CN1235015C - Method for predicting steel plate thickness during rolling process - Google Patents

Abstract

The present invention relates to a method for predicting the thickness of a steel board on line during the rolling process, which comprises the following steps that the affection of the roller-series elastic deformation to the springing of a rolling mill is calculated, a housing window of the rolling mill and the elastic deformation of corresponding mechanical parts are measured, the springing of the rolling mill is calculated, the thickness prediction is affected by the method that the thickness of a steel board is predicted and the zero shift of a springing model of the rolling mill is eliminated, etc. The present invention can stably and quickly predict the thickness of the steel board with very high accuracy, can provide necessary information for the automatic control and the manual intervention during the rolling process, and is suitable for a plate mill, a fine roughing mill during the hot continuous rolling process, a finishing mill, and a non-ferrous metal rolling mill with thicker boards.

Description

The method of operation of rolling prediction steel plate thickness

Technical field

The invention belongs to rolling technical field, particularly a kind of method of in the operation of rolling, predicting steel plate thickness.

Background technology

When at present domestic most of cut deal steel rolling mill transforms at newly-built milling train or to milling train,, be not equipped with online thicknessmeter in order to reduce expenses.To the cut deal steel rolling mill of thicknessmeter is installed, owing to be subjected to the influence of milling train manipulator, the installation site is distant apart from milling train, so can't measure the steel plate thickness of each passage in real time.While Medium and Heavy Plate Rolling process variation in thickness scope big (general sotck thinkness is about 200mm, and finished product thickness is between 6mm～50mm), and the measurement range of thicknessmeter generally is 6mm～50mm, so the steel plate thickness of a lot of passages of the operation of rolling can't be measured.These factors bring very big influence for cut deal process computer setting accuracy and operating personnel's manual intervention.

Research to this respect is to adopt rolling and roller pressing method basically at present.These two kinds of methods are more traditional methods, can't adapt to the continuous variation of as-rolled condition on the one hand, and computational accuracy is lower; Need on the other hand periodically to test, influence the production time, so present domestic most cut deal producer still relies on the artificial jig of taking to measure, this has had a strong impact on rolling stability and automatic control level.

Summary of the invention

At low, the inconvenient defective of classic method precision, the invention provides a kind of method of in the operation of rolling, predicting steel plate thickness, its objective is and improve the thickness of slab precision of prediction, the simplified measurement method, thus improve thickness control accuracy and reduce Measuring Time.

The present invention includes following steps:

1, calculates the influence of roll elastic deformation with influence function method to mill spring

Mill spring is made up of two parts: the elastic deformation of roll elastic deformation, memorial archway and associated mechanical components.The factor that influences roll elastic deformation has: roller diameter, roll crown, rolled piece width and roll-force, and these factors are in the frequent change of the operation of rolling, so need the influence of these factors of accurate Calculation to roll elastic deformation.

The roll elastic deformation that influences thickness prediction is resolved into three parts: flatten and the flattening of working roll center at the center between the deflection deformation of supporting roller bearing center of effort's point, roller, and utilize the high precision solution-influence function method of roll elastic deformation, this three parts distortion is found the solution, drawn of the influence of factors such as working roll radius, backing roll radius, work roll crown, backing roll convexity, rolled piece width and roll-force roll elastic deformation.

Fig. 1 has reflected the influence of roll elastic deformation to thickness prediction.Because pressure is to act directly on the supporting roller bearing seat by housing screw or hydraulic cylinder, so the corresponding change in displacement of the center of effort's point on the supporting roller bearing seat has just reflected the influence of roll elastic deformation to thickness prediction.Clearly, stressed after, the displacement of this stress point is made up of three parts: supporting roller bearing seat center of effort's point deflection deformation Y _r, Y is flattened at the center between roller _{Wb0_flatten}, the working roll center flattens Y _{Ws0_flatten}

The stress model of operation of rolling roller system as shown in Figure 2.Utilize influence function to carry out the calculating of roll elastic deformation to it.The basic thought of influence function method is to change into some unit with roll is discrete, and also by identical cell discretization, the load distribution on each unit is even with load that roll bore and roll elastic deformation.The elastic deformation that body of roll each point produces when determining earlier that about the notion of influence function each unit applied unit force in the applied mathematics physical method, the distortion stack that each unit produces during then with whole loading just draws the deflection of each unit.This method has higher computational accuracy, is suitable for the calculating of roll elastic deformation.

In order to simplify calculating, and guarantee computational accuracy, make following reasonable assumption at the roll elastic deformation:

1. roll elastic deformation rotational symmetry;

2. the correction semi-infinite body theory on island during the flattening of flattening and working roll and rolled piece is adopted between roller;

3. rolled piece isotropy and incompressible;

4. the convexity distribution curve of working roll and backing roll is a second-degree parabola.

Cut apart with influence function method breaker roll, rolled piece, because roll deformation is symmetrical, half of only getting roll analyzed.

(1) input milling train, roll and rolled piece parameter and relevant convergence parameter

Roller parameter comprises:

Working roll and backing roll diameter, working roll and backing roll convexity cross direction profiles, backing roll roll neck diameter, working roll and

Backing roll barrel length and depress stress point centre distance.

The rolled piece parameter comprises:

Inlet thickness central value, exit thickness central value, rolled piece width and rolled piece temperature.

The convergence parameter comprises:

Roll gap pressure convergence permissible error, roll-force convergence permissible error and exit thickness cross direction profiles convergence permissible error.

(2) unit is cut apart, coordinate Calculation, roll distribution convexity are calculated, form the influence matrix G of roll deflection distortion _w, G _b

Needs according to influence function method, roll is carried out the unit vertically to be cut apart, for guaranteeing computational accuracy, the unit is cut apart can not be very little, generally need be more than 20, be about to working roll and backing roll and regard a lot of roundels as and combine, in like manner the rolled piece broad ways is carried out the unit and cut apart, should be noted that the cell width of roll and rolled piece contact portion must equate.

Get roll mandrel to being horizontal ordinate, vertical diameter is to being ordinate, and getting the roll center line is horizontal ordinate zero point, calculates the abscissa value of the central point of each unit, calculates the roll crown value of each unit center point simultaneously.

A cutting unit supposing working roll or backing roll is subjected to unit force, calculate the elastic deflection displacement that this unit force causes all cutting units of roll, and be referred to as the influence coefficient of unit force breaker roll deflection deformation, the influence coefficient of all cutting units is organized into working roll influence matrix G _wWith backing roll influence matrix G _b

(3) back transverse gage distribution vector H is rolled in supposition

Because the exit thickness cross direction profiles of rolled piece can't know accurately that before calculating convergence in order to make iteration carry out smoothly, original state can think that rolling back transverse gage distribution is homogeneous.

(4) calculate roll-force vector F,

Because after knowing that the rolled piece inlet thickness distributes and exit thickness distributes, can calculate the roll-force of each cutting unit of rolled piece according to the roll-force formula.Because dividing elements is thinner, can think that the roll-force of each unit is even distribution, its equivalent stress point is at the unit center point.

Roll-force belongs to external force, and its computing method are a lot, and hot rolling formula commonly used is the Sims model.

(5) supposition roll gap pressure vector Q,

There is a roll gap pressure between working roll and the backing roll, before calculating convergence, cross direction profiles that can't clear understanding roll gap pressure, but according to the working roll stress balance, as formula (1), can obtain the roll gap pressure summation, suppose that then initial roll gap pressure evenly distributes.

Q·Δx＝F·Δx (1)

Wherein, Q is that roller presses force vector indirectly, and F is the roll-force vector, and Δ x is the cell width vector.

(6) calculate roll deflection vector Y _w, Y _b,

Go out working roll and backing roll lateral deflection vector according to working roll and backing roll deflection deformation Equation for Calculating.

Y _w＝G _w·(Q-F) (2)

Y _b＝G _b·Q (3)

Wherein, Y _w, Y _bBe respectively working roll and backing roll deflection vector, G _w, G _bDifference working roll and backing roll flexure effects Jacobian matrix.

(7) the influence function matrix G of flattening deformation between the calculating roller _Wb, and center flattening value amount Y between the hypothesis roller _{Wb0_flatten},

Suppose that cutting unit of roll is subjected to unit force, calculate the elastic flattening displacement that this unit force causes all cutting units of roll, and be referred to as the influence coefficient of unit force breaker roll elastic flattening distortion, the influence coefficient of all cutting units is organized into roll elastic flattening influence matrix G _Wb,

Because flattening amount in center can't directly calculate with formula between roller, so need a given initial value Y _{Wb0_flatten}, and obtain its exact value by the iterative computation of back.

(8) calculate the vectorial Y of flattening between roller _Wb,

Can calculate according to formula (4) and to flatten cross direction profiles between roller,

Y _wb＝G _wb·Q (4)

Wherein, Y _WbBe to flatten vector, G between roller _WbBe to flatten the influence function matrix between roller.

(9) calculate roll gap pressure vector Q '

According to compatibility of deformation equation (5) between roller, can list a system of equations, thereby solve the roll gap pressure distribution Q ' that satisfies this compatibility of deformation equation,

Y _wb＝Y _wb0+Y _b-Y _w-(C _w+C _b) (5)

Wherein, Y _Wb0=[Y _{Wb0_flatten}, Y _{Wb0_flatten}..., Y _{Wb0_flatten}] ^T, be constant vector.C _w, C _bIt is respectively the convexity vector of working roll and backing roll.

(10) judge whether roll gap pressure calculated value Q ' and default restrain,, otherwise adjust roll gap pressure initial value Q, and jump to step (6) and carry out iterative computation if convergence just jumps to step (11),

Roll gap pressure modification method such as formula (6)

Q _n+1＝Q _n+α·(Q _n′-Q _n) (6)

Wherein, n, n+1 represents respectively the n time and the n+1 time iteration that α is a smoothing factor, general value 0.3～0.8.

(11) after the roll gap pressure convergence, judge whether the roll gap pressure summation equates with the roll-force summation,, otherwise adopt secant method to adjust flattening initial value in center between roller, jump to step (8) and carry out iterative computation if convergence jumps to step (12).

(12) the influence function matrix of the flat distortion of evaluation work roll-in,

Suppose a cutting unit unit of being subjected to of working roll roll-force, calculate the elastic flattening displacement that this unit roll-force causes all cutting units of working roll, and the unit's of being referred to as roll-force is to the influence coefficient of working roll elastic flattening distortion, and the influence coefficient of all cutting units is organized into working roll elastic flattening influence matrix G _Ws

(13) the vectorial Y of the flat flattening of evaluation work roll-in _Ws,

The working roll elastic flattening equation that roll-force causes is,

Y _ws＝G _ws·F (7)

Wherein, Y _WsBe that working roll flattens vector, G _WsBe that working roll flattens the influence function matrix.

(14) calculate rolled piece exit thickness cross direction profiles H ',

Utilize the compatibility of deformation equation between working roll and the rolled piece,, obtain rolled piece exit thickness cross direction profiles H ' as formula (8).

H＝H ₀+2·(Y _ws-Y _ws0)+2·(C _w-Y _w) (8)

Wherein, H=[h ₁, h ₂... h _n] ^TBe exit thickness vector, H ₀=[h ₀, h ₀..., h ₀] ^TBeing constant vector, is the thickness vector that rolled piece rolls the rear center place.Y _Ws0=[Y _{Ws0_flatten}, Y _{Ws0_flatten}..., Y _{Ws0_flatten}] ^T, be constant vector.

(15) judge whether rolled piece exit thickness cross direction profiles equates with initial value, if convergence jumps to step (16), otherwise, adjust rolled piece exit thickness cross direction profiles, jump to step (4) and carry out iterative computation,

Rolled piece thickness method of adjustment such as formula (9)

H _n+1＝H _n+β·(H _n′-H _n) (9)

Wherein, n, n+1 represent respectively the n time and the n+1 time iteration that β is a smoothing factor, general value 0.3～0.8.

(16) calculate the roll elastic deformation Ys that influences thickness prediction.

Y _s＝Y _r+Y _{wb0_flatten}+Y _{ws0_faltten} (10)

Y is flattened at the center between roller _{Wb0_flatten}, the working roll center flattens Y _{Ws0_flatten}Calculating according to the front can be easy to obtain.

In addition, the deflection deformation Y of the stress point on the supporting roller bearing seat _rNeed obtain by method of interpolation.Suppose that the backing roll deflection curve can fit to quartic polynomial,

Y _r(x)＝c ₀+c ₁·x ²+c ₂·x ⁴ (12)

In the formula: x is a backing roll axis coordinate figure, and its coordinate zero point is the point of roll center line correspondence.

Then with x=0, x=L _b/ 2, x=L _bThe calculated value substitution formula (12) of/4 correspondences, wherein L _bBe that the backing roll body of roll is long, obtain multinomial coefficient C ₀～C ₂, last deflection deformation according to this polynomial computation stress point.

(2) result of calculation of roll elastic deformation is returned can online use mathematical model

Because it is very big that influence function method relates to calculated amount, computing time is long, be not suitable for on-line setup and calculate, so need by mathematics manipulation with result of calculation fit to online application high precision mathematical model.

By degree of correlation analysis and constantly experiment, the result of calculation that influence function method is drawn returns into the mathematical form of formula (12), can effectively guarantee the Model Calculation precision.

$Y_S=\frac{F*K_F}{1000}---\left(12\right)$

$K_F={\mathrm{<figure-callout\; id="1"\; label="K\; F"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">K</figure-callout>}}_F+{\mathrm{<figure-callout\; id="2"\; label="K\; F"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">K</figure-callout>}}_F\&CenterDot;\left(\frac{\mathrm{Width}}{2000}\right)$

${\mathrm{<figure-callout\; id="1"\; label="K\; F"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">K</figure-callout>}}_F=a_1+a_2\&CenterDot;\left(\frac{R_w}{500}\right)+a_3\&CenterDot;\left(\frac{R_b}{1000}\right)+a_4\&CenterDot;C_w+a_5\&CenterDot;C_b$

${\mathrm{<figure-callout\; id="2"\; label="K\; F"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">K</figure-callout>}}_F={\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">b</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_2\&CenterDot;\left(\frac{R_w}{500}\right)+{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_3\&CenterDot;\left(\frac{R_b}{1000}\right)+{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_4\&CenterDot;C_w+b_5\&CenterDot;C_b$

Wherein, Y _sBe roll elastic deformation, F is a roll-force, and Width is the rolled piece width, R _w, R _bBe respectively working roll, backing roll radius, C _w, C _bBe respectively working roll, backing roll convexity, a ₁～a ₅, b ₁～b ₅It is the mathematical model regression parameter.Analysis mode (12), the elastic deformation of roll is directly proportional with roll-force as can be seen, and scale-up factor K _FForm by two parts, wherein K _F1Be and the irrelevant part of rolled piece width, and K _F2Be the part relevant with the rolled piece width.But K _F1And K _F2It all is the line shape function of working roll roller footpath, backing roll roller footpath, work roll crown and backing roll convexity.

3, measure the elastic deformation of mill housing and associated mechanical components

Because mill housing and associated mechanical components be after milling train builds up, can change, so its elastic deformation of the operation of rolling is only relevant with pressure.

In order to improve computational accuracy, adopt measurement method-roller pressing method: choose a cover with reference to roller system, two rolls are pressed, measure the change curve of pressure and mill spring, i.e. the mill spring curve.The elastic deformation of mill spring curve correspondence has comprised the elastic deformation of roll elastic deformation and memorial archway and associated mechanical components, so utilize calculated with mathematical model to go out corresponding roll elastic deformation, just can obtain the elastic deformation of memorial archway and associated components.

4, will become corresponding mathematical model with reference to the mill spring curvilinear regression that roller system surveys down

The mathematical model of mill spring model adopts the pattern of formula (13).

$S_F={\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">d</figure-callout>}}_1\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^{1/2}+{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_2\&times;(\frac{F}{F_{\mathrm{st}}}-N)+{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_3\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^{3/2}+{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_4\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^2---\left(13\right)$

Wherein, S _FBe mill spring, F is the pressure that is equal to the roll-force of the operation of rolling, F _StBe the step value of measuring process pressure, N is original pressure and F _StRatio, d ₁～d ₄It is the model regression coefficient.In the formula with pressure F divided by F _StCome down to pressure is carried out normalized, it is too big to reduce between each regression coefficient of model order of magnitude difference, influences computational accuracy.

5, set up mill spring forecast model (being also referred to as thickness meter model), the thickness of prediction steel plate

The mill spring forecast model adopts the pattern of formula (14),

$h=(S-S_0)+(S_F-S_{F_0})+\frac{F}{F_{\mathrm{st}}}{\mathrm{\&Delta;K}}_F---\left(14\right)$

In the formula, h is exactly the thickness of steel plate, and S sets roll gap, S ₀Be the roll-force pressure F that equals to return to zero ₀The time gap values between rollers, S _FBe the mill spring when roll-force equals F under reference roller system, S _F0Be the roll-force pressure F that equals to return to zero under reference roller system ₀The time mill spring, Δ K _FBe working roll and backing roll roller footpath, the roll elastic deformation scale-up factor when convexity and rolled piece width change and be the difference of scale-up factor, can calculate very easily by the mill spring under any state of this formula with reference to roller.

6, eliminate the influence of mill spring model zero point drift to thickness prediction

Because operation of rolling roll wear and thermal expansion are difficult to accurate Calculation, so just make the generation at zero point of mill spring model drift about, in order to reduce the influence of zero point drift to thickness prediction, introduce the zero point drift self-learning algorithm and eliminate this influence, further improve intermesh determination precision and THICKNESS CONTROL control precision.

Self-learning algorithm as shown in Equation (15),

Δ _0(n)＝(1-α)·Δ0 _(n-1)+α·(h _n-1′-h _n-1) (0＜α＜1) (15)

Wherein, h _N-1' be the actual measurement thickness of n-1 block plate, h _N-1Be the thickness prediction of n-1 block plate, Δ _{0 (n-1)}Be the modified value at zero point of n-1 block plate, Δ _{0 (n)}It is the modified value at zero point of n block plate.

The mill spring model can be modified to the pattern of formula (16) like this.

$h=(S-S_0)+(S_F-S_{F_0})+\frac{F}{F_{\mathrm{st}}}{\mathrm{\&Delta;K}}_F+{\mathrm{\&Delta;}}_{0\left(n\right)}---\left(16\right)$

7, actually in the steel plate rolling process carry out online thickness prediction and be:

In case steel plate enters milling train and is rolled, then pressure detection instrument and position probing instrument will be measured actual rolling force F and actual roll gap S, calculate with reference to the mill spring S under the roller system according to roll-force then _F, next these parameters are updated to formula (16), can calculate the thickness of steel plate.

The present invention can stablize, predict apace steel plate thickness with very high precision, is that operation of rolling automatic setting and manual intervention improve necessary information; Avoid the operating process of duplicate measurements mill spring curve, increase rolling time; Factors such as roller radius, roll crown, rolled piece width and roll-force can obtain embodying to the influence of mill spring by this model.The present invention is applicable to the roughing mill and the finishing mill of heavy and medium plate mill, hot continuous rolling process.

Description of drawings

Fig. 1 forms synoptic diagram for the roll elastic deformation that influences thickness prediction;

Among Fig. 1: L1 is backing roll axial location and a shape before the stress deformation;

L2 is backing roll axial location and a shape behind the stress deformation;

Fig. 2 is the stress model synoptic diagram of operation of rolling roller system;

Fig. 3 is the roll elastic deformation calculation flow chart;

Fig. 4 is the influence curve of rolled piece width to mill spring;

Fig. 5 is the influence curve of working roll radius to mill spring;

Fig. 6 is the influence curve of backing roll radius to mill spring;

Fig. 7 is worker and the roll crown influence curve to mill spring;

Fig. 8 is the influence of backing roll convexity to mill spring;

Fig. 9 is the computational accuracy of roll elastic deformation regression mathematical model;

Figure 10 is driving side spring curve;

Figure 11 is fore side spring curve.

Embodiment

1, chooses a cover heavy and medium plate mill, utilize influence function method to calculate of the influence of different factors roll elastic deformation.

Wherein table 1 is the heavy and medium plate mill parameter.

Table 1 calculates starting condition

Device name	Parameter	Device name	Parameter
				Work roll diameter/mm backing roll diameter/mm work roll crown/um	900～1100 1900～2100 -300～300	Backing roll convexity/um barrel length/mm depresses centre distance/mm	-200～200 3400 4800

Make the working roll radius equal 450mm, 470mm, 490mm, 510mm, 530mm, 550mm respectively; Make the backing roll radius equal 950mm, 970mm, 1010mm, 1030mm, 1050mm respectively; Make steel plate width equal 1800mm, 2200mm, 2400mm, 2600mm, 2800mm, 3000mm respectively; Make work roll crown be respectively-0.3mm ,-0.2mm ,-0.1mm, 0mm, 0.1mm, 0.2mm; Make the backing roll convexity be respectively-0.3mm ,-0.2mm ,-0.1mm, 0mm, 0.1mm, 0.2mm; So just can obtain 6 ⁵Array mode calculates roll elastic deformation according to the various combination mode.Fig. 4～8th, the different affecting factors that obtains according to influence function method is to the influence of roll elastic deformation.

(2) result of calculation of roll elastic deformation is returned can online use mathematical model.

The mathematical model of corresponding formula (12) is carried out linear regression, obtains the regression coefficient of table 2, and roll elastic deformation forecast of regression model precision is seen Fig. 9.

$Y_s=\frac{F*K_F}{1000}---\left(12\right)$

$K_F={\mathrm{<figure-callout\; id="1"\; label="K\; F"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">K</figure-callout>}}_F+{\mathrm{<figure-callout\; id="2"\; label="K\; F"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">K</figure-callout>}}_F\&CenterDot;\left(\frac{\mathrm{Width}}{2000}\right)$

${\mathrm{<figure-callout\; id="1"\; label="K\; F"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">K</figure-callout>}}_F=a_1+a_2\&CenterDot;\left(\frac{R_w}{500}\right)+a_3\&CenterDot;\left(\frac{R_b}{1000}\right)+a_4\&CenterDot;C_w+a_5\&CenterDot;C_b$

${\mathrm{<figure-callout\; id="2"\; label="K\; F"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">K</figure-callout>}}_F={\mathrm{<figure-callout\; id="1"\; label="b"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">b</figure-callout>}}_1+{\mathrm{<figure-callout\; id="2"\; label="b"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_2\&CenterDot;\left(\frac{R_w}{500}\right)+{\mathrm{<figure-callout\; id="3"\; label="b"\; filenames="C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_3\&CenterDot;\left(\frac{R_b}{1000}\right)+{\mathrm{<figure-callout\; id="4"\; label="b"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">b</figure-callout>}}_4\&CenterDot;C_w+b_5\&CenterDot;C_b$

Table 2 roll elastic deformation model parameter regressand value

Parameter	Parameter value	Parameter	Parameter value
				a 1 a 2 a 3 a 4 a 5 R w0，/mm W 0，/mm	0.23539 -0.04581 -0.10347 0.002115 0.02394 500 2000	b 1 b 2 b 3 b 4 b 5 R b0，/mm	-0.04947 0.023683 0.005905 -0.00475 0.008755001000 1000

3, measure the elastic deformation of mill housing and associated mechanical components.

At this milling train, choose with reference to roller system and measure, be that corresponding parameters is that the working roll radius is 550mm with reference to roller, the backing roll radius is 1050mm, work roll crown is 0, and the backing roll convexity is 0, thinks that simultaneously the process of pressing is equivalent to the steel plate that rolling width equals barrel length.

1. under rotating speed 20r/m, make the hydraulic pressure oil column height equal 10mm, keep the hydraulic pressure invariant position to carry out electronic pressing, reach one-sided 2000kN until snap-in force;

2. keep electronic invariant position, take pressure closed loop to carry out hydraulic pressure and depress, original pressure point is 3500kN, is that step value is adjusted the hydraulic pressure position then with 700kN;

3. under a certain pressure, adopt the pressure closed loop control mode, fine setting hydraulic pressure oil column height is gathered many group pressure signals and roll gap signal;

4. after measurement is finished, need release oil pressure, lift roll gap.

The milling train both sides spring curve such as Figure 10 and 11 that measure.

4, will become corresponding mathematical model with reference to real side mill spring curvilinear regression under the roller system

The measurement curve of milling train both sides is carried out the translation processing, then the both sides roll gap is asked average treatment, obtain the one-to-one relationship of pressure and roll gap, carry out regression treatment, obtain the regression parameter of model (13) according to handling later data, as shown in table 3.

$S_F={\mathrm{<figure-callout\; id="1"\; label="d"\; filenames="C20031011900500151.png,C20031011900500171.png"\; state="\{\{state\}\}">d</figure-callout>}}_1\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^{1/2}+{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_2\&times;(\frac{F}{F_{\mathrm{st}}}-N)+{\mathrm{<figure-callout\; id="3"\; label="d"\; filenames="C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_3\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^{3/2}+{\mathrm{<figure-callout\; id="4"\; label="d"\; filenames="C20031011900500181.png,C20031011900500182.png"\; state="\{\{state\}\}">d</figure-callout>}}_4\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^2---\left(13\right)$

Parameter	Parameter value	Parameter	Parameter value
				d 1 d 2 d 3 d 4	-0.279436 0.236935 -0.024040 0.001344	F St，/KN N R The b_ reference，/mm R The w_ reference，/mm	500 6 1050 525

5, set up mill spring forecast model (being also referred to as thickness meter model)

$h=(S-S_0)+(S_F-S_{F_0})+\frac{F}{F_{\mathrm{st}}}{\mathrm{\&Delta;K}}_F---\left(14\right)$

Wherein, S ₀Be the roll-force pressure F that equals to return to zero _nThe time gap values between rollers, generally with it as relative zero, promptly this numerical value is 0, order zeroing pressure equals 20000kN.Δ K _FBe working roll and backing roll roller footpath, the roll elastic deformation scale-up factor when convexity and rolled piece width change and be the difference of scale-up factor with reference to roller, this can calculate very easily by this formula (12), can calculate the mill spring value S of zeroing pressure correspondence according to regression formula (13) _F0Be 1.2757mm.

6, the zero point drift of mill spring model is to the elimination of thickness prediction influence.

Make the influence Δ of milling train zero point drift to thickness prediction _{0 (n)}Equal zero.

7, online thickness prediction

Utilize pressure transducer to measure and to record pressure and roll gap signal with displacement transducer.

As rolling force F=37965kN, roll gap S=37.8mm, steel plate width Width=2500mm is that parameter substitution formula (12), (13), (16) can obtain steel plate thickness h=39.95mm with these parameters and roller.

If rolling force F=36296kN, roll gap 26.7mm, steel plate width Windth=2500mm is that parameter substitution formula (12), (13), (16) can obtain steel plate thickness h=28.7mm with these parameters and roller.

Claims (3)

1, a kind of in the operation of rolling method of on-line prediction steel plate thickness, it is characterized in that may further comprise the steps:

1. calculate the influence Ys of roll elastic deformation with influence function method to mill spring

$Y_S=\frac{F*K_F}{1000}$ Y in the formula _SBe roll elastic deformation, F is a roll-force;

2. roll elastic deformation result of calculation is returned into online mathematical model

$Y_S=\frac{F*K_F}{1000}---\left(12\right)$

$K_F=K_{F1}+K_{F2}\&CenterDot;\left(\frac{\mathrm{Width}}{2000}\right)$

$K_{F1}=a_1+a_2\&CenterDot;\left(\frac{R_w}{500}\right)+a_3\&CenterDot;\left(\frac{R_b}{1000}\right)+a_4\&CenterDot;C_w+a_5\&CenterDot;C_b$

$K_{F2}=b_1+b_2\&CenterDot;\left(\frac{R_w}{500}\right)+b_3\&CenterDot;\left(\frac{R_b}{1000}\right)+b_4\&CenterDot;C_w+b_5\&CenterDot;C_b$

Wherein, Y _sBe roll elastic deformation, F is a roll-force, and Width is the rolled piece width, R _w, R _bBe respectively working roll, backing roll radius, C _w, C _bBe respectively working roll, backing roll convexity, a ₁～a ₅, b ₁～b ₅Be the mathematical model regression parameter, the elastic deformation of roll is directly proportional with roll-force as can be seen from analysis mode (12), and scale-up factor K _FForm by two parts, wherein K _F1Be and the irrelevant part of rolled piece width, and K _F2Be the part relevant with the rolled piece width, but K _F1And K _F2It all is the line shape function of working roll roller footpath, backing roll roller footpath, work roll crown and backing roll convexity;

3. measure the elastic deformation of mill housing and associated mechanical components;

4. obtain mill spring S _FMathematical model,

$S_F=d_1\&times;{\left(\frac{F}{F_{\mathrm{st}}}\right)}^{1/2}+d_2\&times;(\frac{F}{F_{\mathrm{st}}}-N)d_3\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^{3/2}+d_4\&times;{(\frac{F}{F_{\mathrm{st}}}-N)}^2$

Wherein, S _FBe mill spring, F is the pressure that is equal to the roll-force of the operation of rolling, F _StBe the step value of measuring process pressure, N is original pressure and F _StRatio, d ₁～d ₄It is the model regression coefficient;

5. with the thickness of thickness meter model prediction steel plate,

Thickness meter model is $h=(S-S_0)+(S_F-S_{F_0})+\frac{F}{F_{\mathrm{st}}}\mathrm{\&Delta;}{K}_F,$ H is the thickness of steel plate in the formula, and S sets roll gap, S ₀Be the roll-force pressure F that equals to return to zero ₀The time gap values between rollers, S _FBe the mill spring when roll-force equals F under reference roller system, S _F0Be the roll-force pressure F that equals to return to zero under reference roller system ₀The time mill spring, Δ K _FBe working roll and backing roll roller footpath, the roll elastic deformation scale-up factor when convexity and rolled piece width change and be the difference of scale-up factor with reference to roller;

6. eliminate the influence of mill spring model zero point drift,, dope the thickness of steel plate with mill spring correction calculated with mathematical model to former degree prediction:

$h=(S-S_0)+(S_F-S_{F_0})+\frac{F}{F_{\mathrm{st}}}\mathrm{\&Delta;}{K}_F+{\mathrm{\&Delta;}}_{0\left(n\right)},$ Δ in the formula _{0 (n)}Be the modified value at zero point of n block plate, Δ _{0 (n})=(1-α) Δ _{0 (n-1)}+ α (h ' _N-1-h _N-1) (0＜α＜1)

Wherein, h ' _N-1Be the actual measurement thickness of n-1 block plate,

h _N-1Be the thickness prediction of n-1 block plate,

Δ _{0 (n-1)}It is the modified value at zero point of n-1 block plate.

2, according to claim 1 in the operation of rolling method of on-line prediction steel plate thickness, the elastic deformation that it is characterized in that described measurement mill housing and associated mechanical components is to adopt measurement method, be the roller pressing method: choosing a cover with reference to roller is, two rolls are pressed, measure the change curve of pressure and mill spring, the elastic deformation of mill spring curve correspondence has comprised the elastic deformation of roll elastic deformation and memorial archway and associated mechanical components.

3, according to claim 1 in the operation of rolling method of on-line prediction steel plate thickness, it is characterized in that the step of described calculating roll elastic deformation is:

(1) input milling train, roll and rolled piece parameter and relevant convergence parameter;

(2) the influence matrix G that carry out that the unit is cut apart, the roll deflection distortion is calculated, formed to coordinate Calculation, roll distribution convexity _w, G _b

(3) back transverse gage distribution vector H is rolled in setting;

(4) calculate roll-force vector F;

(5) set roll gap pressure vector Q and center and flatten vectorial Y _{Wb0_flatten}

(6) calculate roll deflection vector Y _w, Y _b

(7) the influence function matrix G of flattening deformation between the calculating roller _Wb

(8) calculate the vectorial Y of flattening between roller _Wb

(9) calculate roll gap pressure vector Q ';

(10) judge whether roll gap pressure calculated value Q ' and default restrain,, otherwise adjust roll gap pressure initial value Q, and jump to step (6) and carry out iterative computation if convergence just jumps to step (11);

(11) after the roll gap pressure convergence, judge whether the roll gap pressure summation equates with the roll-force summation,, otherwise adopt secant method to adjust flattening initial value in center between roller, jump to step (8) and carry out iterative computation if convergence jumps to step (12);

(12) the influence function matrix of the flat distortion of evaluation work roll-in;

(13) the vectorial Y of the flat flattening of evaluation work roll-in _Ws

(14) calculate rolled piece exit thickness cross direction profiles H ';

(15) judge whether rolled piece exit thickness cross direction profiles equates with initial value, if convergence jumps to step (16), otherwise, adjust rolled piece exit thickness cross direction profiles, jump to step (4) and carry out iterative computation;

(16) calculate the roll elastic deformation Y that influences thickness prediction _s

(17) output result of calculation.

Images