CN101927271A - Roll eccentricity compensation method based on on-line recursive parameter estimation and equipment thereof - Google Patents

Abstract

The invention relates to a roll eccentricity compensation method based on on-line recursive parameter estimation and equipment thereof. The method comprises the following steps of: carrying out eccentricity compensation testing, sampling and computing to obtain a roll eccentricity signal delta R1 in eccentricity testing; obtaining the main component forms of all the eccentricity signals of a rolling mill by offline using fast Fourier transform; online sampling and computing to obtain an actual roll eccentricity signal delta R2; calculating by using an on-line recursive algorithm, and continuously correcting a main component amplitude and an initial phase angle parameter to further reconstruct an eccentricity signal estimation value delta R'2; and determining the roll gap eccentricity compensation value delta SC of the delta R'2. The equipment comprises a sampling device, a transformation device, a device for online sampling and computing the roll eccentricity signal in the rolling process, a device for realizing on-line recursive parameter estimation and a roll gap compensation device, which are sequentially connected through electrical signals. The invention can be used for effectively compensating the influence of roll eccentricity on the thickness precision of a rolled material by only using common detection equipment on the rolling mill without being added with special detection equipment or accurately obtaining the rotating angle of a supporting roll.

Description

Roll eccentricity compensation method and equipment thereof based on online recursive parameter estimation

Technical field

The present invention relates to a kind of suppress cold rolling or hot rolling mill produce in roll eccentricities to the method and the equipment thereof of the influence of stocking thickness and precision, particularly relate to a kind of roll eccentricity compensation method and equipment thereof based on online recursive parameter estimation.

Background technology

Because it is not that the situation of just in time coincideing is called roll eccentricities that roll and roll bearing shape irregular caused the axis of rotation of roll and how much axle center, these irregular shapes may result from processes such as the manufacturing, reconditioning, assembling, wearing and tearing, thermal expansion of roll, because it is big many that want with respect to working roll and intermediate calender rolls in the roller of backing roll footpath, thus roll eccentricities mainly be since backing roll itself circle and roller directly produce with other roller disalignment.

In the band steel was produced, roll eccentricities can cause the milling train automatic thickness control system to produce reciprocal adjusting, made the band steel shut out the fluctuation of thickness generating period, and the thickness and precision of being with steel is caused very big influence.Had recent decades the compensation method of many enterprises, research institution's breaker roll off-centre to study, present main method can be divided three classes for this reason:

The one, prevention roll eccentricity control method, this method is to create some conditions as far as possible so that can reduce the influence of roll eccentricities to rolled piece thickness before rolling, and in rolling, do not adopt any corrective action, the method that the dead band is set as early stage employing avoids press down system to be subjected to the High-frequency Interference of roll eccentricities, but raising thickness control accuracy and the disturbance of inhibition roll eccentricities are conflicting to the requirement of dead band size, and the two is difficult to take into account.

The 2nd, passive roll eccentricity control method, the main purpose of these class methods are that the thickness disturbance reponse that roll gap control system breaker roll off-centre is caused is insensitive, and do not need roll gap to proofread and correct according to the roll eccentricities functional relation.

The 3rd, active roll eccentricity control method, these class methods generally draw compensating signal by the roll eccentricities component detection, deliver to then in the roll gap controlled adjuster with the compensation roll eccentricities.This method is the method that present effect is best, precision is the highest.According to the processing mode of detection signal, these class methods can be divided into analytic approach and synthesis again.In analytic approach, the roll eccentricities component extracts from detection signal by mathematical methods (generally adopting the Fourier transformation analysis).In synthesis, the roll eccentricities component obtains by duplicating the roll eccentricities component, signal replication can adopt Mechanical Method or coulometry, a kind of checkout gear that utilizes the cam that rotates simultaneously with backing roll to simulate roll eccentricities that for example German Newman company proposes, the employing roll gap sensor that also has German Krupp to propose detects the influence of roll eccentricities to roll gap.

In these methods, what have needs special detection equipment and instruments such as cam, roll gap meter, and it safeguards that maintenance is installed exists inconvenience.And the employing data analysis method for example uses fourier transform method, because its calculating is a large amount of complex multiplication and additions, is difficult to finish in real time calculating for industrial normally used automatic controller.And the requirement of the service condition of Fourier transformation is very high, and under the situation, Fourier transformation method generally can not accurately calculate amplitude, frequency and the phase angle of each periodic component of eccentric signal at the scene.Also have certain methods to be based on that the backing roll anglec of rotation calculates in addition, this method is very accurate to the measurement or the calculation requirement of the current angle of backing roll, perhaps adopt and on backing roll, install pulse coder additional and determine current angle, perhaps calculate the current angle of backing roll by the encoder on the live-roller.Because backing roll is often changed, there is the inconvenience of installation and maintenance in the method that installs pulse coder in the above additional; And inevitably deposit the inconvenience that has installation and maintenance towards the method for encoder by the angle of live-roller indirect calculation backing roll; And inevitably have error by the angle of live-roller indirect calculation backing roll, along with the rotation of backing roll, the constantly accumulation of this error causes the phase angle meter of eccentricity compensation not calculated accurately really, thereby causes the compensation failure.

The roll eccentricities signal is the signal of one-period, and periodic signal can be decomposed into infinite series, has only the component of some frequencies to account for the principal status of public economy in these progression.At present all suppose fundamental frequency under in the correlative study data or fundamental frequency to add two times of components under the fundamental frequency be main body, but in fact this supposition may not meet reality.

The present invention has adopted the eccentric signal of fast discrete Fourier transfer pair sampling to carry out off-line analysis in the early stage for this reason, but only in order to determine the main body component form of this milling train eccentric signal, is not used for determining the parameter of the main body component in the eccentric signal.And FFT is that off-line carries out on industrial computer, do not have the requirement of real-time operation time.The main body component parameter of eccentric signal uses online recursive algorithm to estimate, its amplitude is all constantly revised in rolling carrying out with angular dimensions mutually, therefore do not need accurately to know the anglec of rotation of backing roll, thereby avoided the problem that exists in the said method.

At present a six-high cluster mill that uses and main detecting element as shown in Figure 1, milling train is by the intermediate calender rolls transmission, last intermediate calender rolls 6 (live-roller) and following intermediate calender rolls 9 (live-roller) rotating speed maintenance basically identical, rotation speed n _sMeasure by the last intermediate calender rolls pulse coder 15 and the following intermediate calender rolls pulse coder 16 that are installed in above the driving motor, calculate the angular velocity omega of backing roll according to this value.Milling train adopts hydraulic cylinder 3 to regulate roll gap positional value between top working roll 7 and the bottom working roll 8, the roll gap actual value is measured by displacement transducer 4, milling train is depressed controller 1 and is regulated the opening degree of servo valve 2 according to the roll gap position setting value, and then controls the displacement of hydraulic cylinder 3.Calibrator 12, back calibrator 13 before being equipped with before and after the milling train, the thick difference of the inlet of stocking 11 and export thick difference and obtain thus.The roll-force actual value is measured by measurement of rolling force element 14.Milling train main transmission governing system 17 can be regulated the rotating speed of milling train.Because the eccentric phenomena of upper support roller 5 and lower support roller 10 has the influence of one-period to exporting thick difference and roll-force, can isolate this periodic signal from export thick poor actual value and roll-force actual value, Here it is eccentric signal.

Summary of the invention

Technical problem to be solved by this invention is: provide a kind of roll eccentricity compensation method based on online recursive parameter estimation that need not to add special detection equipment, to overcome the deficiency of existing eccentric compensation method.Another purpose provides a kind of roll eccentricities compensation arrangement based on online recursive parameter estimation, to improve the stocking thickness control accuracy, enhances productivity.

The present invention solves its technical problem and adopts following technical scheme:

Roll eccentricity compensation method based on online recursive parameter estimation provided by the invention, its step comprises:

Steps A: at first carry out eccentricity compensation test, sampling and calculate the roll eccentricities signal delta R of test in the test ₁

Step B: off-line uses the fast discrete Fourier transformation, tries to achieve the main body component form of all eccentric signals of this milling train;

Step C: on-line sampling also calculates, and obtains the actual roll eccentricities signal delta R in the operation of rolling ₂

Step D: use online recursive algorithm to ask for, and constantly revise the amplitude and the initial phase angle parameter of estimated eccentricity signal, and then reconstruct eccentric signal estimated value Δ R ' ₂

Step e is determined Δ R ' ₂Roll gap eccentricity compensation value Δ S _C:

At definite Δ S _CProcess in, according to the eccentric signal estimated value Δ R ' of above-mentioned estimation ₂, the influence that this eccentric signal causes mill roll-gap is judged by elder generation, should influence oppositely to compensate to fixed value of roller slit S ^*Get on, eliminating the influence that this eccentric signal causes roll gap, and then improve the exit thickness precision of steel;

Through above-mentioned steps, realize roll eccentricities compensation based on online recursive parameter estimation.

Described steps A can adopt and may further comprise the steps:

Steps A ₁: adjust milling train main transmission speed n _s, stably reach the speed of eccentric signal test request;

Steps A ₂: regulate down-pressing system of rolling mill, make roller pressing to eccentric signal gaging pressure or measuring position;

Steps A ₃: rotate in the process of some circles at backing roll, regulate the down-pressing system of rolling mill system that makes and keep the gap values between rollers set constant or keep the roll-force set constant, during gap values between rollers S or general rolling force value F are carried out data sampling, sampling obtains data set S _i, F _i, wherein: S _iBe i gap values between rollers constantly; F _iBe i general rolling force value constantly; 1≤i≤n;

Steps A ₄: if keep gap values between rollers constant, then obtain the mean value of the roll-force of all samplings If keep roll-force constant, then obtain the mean value of the gap values between rollers of all samplings

Steps A ₅: if keep gap values between rollers constant, then deduct roll-force mean value and obtain data set with the roll-force value of all samplings

If keep roll-force constant, then deduct roll gap mean value and obtain data set with the gap values between rollers of all samplings

These two data sets are exactly the roll eccentricities signal delta R of test ₁

Described step B can adopt and may further comprise the steps:

Step B ₁: analyze roll eccentricities signal delta R ₁, try to achieve the fundamental frequency f of this periodic signal, this frequency is the backing roll speed;

Step B ₂: breaker roll eccentric signal Δ R ₁Carry out the fast discrete Fourier conversion, this time-domain signal is transformed to frequency domain analysis;

Step B ₃: observe this roll eccentricities signal delta R at frequency domain ₁The amplitude-phase diagram of signal takes out the main body component of several components of amplitude maximum as the roll eccentricities signal, thereby obtains the main body component form of all roll eccentricities signals of this milling train.

Described step C can adopt and may further comprise the steps:

Step C ₁: the exit thickness that the stocking inlet thickness difference calculating that records according to the inlet calibrator can cause is poor, and computing formula is:

$\mathrm{\Δh}=\mathrm{\ΔH}\×\left(\frac{C_M}{C_G+C_M}\right)$

Wherein: Δ h is that exit thickness is poor; Δ H is that inlet thickness is poor; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient;

Step C ₂: according to the actual milling train entrance velocity that records, muzzle velocity, front and back calibrator distance to the mill roll-gap place, and will be according to the thick difference of outlet that the thick difference of inlet calculates the delay time output of roll gap place and then the outlet calibrator place's output of delaying time;

Step C ₃: will delay time to the thick difference of the outlet that exports the calibrator place through a first-order lag device, its time constant is outlet calibrator time constant, delay time with the sampling of compensation outlet calibrator, finally do not contained the thick difference DELTA h of outlet at the outlet calibrator place of Influence from Eccentric after handling through these _e

Step C ₄: the stocking that the outlet calibrator records exports thick poor Δ h _A1All include eccentric influence among the general rolling force signal F that records with the measurement of rolling force device,, then adopt the thick difference DELTA h of outlet that calculates if select the thick difference of outlet as the eccentric signal source _eDeduct the thick difference DELTA h of the actual outlet that records _A1Obtain actual roll eccentricities signal delta R ₂If selecting general rolling force is the eccentric signal source, then adopt the thick difference DELTA h of outlet that calculates _eDeduct the thick difference DELTA h of outlet that converts by general rolling force _A2Obtain actual roll eccentricities signal delta R ₂, wherein the thick difference computing formula of outlet that is converted by general rolling force is:

Δh _a2＝(F-F _S)/C _M

In the formula: Δ h _A2Be the thick difference of outlet that converts, F is the general rolling force actual value; F _SBe the general rolling force setting value; C _MBe the stocking plastic coefficient.

Described step D can adopt and may further comprise the steps:

Step D ₁: use online recursive algorithm estimated eccentricity signal Δ R ' ₂Amplitude and the initial value of initial phase angle parameter;

Step D ₂: according to the amplitude and the initial phase angle parameter of the eccentric signal that estimates, reconstruct the eccentric signal Δ R ' that current time is in outlet calibrator place ₂Value.With Δ R ' ₂Value is with actual value Δ R ₂Compare, obtain both difference e, use online recursive algorithm constantly to revise the amplitude and the initial phase angle parameter of estimation, become very little, be close to 0 up to e.At this moment can think that estimated value is very approaching or equaled actual value.

Described step e can adopt and may further comprise the steps:

Step e ₁: calculate the phase pushing figure φ of outlet calibrator to mill roll-gap, the eccentric signal Δ R ' that reconstructs according to outlet calibrator to this segment distance and the milling train muzzle velocity of mill roll-gap ₂Value obtains the eccentricity value at current time roll gap place through phase shift;

Step e ₂: with the eccentricity value at the current time roll gap place that obtains by a first-order lag device, its time constant is the down-pressing system of rolling mill time constant, with the time-delay that the action of compensation down-pressing system of rolling mill causes, finally obtain the eccentric signal estimated value Δ R at current time roll gap place " ₂

Step e ₃: calculate the eccentricity compensation amount of roll gap, its computing formula is:

$\mathrm{\&Delta;}{S}_C=-\mathrm{\&Delta;}{\mathrm{<figure-callout\; id="2"\; label="R"\; filenames="HDA0000024877500000011.png,HDA0000024877500000021.png"\; state="\{\{state\}\}">R</figure-callout>}}_2^{\&prime;\&prime;}\&times;\left(\frac{C_G+C_M}{C_G}\right),$

In the formula: Δ S _CBe the roll gap offset; Δ R " ₂Be the eccentric signal estimated value; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient;

Through above-mentioned steps, finally obtain roll gap offset Δ S _C, the roll gap offset of obtaining is added to roll gap position setting value S ^*Get on, just finished eccentricity compensation.

The invention provides roll eccentricities compensation arrangement based on online recursive parameter estimation, this device comprises with on-line sampling in the continuous successively sampling apparatus of the signal of telecommunication, converting means, the operation of rolling and calculates the device of roll eccentricities signal, the device and the roll gap compensation arrangement of realization online recursive parameter estimation, wherein: described sampling apparatus is the device that is used for the eccentric testing sampling and calculates the roll eccentricities signal, promptly regularly carries out data sampling S _i, F _iDevice and calculate roll test eccentric signal Δ R ₁Device, S _iBe i gap values between rollers constantly, F _iBe i general rolling force value constantly, 1≤i≤n; Described converting means is the device that off-line carries out the fast discrete Fourier transformation.

On-line sampling and calculate the device of roll eccentricities signal in the described operation of rolling, it comprises and calculates the thick poor device of outlet; The device of time-delay transmission; First first-order lag device; Thick poor device converts.

The device of described realization online recursive parameter estimation, it comprises online estimation unit and amplitude and initial phase angle calculation element.

Described roll gap compensation arrangement, it is used to calculate the roll gap offset of eccentric signal, and this device comprises phase deviation compensation arrangement, second first-order lag device and roll gap eccentricity compensation device for calculating.

The present invention compared with prior art has following main beneficial effect:

The present invention has managed to overcome the weak point of existing eccentric compensation method.Adopt fast discrete Fourier transfer pair eccentric signal to analyze, accurately obtain the main body component form in the eccentric signal; Only use checkout equipment commonly used on the modern milling train, do not need to increase in addition special equipment; Do not need accurately to detect or calculate the anglec of rotation of backing roll.The present invention can effectively suppress the influence of roll eccentricities to the stocking thickness and precision, and according to result of the test, this method can accurately estimate the main body component of eccentric signal, and the energy of main body component can account for more than 90% of whole eccentric signals.According to the eccentric signal main body component that estimates, calculate reverse rollers seam compensation rate and be applied to positioner, just can eliminate the harmful effect of eccentric signal more than 90%, and then improve thickness control accuracy, enhance productivity.

Description of drawings

A six-high cluster mill and the main detecting element schematic diagram of Fig. 1 for using at present.

Fig. 2 is the schematic diagram of roll eccentricities compensation of the present invention.

Fig. 3 is original eccentric signal amplitude-phase diagram.

Fig. 4 is the eccentric signal amplitude-phase diagram after compensation.

Among the figure: 1. milling train is depressed controller; 2. servo valve; 3. hydraulic cylinder; 4. displacement transducer; 5. upper support roller; 6. go up intermediate calender rolls; 7. top working roll; 8. bottom working roll; 9. descend intermediate calender rolls; 10. lower support roller; 11. stocking; 12. preceding calibrator; 13. back calibrator; 14. roll-force detecting element; 15. last intermediate calender rolls pulse coder; 16. following intermediate calender rolls pulse coder; 17. milling train main transmission governing system; 18. sampling apparatus; 19. converting means; 20. measure thick poor device; 21. time delay output apparatus; 22. first first-order lag device; The thick poor device 23. convert; 24. on-line parameter estimation unit; 25. amplitude and initial phase angle calculation element; 26. phase deviation compensation arrangement; 27. second first-order lag device; 28. roll gap eccentricity compensation device for calculating.

The specific embodiment

The invention will be further described below in conjunction with embodiment and accompanying drawing.

1. 1 kinds of roll eccentricity compensation methods of embodiment based on online recursive parameter estimation:

Referring to Fig. 1, Fig. 2, this compensation method may further comprise the steps:

1. obtain test roll eccentricities signal delta R ₁:

Carry out eccentricity compensation test, sampling and calculate the roll eccentricities signal delta R of test in the test ₁, be specially:

(1) adjusts milling train main transmission speed n _s, stably reach the speed that eccentric signal is measured.

Use milling train main transmission control system 17, the main transmission of control milling train is rotated with the eccentricity compensation test speed, is generally 20% of maximum speed, waits milling train main transmission rotation speed n _sAfter reaching setting speed (for example 20% maximum speed) and settling out, begin to carry out next step.

(2) regulate down-pressing system of rolling mill, make roller pressing to eccentric signal gaging pressure or measuring position.

Rotate in the process of some circles at backing roll, use milling train controller 1 to adjust down-pressing system of rolling mill, the system that makes keeps the gap values between rollers of setting roll-force constant or that maintenance is set constant, Deng down-pressing system of rolling mill reach the setting value and settle out after gap values between rollers S, roll-force value F are carried out data sampling, sampling time, more little then sampled result was accurate more, but high more to the sample devices requirement, generally get a less numerical value, for example 1ms or 4ms.Sampling obtains data set S _i, F _iWherein: S _iBe i gap values between rollers constantly; F _iBe i general rolling force value constantly; 1≤i≤n.

If keep gap values between rollers constant, then obtain the mean value of the roll-force of all samplings

If keep roll-force constant, then obtain the mean value of the gap values between rollers of all samplings

If keep gap values between rollers constant, then deduct roll-force mean value and obtain data set with the roll-force value of all samplings

Roll-force is constant then to be deducted roll gap mean value with the gap values between rollers of all samplings and obtains data set if keep

These two data sets are exactly the roll eccentricities signal delta R of test ₁

2. obtain the main body component form of all roll eccentricities signals of this milling train:

Off-line uses the fast discrete Fourier transformation, tries to achieve the main body component form of all roll eccentricities signals of this milling train, and concrete grammar is:

(1) analyzes roll eccentricities signal delta R ₁, try to achieve the fundamental frequency f (backing roll speed) of this periodic signal;

(2) the Δ R that rapid middle sampling obtains to previous step ₁Carry out the fast discrete Fourier conversion, this time-domain signal is transformed to frequency domain analysis; Observe the amplitude-phase diagram of this eccentric signal at frequency domain, take out the main body component of several components of amplitude maximum, thereby obtain the main body component form of all eccentric signals of this milling train as eccentric signal.

For example, in the eccentric testing, keep gap values between rollers constant, sampling roll-force data calculate test eccentric signal Δ R then according to the method described above ₁, then to Δ R ₁Carry out the fast discrete Fourier conversion, obtain the amplitude-phase diagram under the frequency domain shown in Figure 3.

According to calculating backing roll speed (fundamental frequency) f is 0.116Hz, we from Fig. 3 as can be seen, eccentric signal amplitude under two frequencies is big especially, its energy accounts for more than 90% of gross energy, be 0.232Hz and 0.812Hz, be respectively 2 times and 7 times of fundamental frequency, we are the main body component of the signal below these two frequencies as eccentric signal, and the representation of the main body component of all eccentric signals of this milling train of Que Dinging is like this:

ΔR′＝a ₁sin(4πft+b ₁)+a ₂sin(14πft+b ₂)＝a ₁sin(2ωt+b ₁)+a ₂sin(7ωt+b ₂)

In the formula: Δ R ' is the main body component form of all eccentric signals of this milling train; a ₁, a ₂It is the amplitude of signal under two times of fundamental frequencies and the seven times of fundamental frequencies; b ₁, b ₂It is the initial phase angle of signal under two times of fundamental frequencies and the seven times of fundamental frequencies; ω is a backing roll angular speed; T is a time variable.

Estimated eccentricity signal further can be decomposed into according to triangle formula sin (ω t+ φ)=cos φ sin ω t+sin φ cos ω t:

ΔR′＝[sin?2ωt?cos2ωt?sin?7ωt?cos7ωt]×[a ₁cosb ₁?a ₁sinb ₁?a ₂cosb ₂?a ₂sinb ₂]

So just eccentric signal is divided into two parts, previous section is known, and aft section is the parameter that needs estimation.

3. obtain the actual roll eccentricities signal delta R in the operation of rolling ₂:

On-line sampling, and calculate roll eccentricities signal delta R ₂, concrete grammar is:

Carry out the detailed description of this step in conjunction with Fig. 2.Fig. 2 is the schematic diagram of eccentric compensation method.

The exit thickness that the stocking inlet thickness difference calculating that records according to the inlet calibrator can cause is poor, and computing formula is:

$\mathrm{\&Delta;h}=\mathrm{\&Delta;H}\&times;\left(\frac{C_M}{C_G+C_M}\right)$

In the formula: Δ h is poor for the exit thickness that calculates; Δ H is poor for the inlet thickness of measuring; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient.

According to the actual milling train entrance velocity that records, muzzle velocity, front and back calibrator distance to the mill roll-gap place, with the thick difference of the outlet that calculates by the outlet calibrator place's output of delaying time of time-delay transmission link 21.

With the time-delay to the thick difference of the outlet that exports the calibrator place through a first-order lag device 22, its time constant is outlet calibrator time constant, delay time with the sampling of compensation outlet calibrator, finally do not contained the thick difference DELTA h of outlet at the outlet calibrator place of Influence from Eccentric after handling through these _e

The stocking that the outlet calibrator records exports thick poor Δ h _A1The general rolling force signal F that records with the measurement of rolling force device _aIn all include eccentric influence, if select the thick difference of outlet, then adopt the thick difference DELTA h of outlet that calculates as the eccentric signal source _eDeduct the thick difference DELTA h of the actual outlet that records _A1Obtain eccentric signal Δ R ₂If selecting general rolling force is the eccentric signal source, then adopt the thick difference DELTA h of outlet that calculates _eDeduct the thick difference DELTA h of outlet that converts by general rolling force _A2Obtain actual roll eccentricities signal delta R ₂Wherein the thick difference computing formula of outlet that is converted by general rolling force is:

Δh _a2＝(F-F _S)/C _M

In the formula: F is the general rolling force actual value; F _SBe the general rolling force setting value; C _MBe the stocking plastic coefficient.

4. reconstruct eccentric estimated signal value Δ R ' ₂:

Use online recursive algorithm to ask for estimated parameter, and constantly revise, and then reconstruct Δ R ' ₂, concrete grammar is as follows:

Use online recursive algorithm to estimate wherein unknown parameter [a ₁Cosb ₁a ₁Sinb ₁a ₂Cosb ₂a ₂Sinb ₂], popular online recursive algorithm has RLS, Minimum Mean Square Error recursive algorithm etc. at present, can use.And then calculate the amplitude and the initial phase angle of eccentric signal with following formula:

$A=\sqrt{{\left(A\cos b\right)}^2+{\left(A\sin b\right)}^2},$

$b=\arctan \left(\frac{A\sin b}{A\cos b}\right)$

According to the amplitude and the initial phase angle of the eccentric signal that estimates, reconstruct the eccentric signal Δ R ' that current time is in outlet calibrator place ₂Value is with Δ R ' ₂Value is with actual value Δ R ₂Compare, obtain both difference e, use online recursive algorithm constantly to revise the amplitude and the initial phase angle parameter of estimation, become very little, be close to 0 up to e.At this moment can think that estimated value is very approaching or equaled actual value, the precision of estimated parameter is very high, and the back just can have been calculated with estimated parameter.

For example, adopt the eccentric estimated signal form in the previous example, use RLS to estimate unknown parameter, calculate amplitude and initial phase angle parameter according to following formula then.Calculate a period of time through online recursion, it is very little that error e becomes, and less than 1 micron, reached the condition of convergence.At this moment the current time that can reconstruct is in the Δ R ' at outlet calibrator place ₂For:

ΔR′ ₂＝[7.792×sin(1.457t-0.917)+8.434×sin(5.062t-0.361)]/C _M

In the formula: t is a time variable, C _MBe the stocking plastic coefficient.Because original eccentric signal uses the roll-force signal, here divided by C _MThe roll-force signal is converted to the thickness difference signal.The amplitude-phase diagram of original eccentric signal is seen Fig. 3.

Original eccentric signal is after the eccentric signal that has deducted above-mentioned estimation, and we carry out discrete fast Fourier transform once more, obtain amplitude-phase diagram shown in Figure 4.

Compare with original eccentric signal, can be seen by Fig. 4, the main body component of eccentric signal has obtained eliminating well, and this method can be eliminated more than 90% of original eccentric signal energy, backing roll off-centre can be reduced to minimum to the influence of thickness of steel product precision.

5. determine Δ R ' ₂Roll gap eccentricity compensation value Δ S _C:

After the eccentric signal that obtains estimating, we judge the influence that this eccentric signal causes mill roll-gap earlier, and then should influence and oppositely compensate to fixed value of roller slit and get on, thereby eliminate the influence that eccentric signal causes roll gap, to improve the exit thickness precision of steel.Calculate Δ S _CConcrete grammar be:

(1) obtains the eccentric signal value Δ R at roll gap place " ₂:

Calculate the phase pushing figure φ of outlet calibrator to mill roll-gap, the eccentric signal Δ R ' that reconstructs according to outlet calibrator to this segment distance and the milling train muzzle velocity of mill roll-gap ₂Value obtains the eccentricity value at current time roll gap place through phase deviation compensation arrangement 26;

With the eccentricity value at the current time roll gap place that obtains by second first-order lag device 27, its time constant is the down-pressing system of rolling mill time constant, with the time-delay that the action of compensation down-pressing system of rolling mill causes, finally obtain the eccentric signal estimated value Δ R at current time roll gap place " ₂

(2) obtain roll gap offset Δ S _C:

Adopt following computing formula to calculate, obtain Δ S _C,

$\mathrm{\&Delta;}{S}_C=-\mathrm{\&Delta;}{\mathrm{<figure-callout\; id="2"\; label="R"\; filenames="HDA0000024877500000011.png,HDA0000024877500000021.png"\; state="\{\{state\}\}">R</figure-callout>}}_2^{\&prime;\&prime;}\&times;\left(\frac{C_G+C_M}{C_G}\right)$

In the formula: Δ S _CBe the roll gap offset; Δ R " ₂Be the eccentric signal estimated value; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient.

Through above-mentioned steps, finally obtain roll gap offset Δ S _C, then with Δ S _CBe added to the setting value S of roll gap positioner ^*Get on, just realized the compensation of breaker roll off-centre.

Embodiment 2. is based on the roll eccentricities compensation arrangement of online recursive parameter estimation:

The structure of this roll eccentricities compensation arrangement as shown in Figure 2, comprise on-line sampling in the sampling apparatus 18 that links to each other successively with the signal of telecommunication, converting means 19, the operation of rolling and calculate the roll eccentricities signal device, realize the device and the roll gap compensation arrangement of online recursive parameter estimation.

Described sampling apparatus 18 is a kind of devices that are used for the eccentric testing sampling and calculate the roll eccentricities signal, and this device is realized by programming on programmable logic controller (PLC).This device is regularly to carry out data sampling S _i, F _iDevice and calculate roll test eccentric signal Δ R ₁Device.Wherein, S _iBe i gap values between rollers constantly, F _iBe i general rolling force value constantly, 1≤i≤n, n are natural number.

Described converting means 19 is industrial control computer or personal computer, by programming on industrial computer or utilizing personal computer to carry out data and handle, can realize that all off-line carries out the fast discrete Fourier transformation.

On-line sampling and calculate the device of roll eccentricities signal in the described operation of rolling, it comprises device 21, first first-order lag device 22 and the thick poor device 23 that converts that calculates the thick poor device 20 of outlet, time-delay transmission, wherein: calculate the thick poor device 20 of outlet by the signal of telecommunication through the device 21 of time-delay transmission, first first-order lag device 22 back with the actual thick difference DELTA h of outlet _A1Or the thick difference DELTA h of outlet of roll-force conversion _A2Relatively, the output that feeds input, amplitude and the initial phase angle calculation element 25 of On-line Estimation device 24 is then respectively compared and is obtained evaluated error e.

It is a kind of devices that are used for being calculated by the thick difference of the inlet of measuring the thick difference of outlet that described calculating exports thick poor device 20, and this device utilizes following formula to calculate:

$\mathrm{\&Delta;h}=\mathrm{\&Delta;H}\&times;\left(\frac{C_M}{C_G+C_M}\right)$

In the formula: Δ h is poor for the exit thickness that calculates; Δ H is poor for the inlet thickness of measuring; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient.

The device 21 of described time-delay transmission is a kind of device that input value is delayed time and exported after a period of time of being used for.

Described first first-order lag device 22, it utilizes following formula to calculate:

$Y_n=Y_{n-1}+\frac{\mathrm{TA}}{T}(X_n-X_{n-1})$

In the formula: Y is a first-order lag device output valve; TA is the controller sampling period; T is the time constant of first-order lag device; X is a first-order lag device input value; N is a natural number.

The thick poor device 23 of described conversion is a kind of devices that roll-force are converted to thick difference; This device utilizes following formula to calculate:

Δh _a2＝(F-F _S)/C _M

In the formula: Δ h _A2Be the thick difference of outlet that converts, F is the general rolling force actual value; F _SBe the general rolling force setting value; C _MBe the stocking plastic coefficient.

The device of described realization online recursive parameter estimation is the industry spot intelligent controller, and this controller is provided with on-line parameter estimation unit 24, amplitude and the initial phase angle calculation element 25 that links to each other with the signal of telecommunication.

Described on-line parameter estimation unit 24, it is used to realize the device of online recursive parameter estimation algorithm, uses least square method of recursion to estimate unknown parameter [a in this example ₁Cosb ₁a ₁Sinb ₁a ₂Cosb ₂a ₂Sinb ₂], also can adopt other online recursive algorithm to carry out parameter Estimation, for example lowest mean square root algorithm.

Described amplitude and initial phase angle calculation element 25, it is used to calculate the device of eccentric signal amplitude A and phase angle b, and this device utilizes following formula to calculate:

$A=\sqrt{{\left(A\cos b\right)}^2+{\left(A\sin b\right)}^2},$

$b=\arctan \left(\frac{A\sin b}{A\cos b}\right).$

In the formula: A is the amplitude of the component of signal under certain frequency; B is the initial phase angle of the component of signal under certain frequency.

Described roll gap compensation arrangement, it is used to calculate the roll gap offset of eccentric signal, and this device comprises: phase deviation compensation arrangement 26; Second first-order lag device 27; Roll gap eccentricity compensation device for calculating 28.Described second first-order lag device 27, its input links to each other by the output of the signal of telecommunication with phase deviation compensation arrangement 26, and its output links to each other by the input of the signal of telecommunication with roll gap eccentricity compensation device for calculating 28.

Above-mentioned phase deviation compensation arrangement 26, it is used to calculate phase pushing figure, and this device calculates the backing roll phase pushing figure φ of outlet calibrator to mill roll-gap according to outlet calibrator to this segment distance and the milling train muzzle velocity of mill roll-gap.

Above-mentioned second first-order lag device 27 is used to compensate the action delay of down-pressing system of rolling mill.This device computing formula is with first first-order lag device 22.

Above-mentioned roll gap eccentricity compensation device for calculating 28, it is used to calculate roll gap eccentricity compensation value, and this device utilizes following formula to calculate:

$\mathrm{\&Delta;}{S}_C=-\mathrm{\&Delta;}{\mathrm{<figure-callout\; id="2"\; label="R"\; filenames="HDA0000024877500000011.png,HDA0000024877500000021.png"\; state="\{\{state\}\}">R</figure-callout>}}_2^{\&prime;\&prime;}\&times;\left(\frac{C_G+C_M}{C_G}\right)$

In the formula: Δ S _CBe the roll gap offset; Δ R " ₂Be the eccentric signal estimated value; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient.

The roll eccentricities compensation arrangement that present embodiment provides based on online recursive parameter estimation, its course of work is:

At first carry out the eccentricity compensation test, that is: use gap values between rollers and roll-force value under the sampling apparatus 18 sampling eccentric testing conditions, through handling the eccentric signal that obtains in the eccentric testing, the gained eccentric signal is fed converting means 19 carry out the fast discrete Fourier transformation, obtain the main body component form of all eccentric signals of this milling train by analysis, this form feeds on-line parameter estimation unit 24.

Then in the operation of rolling, it is thick poor that the thick difference of inlet of using the thick poor device 20 of calculating outlet to record is converted to outlet, and by the time delay output apparatus 21 thick difference at the calibrator place outlet calibrator place that delays time that will enter the mouth, re-use the sampling time-delay of first first-order lag device 22 compensation outlet calibrators, obtain exporting the thick difference at calibrator place like this; Perhaps use the thick poor device 23 that converts to convert, obtain exporting the thick difference at calibrator place according to the roll-force signal.The thick difference at the outlet calibrator place that use calculates deducts outlet calibrator measured value can obtain the eccentric signal actual value.We use on-line parameter estimation unit 24 and amplitude and initial phase angle calculation element 25 to estimate unknown parameter respectively, and calculate amplitude and initial phase angle parameter, the eccentric signal actual value is constantly revised estimated parameter with the poor e between the estimated value by online recursive algorithm, through recursion after a while, poor e between the two reaches minimum, and estimated value equals actual value gradually.

At last, use the eccentric signal estimated value to calculate the roll gap compensation rate.Specifically, use earlier the phase difference of phase deviation compensation arrangement 26 compensation outlet calibrators to this segment distance of mill roll-gap, and the time-delay of using second first-order lag device, 27 compensation press down systems to cause, use roll gap eccentricity compensation device for calculating 28 to calculate the roll gap compensation rate then, oppositely compensate to the roll gap position setting value and get on, the whole process of eccentricity compensation is accomplished.

Below in conjunction with the embodiments the present invention has been done concrete description, but not as qualification of the present invention, all modifications and variations in claim scope of the present invention all drop on and are subjected within protection scope of the present invention.

Claims (10)

1. roll eccentricity compensation method based on online recursive parameter estimation is characterized in that adopting the method that may further comprise the steps:

Steps A: at first carry out eccentricity compensation test, sampling and calculate the roll eccentricities signal delta R of test in the test ₁

Step B: off-line uses the fast discrete Fourier transformation, tries to achieve the main body component form of all eccentric signals of this milling train;

Step C: on-line sampling also calculates, and obtains the actual roll eccentricities signal delta R in the operation of rolling ₂

Step D: use online recursive algorithm to ask for, and constantly revise the amplitude and the initial phase angle parameter of estimated eccentricity signal, and then reconstruct eccentric signal estimated value Δ R ' ₂

Step e is determined Δ R ' ₂Roll gap eccentricity compensation value Δ S _C:

At definite Δ S _CProcess in, according to the eccentric signal estimated value Δ R ' of above-mentioned estimation ₂, judge earlier the influence that this eccentric signal causes mill roll-gap, should influence and oppositely compensate to fixed value of roller slit and get on, eliminating the influence that this eccentric signal causes roll gap, and then the exit thickness precision of raising steel;

Through above-mentioned steps, realize roll eccentricities compensation based on online recursive parameter estimation.

2. roll eccentricity compensation method according to claim 1 is characterized in that concrete employing of described steps A may further comprise the steps:

Steps A ₁: adjust milling train main transmission speed n _s, stably reach the speed of eccentric signal test request;

Steps A ₂: regulate down-pressing system of rolling mill, make roller pressing to eccentric signal gaging pressure or measuring position;

Steps A ₃: rotate in the process of some circles at backing roll, regulate the down-pressing system of rolling mill system that makes and keep the gap values between rollers set constant or keep the roll-force set constant, during gap values between rollers S or general rolling force value F are carried out data sampling, sampling obtains data set S _i, F _i, wherein: S _iBe i gap values between rollers constantly; F _iBe i general rolling force value constantly; 1≤i≤n;

Steps A ₄: if keep gap values between rollers constant, then obtain the mean value of the roll-force of all samplings

If keep roll-force constant, then obtain the mean value of the gap values between rollers of all samplings

Steps A ₅: if keep gap values between rollers constant, then deduct roll-force mean value and obtain data set with the roll-force value of all samplings

If keep roll-force constant, then deduct roll gap mean value and obtain data set with the gap values between rollers of all samplings

These two data sets are exactly the roll eccentricities signal delta R of test ₁

3. roll eccentricity compensation method according to claim 1 is characterized in that concrete employing of described step B may further comprise the steps:

Step B ₁: analyze roll eccentricities signal delta R ₁, try to achieve the fundamental frequency f of this periodic signal, this frequency is the backing roll speed;

Step B ₂: breaker roll eccentric signal Δ R ₁Carry out the fast discrete Fourier conversion, this time-domain signal is transformed to frequency domain analysis;

Step B ₃: observe this roll eccentricities signal delta R at frequency domain ₁The amplitude-phase diagram of signal takes out the main body component of several components of amplitude maximum as the roll eccentricities signal, thereby obtains the main body component form of all roll eccentricities signals of this milling train.

4. roll eccentricity compensation method according to claim 1 is characterized in that concrete employing of described step C may further comprise the steps:

Step C ₁: the exit thickness that the stocking inlet thickness difference calculating that records according to the inlet calibrator can cause is poor, and computing formula is:

$\mathrm{\&Delta;h}=\mathrm{\&Delta;H}\&times;\left(\frac{C_M}{C_G+C_M}\right)$

Wherein: Δ h is that exit thickness is poor; Δ H is that inlet thickness is poor; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient;

Step C ₂: according to the actual milling train entrance velocity that records, muzzle velocity, front and back calibrator distance to the mill roll-gap place, and will be according to the thick difference of outlet that the thick difference of inlet calculates the delay time output of roll gap place and then the outlet calibrator place's output of delaying time;

Step C ₃: will delay time to the thick difference of the outlet that exports the calibrator place through a first-order lag device, its time constant is outlet calibrator time constant, delay time with the sampling of compensation outlet calibrator, finally do not contained the thick difference DELTA h of outlet at the outlet calibrator place of Influence from Eccentric after handling through these _e

Step C ₄: the stocking that the outlet calibrator records exports thick poor Δ h _A1All include eccentric influence among the general rolling force signal F that records with the measurement of rolling force device,, then adopt the thick difference DELTA h of outlet that calculates if select the thick difference of outlet as the eccentric signal source _eDeduct the thick difference DELTA h of the actual outlet that records _A1Obtain actual roll eccentricities signal delta R ₂If selecting general rolling force is the eccentric signal source, then adopt the thick difference DELTA h of outlet that calculates _eDeduct the thick difference DELTA h of outlet that converts by general rolling force _A2Obtain actual roll eccentricities signal delta R ₂, wherein the thick difference computing formula of outlet that is converted by general rolling force is:

Δh _a2＝(F-F _S)/C _M

In the formula: Δ h _A2Be the thick difference of outlet that converts, F is the general rolling force actual value; F _SBe the general rolling force setting value; C _MBe the stocking plastic coefficient.

5. roll eccentricity compensation method according to claim 1 is characterized in that described step D specifically may further comprise the steps:

Step D ₁: use online recursive algorithm estimated eccentricity signal Δ R ' ₂Amplitude and the initial value of initial phase angle parameter;

Step D ₂: according to the amplitude and the initial phase angle parameter of the eccentric signal that estimates, reconstruct the eccentric signal Δ R ' that current time is in outlet calibrator place ₂Value.With Δ R ' ₂Value is with actual value Δ R ₂Compare, obtain both difference e, use online recursive algorithm constantly to revise the amplitude and the initial phase angle parameter of estimation, become very little, be close to 0 up to e.At this moment can think that estimated value is very approaching or equaled actual value.

6. roll eccentricity compensation method according to claim 1 is characterized in that concrete employing of described step e may further comprise the steps:

Step e ₁: calculate the phase pushing figure φ of outlet calibrator to mill roll-gap, the eccentric signal Δ R ' that reconstructs according to outlet calibrator to this segment distance and the milling train muzzle velocity of mill roll-gap ₂Value obtains the eccentricity value at current time roll gap place through phase shift;

Step e ₂: with the eccentricity value at the current time roll gap place that obtains by a first-order lag device, its time constant is the down-pressing system of rolling mill time constant, with the time-delay that the action of compensation down-pressing system of rolling mill causes, finally obtain the eccentric signal estimated value Δ R at current time roll gap place " ₂

Step e ₃: calculate the eccentricity compensation amount of roll gap, its computing formula is:

$\mathrm{\&Delta;}{S}_C=-\mathrm{\&Delta;}{R}_2^{\&prime;\&prime;}\&times;\left(\frac{C_G+C_M}{C_G}\right),$

In the formula: Δ S _CBe the roll gap offset; Δ R " ₂Be the eccentric signal estimated value; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient;

Through above-mentioned steps, finally obtain roll gap offset Δ S _C, the roll gap offset of obtaining is added to roll gap position setting value S ^*Get on just to have finished eccentricity compensation.

7. roll eccentricities compensation arrangement based on online recursive parameter estimation, it is characterized in that described roll eccentricities compensation arrangement comprise on-line sampling in the sampling apparatus (18) that links to each other successively with the signal of telecommunication, converting means (19), the operation of rolling and calculate the roll eccentricities signal device, realize the device and the roll gap compensation arrangement of online recursive parameter estimation, wherein: described sampling apparatus (18) is the device that is used for the eccentric testing sampling and calculates the roll eccentricities signal, promptly regularly carries out data sampling S _i, F _iDevice and calculate roll test eccentric signal Δ R ₁Device, S _iBe i gap values between rollers constantly, F _iBe i general rolling force value constantly, 1≤i≤n; Described converting means (19) is the device that off-line carries out the fast discrete Fourier transformation.

8. the roll eccentricities compensation arrangement based on online recursive parameter estimation according to claim 7 is characterized in that on-line sampling in the described operation of rolling and calculates the device of roll eccentricities signal, and it comprises and calculates outlet thick poor device (20); The device (21) of time-delay transmission; First first-order lag device (22); The thick poor device (23) that converts, wherein:

It is a kind of device that is used for being calculated by the thick difference of the inlet of measuring the thick difference of outlet that described calculating exports thick poor device (20), and this device utilizes following formula to calculate:

$\mathrm{\&Delta;h}=\mathrm{\&Delta;H}\&times;\left(\frac{C_M}{C_G+C_M}\right)$

In the formula: Δ h is poor for the exit thickness that calculates; Δ H is poor for the inlet thickness of measuring; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient,

The device (21) of described time-delay transmission is a kind of device that input value is delayed time and exported after a period of time of being used for,

Described first first-order lag device (22) is used to compensate the action delay of calibrator, and this device utilizes following formula to calculate:

$Y_n=Y_{n-1}+\frac{\mathrm{TA}}{T}(X_n-X_{n-1})$

In the formula: Y is a first-order lag device output valve; TA is the controller sampling period; T is the time constant of first-order lag device; X is a first-order lag device input value; N is a natural number,

The thick poor device of described conversion (23) is a kind of device that roll-force is converted to thick difference; This device utilizes following formula to calculate:

Δh _a2＝(F-F _S)/C _M

In the formula: Δ h _A2Be the thick difference of outlet that converts, F is the general rolling force actual value; F _SBe the general rolling force setting value; C _MBe the stocking plastic coefficient.

9. the roll eccentricities compensation arrangement based on online recursive parameter estimation according to claim 7, the device that it is characterized in that described realization online recursive parameter estimation, it comprises online parameter estimation apparatus (24) and amplitude and initial phase angle calculation element (25), wherein:

Described on-line parameter estimation unit (24), it is used to realize the device of online recursive parameter estimation algorithm, this device uses least square method of recursion to estimate unknown parameter [a ₁Cosb ₁a ₁Sinb ₁a ₂Cosb ₂a ₂Sinb ₂], or adopt lowest mean square root algorithm to carry out parameter Estimation, in the formula: a ₁, a ₂Amplitude for estimated eccentricity signal main body component; b ₁, b ₂Initial phase angle for estimated eccentricity signal main body component;

Described amplitude and initial phase angle calculation element (25), it is used to calculate the device of eccentric signal amplitude A and phase angle b, and this device utilizes following formula to calculate:

$A=\sqrt{{\left(A\cos b\right)}^2+{\left(A\sin b\right)}^2},$

$b=\arctan \left(\frac{A\sin b}{A\cos b}\right),$

In the formula: A is the amplitude of the component of signal under certain frequency; B is the initial phase angle of the component of signal under certain frequency.

10. the roll eccentricities compensation arrangement based on online recursive parameter estimation according to claim 7, it is characterized in that: described roll gap compensation arrangement, it is used to calculate the roll gap offset of eccentric signal, this device comprises: phase deviation compensation arrangement (26), second first-order lag device (27), roll gap eccentricity compensation device for calculating (28)

Described phase deviation compensation arrangement (26), it is used to calculate phase pushing figure, and this device calculates the backing roll phase pushing figure φ of outlet calibrator to mill roll-gap according to outlet calibrator to this segment distance and the milling train muzzle velocity of mill roll-gap,

Described second first-order lag device (27) is used to compensate the action delay of down-pressing system of rolling mill, and this device computing formula is with first first-order lag device (22),

Described roll gap eccentricity compensation device for calculating (28), it is used to calculate roll gap eccentricity compensation value, and this device utilizes following formula to calculate:

$\mathrm{\&Delta;}{S}_C=-\mathrm{\&Delta;}{R}_2^{\&prime;\&prime;}\&times;\left(\frac{C_G+C_M}{C_G}\right)$

In the formula: Δ S _CBe the roll gap offset; Δ R " ₂Be the eccentric signal estimated value; C _GBe the mill stiffness coefficient; C _MBe the stocking plastic coefficient.

Images