013 includes:
The calculation result CrWRi (x) is stored in the plate profile data table 8. This processing 9011 to processing 9013
Is repeated as many times as necessary, with data corresponding to one divided element of the rolled plate in the tracking processing as one unit.

Next, the parameter learning process (1) will be described. There are various parameter learning methods. Here, a method of regression estimation using a matrix will be briefly described. This method is well known, for example, in the literature "Basic of regression analysis (Takeshi Hayakawa, Asakura Shoten, December 1986, p52-p80)".

The variable y _i has n observations (x _i1 ,
x _i2 ,..., x _in ) and Equation 12;
Suppose that it can be described by the determinant of 3.

[0065]

(Equation 12)

[0066]

(Equation 13)

Here, y in Equation 13 is called a dependent variable vector, X is an independent variable matrix, e is an error variable vector, and b is a regression coefficient vector.

The least squares method for obtaining the regression estimator is to minimize the sum of the squares of the length of the error vector. A regression coefficient vector that minimizes the sum of squares of the length of the error vector is represented by a regression coefficient estimation vector b * in Expression 14.

[0069]

[Equation 14]

The learning described below means that a dependent variable vector y and an independent variable matrix X are created from various data, and a regression coefficient estimation vector b * is obtained from Expression 14.

The expression to be learned in the parameter learning process (1) is Expression 15 in which Expression 1 is set to x = xo and Expression 12 is adjusted.

[0072]

(Equation 15)

The subscript j of the variable in the expression is the variable i in the equation 12.
Indicates that the value is the j-th observation value obtained by the tracking processing.

Expression 15 can be described by Expressions 16 to 18 according to the formats of Expressions 13 and 14.

[0075]

(Equation 16)

[0076]

[Equation 17]

[0077]

(Equation 18)

FIG. 11 shows a processing flow of parameter learning (1). In the process 9021, the board width direction coordinate x = xo is determined when performing the learning calculation for the parameter of Expression 15. In the process 9022, the data of Cre (x) and Cro (x) at x = xo from the tracking data table 5 and the data of CrWRi (x) at x = xo from the plate profile data table 8 are read.

In the process 9023, the dependent variable vector y in the regression estimation calculation of the parameter of the equation (15) is determined by the equation (16). Similarly, the process 9024 determines the independent variable matrix X by Expression 17.

Processing 9025 is based on Equations 16 and 17,
A matrix calculation process of Expression 18 for obtaining a regression coefficient estimation vector using Expression 15 is performed. Thereby, the parameters Ai and B in Expression 15 are obtained.

In the data writing process 9026, the parameters Ai and B, which are the results of the parameter learning processes 9023 to 9025, are stored in the parameter data table 10.

Next, the process of calculating the thermal crown influence component Cro '(x) in the parameter learning mechanism 9 will be described.
FIG. 12 shows a processing flow of the calculation processing 903. In the data reading process 9031, Cre (x) and Cro (x) are read from the tracking data table 5, CrWRi (x) is read from the plate profile data table 8, and parameters Ai and B are read from the parameter data table 10.

The calculation process 9032 calculates Cro ′ (x) by using the data read in the process 9031 according to Expression 10. The process 9033 stores the calculation result Cro '(x) in the plate profile data table 8. This processing 903
Steps 1 to 9033 are repeated as many times as necessary, using data corresponding to one of the above-described divided elements of the rolled sheet as one unit.

Next, the parameter learning process (2) in the parameter learning mechanism 9 will be described. In the parameter learning process (2) 904, the expression to be learned is expressed by the following equation (11).
Is substituted into Expression 8 and Expression 19 is obtained by setting x = x _k in Expression 19 and Expression 1
Expression 20 converted to Expression 2 and Expressions 21 to 23 described according to Expressions 13 and 14 are also provided. In addition,
J in Equation 20 indicates the order when the tracking data is viewed in time series.

[0085]

[Equation 19]

[0086]

(Equation 20)

[0087]

(Equation 21)

[0088]

(Equation 22)

[0089]

(Equation 23)

FIG. 13 shows a processing flow of the parameter learning processing (2). The data reading process 9041 is based on the plate profile data table 8 to set STi, Cro '(x),
Ai is read from the parameter data table 10.

In the process 9043, the dependent variable vector y in the regression estimation calculation of the parameter of Expression 20 is determined by Expression 21. The process 9044 is performed in the same manner as the independent variable matrix X
Is determined by Expression 22.

The processing 9045 is based on the equations (21) and (22).
This is a matrix calculation process for obtaining a regression coefficient estimation vector according to equation (23).
(X _k ) is required.

Processing 9046 is a parameter learning processing 90
Parameters Ai (x _k ) resulting from 42 to 9045
Is stored in the parameter data table 10. here,
From the processing 9041 to the processing 9045, the above-mentioned x = x _k
Is repeated as many times as necessary, with the learning process for as one unit.

As described above, the data of the control parameters Ai and B that have been learned by the parameter learning mechanism 9 and stored in the parameter data table 10 are passed to the setup mechanism 11 and the control system design tool 12. Thereby, the feedback control mechanism 13 of the rolling mill system 1 can execute control in which the control target is represented by the model of Expression 1.

Here, a specific example of the processing of the mechanical change profile model calculation mechanism 6, the thermal crown calculation mechanism 7, and the parameter learning mechanism 8 will be described. To avoid complicating the description, only one of the processes that repeat the same operation, such as the process of each stand, the process of each divided element, and the process at each width direction coordinate position, will be described.
It is assumed that the rolling mill of five stands operates and data of tracking division elements 1 to 6 is obtained by the tracking processing.

FIG. 14 shows a specific example of the processing by the mechanical change profile model calculation mechanism (FIG. 7). The illustrated example is a processing example for i = 2 (two stands) and j = 6 (divided element 6).

Each data of the actuator operation amount deviation is read from the tracking data table 5 (60).
1) The profile CrM2 (x) generated by the mechanical action of the work roll is calculated by a formula 6020 according to Equation 3 (602) and stored in the plate profile data table 8 (603). The plate width direction x is a plurality of points of −650 to 650, for example, a profile CrM2 (0) = 1.
242. This process is performed in the same manner for the data of the divided elements 1 to 6 of 1 to 5 stands.

FIG. 15 shows a specific example of the processing by the thermal crown calculation mechanism (FIG. 8). Each data of the exit side plate thickness h2, the load P2, and the roll gap measurement value Sm2 is read from the tracking data table 5 (701), and the roll gap set value S2 is obtained by the calculation formula 7020 by the equation (7) (7).
02), the thermal crown STC2 is calculated by a formula 7030 according to Equation 7 (703), and the thermal crown STC2 is stored in the profile data table 8 (704). In this specific example, from the data of the split element 6 of two stands,
The average thermal crown amount of the stand 2 in the plate width direction is STC2 = −0.018. Similarly, this processing is performed on the data of the divided elements 1 to 6 of 1 to 5 stands.

FIG. 16 shows a specific example of the work roll profile calculation processing of the parameter learning mechanism (FIG. 9). This calculation processing 901 corresponding to FIG. 10 reads the sixth data of the tracking division element for each of CrM2 (x) and STC2 of two stands from the plate profile data table 8 (9011), and calculates
The work roll profile CrWR2 (x) of the stand is obtained (9012) and stored in the plate profile data table 8 as the sixth element of the table. This process is repeated for the data of the divided elements 1 to 6 of the tracking data table 5.

FIG. 17 shows a specific example of the parameter learning process (1) of the parameter learning mechanism. In the learning process (1) corresponding to FIG. 11, first, a position in the width direction serving as a reference for determining a parameter is determined as x = xo = 0 (9021). Next, from the tracking data table 5, the input side plate profile Cre (0) and the output side plate profile C
ro (0), for the work roll profile CrWRi (0) of all five stands from the plate profile data table 8, read the data of six tracking elements at the position of x = 0 (9022) and perform the regression estimation calculation. Dependent variable vector y (Equation 16) and independent variable matrix X
(Equation 17) is set as the determinants 922 and 923 (9023 and 9024), and the regression estimation vector b * by the equation 14 and the parameter A by the result and the equation 18
i and B are obtained by the calculation formula 924 (9025), and the parameters A1 to A5 and the parameter B are stored in the parameter data table 10 (9026).

FIG. 18 shows a specific example of the calculation processing 903 of the parameter learning mechanism. This calculation process 903 corresponding to FIG. 12 is performed by using the tracking data table 5 to obtain data on all the measurement points in the width direction n of the incoming side plate profile Cre (x) and the outgoing side plate profile Cro (x), and the plate profile data table 8. From work roll profile CrW
The data of all the measurement points in the width direction n of the sixth tracking division element of Ri (x) and the data of the parameters Ai and B are read from the parameter data table 10 (9031), and are calculated by Expression 931 according to Expression 10. The thermal crown influence portion Cro '(x) of the exit side plate profile is calculated for all the measurement points (9032), and stored in the plate profile data table 8 as the sixth element of the table. This process is repeated a number of times corresponding to the number of divided data in the tracking data table 5, that is, six times.

FIG. 19 shows a specific example of the parameter learning process (2) of the parameter learning mechanism. The parameter learning process (2) 904 corresponding to FIG.
i, thermal crown influence Cr of the outlet plate profile
The data at the position of x = xk of o '(x) and the data of the parameters Ai and B are read from the parameter data table 10 (9041), and the dependent variable vector y (equation 21) for performing the regression estimation calculation. And the independent variable matrix X
(Equation 22) is set as determinants 942 and 943, and (9)
042, 9043), a regression estimation vector is calculated from Equations 14 and 23 (9044), and the parameter Ai (x; x = xk) as the calculation result is stored in the parameter data table 10.
To be stored.

Here, xk indicating the coordinates in the width direction is set in advance at a plurality of values, and a series of processing is performed for all positions defined by xk, and the parameter Ai is defined as a distribution in the width direction in the parameter data table. 10 is repeated.

According to the above embodiment, the exit side plate profile Cro (x) of the final stand is set to the sum of the influence of the work roll profile CrWRi (x) of each stand and the influence of the first stand entrance side profile Cre (x). In the rolling control model (Equation 1) based on the above, the profile CrMi (x) generated by the mechanical action of the work roll calculated from the deviation of the operation amount of each actuator from the CrWRi (x) of each stand, the roll gap measurement value and the gauge meter Thermal crown amount S calculated as the roll gap variation from the measured value of
From TCi (Equation 2), the influence coefficients Ai and B of Equation 1 as control parameters are first learned for one point (center) in the width direction, and then expanded in the width direction for the thermal crown. The influence coefficient Ai (x) is obtained by learning.

Thus, the work roll profile of each stand, that is, the plate profile on the stand exit side, is
An observer that can be estimated without using measurement can be provided. In addition, the error of the estimated value with respect to the measured value of the profile of the final stand is learned as the influence of the thermal crown, and the resulting coefficient of influence is fed back to the work roll profile calculation mechanism. Can be obtained with high accuracy.

Next, a description will be given of an embodiment in which the thermal crown calculating mechanism 7 takes into account the variation in the thickness of the inlet side plate. Here, the measured value of the roll gap Smi by the rolling-down position sensor is considered to be the sum of the roll gap set value Si, the variation SCTCi due to the thermal crown, and the entrance thickness variation SDHi, and can be expressed by Equation 24.

[0107]

(Equation 24)

When the entry side plate thickness Hi changes, the load Pi changes and a load deviation PHi is added. Roll gap entry side sheet thickness variation S for canceling this load deviation PHi
DHi can be obtained from Equation 25.

[0109]

(Equation 25)

The roll gap set value Si is obtained from the gauge meter equation as described above. Therefore, the thermal crown STCi in consideration of the variation of the entry side plate thickness can be obtained by Expression 26.

[0111]

(Equation 26)

FIG. 21 shows a processing flow of the thermal crown calculation mechanism according to this embodiment. The difference from FIG. 8 is that a process 7022 is added. In step 7011, the exit side plate thickness hi, the load Pi, and the roll gap measurement value Smi required for calculating the thermal crown STCi are read from the tracking data table 5.

In the calculation processing 7021, the equation 6 is calculated, and
The roll gap set value Si is obtained. Calculation processing 7022
Now, remember the previous load value Pi-1 and the current load P
The load variation PHi is determined from i, and the roll gap entry side thickness variation SDHi is determined from Equation 25. In the calculation process 7031, using the results of the calculation processes 7021 and 2022,
Is calculated. In the data writing process 7041, the STCi of the calculation result is stored in the plate profile data table 8.

According to this embodiment, the thermal crown STC
In the calculation of i, separation of the entry side plate thickness variation is performed together with the set value of the roll gap, so that highly accurate profile data can be estimated.

Finally, profile control to which the rolling strip profile estimation method of the present invention is applied will be described. FIG. 22 shows the configuration of a profile control system according to one embodiment. The same components as those in FIG. 1 are denoted by the same reference numerals.

The rolling mills 100 and 101 are part of a rolling mill system including a plurality of stands. A total of 3 profiles are provided on the entrance side of the first stand and the exit side of the final stand.
0, 31 are installed. Profile meter 30, 31
Plate profile estimating device 14 that collects data from
Estimates and outputs a plate profile (work roll profile) on the exit side of the rolling mills 100 and 101.

The control system design tool 12 obtains an inverse function of a function representing the correspondence between the actuator operation amount and the plate thickness deviation in advance, and defines the operations of the feedback control mechanisms 130 and 131. The setup mechanism 11 outputs a predetermined target profile on the exit side of each stand based on the correspondence between the operation condition data and the result data. The deviation between the target profile and the estimated profile is input to the corresponding feedback control mechanisms 130 and 131. The feedback control mechanisms 130 and 131 calculate the actuator operation amount based on the input profile deviation, and control the corresponding rolling mills 100 and 101.

As described above, without using a profile meter for the intermediate stand, each stand is controlled independently by the estimated value of the plate profile on the exit side of each stand, and the profile deviation in the entire sheet width direction using the roll bender is reduced. Reduce or
The work roll shift can be changed at each stand by changing the shift amount to reduce the deviation of the profile at the plate edge, thereby improving the controllability of the profile.

[0119]

According to the present invention, it is possible to provide an observer for estimating the shape of the outlet side plate thickness of each stand by using a mathematical model without using an expensive profile meter. Furthermore, since profile control by feedback control can be realized for each stand using the estimated profile information,
This has the effect of maximizing the ability of each stand to modify the profile and greatly improving the profile accuracy of the rolled plate.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a rolling control system according to one embodiment of the present invention.

FIG. 2 is a configuration diagram of a six-high rolling mill.

FIG. 3 is an explanatory diagram showing movement of a work roll by an actuator.

FIG. 4 is an explanatory diagram showing setting of coordinates of a mathematical model.

FIG. 5 is a conceptual diagram showing a mathematical model of rolling mill control.

FIG. 6 is an explanatory diagram showing a flow of processing for estimating plate profile data based on sensor information.

FIG. 7 is a flowchart illustrating processing of a role profile model calculation mechanism.

FIG. 8 is a flowchart showing processing of a thermal crown calculation mechanism.

FIG. 9 is a flowchart illustrating a schematic process of a parameter learning mechanism.

FIG. 10 is a flowchart showing a work role profile calculation process as a part of the process of the parameter learning mechanism.

FIG. 11 is a flowchart showing a parameter learning process (1) as a part of the process of the parameter learning mechanism.

FIG. 12 is a flowchart showing a calculation process of the thermal crown effect of the plate profile as a part of the process of the parameter learning mechanism.

FIG. 13 is a flowchart showing a parameter learning process (2) as a part of the process of the parameter learning mechanism.

FIG. 14 is an explanatory diagram showing a specific example of the processing of the role profile model calculation mechanism.

FIG. 15 is an explanatory diagram showing a specific example of processing of a thermal crown calculation mechanism.

FIG. 16 is an explanatory diagram showing a specific example of a work role profile calculation process as a part of the process of the parameter learning mechanism.

FIG. 17 is an explanatory diagram showing a specific example of a parameter learning process (1), which is a part of the process of the parameter learning mechanism.

FIG. 18 is an explanatory diagram showing a specific example of thermal crown influence calculation processing of a plate profile as a part of the processing of the parameter learning mechanism.

FIG. 19 is an explanatory diagram showing a specific example of a parameter learning process (2), which is a part of the process of the parameter learning mechanism.

FIG. 20 is a conceptual diagram showing a basic concept of the present invention, defining a thermal crown amount to be estimated, and explaining a processing cycle of plate profile calculation and learning.

FIG. 21 is a flowchart illustrating processing of a thermal crown calculation mechanism according to another embodiment.

FIG. 22 is a configuration diagram of a profile control system to which the rolling strip profile estimation method of the present invention is applied.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Rolling mill system, 2 ... Actuator, 3 ... Sensor, 4 ... Tracking processing mechanism, 5 ... Tracking data table, 6 ... Roll profile model calculation mechanism,
7: thermal crown calculation mechanism, 8: plate profile data table, 9: parameter learning mechanism, 10: parameter data table, 11: setup mechanism, 12 ...
Control system design tool, 13, 130, 131: feedback control mechanism, 14: rolled plate profile estimation mechanism, 2
0, 21 ... actuator, 30, 31 ... profile meter, 100, 101 ... rolling mill.

Continued on the front page (51) Int.Cl. ⁶ Identification symbol FI B21B 37/00 116F (72) Inventor Susumu Susumu Kitani 5-2-1, Omika-cho, Hitachi City, Ibaraki Pref. Hitachi Information & Control Systems Co., Ltd. (72) Inventor Gen Tobita 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture Inside Hitachi Information & Control Systems Co., Ltd. (72) Inventor Tetsu Hattori 5-2-1 Omika-cho, Hitachi City, Ibaraki Prefecture Omika Plant, Hitachi, Ltd. (72) Inventor Hiroshi Saito 5-2-1 Omikacho, Hitachi City, Ibaraki Prefecture Inside the Omika Plant, Hitachi, Ltd. (72) Inventor Hiromi Inaba 7-2-1, Omikamachi, Hitachi City, Ibaraki Prefecture Co., Ltd. Within the Hitachi, Ltd. Power & Electricity Development Division (72) Inventor Kenichi Yasuda 7-2-1, Omika-cho, Hitachi City, Ibaraki Prefecture Inside the Hitachi, Ltd. Electric Power & Electricity Development Division

Claims (8)

[Claims] 1. A cross-sectional shape (hereinafter referred to as a profile) of a rolled plate in a width direction at an entrance side of a first stand and an exit side of a final stand of a rolling mill having a plurality of stands is measured by a profile meter, and the other stands are measured. In the method of estimating the exit side profile by calculation, for the work roll for each stand, a thermal crown caused by a temperature change and a mechanical change profile caused by an actuator operation are obtained, and a work roll profile based on the addition result is obtained. A method for estimating the profile of a rolled plate, wherein the profile is regarded as the plate profile on the exit side of each stand. 2. The thermal crown according to claim 1, wherein the thermal crown corresponds to a measurement position of the sheet pressure based on a measurement value of a rolling position sensor, a load meter, and a thickness gauge as a variation with respect to a set value of a roll gap. A method for estimating a profile of a rolled sheet, characterized by estimating a representative value at a position to be rolled. 3. The method according to claim 2, wherein, in order to expand the thermal crown based on the representative value in the width direction, an error of an estimated value with respect to a measured value of the profile of the final stand is learned as an influence of the thermal crown. And estimating an influence coefficient having the distribution in the width direction as a result. 4. The method according to claim 1, wherein the mechanical change profile is estimated as a function of an operation amount deviation such as a roll bender or a roll shift of the actuator. . 5. A profile in the strip width direction at the entrance of the first stand and the exit of the last stand of a rolling mill comprising a plurality of stands is measured by a profile meter, and the exit profiles of the other stands are calculated. In the estimating method, the exit profile of the final stand is changed by the influence of the profile on the entrance of the first stand and the influence of the profile on the exit of each of the other stands, and is calculated by a mathematical model including an influence coefficient. In this case, for the work roll for each stand, a thermal crown caused by a temperature change and a mechanical change profile caused by an actuator operation are obtained, and a plate profile on the exit side of each stand is obtained from the addition result. Estimating, further, in the mathematical model, the measured value of the profile meter and A method for estimating a rolled plate profile, wherein learning is performed by developing an influence coefficient in the width direction using an error when an estimated value of the plate profile of each stand is given as an influence of the thermal crown. 6. A rolling mill comprising a plurality of stands, an actuator for driving the rolling mill, a rolling position sensor, a load meter, a thickness gauge, a width of a rolled plate on an entrance side of the first stand and an exit side of the final stand. A rolling system including a profile meter for measuring a profile in each direction, such as various sensors for detecting the state of the rolling mill and the actuator, and a tracking device for tracking the values of the various sensors of the rolling mill; A thermal crown mathematical model for estimating a thermal crown caused by a temperature change, based on
A work roll profile calculation mechanism having a mechanical change profile mathematical model for estimating a mechanical change due to the actuator operation is provided, and a plate profile on an exit side of each stand is obtained by a sum of the mechanical change profile and the thermal crown. An apparatus for estimating a profile of a rolled sheet characterized by being able to estimate. 7. The learning process according to claim 6, wherein an error of an estimated value by the calculation mechanism with respect to a measured value of the profile meter of the last stand is subjected to learning processing as an influence by the thermal crown, and an influence coefficient obtained as a result is obtained. An apparatus for estimating a profile of a rolled sheet, comprising: a parameter learning mechanism that enables the thermal crown to be expanded substantially in the width direction. 8. A rolling mill with a plurality of stands, an actuator for driving the rolling mill, a rolling position sensor, a load meter, a thickness gauge, and various sensors for detecting the state of the rolling mill and the actuator, and various types of the rolling mill. In a rolling mill system equipped with a tracking device that tracks the value of a sensor, a thermal crown caused by a temperature change of a work roll of each stand and a mechanical change profile caused by operation of the actuator are obtained based on the tracking data. A rolled strip profile estimating apparatus for estimating the strip profile on the delivery side based on the sum thereof is provided, and a feedback control mechanism performs control so as to eliminate a deviation between the target professional plate file on the delivery side and the estimated value for each stand. Rolling control device.