CN103357669A - Plate model prediction control method - Google Patents

Abstract

The invention relates to a plate model prediction control method and belongs to the technical field of steel rolling quality control. The plate model prediction control method includes: when a rolling control computer detects that a band steel speed is lower than a preset value, taking a secondary plate shape component and a first plate shape component in an actual plate shape curve as target values of model parameter optimization; obtaining a static secondary plate shape component, a static first plate shape component, a dynamic secondary plate shape component and a dynamic first plate shape component according to a static prediction model and a dynamic prediction model; after the static secondary plate shape component and the static first plate shape component are taken as control initial values, respectively comparing the dynamic secondary plate shape component and the dynamic first plate shape component with the corresponding target values, and outputting corresponding control parameters according to comparing results. Reasonable prediction parameters are adopted to substitute for actual measuring parameters for comparison and control, so that control hysteresis when the band steel rotating speed is low can be avoided, existing plate shape feedback control can be supplemented, and cold rolling quality is guaranteed.

Description

A kind of shape models forecast Control Algorithm

Technical field

The present invention relates to a kind of method of steel rolling quality control, especially a kind of shape models forecast Control Algorithm belongs to steel rolling Quality Control Technology field.

Background technology

The plate shape of cold-strip steel directly has influence on the height of productivity ratio, lumber recovery and cost of downstream industry and the outward appearance of product, so the control of plate shape has great importance for the quality that improves cold-strip steel.

Plate shape with steel comprises the in length and breadth size index of two aspects, and with regard to the band steel was vertical, plate shape referred to glacing flatness usually, is commonly called as shape wave, and namely the edge is with the smooth degree on the steel length direction; With regard to the band steel laterally with regard to, plate shape refer to is with the section configuration of steel, i.e. thickness distribution on the plate width direction, convexity is that the transverse plate shape of commonly using the most represents index.

Understand according to the applicant, what adopt in the automatic strip shape control system at present is plate shape FEEDBACK CONTROL, namely detect actual plate shape value with steel by plate shape measurement roller and plate shape measurement system, automatic strip shape control in the controller calculates the setting value of executing agency according to the deviation of target flatness and actual plate shape value.Be that the disclosed plat control system technical scheme of Chinese patent application of CN200510028316.2 is exactly the processing according to the actual measurement board form data such as application number, obtain after treatment surveying Shape signal; Calculate by plate shape deviation, deduct target flatness with actual measurement plate shape, obtain the deviation Shape signal, thereby carry out the control of plate shape.But less than the plate shape closed-loop control cycle, namely its controlled condition is that strip speed just can come into operation greater than certain value based on the sense cycle of plate shape measurement system for it.Because plate profile instrument has certain distance apart from rolling mill 5 frames (last frame) outlet, when strip speed was lower than certain value, plate shape detection signal τ lag time will be greater than plate shape closed-loop control cycle T, and control system will be difficult to stable control.In order to ensure the stability of closed-loop control link, control system gain value is had to less, and the result causes the deleterious of plate shape closed-loop control.Further retrieval is found, application number is 200810207919.2 Chinese patent application---in the sheet shape prediction and control method of model adaptation, by real data plate shape control model is revised, but because this plate shape forecast model is not considered the variation of previous moment board form data, when real data changes in the Dynamic Rolling Process process more violent the time, plate shape control model causes the unstable of system easily.

Application number is the method for designing that the patent of invention of CN200810011561.6 discloses a kind of cold-rolled strip steel shape control object module, determines the Mathematical Modeling that the description strip profile and flatness is controlled according to the mathematics constraints that structure and operation principle and the target flatness of cold-rolled strip steel shape characteristics, Mill shape control actuator should satisfy; According to the different process quality requirement of the kind of rolled band steel and specification, roll rear different disposal operation roll wear and hot convexity in the requirement of belt plate shape and the operation of rolling changed to determine different control parameters in the plate shape object module, form different target flatness curves, be used for the real-time plate shape control of process control of cold rolling calculated with mathematical model and basic automatization.Its object module mainly solves the desired value of plate shape flatness control, and off-line simulation is static to be set, rather than Dynamic Regulating Process.

In addition, application number is that the patent of invention of CN200910011950.3 discloses a kind of optimizing regulating and controlling efficiency coefficient of board shape controlling actuator of cold rolling mill method, set up the plate shape regulation and control efficiency coefficient priori value table under the different rolling operating points, in table, the corresponding rolling operating point of one group of strip width value and roll-force value, according to its operating point, border of the location positioning in table, the rolling operating point of reality, set weight factor by the similarity degree of operating point, border parameter and actual rolling operating point parameter, the priori efficiency coefficient weighted superposition by boundary point obtains the plate shape regulation and control efficiency coefficient under the actual rolling operating point; The plate shape of using online self learning model and actual measurement board form data to update in the table is regulated and control the efficiency coefficient precision, can obtain the regulation and control efficiency coefficient of accurate profile regulation mechanism, and be applied in the closed loop plat control system, have higher plate shape control accuracy.The problem of its existence is not have to solve because the control of open loop plate shape or the unsettled problem of closed-loop control that actual measurement plate shape lags behind and carries out under the low-speed conditions.

Summary of the invention

The object of the invention is to: for the deficiency of above-mentioned prior art existence, a kind of plate shape shape models forecast Control Algorithm that can effectively solve strip speed control hysteresis problem when low is proposed, thereby as replenishing of existing plate shape FEEDBACK CONTROL, for the quality that guarantees cold-rolled strip steel shape is created more favorably condition.

By background technology as can be known, in the present cold-strip steel automatic control system, determining target flatness curve y (x)=A ₀+ A ₁X+A ₂x ², be A ₀, A ₁, A ₂After determining, when actual measurement plate shape curve once, when quadratic term coefficient and aim curve are not inconsistent, computer can be by the existing program of operation, comparative result according to actual measurement plate shape component and target flatness component (once item, quadratic term coefficient), the controlled quentity controlled variables such as the roller declination of electronically operated mill and bending roller force, try hard to make actual measurement coefficient convergence target factor, just since this feedback closed loop be controlled at strip speed can't simultaneously match when low, be difficult to play required feedback modifiers effect.

Therefore, in order to realize the above object, shape models forecast Control Algorithm of the present invention is mainly by the roll control computer of control five stand mill and be arranged in the control system that the plate profile instrument of final milling train output consists of, the signal output part of described plate profile instrument connects the corresponding ports of roll control computer, and the control output end of described roll control computer is connected to respectively the controlled end of each roll; When the roll control COMPUTER DETECTION is lower than predetermined value to strip speed, carry out according to the following steps a basic Recurrence Process control:

The first step, one group of standard actual measurement board form data of plate profile instrument collection is carried out data process, obtain secondary plate shape component and a plate shape component in the actual plate shape curve, as the desired value of Model Parameter Optimization;

Second step, according to the static prediction model, try to achieve the static secondary plate shape component A in the desirable rolling situation ₂(t) and static plate shape component A ₁(t);

Secondary plate shape static prediction model is:

$A_2\left(t\right)=F_I\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; I"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_I}+F_W\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; W"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_W}+I_{\mathrm{MR}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; I\; MR"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_{\mathrm{MR}}}+P\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_P+T_5\left(t\right)\&times;{\mathrm{<figure-callout\; id="5"\; label="K\; T"\; filenames="DEST\_PATH\_120606113620.png,HDA0000147268170000032.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}+T_{45}\left(t\right)\&times;{\mathrm{<figure-callout\; id="45"\; label="K\; T"\; filenames="HDA0000147268170000032.png,HDA0000147268170000041.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}$ In the formula:

A ₂(t)---be t static secondary plate shape component constantly;

F _I(t)---be t moment intermediate calender rolls bending roller force mean value, unit: ton;

F _W(t)---be t moment work roll bending power mean value, unit: ton;

I _MR(t)---be t upper and lower intermediate calender rolls string roller amount mean value of the moment, unit: millimeter;

P (t)---be t moment roll-force, unit: ton;

T ₅(t)---be constantly 5 frames outlet of t tension force, unit: ton;

T ₄₅(t)---be the t moment 4～5 interstand tensions, unit: ton;

K _P2,

--be respectively intermediate calender rolls bending roller force, work roll bending power, intermediate roll shifting amount, roll-force, 5 frames outlet tension force, 4～5 interstand tensions to the influence coefficient of static secondary plate shape component;

One time plate shape static prediction model is:

$A_1\left(t\right)=F_{\mathrm{ID}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; ID"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{ID}}}+F_{\mathrm{WD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; WD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{WD}}}+I_{\mathrm{MRD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; I\; MRD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_{\mathrm{MRD}}}+I_P\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; I\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_P}$

In the formula:

A ₁(t)---be a t plate shape of static state component constantly;

F _ID(t)---the intermediate calender rolls bending roller force is poor constantly for t, unit: ton

F _WD(t)---work roll bending power is poor constantly for t, unit: ton;

I _MRD(t)---the intermediate calender rolls shifting amount is poor up and down constantly for t, unit: millimeter;

I _P(t)---be t moment roller declination amount, unit: micron;

---be respectively that the intermediate calender rolls bending roller force is poor, work roll bending power is poor, the intermediate calender rolls shifting amount is poor, the roller declination amount is to the influence coefficient of a static plate shape component;

The 3rd step, according to the dynamic forecasting model, try to achieve the secondary plate shape component A of the dynamic prediction model prediction in the actual rolling situation ₂(t) and a plate shape component A ₁(t);

Secondary plate shape dynamic forecasting model is:

$A_2\left(t\right)=A_2(t-1)\&times;{\mathrm{<figure-callout\; id="2"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_2+{\mathrm{\&Delta;F}}_I\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; I"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_I}+{\mathrm{\&Delta;F}}_W\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; W"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_W}+{\mathrm{\&Delta;I}}_{\mathrm{MR}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; I\; MR"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_{\mathrm{MR}}}+\mathrm{\&Delta;P}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_P+$

${\mathrm{\&Delta;T}}_5\left(t\right)\&times;{\mathrm{<figure-callout\; id="5"\; label="K\; T"\; filenames="DEST\_PATH\_120606113620.png,HDA0000147268170000032.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}+{\mathrm{\&Delta;T}}_{45}\left(t\right)\&times;{\mathrm{<figure-callout\; id="45"\; label="K\; T"\; filenames="HDA0000147268170000032.png,HDA0000147268170000041.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}$

In the formula,

A ₂(t)---be the dynamic secondary plate shape component when previous moment;

A ₂(t-1)---be the dynamic secondary plate shape component of previous moment (being the t-1 moment);

Δ F _I(t)---be t moment intermediate calender rolls bending roller force increment, unit: ton;

Δ F _W(t)---be t moment work roll bending power increment, unit: ton;

Δ P (t)---be t moment roll-force increment, unit: ton;

Δ T ₅(t)---be constantly 5 frames outlet of t tension increment, unit: ton;

Δ T ₄₅(t)---be the tension increment between the t moment 4～5 frames, unit: ton;

One time plate shape dynamic forecasting model is:

$A_1\left(t\right)=A_1(t-1)\&times;{\mathrm{<figure-callout\; id="1"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_1+{\mathrm{\&Delta;F}}_{\mathrm{ID}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; ID"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{ID}}}+{\mathrm{\&Delta;F}}_{\mathrm{WD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; WD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{WD}}}+\mathrm{\&Delta;}{I}_P\left(t\right)\&times;K_{I_{\mathrm{<figure-callout\; id="1"\; label="P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">P</figure-callout>}}1}$

In the formula,

A ₁(t-1)---be the dynamic plate shape component of previous moment (being the t-1 moment);

Δ F _ID(t)---be the poor increment of t moment intermediate calender rolls bending roller force, Δ F _ID(t)=F _ID(t)-F _ID(t-1);

Δ F _WD(t)---be the poor increment of t moment work roll bending power, Δ F _WD(t)=F _WD(t)-F _WD(t-1);

Δ I _p(t)---be t moment roller declination amount increment, Δ I _p(t)=I _p(t)-I _p(t-1);

The 4th step, with static secondary plate shape component and a static plate shape component respectively with the comparative result of corresponding described desired value after each roll output initial control signal, compare with corresponding described desired value with dynamic secondary plate shape component and a dynamic plate shape component respectively, and export dynamic control signal according to comparative result to each roll.

After adopting the present invention, compare control owing to replace the actual measurement parameter with rational Prediction Parameters, therefore the control hysteresis in the time of can avoiding strip speed low, thereby as having replenishing of plate shape FEEDBACK CONTROL now, for the quality that guarantees cold-rolled strip steel shape is created more favorably condition.

Description of drawings

Fig. 1 is the plat control system schematic diagram of one embodiment of the invention.

Fig. 2 is the flow chart of one embodiment of the invention control procedure.

Fig. 3 is dynamic sheet shape asymmetry part deviation measured waveform figure.Abscissa is sampling instant among the figure, and ordinate is the deviation of asymmetry part A1.

Fig. 4 is dynamic sheet shape symmetrical components deviation measured waveform figure.Abscissa is sampling instant among the figure, and ordinate is the deviation of symmetrical components A2.

Fig. 5 is the forecast bending roller force increment changing trend diagram that symmetrical plate shape component deviation causes.

The specific embodiment

The plat control system of present embodiment comprises that mainly the control output end is connected to the roll control computer 1 of each roll and the plate profile instrument 2 that signal output part connects roll control computer corresponding ports by interface unit respectively as shown in Figure 1.Testing agency's plate shape measurement roll spacing in the plate profile instrument 2 exports apart from 1=3.2m from 5# frame (also being called for short 5 frames), when plate shape closed-loop control cycle T=0.65s, when 5# frame outlet during with the speed of service V of steel＜282m/min, plate shape detection signal time-delay τ=1/V=3.2 ÷ 282 * 60=0.68s appears greater than the situation of plate shape control cycle T=0.65s then.At this moment, the control computer automatically switches to shape models forecast Control Algorithm of the present invention when the roll control COMPUTER DETECTION is lower than predetermined value to strip speed, controls according to the following steps (referring to Fig. 2):

The first step, adopt least square method to return with conic section to one group of standard actual measurement board form data of plate profile instrument collection, the secondary plate shape component that obtains in the actual plate shape curve (is quadratic term coefficient A ₂---reflect symmetrical plate shape) and a plate shape component (be Monomial coefficient A ₁---reflection asymmetric plate shape), as the desired value of Model Parameter Optimization;

This enforcement plate profile instrument adopts 40 passages, wherein in the middle of 12 passages, each passage 52mm of being separated by; Everybody 14 passages of both sides, each passage 26mm of being separated by.Each passage upward pressure measuring transducer is converted to the signal of telecommunication with radial load, and the plate shape measurement system is processed and calculates the signal of telecommunication that sends, and obtains actual Shape signal.After removing noise processed, adopt least square method to return with conic section to the plate profile instrument detection signal.Actual measurement plate shape can be expressed from the next:

y(x)＝A ₀+A ₁x+A ₂x ²

In the formula, y (x) is actual measurement plate shape pattern-recognition fitting function, A ₀～A ₂Be constant, wherein A ₂, A ₁Secondary plate shape component (being symmetrical plate shape) and a plate shape component (being asymmetric plate shape) of rolled band steel have been represented respectively, it is respective panels shape executing agency characteristic respectively, the bending roller force of the corresponding working roll of symmetrical plate shape component and intermediate calender rolls, the corresponding roller declination of asymmetric plate shape component.A ₁X represents to survey the linear segment in the plate shape; A ₂x ²Parabola part in the expression actual measurement plate shape.

Native system adopts plate profile instrument for detecting plate profile instrument along the tension force that laterally is divided into 40 sections with steel, in regression process, at first carry out the normalization on the plate width direction, be about to plate profile instrument and regard a reference axis as, take the center of plate profile instrument as initial point, then the position of each measuring section on number axis is known, as shown in table 1.

The position of measuring section on reference axis when table 1 covers fully with steel

i	1	2	3	......	20	......	38	39	40
										x i	-1.000	-0.950	-0.900	......	-0.050	......	0.900	0.950	1.000
y i	y 1	y 2	y 3	......	y 20	......	y 38	y 39	y 40

In the table 1, i is the measuring section numbering, x _iBe the coordinate position of i section on number axis, namely leave the distance at initial point (band steel center), y _iBe the plate shape measured value on the i section.If actual measurement can not cover all passages of plate profile instrument with the plate of steel is wide, then need to calculate smallest passage min and the largest passages max of effective overlay area.Take the wide 847mm of plate as example, in fact drafting board shape as shown in Figure 2, its smallest passage min=7, largest passages max=31.Can get thus:

$y_{\min }=A_0+A_1{x}_{\min }+A_2{\mathrm{<figure-callout\; id="2"\; label="x\; min"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">x</figure-callout>}}_{\min }^{}+e_{\min }$

. .

. .

. .

$y_{\max }=A_0+A_1{x}_{\max }+A_2{\mathrm{<figure-callout\; id="2"\; label="x\; max"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">x</figure-callout>}}_{\max }^{}+e_{\max }$

And note:

$Y=\left[\begin{array}{c}{y}_{\min }\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ y_{\max }\end{array}\right],$ $X=\left[\begin{array}{ccc}1& x_{\min }& x_{\mathrm{<figure-callout\; id="2"\; label="min"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">min</figure-callout>}}^2\\ \&CenterDot;& \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;\\ 1& x_{\mathrm{<figure-callout\; id="2"\; label="max\; x\; max"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">max</figure-callout>}}& x_{\max }^{}{}_2^{}\end{array}\right],$ $\mathrm{\&theta;}=\left[\begin{array}{c}{A}_0\\ A_1\\ {\mathrm{<figure-callout\; id="2"\; label="A"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">A</figure-callout>}}_2\end{array}\right]$

Plate shape characteristic coefficient A ₀～A ₂Error of fitting is satisfied:

Value minimum, namely to this function Q (θ)=(Y-X θ) ^T(Y-X θ) minimizing can be tried to achieve A ₀～A ₂

Second step, according to the static prediction model, try to achieve the static secondary plate shape component A in the desirable rolling situation ₂(t) and static plate shape component A ₁(t), compare with corresponding target flatness component respectively, as initial control foundation;

Secondary plate shape static prediction model is:

$A_2\left(t\right)=F_I\left(t\right)\&times;K_{F_{I}2}+F_W\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; W"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_W}+I_{\mathrm{MR}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; I\; MR"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_{\mathrm{MR}}}+P\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_P+T_5\left(t\right)\&times;{\mathrm{<figure-callout\; id="5"\; label="K\; T"\; filenames="DEST\_PATH\_120606113620.png,HDA0000147268170000032.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}+T_{45}\left(t\right)\&times;{\mathrm{<figure-callout\; id="45"\; label="K\; T"\; filenames="HDA0000147268170000032.png,HDA0000147268170000041.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}$ In the formula:

A ₂(t)---be t static secondary plate shape component constantly;

F _I(t)---be t moment intermediate calender rolls bending roller force mean value,

Namely equal WS side (active side) and DS side (transmission side) intermediate calender rolls bending roller force

Half of sum, unit: ton (10KN);

F _W(t)---be t moment work roll bending power mean value,

Namely equal WS side and DS side work roll bending power Half of sum, unit: ton (10KN);

I _MR(t)---be t upper and lower intermediate calender rolls string roller amount mean value of the moment,

It is upper and lower intermediate calender rolls string roller amount

Half of sum, unit: millimeter (mm);

P (t)---be t moment roll-force, unit: ton (10KN);

T ₅(t)---be constantly 5 frames outlet of t tension force, unit: ton (10KN);

T ₄₅(t)---be the t moment 4～5 interstand tensions, unit: ton (10KN);

K _P2,

--be respectively intermediate calender rolls bending roller force, work roll bending power, intermediate roll shifting amount, roll-force, 5 frames outlet tension force, 4～5 interstand tensions to the influence coefficient of above-mentioned secondary plate shape.(some groups of real data x values of rolling same size band steel gained and A under the identical rolling condition before obtaining ₂Adopt young waiter in a wineshop or an inn's method to take advantage of match to obtain);

Consider the roller declination amount, work roll bending power is poor, the intermediate calender rolls bending roller force is poor and intermediate calender rolls string roller amount is poor to a plate shape (asymmetric plate shape) component A ₁(t), setting up a plate shape static prediction model is:

$A_1\left(t\right)=F_{\mathrm{ID}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; ID"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{ID}}}+F_{\mathrm{WD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; WD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{WD}}}+I_{\mathrm{MRD}}\left(t\right)\&times;K_{I_{\mathrm{MRD}}1}+I_P\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; I\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_P}$

In the formula:

A ₁(t)---be a t plate shape of static state component constantly;

F _ID(t)---the intermediate calender rolls bending roller force is poor constantly for t, unit: ton (10KN);

F _WD(t)---work roll bending power is poor constantly for t, unit: ton (10KN);

I _MRD(t)---the intermediate calender rolls shifting amount is poor up and down constantly for t, unit: millimeter (mm);

I _P(t)---be t moment roller declination amount, unit: micron (um);

---be respectively that the intermediate calender rolls bending roller force is poor, work roll bending power is poor, the intermediate calender rolls shifting amount is poor, the roller declination amount is to the influence coefficient of an above-mentioned plate shape (some groups of real data x of rolling same size band steel gained and A under the identical rolling condition before obtaining ₁Adopt young waiter in a wineshop or an inn's method to take advantage of match to obtain);

The 3rd step, next moment prediction of plate shape value precision that obtains owing to plate shape static prediction model can not satisfy actual requirement, are incorporated herein previous moment plate shape component, make up plate shape dynamic forecasting model.After introducing the plate shape component of previous moment, subtract each other with the plate shape component of current time, quite and the increment of lead-in plate shape component.According to the dynamic forecasting model, try to achieve the secondary plate shape component A of the dynamic prediction model prediction in the actual rolling situation ₂(t) and a plate shape component A ₁(t), and compare with corresponding actual measurement plate shape component respectively, as the foundation of dynamic prediction model parameter correction;

Secondary plate shape dynamic forecasting model is:

$A_2\left(t\right)=A_2(t-1)\&times;{\mathrm{<figure-callout\; id="2"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_2+{\mathrm{\&Delta;F}}_I\left(t\right)\&times;K_{F_{I}2}+{\mathrm{\&Delta;F}}_W\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F\; W"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_W}+{\mathrm{\&Delta;I}}_{\mathrm{MR}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; I\; MR"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_{\mathrm{MR}}}+\mathrm{\&Delta;P}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_P+$

${\mathrm{\&Delta;T}}_5\left(t\right)\&times;{\mathrm{<figure-callout\; id="5"\; label="K\; T"\; filenames="DEST\_PATH\_120606113620.png,HDA0000147268170000032.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}+{\mathrm{\&Delta;T}}_{45}\left(t\right)\&times;{\mathrm{<figure-callout\; id="45"\; label="K\; T"\; filenames="HDA0000147268170000032.png,HDA0000147268170000041.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}$

In the formula,

A ₂(t)---be the dynamic secondary plate shape component when previous moment;

A ₂(t-1)---be the dynamic secondary plate shape component of previous moment (being the t-1 moment);

Δ F _I(t)---be t moment intermediate calender rolls bending roller force increment;

Δ F _W(t)---be t moment work roll bending power increment;

Δ P (t)---be t moment roll-force increment;

Δ T ₅(t)---be constantly 5 frames outlet of t tension increment;

Δ T ₄₅(t)---be the tension increment between the t moment 4～5 frames;

One time plate shape dynamic forecasting model is:

$A_1\left(t\right)=A_1(t-1)\&times;{\mathrm{<figure-callout\; id="1"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_1+{\mathrm{\&Delta;F}}_{\mathrm{ID}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; ID"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{ID}}}+{\mathrm{\&Delta;F}}_{\mathrm{WD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; WD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{WD}}}+\mathrm{\&Delta;}{I}_P\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; I\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_P}$

In the formula,

A ₁(t-1)---be the dynamic plate shape component of previous moment (being the t-1 moment);

Δ F _ID(t)---be the poor increment of t moment intermediate calender rolls bending roller force, Δ F _ID(t)=F _ID(t)-F _ID(t-1);

Δ F _WD(t)---be the poor increment of t moment work roll bending power, Δ F _WD(t)=F _WD(t)-F _WD(t-1);

Δ I _p(t)---be t moment roller declination amount increment, Δ I _p(t)=I _p(t)-I _p(t-1);

The 4th step, with static secondary plate shape component and a static plate shape component as the control initial value after, compare with corresponding desired value with dynamic secondary plate shape component and a dynamic plate shape component respectively, and control accordingly parameter according to comparative result output.

Because the increment of work roll bending power and the action effect of the increment of intermediate calender rolls bending roller force all are consistent with variation tendency, these two amounts can be merged into a controlled quentity controlled variable increment: Δ F (t)=α Δ F _I(t)+(1-α) Δ F _W(t), wherein α=0～1 gets 0.5 in the current procedure, and this can adjust according to on-the-spot plate shape situation.Simultaneously, because in the actual operation of rolling, the intermediate roll shifting amount is basically constant generally speaking, so secondary plate shape dynamic forecasting model can be reduced to:

$A_2\left(t\right)=A_2(t-1)\&times;{\mathrm{<figure-callout\; id="2"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_2+\mathrm{\&Delta;F}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; F"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_F+\mathrm{\&Delta;P}\left(t\right)\&times;{\mathrm{<figure-callout\; id="2"\; label="K\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_P+\mathrm{\&Delta;}{T}_5\left(t\right)\&times;{\mathrm{<figure-callout\; id="5"\; label="K\; T"\; filenames="DEST\_PATH\_120606113620.png,HDA0000147268170000032.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}+{\mathrm{\&Delta;T}}_{45}\left(t\right)\&times;{\mathrm{<figure-callout\; id="45"\; label="K\; T"\; filenames="HDA0000147268170000032.png,HDA0000147268170000041.png"\; state="\{\{state\}\}">K</figure-callout>}}_{T_{}}.$

In like manner, its plate shape dynamic forecasting model is:

$A_1\left(t\right)=A_1(t-1)\&times;{\mathrm{<figure-callout\; id="1"\; label="K"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_1+{\mathrm{\&Delta;F}}_{\mathrm{ID}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; ID"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{ID}}}+{\mathrm{\&Delta;F}}_{\mathrm{WD}}\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; F\; WD"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{F_{\mathrm{WD}}}+{\mathrm{\&Delta;I}}_P\left(t\right)\&times;{\mathrm{<figure-callout\; id="1"\; label="K\; I\; P"\; filenames="DEST\_PATH\_120606113620.png"\; state="\{\{state\}\}">K</figure-callout>}}_{I_P}$

The 4th step, with static secondary plate shape component and a static plate shape component as the control initial value after, compare with corresponding desired value with dynamic secondary plate shape component and a dynamic plate shape component respectively, and control accordingly parameter according to comparative result output.

After this, the above control procedure that circulates, thus the control of plate shape constantly is optimized.

The parameter on above-mentioned each formula equal sign the right of present embodiment can obtain by prior art means such as actual measurement, data processing respectively, so after bringing formula into, can obtain required calculated value.

Utilize the actual production data of a certain coiled strip steel, adopt the method for present embodiment to carry out simulation calculation.Under former and later two speed constantly differ condition among the 10m/min, Fig. 3 and Fig. 4 provided respectively asymmetric plate shape component deviation and the model prediction value of symmetrical plate shape symmetrical components deviation and the relative error of actual value, and Fig. 5 has provided the change trend curve of the work roll bending power increment that rolling optimization as a result obtains.Can find out from Fig. 3 and Fig. 4, the relative error of asymmetric plate shape component and symmetrical plate shape component all illustrates that less than 10-1 (namely 10%) plate shape dynamic forecasting model can improve the prediction of plate shape precision effectively, satisfies the required precision of plat control system basically.

In a word, present embodiment returns decomposition through carrying out least square method after the plate shape with the steel transmission delay with conic section by the sheet shape measurer actual measurement, then with the mode input amount in the corresponding moment forecast model coefficient is revised, through revised model coefficient, be used for by plate shape controlled quentity controlled variable on the one hand, roll-force and tension force etc. is pre-drafting board shape more accurately, be used on the other hand dynamic optimal ground and determine plate shape controlled quentity controlled variable, realize the Accurate Prediction of plate shape and the dynamic optimal control of plate shape, improve strip shape quality and lumber recovery with steel, improve simultaneously stability and the reliability of milling train operation.

Claims (6)

1. shape models forecast Control Algorithm, mainly by the roll control computer of control five stand mill and be arranged in the control system that the plate profile instrument of final milling train output consists of, the signal output part of described plate profile instrument connects the corresponding ports of roll control computer, and the control output end of described roll control computer is connected to respectively the controlled end of each roll; When the roll control COMPUTER DETECTION is lower than predetermined value to strip speed, carry out according to the following steps a basic Recurrence Process control:

The first step, one group of standard actual measurement board form data of plate profile instrument collection is carried out data process, obtain secondary plate shape component and a plate shape component in the actual plate shape curve, as the desired value of Model Parameter Optimization;

Second step, according to the static prediction model, try to achieve the static secondary plate shape component A in the desirable rolling situation ₂(t) and static plate shape component A ₁(t);

Secondary plate shape static prediction model is:

$A_2\left(t\right)=F_I\left(t\right)\&times;K_{F_{I}2}+F_W\left(t\right)\&times;K_{F_{W}2}+I_{\mathrm{MR}}\left(t\right)\&times;K_{I_{\mathrm{MR}}2}+P\left(t\right)\&times;K_{P2}+T_5\left(t\right)\&times;K_{T_{5}2}+T_{45}\left(t\right)\&times;K_{T_{45}2}$

In the formula:

A ₂(t)---be t static secondary plate shape component constantly;

F _I(t)---be t moment intermediate calender rolls bending roller force mean value, unit: ton;

F _W(t)---be t moment work roll bending power mean value, unit: ton;

I _MR(t)---be t upper and lower intermediate calender rolls string roller amount mean value of the moment, unit: millimeter;

P (t)---be t moment roll-force, unit: ton;

T ₅(t)---be constantly 5 frames outlet of t tension force, unit: ton;

T ₄₅(t)---be the t moment 4～5 interstand tensions, unit: ton;

K _P2,

--be respectively intermediate calender rolls bending roller force, work roll bending power, intermediate roll shifting amount, roll-force, 5 frames outlet tension force, 4～5 interstand tensions to the influence coefficient of static secondary plate shape component; One time plate shape static prediction model is:

$A_1\left(t\right)=F_{\mathrm{ID}}\left(t\right)\&times;K_{F_{\mathrm{ID}}1}+F_{\mathrm{WD}}\left(t\right)\&times;K_{F_{\mathrm{WD}}1}+I_{\mathrm{MRD}}\left(t\right)\&times;K_{I_{\mathrm{MRD}}1}+I_P\left(t\right)\&times;K_{I_{P}1}$

In the formula:

A ₁(t)---be a t plate shape of static state component constantly;

F _ID(t)---the intermediate calender rolls bending roller force is poor constantly for t, unit: ton

F _WD(t)---work roll bending power is poor constantly for t, unit: ton;

I _MRD(t)---the intermediate calender rolls shifting amount is poor up and down constantly for t, unit: millimeter;

I _P(t)---be t moment roller declination amount, unit: micron;

---be respectively that the intermediate calender rolls bending roller force is poor, work roll bending power is poor, the intermediate calender rolls shifting amount is poor, the roller declination amount is to the influence coefficient of a static plate shape component;

The 3rd step, according to the dynamic forecasting model, try to achieve the secondary plate shape component A of the dynamic prediction model prediction in the actual rolling situation ₂(t) and a plate shape component A ₁(t);

Secondary plate shape dynamic forecasting model is:

$A_2\left(t\right)=A_2(t-1)\&times;K_2+{\mathrm{\&Delta;F}}_I\left(t\right)\&times;K_{F_{I}2}+{\mathrm{\&Delta;F}}_W\left(t\right)\&times;K_{F_{W}2}+{\mathrm{\&Delta;I}}_{\mathrm{MR}}\left(t\right)\&times;K_{I_{\mathrm{MR}}2}+\mathrm{\&Delta;P}\left(t\right)\&times;K_{P2}+$

${\mathrm{\&Delta;T}}_5\left(t\right)\&times;K_{T_{5}2}+{\mathrm{\&Delta;T}}_{45}\left(t\right)\&times;K_{T_{45}2}$

In the formula,

A ₂(t)---be the dynamic secondary plate shape component when previous moment;

A ₂(t-1)---be the dynamic secondary plate shape component of previous moment (being the t-1 moment);

Δ F _I(t)---be t moment intermediate calender rolls bending roller force increment, unit: ton;

Δ F _W(t)---be t moment work roll bending power increment, unit: ton;

Δ P (t)---be t moment roll-force increment, unit: ton;

Δ T ₅(t)---be constantly 5 frames outlet of t tension increment, unit: ton;

Δ T ₄₅(t)---be the tension increment between the t moment 4～5 frames, unit: ton; One time plate shape dynamic forecasting model is:

$A_1\left(t\right)=A_1(t-1)\&times;K_1+{\mathrm{\&Delta;F}}_{\mathrm{ID}}\left(t\right)\&times;K_{F_{\mathrm{ID}}1}+{\mathrm{\&Delta;F}}_{\mathrm{WD}}\left(t\right)\&times;K_{F_{\mathrm{WD}}1}+\mathrm{\&Delta;}{I}_P\left(t\right)\&times;K_{I_{P}1}$

In the formula,

A ₁(t-1)---be the dynamic plate shape component of previous moment (being the t-1 moment);

Δ F _ID(t)---be the poor increment of t moment intermediate calender rolls bending roller force, Δ F _ID(t)=F _ID(t)-F _ID(t-1);

Δ F _WD(t)---be the poor increment of t moment work roll bending power, Δ F _WD(t)=F _WD(t)-F _WD(t-1);

Δ I _p(t)---be t moment roller declination amount increment, Δ I _p(t)=I _p(t)-I _p(t-1);

The 4th step, with static secondary plate shape component and a static plate shape component respectively with the comparative result of corresponding described desired value after each roll output initial control signal, compare with corresponding described desired value with dynamic secondary plate shape component and a dynamic plate shape component respectively, and export dynamic control signal according to comparative result to each roll.

2. shape models forecast Control Algorithm according to claim 1, it is characterized in that: described predetermined value is the plate shape closed-loop control cycle.

3. shape models forecast Control Algorithm according to claim 2 is characterized in that: the data described in the described second step in the first step are treated to and adopt least square method to return with conic section.

4. shape models forecast Control Algorithm according to claim 3 is characterized in that: the intermediate calender rolls bending roller force in the described second step, work roll bending power, intermediate roll shifting amount, roll-force, 5 frames outlet tension force, 4～5 interstand tensions are to the influence coefficient of static secondary plate shape component some groups of real data x values and the A by rolling same size band steel gained under the identical rolling condition before obtaining ₂, adopt young waiter in a wineshop or an inn's method to take advantage of match to obtain;

5. shape models forecast Control Algorithm according to claim 4 is characterized in that: the intermediate calender rolls bending roller force in the described second step is poor, work roll bending power is poor, the intermediate calender rolls shifting amount is poor, the roller declination amount is to the influence coefficient of a static plate shape component some groups of real data x and the A by rolling same size band steel gained under the identical rolling condition before obtaining ₁Adopt young waiter in a wineshop or an inn's method to take advantage of match to obtain.

6. shape models forecast Control Algorithm according to claim 5, it is characterized in that: described secondary plate shape dynamic forecasting model and a plate shape dynamic forecasting model are reduced to respectively:

$A_2\left(t\right)=A_2(t-1)\&times;K_2+\mathrm{\&Delta;F}\left(t\right)\&times;K_{F2}+\mathrm{\&Delta;P}\left(t\right)\&times;K_{P2}+\mathrm{\&Delta;}{T}_5\left(t\right)\&times;K_{T_{5}2}+{\mathrm{\&Delta;T}}_{45}\left(t\right)\&times;K_{T_{45}2}$

$A_1\left(t\right)=A_1(t-1)\&times;K_1+{\mathrm{\&Delta;F}}_{\mathrm{ID}}\left(t\right)\&times;K_{F_{\mathrm{ID}}1}+{\mathrm{\&Delta;F}}_{\mathrm{WD}}\left(t\right)\&times;K_{F_{\mathrm{WD}}1}+\mathrm{\&Delta;}{I}_P\left(t\right)\&times;K_{I_{P}1}.$

Images