CN1270838C - Automatic controlling technical parameter optimization method of metal plate rolling - Google Patents

Abstract

The present invention belongs to the field of metal plate band rolling, which is suitable for automatically controlled process parameter optimization in the rolling process of metal (which comprises metal and alloy, such as ferrum, aluminum, copper, molybdenum, titanium, etc.) plate bands. The on-line automatic measurement of rolling technology parameters is used, a post-calculated value of deformation resistance F and a friction coefficient mu in the process of metal plate band rolling is accurately calculated by collected data in a mode of combining an automatic control system and plate band rolling technology techniques, and the automatic control system is composed of network communication transmission and computer data processing; practically measured rolling force F and a forward slip value fs are substituted to a binary high order nonlinear equation system in which deformation resistance K and the friction coefficient mu are used as unknown quantity; simultaneously, post-calculated values of the two unknown quality are solved, and adaptive learning optimization is carried out to calculation models of the deformation resistance K and the friction coefficient mu by the post-calculated values. Thus, the set precision of a rolling force F model of the metal plate band is improved, and the deviation of set values and practically measured values is controlled within +/-5%.

Description

Rolling metal plate and tape is controlled parameter optimizing method automatically

Technical field

The invention belongs to metal rolled technical field, control parameter optimizing method in particularly a kind of rolling metal plate and tape process automatically.

Background technology

The core of metal (comprising metal and alloys such as iron, aluminium, copper, molybdenum, titanium) plate strip rolling process high accuracy control is to obtain rolling force setup value accurately.Seek out rolling force setup value accurately, must at first obtain resistance of deformation and coefficient of friction calculated value accurately.Rolling metal plate and tape is the process of a multivariable, non-linear, close coupling, singly is difficult to calculate rolling technological parameter accurately from point of theory, and the control accuracy that this has influenced board rolling causes product quality defect.

For this reason, adopt the technological parameter of the instrumentation actual measurement operation of rolling both at home and abroad, the method for model adaptation being learnt by measured value improves the model computational accuracy.But resistance of deformation and coefficient of friction are to carry out continuous measurement in process of production, therefore can only carry out back calculating indirectly to its actual value by the actual measurement roll-force.Because roll-force is the function of resistance of deformation and coefficient of friction, can't obtain calculated value behind two of resistance of deformation and the coefficient of frictions simultaneously by the roll-force measured value, so can only take in the past to fix a value, calculate the method for another value, this can reduce the CALCULATION OF PARAMETERS precision.

Summary of the invention

At problems of the prior art, the present invention proposes a kind of rolling metal plate and tape process and controls process parameter optimizing method automatically, calculates actual value after can calculating the resistance of deformation of plate strip rolling process and coefficient of friction simultaneously, exactly.And Mathematical Modeling is carried out the precision that adaptive learning improves rolling control model by measured value.

The present invention adopts by the on-line automatic measurement of rolling technological parameter, and test data is handled the control system of forming by network communication transmission, data computer, combines with rolling mill practice accurately to calculate calculating actual value behind resistance of deformation and the coefficient of friction.Technical scheme of the present invention: as shown in Figure 1, the data transmission network by the strip-mill strip control system is sent to the speed of the milling train exit strip of the roll-force of power sensor actual measurement in the production process, laser velocimeter actual measurement and the roll rotational speed of driving motor encoder actual measurement in the process computer measured value database.Enter into computer processing afterwards, as shown in Figure 2, the process computer data processor is handled measured data, removes unreasonable measured value, and rational measured value is averaged.Can use as data, enter next step program, otherwise quit a program, restart.Actual measurement strip muzzle velocity v after the utilization of Mathematical Modeling programming system is handled and actual measurement speed of rolls v _RCalculate the actual advancing slip value f of plate strip rolling process _s=(v-v _R)/v _RActual advancing slip value f _s, actual measurement roll-force value F and the corresponding advancing slip theoretical calculation model formula of other parameter substitution (1).

$f_s=(\frac{{2R}^{\′}}{h_{\mathrm{out}}}\cos {\mathrm{\Φ}}_n-1)(1-\cos {\mathrm{\Φ}}_n)---\left(1\right)$

${\mathrm{\Φ}}_n=\sqrt{\frac{h_{\mathrm{out}}}{R^{\′}}}\tan \left[\frac{1}{2}{\tan }^{-1}\right\{\sqrt{\frac{h_{\mathrm{in}}-h_{\mathrm{out}}}{h_{\mathrm{out}}}\}}-\frac{1}{4\mathrm{\μ}\sqrt{\frac{R^{\′}}{h_{\mathrm{out}}}}}\mathrm{In}\left\{\frac{h_{\mathrm{in}}}{h_{\mathrm{out}}}\frac{1-\frac{t_{\mathrm{out}}}{k\·{{\mathrm{\η}}_{\mathrm{kout}}}^n\·}}{1-\frac{t_{\mathrm{in}}}{k\·{{\mathrm{\η}}_{\mathrm{kin}}}^n}}\right\}]$

In the formula: f _s-advancing slip value

h _In, h _Out-inlet, outlet belt steel thickness

t _In, t _Out-inlet, exporting unit's tension force

η k _In, η k _Out-inlet, outlet resistance of deformation mean coefficient

K ,-resistance of deformation

μ-coefficient of friction

Φ _n-neutral angle

R '-roll flattening radius

The n-model coefficient

With roll-force theoretical calculation model formula (2)

$F=Q_F(k-\mathrm{\ξ})W\sqrt{R^{\′}(h_{\mathrm{in}}-h_{\mathrm{out}})}+\frac{2}{3}\sqrt{\frac{1-v^2}{E}k\frac{h_{\mathrm{out}}}{h_{\mathrm{in}}-h_{\mathrm{out}}}}(k-\mathrm{\ξ})W\sqrt{R^{\′}(h_{\mathrm{in}}-h_{\mathrm{out}})}---\left(2\right)$

ξ＝α·t _in+β·t _out

$Q_F=1.08-1.02\·r+1.79\·r\·\mathrm{\μ}\·\sqrt{1-r}\·\sqrt{\frac{R^{\′}}{h_{\mathrm{out}}}}$

Formula: F-roll-force

F ^p, F ^e-plasticity, elastic region roll-force

F ^Ein, F ^Eout-elastic compression, recovery district roll-force

h _In, h _Out-inlet, exit thickness

The W-strip width

t _In, t _Out-inlet, exporting unit's tension force

The k-resistance of deformation

μ-coefficient of friction

R '-roll flattening radius

Q _F-roll-force external friction influence coefficient

The r-reduction ratio

The v-Poisson's ratio

The E-Young's modulus

α, β-inlet, outlet tension force influence coefficient

Formation with resistance of deformation K after behind calculated value and the coefficientoffriction calculated value be the binary high order Nonlinear System of Equations formula (3) of unknown quantity

$\left.\begin{array}{c}{f}_s(k,\mathrm{\μ})=0\\ F(k,\mathrm{\μ})=0\end{array}\right\}---\left(3\right)$

Find the solution Nonlinear System of Equations formula (3) by the Newton-Raphson method, draw the back calculated value of resistance of deformation K and coefficientoffriction.Above-mentioned solution procedure is to adopt iterative method to calculate, if equation group does not restrain, calculating just quits a program, otherwise just result of calculation is stored in the database of back calculated value.The adaptive learning program with resistance of deformation K after the calculated value theoretical model formula (4) of bringing resistance of deformation K into.

$k=\frac{2}{\sqrt{3}}{C}_{k0}{k}_0{(\frac{2}{\sqrt{3}}\mathrm{In}\frac{100}{100-\mathrm{\ξ}\·r_{\mathrm{tin}}}+{\mathrm{\ϵ}}_0)}^n---\left(4\right)$

$r_{\mathrm{tin}}=\frac{h_0-h_{\mathrm{tin}}}{h_0}\×100$

h _m＝(1-γ)·h _in+γ·h _out

In the formula

The k-resistance of deformation

h ₀-original depth

h _m-average thickness

h _In, h _Out-inlet, exit thickness

k ₀, ε ₀, n, ξ, γ-model parameter

C _K0-model adaptation learning coefficient

The calculated value of coefficientoffriction is brought into coefficientoffriction theoretical model formula (5)

$\mathrm{\μ}=C_1\·e^{(b_L\·L+C_L\·L^2+d_L\·L^3+e_L\·L^4)+b_v(v-v_0)+c_v{(v-v_0)}^2+d_v{(v-v_0)}^3+e_v{(v-v_0)}^4---\left(5\right)}$

μ-coefficient of friction in the formula

L-working roll rolled band steel length

V-band steel exports speed setting value

b _L～e _L, b _v～f _v-model parameter

v ₀-band steel exports velocity standard value

C ₁-model adaptation learning coefficient

Calculated value is carried out adaptive equalization, optimize adaptive learning coefficient C _K0, C ₁Improve the computational accuracy of model.

The present invention can be by actual measurement rolling force F and actual advancing slip value f, calculate actual value after calculating resistance of deformation K and coefficientoffriction simultaneously, solved and only used the actual measurement rolling force F to carry out the coupled problem in the computational process behind resistance of deformation K and the coefficientoffriction, can improve the adaptive learning of resistance of deformation K model and coefficientoffriction model effectively and optimize effect, with its result of calculation substitution rolling force F modular form (2), rolling force F setting value can be controlled in the measured value deviation ± 5% in.

Description of drawings

Fig. 1 gathers schematic diagram for the rolling metal plate and tape measured data,

Fig. 2 is a computational process flow chart behind resistance of deformation K and the coefficientoffriction,

Fig. 3 is embodiment industrial experiment 75 volume cold-strip steel rolling force F calculated value and measured value comparison diagrams.

The specific embodiment

The present invention is Success in Experiment on cold-strip steel 1220mm cold-rolling mill.Thickness specification: 1.5～3.5m; Finished product thickness: 0.2～0.8m; Width: 550～1020m; Steel grade: SPHC, MRT, Stw22; Work roll diameter: 550mm; Backing roll diameter: 1194mm; Main motor is for handing over the orthogonal vector converter, power 3800kW, the maximum measurement category 1100r/min of digital revolution speed measuring encoder.Pressure head (rolling force sensor) adopts piezomagnetic, maximum measurement category 10MN.Laser speedometer scope 0.6～2520mpm, certainty of measurement ± 0.025%.The process optimization computer adopts the P4-1.8GHz micro computer, and program is write by visual c++.

With belt steel rolling steel grade SPHC, inlet thickness 2.3mm, exit thickness 1.67mm, the wide 1020mm of strip is an example.Recording motor speed by motor encoder is v _R=769m/min, laser velocimeter record strip muzzle velocity v=787m/min, calculate actual advancing slip value f by model program _s=2.34%.By pressure head record and by the actual measurement handling procedure get roll-force measured value F=8637.4KN.With the advancing slip value of reality f _sIn its theoretical model of actual measurement rolling force F value difference substitution, formation is the Nonlinear System of Equations formula (3) of unknown quantity with resistance of deformation K and coefficientoffriction, find the solution this Nonlinear System of Equations by the Newton-Raphson method and get that calculated value is 538.6Mpa behind the resistance of deformation K, calculated value is 0.107 behind the coefficientoffriction.Above-mentioned back calculated value is carried out after adaptive learning optimizes resistance of deformation K model and coefficientoffriction modular form (4), formula (5), resistance of deformation K and coefficientoffriction model calculated value substitution rolling force F modular form (2) are obtained calculated value 8541.2KN, the actual measurement roll-force 8805.7KN of next volume same specification band steel, both deviations are-3.0%.

Claims (2)

1, a kind of rolling metal plate and tape is controlled parameter optimizing method automatically, it is characterized in that comprising following four steps:

(1) automatic on-line actual measurement rolling metal plate and tape process parameter comprises: speed of rolls v _R, metal plate and belt muzzle velocity v, rolling force F and calculate resistance of deformation and the required parameter of coefficient of friction;

(2) data that record are handled, calculated the actual advancing slip value f of plate strip rolling process according to following formula _s=(v-v _R)/v _RIn the formula: f _sActual advancing slip value for plate strip rolling process

V is actual measurement strip muzzle velocity

v _RBe the actual measurement speed of rolls;

(3) by actual measurement rolling force F and actual advancing slip value f _sCalculate the back calculated value of resistance of deformation and coefficient of friction, adopt the Newton-Raphson method to find the solution following nonlinear way formula:

$\left.\begin{array}{c}{f}_s(k,\mathrm{\μ})=0\\ F(k,\mathrm{\μ})=0\end{array}\right\}$

Draw the back calculated value of resistance of deformation K and coefficientoffriction;

(4) with the back calculated value substitution resistance of deformation and the coefficientoffriction theoretical model of resistance of deformation K and coefficientoffriction, calculated value is carried out adaptive equalization, optimize the adaptive learning coefficient, wherein the resistance of deformation theoretical model

$k=\frac{2}{\sqrt{3}}{C}_{k0}{k}_0{(\frac{2}{\sqrt{3}}1n\frac{100}{100-\mathrm{\ξ}\·r_{\mathrm{tm}}}+{\mathrm{\ϵ}}_0)}^n$

$r_{\mathrm{tm}}=\frac{h_0-h_m}{h_0}\×100$

h _m＝(1-γ)·h _in+γ·h _out

In the formula

K-resistance of deformation

h ₀-original depth

h _m-average thickness

h _In, h _Out-inlet, exit thickness

k ₀, ε ₀, n, ξ, γ-model parameter

C _K0-model adaptation learning coefficient

Coefficientoffriction theoretical model formula

$\mathrm{\μ}=C_1\·e^{(b_L\·L+C_L\·L^2+d_L\·L^3+e_L\·L^4)+b_v(v-v_0)+c_v{(v-v_0)}^2+d_v{(v-v_0)}^3+e_v{(v-v_0)}^4}$

μ-coefficient of friction in the formula

L-working roll rolled band steel length

V-band steel exports speed setting value

b _L～e _L, b _v～f _v-model parameter

v ₀-band steel exports velocity standard value

C ₁-model adaptation learning coefficient.

2, the described rolling metal plate and tape of claim 1 is controlled parameter optimizing method automatically and be it is characterized in that described step (2), (3), (4) finish by computer program, and its implementation is as follows: beginning, and measured data is read in; Measured data is handled, and removes unreasonable measured value, and to reasonable measured value average make data can be with judgement not; Calculate advancing slip value f _s,, survey advancing slip value f the actual measurement rolling force F substitution theoretical model that takes out _sThe substitution theoretical model is found the solution the binary equation of higher degree group by resistance of deformation K and coefficientoffriction two unknown quantitys, carries out iterative computation, draws the back calculated value of resistance of deformation K and coefficient of friction u as convergence, quits a program as not restraining; The theoretical model of calculated value substitution resistance of deformation K and coefficientoffriction behind resistance of deformation K and the coefficientoffriction is carried out the optimization of adaptive learning coefficient, the rolling force setup value that the resistance of deformation K that has optimized and coefficientoffriction numerical value substitution rolling force F theoretical model are optimized.

Images