CN104741388B - A kind of Rolling Thickness control method - Google Patents

Abstract

The present invention provides a kind of Rolling Thickness control method, including obtaining milling equipment parameter and strip steel specification parameter；End machine frame rolling mill is carried out unit-step response test, determines the unit-step response cycle i.e. hydraulic cylinder transmission time parameter of function, the control cycle of Monitor Process system and the number of unit-step response late samples discrete point；End machine frame rolling mill is controlled by the Smith Prediction Control strategy using the pi controller of band inertial element；Utilize hot continuous rolling finish rolling Monitor Process system control model, carry out next periodic thickness control by regulation hydraulic cylinder.The control process of Monitor Process is equal to a control object with purely retarded by the present invention, Smith predictive compensation is introduced Monitor Process control system, directly the roll gap of milling train to be carried out hard measurement by GM method, avoid owing to HGC transmission function is forbidden issuable calculating error, significantly improve the response speed of control system, stability and control accuracy.

Description

A kind of Rolling Thickness control method

Technical field

The invention belongs to operation of rolling automatic control technology field, be specifically related to a kind of Rolling Thickness control method.

Background technology

In plate strip rolling process, a kind of the most frequently used method for controlling thickness is by the rack outlet calibrator actual thickness to strip steel Degree measures, and and then by the Hydraulic Roll Gap of regulation milling train, thickness is carried out feedback control.Due to rolling mill structure restriction, The maintenance of calibrator and in order to prevent strip steel broken belt from damaging calibrator, calibrator is typically mounted at from directly producing thickness change Roll gap place farther out, such as, the outlet calibrator of continuous hot-rolling mill be just arranged on from working roll distance between center line about 2000～ Between 4000mm, as shown in Figure 1.

The amounts of thickness variation detected due to calibrator and the gap preset amount producing thickness change are not to send out within the same time Raw, so the actual fluctuation shutting out thickness can not be reflected timely, result make thickness MN-AGC system have one delayed Time τ, represents by (1) formula:

$\mathrm{\τ}=\frac{L}{v}---\left(1\right)$

In formula, τ-lag time；

V-mill speed；

L-roller central line is to the distance of calibrator.

It is typically based on milling train outlet calibrator and measures the feedback control of thickness, referred to as MN-AGC (Monitioring Automatic Gauge Control).Knowable to control theory, the existence of object pure delay time τ is totally unfavorable to control system 's.It makes the stability of control system reduce, and particularly weighs the purely retarded characterisitic parameter to systematic influence degreeRight As (T is the time constant of object here), it is to be difficult to obtain good control quality according to regulatory PID control.

In prior art, in terms of lag control system, although domestic made many research work, but in board rolling Actual control effect is the most unsatisfactory.Control strategy about MN-AGC has a variety of, but, existing these control plan Slightly being difficult to the rapidity of the dynamically regulation of system of taking into account and static high accuracy, and regulator parameter selection is improper, system is easily produced Raw vibration.Therefore, find a kind of rapidity having dynamically regulation concurrently and static state high-precision MN-AGC strategy, and provide one Distinct optimum control rate, to substitute the Traditional control strategy that current board rolling uses, thus improves the thickness control of strip product Precision processed, is a technical problem urgently to be resolved hurrily.

Summary of the invention

The deficiency existed for prior art, the present invention provides a kind of Rolling Thickness control method.

The technical scheme is that

A kind of Rolling Thickness control method, comprises the following steps:

Step 1: obtain milling equipment parameter and strip steel specification parameter；

Milling equipment parameter includes: mill speed, X-ray thickness gauge from end the distance of frame central line, calibrator response time Between, working roll actual roller footpath, working roll original roller diameter, support roller actual roller footpath, support roller original roller diameter, work roll crown, Support roll crown；

Strip steel specification parameter includes: rolling steel grade, supplied materials thickness, finished product thickness；

Step 2: end machine frame rolling mill is carried out unit-step response test, determines that unit-step response cycle i.e. hydraulic cylinder transmits letter The time parameter of number, the control cycle of Monitor Process system and the number of unit-step response late samples discrete point；

Step 3: use the Smith Prediction Control strategy of the pi controller of band inertial element that end machine frame rolling mill is controlled System；

Step 4: utilize hot continuous rolling finish rolling Monitor Process system control model, carries out next periodic thickness control by regulation hydraulic cylinder System；

Described hot continuous rolling finish rolling Monitor Process system control model determines the adjustment amount of roll gap in next cycle for using GM mode.

Described hot continuous rolling finish rolling Monitor Process system control model obtains as follows:

Step 4.1:X gamma thickness gage carries out multi-point sampling to the thickness deviation measured value in each control cycle, and determines each The mean thickness variation of sampling instant strip sample；

Step 4.2: the fore side pressure actual value in each control cycle is carried out by pressure transducer with transmission side pressure actual value Multi-point sampling, and determine the average pressure value of each sampling instant strip sample；

Step 4.3: the thickness deviation that calculating mill spring causes bends, with roller system, the thickness deviation caused, and receives process computer Oil membrane thickness compensation amount, thermal expansion of rollers compensation dosage and the roll wear compensation dosage issued, after null offset correction, uses GM Mode obtains the hard measurement thickness deviation of strip；

Step 4.4: use inertia scaling factor filtering method to be filtered processing the hard measurement thickness deviation of strip, use GM mode Hard measurement thickness deviation according to strip determines the adjustment amount of roll gap in next cycle, obtains Rolling Thickness Controlling model i.e. heat Tandem rolling finish rolling Monitor Process system control model.

/ 10th of the lag time that lower limit is Monitor Process system controlling the cycle of Monitor Process system described in step 2, The upper limit is 1/2nd of the lag time of Monitor Process system.

Described employing GM mode obtains the hard measurement thickness deviation of strip, specific as follows:

Δh_g=h^*-(S_act-S_zero+Δh_std+Δh_roll-Δh_ocm-Δh_tcm+Δh_dec)+Δh₀

In formula: Δ h_gThe hard measurement thickness deviation of-strip, mm；

h^*The thickness setting value of-strip, mm；

S_actThe actual gap values between rollers of-milling train, mm；

S_zero-roll gap zero point, mm；

Δh_stdThe thickness deviation that-mill spring causes, mm；

Δh_roll-roller system of rolling mill bends the thickness deviation caused, mm；

Δh_ocm-oil membrane thickness compensation amount, is given by process computer, mm；

Δh_tcm-thermal expansion of rollers compensation dosage, is given by process computer, mm；

Δh_dec-roll wear compensation dosage, is given by process computer, mm；

Δh₀-null offset correction, mm.

Beneficial effect:

The control process of Monitor Process is equal to a control object with purely retarded by the present invention, is drawn by Smith predictive compensation Enter Monitor Process control system, directly the roll gap of milling train to be carried out hard measurement by GM method, avoid owing to HGC passes Delivery function is forbidden issuable calculating error, improves the precision of prediction of Smith predictor；Proposing should to process variable The means of filtering taked, gives simple and practical filtering method and derives brand-new Monitor Process control plan based on timing sampling Slightly.Compared with the Monitor Process control strategy that an original segment length is tracked as benchmark, the Monitor Process control strategy of the present invention Significantly improve the response speed of control system, stability and control accuracy.The present invention can also promote the use of similar with In the control object of purely retarded.

Accompanying drawing explanation

Fig. 1 is the outlet calibrator installation site schematic diagram of continuous hot-rolling mill in prior art；

Fig. 2 is the Monitor Process structured flowchart of the specific embodiment of the invention；

Fig. 3 is the Monitor Process system construction drawing with Smith predictive compensation in the specific embodiment of the invention；

Fig. 4 is the based on GM mode with the fixed sample time for controlling the Monitor Process of sample of the specific embodiment of the invention Control block diagram；

Fig. 5 is that in the specific embodiment of the invention, sample time is T_sTime Monitor Process control system block diagram；

Fig. 6 is end frame spring curve in the specific embodiment of the invention；

Fig. 7 is end frame roll gap step response test curve in the specific embodiment of the invention；

Fig. 8 is the strip steel ideal format thickness deviation curve using Novel monitoring AGC to control in the specific embodiment of the invention；

Fig. 9 is a kind of Rolling Thickness control method flow chart of the specific embodiment of the invention；

Figure 10 be the specific embodiment of the invention set up hot continuous rolling finish rolling Monitor Process system control model flow chart.

Detailed description of the invention

Below in conjunction with the accompanying drawings the detailed description of the invention of the present invention is elaborated.

Implement the Rolling Thickness control method of the present invention for the strip that steel grade is Q235B, present embodiment provides one Monitor Process based on GM mode (Automatic Gauge Control) control strategy, with a length of control cycle regular time, Set up Monitor Process based on Smith predictive compensation and control process transmission function, and draw that there is typical second-order engineering best features The structure and parameter of Monitor Process controller.Directly milling train to be had load roll gap to enter by GM (Gauge Meter) method The method of row hard measurement, substitutes the traditional control method that current MN-AGC controls to use, thus effectively improves the thickness of strip steel Degree control accuracy.

A kind of Rolling Thickness control method, as it is shown in figure 9, comprise the following steps:

Step 1: obtain milling equipment parameter and strip steel specification parameter；

Milling equipment parameter includes: mill speed is 6.0m/s, X-ray thickness gauge from end frame central line distance L=3.77m, The response time T of calibrator_x=0.05s, working roll actual roller footpath D_wr=450mm, working roll original roller diameter D_wr0=455mm, Support roller actual roller footpath D_br=1010mm, support roller original roller diameter D_br0=1018mm, work roll crown C_wr=0.15mm, support Roll crown C_br=-0.07mm；

Strip steel specification parameter includes: rolling steel grade is Q235B, and supplied materials thickness is 6.0mm, and finished product thickness is 3.5mm；

Step 2: end machine frame rolling mill is carried out unit-step response test, as it is shown in fig. 7, determine that the unit-step response cycle is i.e. Hydraulic cylinder transmission the time parameter T of function, the control cycle T of Monitor Process system_sAnd unit-step response late samples is discrete Number d of point；

Step 2.1: determine unit-step response cycle T i.e. hydraulic cylinder transmission function time parameter:

In the transmission function of Position of Hydraulic Cylinder closed loop, T approves the following way of employing really, i.e. in debugging hydraulic roll gap control system During system, when the overshoot of system is 4.3%, record now unit step rise time t_r, according to typical case's second-order optimum Theory, T is shown below with the relation of unit step rise time:

$T=\frac{t_r}{4.3}---\left(2\right)$

In present embodiment, end machine frame rolling mill is carried out in unit-step response test process, when the unit step of end frame rises Between t_r×10³=23ms, then according to formulaObtain the unit-step response cycle

Step 2.2: determine the control cycle T of Monitor Process system_s:

Lag time, τ was mainly determined by the distance of calibrator to end frame, and its lag time is as shown in formula (1).Except this In addition, strictly speaking, also the response time of thickness measuring and Position of Hydraulic Cylinder closed loop produce delayed.Stagnant due to hydraulic cylinder part The rear time compare X-ray thickness gauge for the least, be negligible.But the response time of X-ray thickness gauge generally exists Between 0.05～0.1s, it is impossible to ignore.Therefore, lag timeWherein T_xFor thickness measuring The response time of instrument, s；

When mm finishing mill unit uses increasing speed rolling, mill speedWherein v_refFor mm finishing mill unit end frame threading speed Degree, m/s, v_maxFor mm finishing mill unit increasing speed rolling speed, m/s；Present embodiment is v=10.0m/s rolling under constant speed.

By the formula of lag timeUnderstanding, lag time and mill speed v are linear positive correlation.Control strategy is with former The control thinking difference being realized Monitor Process by strip steel segmentation proposed, control strategy uses timing sampling control mode, Thus, in this case, the most correctly readily obtain Δ h_g(i-d) most important.

Due to Monitor Process system only undertake elimination low frequency tendency thickness deviation, thus the control cycle there is no practical significance the soonest, Simulative and actual control all shows to control cycle T_sMeet following condition:

$\frac{\mathrm{\τ}}{10}\≤T_s\≤\frac{\mathrm{\τ}}{2}---\left(3\right)$

Certain 1700mm hot continuous rolling outlet X-ray thickness gauge is 3.77m from the distance of end frame central line, the rolling speed of end frame Degree is 6～19m/s, then be apparent from pure delay time in the range of 0.198～0.62s, add the response time of X-ray thickness gauge 0.05s, typical lag time minimum 200ms to be exceeded.Thus, a typical hot continuous rolling system, control cycle T_s=20ms Its response speed will be satisfactory.

Step 2.3: determine the number of unit-step response late samples discrete point:

$d=\mathrm{int}\frac{\mathrm{\τ}}{T_s}=\mathrm{int}\frac{\frac{L}{v}+T_x}{T_s}---\left(4\right)$

Wherein, T_xFor the response time of X-ray thickness gauge, s, v are mill speed, and L is that X-ray thickness gauge is from end frame central line Distance；Present embodiment

Based on GM mode during the Monitor Process control that the fixed sample time is control sample, by continuous for the series moment N sampled point Δ h_g(i)、Δh_g(i-1)、...Δh_g(i-n+1) storage that computer is opened up it is stored in exclusively for Monitor Process In device, and often sampling once, pointwise moves down and is updated the most in a serial fashion, by a upper Δ h_g(i-n+1) depositor is released, By this way, depositor is made to be always maintained at n up-to-date Δ h_gThe sampled value of (i), as shown in Figure 4, the d in formula (2) Value is it can be understood as find Δ h_g(i-d) pointer, d value determines Δ h in Smith predictor controller_g(i-d) in memorizer Particular location.

Step 3: the Smith Prediction Control strategy of the pi controller of design band inertial element combines and carries out end machine frame rolling mill Control；

The determination of step 3.1:MN-AGC thickness control system transmission function:

The control block diagram of MN-AGC is as in figure 2 it is shown, G in figure_cS () represents the transmission function of caliper profiler, G_p(s)e^-τsTable Show the transmission function of THICKNESS CONTROL object, the i.e. position-force control of hydraulic cylinder to the transmission function between X-ray thickness gauge, its Middle G_pS () is the transmission function that object does not comprise purely retarded part, e^-τsTransmission function for object purely retarded part.

In fig. 2, input signal Δ h^*T () (Laplace transformation is Δ H^*(s)) for setting thickness；(Laplace transformation is Δ S (t) Δ S (s)) it is the added value of hydraulic cylinder setting position；Δ h (t) (Laplace transformation is Δ H (s)) is that X-ray thickness gauge records Strip steel actual (real) thickness, its closed loop transfer function is:

$G_S\left(s\right)=\frac{\mathrm{\Δ}H\left(s\right)}{{\mathrm{\ΔH}}^{*}\left(s\right)}=\frac{G_c\left(s\right)G_p\left(s\right)e^{-\mathrm{\τ}s}}{1+G_c\left(s\right)G_p\left(s\right)e^{-\mathrm{\τ}s}}---\left(5\right)$

MN-AGC thickness control system transmission function denominator includes pure lag system e^-τs, make the stability of system reduce, If τ is sufficiently large, system is unstable.For the control of this dead-time system, as far back as the beginning of the fifties, Smith Just propose generalized object link shunt compensation link, the impact of purely retarded during regulation can be eliminated.Its accuracy is the most resonable Opinion is confirmed in practice.Due to instrument in modern age and the high speed development of computer technology, make the method in Distributed Control System Extensively applied.

During Monitor Process control, the control of belt steel thickness mainly realizes exporting strip steel with the roll gap of regulation end frame THICKNESS CONTROL, i.e. in the control block diagram of Fig. 2, G_gS () mainly reflection is the dynamic of end frame hydraulic position closed loop HGC Transmission function, K is rolling mill screwdown efficiency, relevant with the plastic coefficient of mill stiffness and rolling stock.Different from Fig. 2, if This transmission function and e^-τsSeparately, and after carrying out Smith predictive compensation, then can obtain following Monitor Process and control system System structured flowchart, as shown in Figure 3.

Δ H in Fig. 3^*S () is that milling train exports strip thickness deviation setting value Δ h^*T the Laplace transformation of (), Δ H (s) is milling train outlet band The Laplace transformation of measured value Δ h (t) of steel calibrator, Δ H_sS () is the output Δ h of Smith lead compensation part_sThe Laplace transformation of (t)； ΔH_τS () is stripped deviation or the controller G of system_cThe input value Δ h of (s)_τThe Laplace transformation of (t).Δ S (s) is that mill roll-gap sets The Laplace transformation of value Δ S (t), Δ S_gS () is mill roll-gap actual value Δ S_gThe Laplace transformation of (t), Δ H_gS () is that milling train exports strip steel The band of no-delay actual (real) thickness or milling train carries gap values between rollers Δ h_gThe Laplace transformation of (t).Dead time compensation can be obtained by Fig. 3 The closed loop transfer function of MN-AGC system:

$G_{st}\left(s\right)=\frac{\mathrm{\Δ}H\left(s\right)}{{\mathrm{\ΔH}}^{*}\left(s\right)}=\frac{\frac{G_c\left(s\right)G_g\left(s\right){\mathrm{Ke}}^{-\mathrm{\τ}s}}{1+G_c\left(s\right)G_g\left(s\right)K(1-e^{-\mathrm{\τ}s})}}{1+\frac{G_c\left(s\right)G_g\left(s\right){\mathrm{Ke}}^{-\mathrm{\τ}s}}{1+G_c\left(s\right)G_g\left(s\right)K(1-e^{-\mathrm{\τ}s})}}=\[\frac{G_c\left(s\right)G_g\left(s\right)K}{1+G_c\left(s\right)G_g\left(s\right)K}\]e^{-\mathrm{\τ}s}---\left(6\right)$

The closed loop transfer function of the dead time compensation MN-AGC system be given from above formula, after dead time compensation, eliminates The impact on system of the purely retarded part, the i.e. e of closed loop transfer function^-τsOutside close loop control circuit, do not affect the steady of system Qualitative；Be may certify that by the shifting theorem of Laplace transformation, control process has only been elapsed a time τ on time coordinate by it, The shape of its transient process and other all quality index are identical when being all to there is not purely retarded part with plant characteristic.So, To any large time delay time τ, system is all stable.

The determination of step 3.2:Smith Prediction Control system controller:

During regulation Hydraulic Roll Gap Control System, it will usually making the corresponding overshoot of system unit step is 4%～5% left side The right side, i.e. substantially coincide with two grades of engineering optimal transfer function of typical case (typical two grades of engineerings optimal overshoot be 4.3%), this Assume can meet completely engine request.Thus G_gThe transmission function such as following formula of (s):

$G_g\left(s\right)=\frac{1}{2T^2{S}^2+2TS+1}---\left(7\right)$

Its proportionality coefficient sets K such as following formula, is also called the pressure efficiency of milling train, such as following formula:

$K=\frac{M}{M+Q}---\left(8\right)$

M-mill stiffness in formula, kN/mm；

The plastic coefficient of Q-strip steel, kN/mm.

In this case, setting controller G_cS the parameter of the transmission function of (), i.e. allows whole closed loop transfer function or one Typical case's second order is optimal.After so supposing, from the figure 3, it may be seen that the transmission function of the anhysteretic part after Smith predictive compensation should This meets following formula:

$\frac{G_c\left(s\right)G_g\left(s\right)K}{1+G_c\left(s\right)G_g\left(s\right)K}=\frac{1}{2T^2{S}^2+2TS+1}---\left(9\right)$

Formula (8) formula is brought in formula (9), collated obtain G_cS the transmission function of () is as follows:

$G_c\left(s\right)=\frac{2T^2{S}^2+2TS+1}{K(2T^2{S}^2+2TS)}=\frac{1}{K}(1+\frac{1}{2TS}\frac{1}{TS+1})---\left(10\right)$

By the transmission function (10) of controller Suo Shi, controller is the proportional+integral controller of a band inertia, is very easy in work Realizing in journey, it is relevant with the time parameter T of hydraulic cylinder transmission function that it controls parameter, and its proportional parts is rolling mill screwdown efficiency Reciprocal.

In present embodiment, the control process of Monitor Process system is equal to a control object with purely retarded, gives The transmission function of links, and derive structure and the ginseng of the Monitor Process controller with typical second-order engineering best features Number；Smith predictive compensation is introduced Monitor Process control system, specifies position and the physical significance of compensator input variable, And give the computational methods of system pure delay time；Use GM method to come directly has load roll gap to carry out hard measurement to milling train, Avoid owing to HGC transmission function is forbidden issuable calculating error, improve the precision of prediction of Smith predictor；Propose The means of filtering should taked process variable, gives simple and practical filtering method and derives based on timing sampling brand-new Monitor Process control strategy, and give the detailed control system block diagram how completing Project Realization.With an original segment length The Monitor Process control strategy being tracked as benchmark is compared, and the Monitor Process control strategy of present embodiment significantly improves control system Response speed, stability and the control accuracy of system.This method is become in 7 actual frame 1700mm hot continuous rolling systems Merit is applied, and achieves very satisfied control effect.It is right that the method can also promote the use of the similar control with purely retarded In as.The Monitor Process control strategy using this method to provide, randomly draws the board rolling history note of several typical thickness product As shown in Figure 8, wherein (a) is 2.5mm specification strip steel THICKNESS CONTROL effect, and (b) is 3.65mm specification strip steel thickness control in record Effect processed, (c) is 4.5mm specification strip steel THICKNESS CONTROL effect, and (d) is 5.65mm specification strip steel THICKNESS CONTROL effect.

Step 4: utilize hot continuous rolling finish rolling Monitor Process system control model, carries out next periodic thickness control by regulation hydraulic cylinder System；

Hot continuous rolling finish rolling Monitor Process system control model determines the adjustment amount of roll gap in next cycle for using GM mode.

As shown in Figure 10, hot continuous rolling finish rolling Monitor Process system control model obtains as follows:

Step 4.1:X gamma thickness gage controls cycle T to each_sInterior thickness deviation measured value Δ h_eCarry out multi-point sampling, and Determine the mean thickness variation Δ h of sampling instant i strip sample_e(i)；

Step 4.2: pressure transducer controls cycle T to each_sInterior fore side pressure actual value F_osActual with transmission side pressure Value F_dsCarry out multi-point sampling, and determine average pressure value F (i) of sampling instant i strip sample；

$F\left(i\right)=\frac{F_{os}\left(i\right)+F_{ds}\left(i\right)}{2}---\left(11\right)$

Step 4.3: utilize and calculate the thickness deviation that mill spring causes shown in mill spring characteristic curve equation such as formula (12), profit Bend, with calculating roller system roller system of rolling mill flexural property curvilinear equation such as formula (13) Suo Shi, the thickness deviation caused, and receive process meter The oil membrane thickness compensation amount Δ h that calculation machine issues_ocm, thermal expansion of rollers compensation dosage Δ h_tcmWith roll wear compensation dosage Δ h_dec, through zero point After drift correction, GM mode is used to obtain the hard measurement thickness deviation Δ h of strip_g(i)；

$f_H\left(F\right)={\mathrm{\α}}_0+{\mathrm{\α}}_1{(\frac{F}{F_s}-\frac{F_{S0}}{F_s})}^{0.5}+{\mathrm{\α}}_2{(\frac{F}{F_s}-\frac{F_{S0}}{F_s})}^{1.0}+{\mathrm{\α}}_3{(\frac{F}{F_s}-\frac{F_{S0}}{F_s})}^{1.5}+{\mathrm{\α}}_4{(\frac{F}{F_s}-\frac{F_{S0}}{F_s})}^2---\left(12\right)$

F in formula_S0-set and be pressed against the initial roll-force of experiment, kN, on-the-spot setting F at this_S0=4000kN；

F_S-it is pressed against experiment step pitch, kN, on-the-spot setting F at this_S=500kN；

Mill spring curvilinear equation factor alpha after being pressed against experimental fit₁=0.25166, α₂=-0.05828, α₃=0.04326, α₄=-0.00497；Mill spring matched curve is as shown in Figure 6.

The flexure of roller system is calculated by the off-line mechanistic model about width.Roller system of rolling mill flexural property curve such as formula (13):

f_M(B)=α_Dwr·D_wr+α_Dbr·D_br+α_Cwr·C_wr+α_Cbr·C_br+α_B·B (13) In formula: α_Dwr-working roll roller footpath affects coefficient to formed bits for mill roller；

D_wr-working roll roller footpath, mm；

α_Dbr-support roller roller footpath formed bits for mill roller affected coefficient；

D_br-support roller roller footpath, mm；

α_Cwr-work roll crown affects coefficient to formed bits for mill roller；

C_wr-work roll crown, mm；

α_Cbr-support roll crown formed bits for mill roller affected coefficient；

C_br-support roll crown, mm；

α_B-rolled piece broadband affects coefficient to formed bits for mill roller；

B-rolled piece broadband, mm；

Through the roller system of rolling mill Flexural Equation factor alpha that off-line mechanistic model calculates_Dwr=1.02, α_Dbr=1.15, α_Cwr=1.07, α_Cbr=1.32, α_B=0.9；

GM mode is used to obtain the hard measurement thickness deviation of strip, specific as follows:

Δh_g=h^*-(S_act-S_zero+Δh_std+Δh_roll-Δh_ocm-Δh_tcm+Δh_dec)+Δh₀ (14)

In formula: Δ h_gThe hard measurement thickness deviation of-strip, mm；

h^*The thickness setting value of-strip, mm；

S_actThe actual gap values between rollers of-milling train, mm；

S_zero-roll gap zero point, mm；

Δh_stdThe thickness deviation that-mill spring causes, mm；

Δh_roll-roller system of rolling mill bends the thickness deviation caused, mm；

Δh_ocm-oil membrane thickness compensation amount, is given by process computer, mm；

Δh_tcm-thermal expansion of rollers compensation dosage, is given by process computer, mm；

Δh_dec-roll wear compensation dosage, is given by process computer, mm；

Δh₀-null offset correction, mm.

Step 4.4: use inertia scaling factor filtering method to be filtered processing the hard measurement thickness of strip, utilize filtered soft survey Amount thickness, the strip actual (real) thickness in front several cycles and adjustment amount of roll gap, determine the adjustment amount of roll gap in next cycle, i.e. hot continuous rolling Finish rolling Monitor Process system control model；

Step 4.4.1: the hard measurement thickness deviation Δ h to strip_gI () uses inertia scaling factor filtering method to be filtered processing；

$\mathrm{\Δ}{\stackrel{\‾}{h}}_g\left(i\right)=\frac{3{\mathrm{\Δh}}_g\left(i\right)+2{\mathrm{\Δh}}_g(i-1)+{\mathrm{\Δh}}_g(i-2)}{6}---\left(15\right)$

Low frequency tendency thickness deviation is only adjusted, when implementing monitoring AGC controls, to X-ray ga(u)ging by Monitor Process The variablees such as the exit thickness of instrument, roll-force, mill roll-gap are desirable to suitably filter, to eliminate these variable high-frequency fluctuations pair The impact of Monitor Process.Therefore, present embodiment carries out low-pass filtering to these process variables.

For roll-force, before using roll-force, the filtered roll-force of roll eccentricities, the filtering of milling train bias should be used to have A lot of methods, are not repeating at this.

Present embodiment uses a kind of simple inertia scaling factor filtering method, is used for solving Δ h_g(i-n), Δ h has been eliminated_g The high frequency components impact on Monitor Process.Its filtering method is shown below:

$\mathrm{\Δ}{\stackrel{\‾}{h}}_g\left(i\right)=\frac{3{\mathrm{\Δh}}_g\left(i\right)+2{\mathrm{\Δh}}_g(i-1)+{\mathrm{\Δh}}_g(i-2)}{6}---\left(16\right)$

Seeking initial value Δ h_g(1), time, Δ h is made_g(-1)=Δ h_g(0)=Δ h_g(1) Δ h, is sought_g(2), time, Δ h is made_g(0)=Δ h_g(1).? In Fig. 4, by filteredReplace Δ h_g(i).The most also thickness measurements Δ h (i) of X-ray thickness gauge is also carried out Such Filtering Processing.Monitor Process after thus having obtained a weight is brand-new, process variable having carried out Filtering Processing controls plan Slightly.

Step 4.4.2: the calculating of hot continuous rolling finish rolling Monitor Process system control model；

As shown in Figure 4, it is fixed to will be given below based on strip steel for the structure of Monitor Process control strategy control system based on GM mode The optimum control rate of the inertia time lag system of duration sample.As shown in Figure 4:

${\mathrm{\Δh}}_t\left(i\right)(\frac{K_i}{2TS+1}\·\frac{K_{in}}{2TS}+K_p)\·\frac{M+Q}{M}={\mathrm{\ΔS}}_i---\left(17\right)$

Δh_t(i)=Δ h_e(i)-Δh_s(i) (18)

Δh_s(i)=Δ h_g(i)-Δh_g(i-d) (19)

Formula (18) and formula (19) are updated to formula (17) can obtain:

[Δh_e(i)+Δh_g(i-d)-Δh_g(i)]·(K_i·K_in·4T²S²+K_p2TS)=Δ S (i) K_z·(4T²S²+2TS) (20)

The sampling time T of control system_s(i)=T_s(i-1)==T_s(1)=20ms, formula (19) is timed long sample from Dispersion, and be (21) formula, (22) formula, (23) formula, (24) formula by single order and secondary differential element approximate processing:

$s\·{\mathrm{\Δh}}_e\left(i\right)\⇒\frac{{\mathrm{\Δh}}_e\left(i\right)-{\mathrm{\Δh}}_e(i-1)}{T_s\left(i\right)}---\left(21\right)$

$s\·{\mathrm{\Δh}}_g\left(i\right)\⇒\frac{{\mathrm{\Δh}}_g(i-d)-{\mathrm{\Δh}}_g(i-d-1)}{T_s\left(i\right)}---\left(22\right)$

$\begin{array}{c}{s}^2\·{\mathrm{\Δh}}_e\left(i\right)\⇒\frac{\frac{{\mathrm{\Δh}}_e\left(i\right)-{\mathrm{\Δh}}_e(i-1)}{T_s\left(i\right)}-\frac{{\mathrm{\Δh}}_e(i-1)-{\mathrm{\Δh}}_e(i-2)}{T_s(i-1)}}{T_s\left(i\right)}\\ =\frac{{\mathrm{\Δh}}_e\left(i\right)-{\mathrm{\Δh}}_e(i-1)}{T_s{\left(i\right)}^2}-\frac{{\mathrm{\Δh}}_e(i-1)-{\mathrm{\Δh}}_e(i-2)}{T_s\left(i\right)T_s(i-1)}\end{array}---\left(23\right)$

$\begin{array}{c}{s}^2\·{\mathrm{\Δh}}_g(i-d)\⇒\frac{\frac{{\mathrm{\Δh}}_g(i-d)-{\mathrm{\Δh}}_g(i-d-1)}{T_s\left(i\right)}-\frac{{\mathrm{\Δh}}_g(i-d-1)-{\mathrm{\Δh}}_g(i-d-2)}{T_s(i-1)}}{T_s\left(i\right)}\\ =\frac{{\mathrm{\Δh}}_g(i-d)-{\mathrm{\Δh}}_g(i-d-1)}{T_s{\left(i\right)}^2}-\frac{{\mathrm{\Δh}}_g(i-d-1)-{\mathrm{\Δh}}_g(i-d-2)}{T_s\left(i\right)T_s(i-1)}\end{array}---\left(24\right)$

By (21) formula, (22) formula, (23) formula, (24) formula is updated to formula (20) and arrangement has then obtained final Monitor Process feedback control The explicit control rate of system processed:

$\begin{array}{c}\mathrm{\Δ}S\left(i\right)=\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(i\right)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-1)\\ +\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-1)\\ -\frac{K_i\·K_m\·T_i^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(i\right)-\frac{2T\·K_p}{2T+T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-2)+\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-1)\\ -\frac{2T}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-2)+\frac{4T+T_s\left(i\right)}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-1)\end{array}---\left(25\right)$

From control rate (25) formula, that affect control rate is not only current actual (real) thickness deviation signal Δ h_eI (), last time are real Border thickness deviation signal delta h_eAnd the actual (real) thickness deviation signal Δ h of upper twice (i-1)_e(i-2) relevant, also the softest with current Measure thickness deviation signal delta h_g(i), last hard measurement thickness deviation signal delta h_gAnd the actual (real) thickness deviation of upper twice (i-1) Signal delta h_g(i-2), the d time hard measurement thickness deviation signal delta h_g(i-d), the d+1 time hard measurement thickness deviation signal Δh_g(i-d-1) and the hard measurement thickness deviation signal of the d+2 time is relevant, the most also part amount is adjusted with previous roll gap Δ S (i-1), adjustment amount of roll gap Δ S (i-2) of first twice are relevant.With control rate Δ S (i) in the i-th moment as adjustment amount of roll gap, really Determining Δ S (i) is:

$\begin{array}{c}\mathrm{\Δ}S\left(i\right)=\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(i\right)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-1)\\ +\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-1)\\ -\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(i\right)-\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-2)+\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-1)\\ -\frac{2T}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-2)+\frac{4T+T_s}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-1)\end{array}$

The determination step of Δ S (i) is as follows:

The first step:

$\mathrm{\Δ}S\left(1\right)=\frac{K_i\·K_m\·T_s^2+K_p\·4T^2+K_p\·2T\·T_s}{4T^2+2T\·T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(1\right)-\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(1\right)$

Second step:

$\begin{array}{c}\mathrm{\Δ}S\left(2\right)=\frac{K_i\·K_m\·T_s+K_p\·4T^2+K_p\·2T\·T_s}{4T^2+2T\·T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(2\right)-\frac{4T\·K_p\·T_s+K_p\·T_s^2}{2T+T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(1\right)\\ -\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(2\right)\\ +\frac{4T\·K_p\·T_s+K_p\·{T_s}^2}{2T+T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(1\right)+\frac{4T+T_s}{2T+T_s}\·\mathrm{\Δ}S\left(1\right)\end{array}$

I-th step (n >=2, d >=i >=3)

$\begin{array}{c}\mathrm{\Δ}S\left(i\right)=\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(i\right)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-2)\\ -\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-1)-\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(i\right)\\ -\frac{2T\·K_p}{2T+T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-2)+\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-1)\\ -\frac{2T}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-2)+\frac{4T+T_s\left(i\right)}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-1)\end{array}$

I-th step (d+2 >=i >=d+1)

$\begin{array}{c}\mathrm{\Δ}S\left(i\right)=\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(i\right)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-1)\\ +\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-1)\\ -\frac{K_i\·K_m\·T_i^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(i\right)-\frac{2T\·K_p}{2T+T_s}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-2)+\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-1)\\ -\frac{2T}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-2)+\frac{4T+T_s\left(i\right)}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-1)\end{array}$

I-th step (i >=d+2)

$\begin{array}{c}\mathrm{\Δ}S\left(i\right)=\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e\left(i\right)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_e(i-1)\\ +\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d)+\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-2)-\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-d-1)\\ -\frac{K_i\·K_m\·T_s^2\left(i\right)+K_p\·4T^2+K_p\·2T\·T_s\left(i\right)}{4T^2+2T\·T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g\left(i\right)-\frac{2T\·K_p}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-2)+\frac{4T\·K_p\·T_s\left(i\right)+K_p\·T_s^2\left(i\right)}{2T+T_s\left(i\right)}\·\frac{M}{M+Q}\·{\mathrm{\Δh}}_g(i-1)\\ -\frac{2T}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-2)+\frac{4T+T_s}{2T+T_s\left(i\right)}\·\mathrm{\Δ}S(i-1)\end{array}$

Its control block diagram is as shown in Figure 5.

Present embodiment uses TDC controller that Siemens Company produces for the control to press down system at the scene, it is possible to use CFC language has been write " Monitor Process based on GM mode " and has been controlled functional module, calculates the roll gap adjustment in each control cycle Amount, sends into adjustment amount of roll gap in pi controller, and the output valve of pi controller is defeated as servo valve amplifier Enter value, drive servo valve, control Position of Hydraulic Cylinder to move up and down thus thickness is controlled by servo valve amplifier.

Controlling to realize the long sample Monitor Process of timing based on GM mode, the configuration to hardware device to meet following wanting Ask:

1) being provided with calibrator after mm finishing mill unit, calibrator can export the voltage proportional to thickness deviation or current analog letter Number, MTS displacement transducer and pressure transducer are installed on finishing mill simultaneously, MTS displacement transducer can export and position Voltage that actual value is proportional or current analog signal, pressure transducer can export the voltage proportional to pressure actual value or electricity Flow field simulation signal；

2) have one with simulation input/output interface board, the computer system that can perform mathematical calculations or TDC, there is simulation Input and the SIEMENS TDC of output interface plate, to read the actual (real) thickness signal of calibrator output, thus realize Monitor Process The determination of closed loop control rate, store and export.

Claims (4)

1. a Rolling Thickness control method, it is characterised in that comprise the following steps:

Step 1: obtain milling equipment parameter and strip steel specification parameter；

Milling equipment parameter includes: mill speed, X-ray thickness gauge from end the distance of frame central line, calibrator response time Between, working roll actual roller footpath, working roll original roller diameter, support roller actual roller footpath, support roller original roller diameter, work roll crown, Support roll crown；

Strip steel specification parameter includes: rolling steel grade, supplied materials thickness, finished product thickness；

Step 2: end machine frame rolling mill is carried out unit-step response test, determines that unit-step response cycle i.e. hydraulic cylinder transmits letter The time parameter of number, the control cycle of Monitor Process system and the number of unit-step response late samples discrete point；

Step 3: use the Smith Prediction Control strategy of the pi controller of band inertial element that end machine frame rolling mill is controlled System；

Step 4: utilize hot continuous rolling finish rolling Monitor Process system control model, carries out next periodic thickness control by regulation hydraulic cylinder System；

Described hot continuous rolling finish rolling Monitor Process system control model determines the adjustment amount of roll gap in next cycle for using GM mode.

Rolling Thickness control method the most according to claim 1, it is characterised in that described hot continuous rolling finish rolling Monitor Process System control model obtains as follows:

Step 4.1:X gamma thickness gage carries out multi-point sampling to the thickness deviation measured value in each control cycle, and determines each The mean thickness variation of sampling instant strip sample；

Step 4.2: the fore side pressure actual value in each control cycle is carried out by pressure transducer with transmission side pressure actual value Multi-point sampling, and determine the average pressure value of each sampling instant strip sample；

Step 4.3: the thickness deviation that calculating mill spring causes bends, with roller system, the thickness deviation caused, and receives process computer Oil membrane thickness compensation amount, thermal expansion of rollers compensation dosage and the roll wear compensation dosage issued, after null offset correction, uses GM Mode obtains the hard measurement thickness deviation of strip；

Step 4.4: use inertia scaling factor filtering method to be filtered processing the hard measurement thickness deviation of strip, use GM mode Hard measurement thickness deviation according to strip determines the adjustment amount of roll gap in next cycle, obtains Rolling Thickness Controlling model i.e. heat Tandem rolling finish rolling Monitor Process system control model.

Rolling Thickness control method the most according to claim 1, it is characterised in that Monitor Process described in step 2 / 10th of the lag time that lower limit is Monitor Process system controlling the cycle of system, the upper limit is the stagnant of Monitor Process system / 2nd of rear time.

Rolling Thickness control method the most according to claim 1, it is characterised in that described employing GM mode obtains plate The hard measurement thickness deviation of band, specific as follows:

Δh_g=h^*-(S_act-S_zero+Δh_std+Δh_roll-Δh_ocm-Δh_tcm+Δh_dec)+Δh₀

In formula: Δ h_gThe hard measurement thickness deviation of-strip, mm；

h^*The thickness setting value of-strip, mm；

S_actThe actual gap values between rollers of-milling train, mm；

S_zero-roll gap zero point, mm；

Δh_stdThe thickness deviation that-mill spring causes, mm；

Δh_roll-roller system of rolling mill bends the thickness deviation caused, mm；

Δh_ocm-oil membrane thickness compensation amount, is given by process computer, mm；

Δh_tcm-thermal expansion of rollers compensation dosage, is given by process computer, mm；

Δh_dec-roll wear compensation dosage, is given by process computer, mm；

Δh₀-null offset correction, mm.