EP1181992B1 - Multivariable flatness control system - Google Patents

Abstract

The system detects band flatness with a measurement system and separates the planarity error into orthogonal components. A real-time model takes all rolling process parameters into account. Another model computes planarity regulation demand values. A multi-parameter regulator regulates band planarity. Control parameters for inclusion in dynamic optimization are predicted from the dead time. Noise parameter mixing is performed. The system detects the flatness of a band with a measurement system and separates the planarity error into orthogonal components. A real-time capability model takes all parameters involved in the rolling process in to account and another model computes demand values for planarity regulation. A multi-parameter regulator regulates band planarity. Control parameters for inclusion in dynamic optimization are predicted from the dead time. Noise parameter mixing takes into account the characteristics of the incoming band, the variation of the roller force and thermal crowning. Independent claims are also included for the following: a method of measuring and/or regulating flatness in when rolling material, and a device for measuring and/or regulating the flatness of a band.

Description

The invention relates to a method for measuring and / or regulating the Flatness of strips during rolling.

When rolling tapes, it is a particular problem, a to obtain optimum flatness and shape of the band. For this it is advantageous that the preliminary band already largely the intended band profile owns and runs centric into the finishing train. Furthermore, the should Stichabnahmen in the individual scaffolding done so that in all Scaffolding a uniform band length over the entire Bandwidth is obtained. In addition, a reduction in the Strip length (in the finished product) sought, whose flatness outside the Tolerance is. This is especially true for the tape head and band foot.

For this purpose, it is known, the length distribution of the rolled sheet by means a Planheitsmeßsystems to determine (see Figure 1). The different Types of errors - e.g. Center waves, edge waves, quarter waves or unevennesses of higher order - are determined by a mathematical Analysis of the measured length distribution determined to specifically the to use suitable actuators for error correction.

The representation of the length distribution takes place with the help of a usual polynomial: p ( x ) = a 0 + a 1 x + a 2 x 2 + a 3 x 3 + a 4 x 4

In this case, one-sided edge waves on the left or right side of the band are described by the coefficients a ₁ and a ₃ . The coefficients a ₂ and a ₄ describe either symmetrical center waves or symmetrical edge waves on the left and right sides of the band. The coefficients a ₁ and a ₃ or a ₂ and a ₄ thus contain common information components.

So far, at least in most practical implementations for flatness control, primarily the coefficients (hereinafter also called components) a ₁ and a _{2 are} used.

For regulating the flatness of the finishing scale usually the manipulated variables of the work roll bending for the compensation of the components a ₂ and a ₄ and the hydraulic positions on the operator and drive side (panning) are used to eliminate the error components a ₁ and a ₃ . For the purpose of regulation, therefore, the coefficients a ₁ and a ₃ are used for the pivoting , and the coef fi cients a ₂ and a _{4 are used} as the controlled variable for the bending.

In some stands, the axial displacement of the work rolls is used primarily for presetting the roll gap contour and only occasionally within of the control loop in combination with the bend to correct the Quarter wave. Finally, the selective multi - zone cooling of Work rolls allow the correction of the flatness errors higher order. Such a regulation is for example from the German Patent application DE 197 58 466 A1 known.

In each case, a predetermined by a setup calculation Setting the rolling force and the bending force calculates the manipulated variables. As regulators known PI controllers are used, however, the Dead times of the route can not explicitly take into account. Consequently, must a weak setting of the controller gains, in particular the I component, be made to the instabilities of the control loop avoid.

This control system can meet the increasing quality requirements to the flatness do not do justice, since the flatness regulation only after relatively high time reaches its target curve. This has the consequence that First, a large tape length must be taken into account, whose Flatness lies outside the tolerance. Often, however, the setpoint curve not at all, but achieved only in approximation, so that large marginal or Center waves can arise.

Furthermore, it is disadvantageous that the components a ₀ , a ₁ , a ₂ , a ₃ and a ₄ influence one another and do not take into account the dead times, ie they are not compensated. Moreover, the actuator characteristics (influence functions) are calculated and assumed to be constant only once per band since iterative model equations are used for the calculation.

Based on the previously described classic flatness regulation Extensions of the classical control concept have already been proposed, partially to fix the existing disadvantages.

A decomposition of the measured flatness in the direction of influence functions, which are not orthogonal to each other, is in Schneider, A .; Kern, P .; Steffens, M .: Model Supported Profile and Flatness Control Systems, Proc. of 49 ° Congresso Internaciona de Tecnologia Metalurgica e de Materials - International Conference, 9-14 October 1994, S ± o Paulo, Vol. 6, p. 49/60 and McDonald, I.R .; Mason, J. D .: Advances in flatness control technology, Proc. of the Conf. on the control of profiles and Flatness, 25-27 March 1996, The Institute of Materials, Birmingham, p. 161/70 described. As a result, improved results can be achieved but with redundant and very similar control variables due to the poor conditioning of the system (poorly invertible Systems) very large manipulated variables. A very heavy duty can result from this.

In Grimble, M.J .; Fotakis, J .: The Design of Strip shape Control Systems for Sendzimir Mills, IEEE Trans. on Automatic Control 27 (1982) no. 3, p. 656/666 and Ringwood, J.V .: Shape Control Systems for Sendzimir Steel Mills, IEEE Trans. On Control Systems Tegnology 8 (2000) no. 1, p. 70/86 becomes a flatness control for Sendzimiergerüste with an orthogonal Decomposition of flatness values proposed in Chebyshev polynomials, in order to improve the flatness regulation, but there will be Dead time compensation and manipulated variable restrictions are not taken into account. The manipulated variables are using a multi-variable controller certainly. The multi-size controller is not dynamic for an online capable Optimization designed.

A planarity control by means of observer and classic state controller is in Hoshino, I .; Kimura, H .: Observer-based multivariable control of Rolling Mills, Preprints of the IFAC Workshop on Automation in Mining, Mineral and Metal Processing, 1-3 September 1998, Cologne, p. 251/256 presented. An extension to nonlinear models and dynamic optimization is in Pu, H .: Nern, H.-J .; Roemer, R .; Nour Eldin, H. A .; Kern, P .; Jelali, M .: State observer design and verification towards developing an integrated flatness-thickness control system for the 20 roll sendzimir cluster mill, proc. Intern. Conf. on Steel Rolling (Steel Rolling '98), 9-11 November 1998, The Iron and Steel Institute of Japan, Chiba, p. 124/29 and Pu, H .; Nern, H.-J .; Nour Eldin, H. A .; Jelali, M .; Totz, O .; Kern, P .: The hardware-in-loop simulation and on-line Tests on an integrated thickness and flatness control system for the 20 rolls sendzimir cold rolling mill, Proc. Intern. Conf. on Modeling of Metal Rolling Processes, 13-15 December 1999, London, p. 208/16 to find. In these solutions, however, no decomposition of the planarity in orthogonal Polynomials made. The dead times are also in these Approaches not compensated.

An improvement of the flatness control by compensation of the dead time Smith predicator is used in Soda, K .; Amanuma, Y .; Tsuchii, K .; Ohno, S .; N .: Improvement in Flatness Control Response for Tandem Cold Strip Mill, Proc. Intern. Conf. on Steel Rolling (Steel Rolling '98), 9-11 November 1998, The Iron and Steel Institute of Japan, Chiba, p. 760/765. The predicator calculates the control variables, which occur in the first sampling step after the dead time has passed and compensates for the dead time. The flatness becomes along the influence functions disassembled. For redundant and very similar manipulated variables occur due to the bad conditioning of the system (bad invertible systems) very large manipulated variables. This allows the System are overstressed. The Smith predicator is a classic one Multi-variable controller (PID controller) integrated. It lacks a dynamic Optimization with prediction beyond the dead time Controlled variable course. Here is a rule predicted, the occurs immediately after the dead time in the first scanning step.

In many conventional flatness measuring systems, measured values become delivered predetermined sampling times. When using discrete-time Controllers, such as PI controllers, it is assumed that the Sampling time is constant. If the sampling time is not kept constant, it deteriorates the control result or the control loop becomes even unstable.

The invention is therefore based on the object to provide a method that measuring and / or regulating the flatness of a band reliable when rolling.

This object is achieved by a method according to claim 1. Advantageous Further developments are the subject of the dependent claims.

The invention is based on the idea, the flatness control in the prior art by an orthogonal model-based Multi-size flatness control system with registration of flatness and their decomposition into orthogonal components to improve and the Stability of the control loop through consideration of measuring systems with time-variant sampling time using an IMC (Internal Model Control) approach is improved with event-triggered scanning elements. Before zugt, the multi-size flatness control system has a determination the manipulated variables by means of a dynamic online optimization under consideration of manipulated variables restrictions on and a prediction of the controlled variables (flatness values), which result in a dynamic optimization is included. The prediction of the controlled variables goes beyond the Dead time out. The model-based predictive approach becomes a Prediction of the controlled variables from the first sampling step to the dead time used up to a prediction horizon. This will become everyone Time optimal manipulated variables calculated, even if the time constants the individual actuators are very different. This information advantageously also participate in dynamic optimization one.

Advantageously, the components can be compared with values which provides an online-capable model of the plant. The resulting Difference can serve as a control variable and then with the in unabgängige Components decomposed Sollplanheitskurve be compared. The resulting control difference can be achieved via optimal decoupling fed to a multi-variable controller.

In the method according to the invention it is particularly advantageous that Dead time taken into account by the Internal Model Control (IMC) approach can be. This can shorten the settling time and the tape length be reduced, which is outside the tolerance range.

By breaking down the flatness measurements into independent components Furthermore, the respective flatness errors can be clearly identified be, whereby a significant improvement in the control quality allows becomes.

Likewise, it is an advantage that by the orthogonal decomposition the flatness of the number and shape of the required manipulated variables determinable become.

The inventive method further allows the consideration the change of rolling force, thermal crowning and the incoming tape properties at each time step by a Feedforward.

In addition, it has the advantage of being an explicit online consideration the nonlinearities of the plant by a parallel to the track online investment model of the plant.

• Erfassung der Planheit des Bandes mit einem Meßsystem,
• Zerlegen der Planheitsfehler (Längenverteilung) in orthogonale Komponenten,
• einem expliziten, linearen oder nichtlinearen online-fähigen Profil- und Planheitsmodell, das alle wesentlichen am Walzprozeß beteiligten Größen (Biegen, Schwenken, Verschieben thermische Bombierung usw.) berücksichtigt;
• einem expliziten online-fähigen Modell, das Sollwerte für die Planheitsregelung berechnet,
• einen Mehrgrößenregler für die Regelung der Planheit des Bandes
• einer Prädiktion der Regelgrößen (Planheitswerte), die in die dynamische Optimierung einbezogen wird, die über die Totzeit hinausgeht,
• einer Störgrößenaufschaltung, die die Eigenschaften des einlaufenden Bandes, die Variation von Walzkraft und thermischer Bombierung berücksichtigt und
• einem ereignisgetriggerten Abtastsystem zur Berücksichtigung von Planheitsmeßsystemen mit variabler Abtastzeit.

In an advantageous embodiment, the multi-size flatness control system according to the invention comprises the following steps:

• Recording the flatness of the strip with a measuring system,
• Decomposing the flatness error (length distribution) into orthogonal components,
• an explicit, linear or non-linear on-line profile and planarity model that takes into account all major factors involved in the rolling process (bending, pivoting, thermal crowning, etc.);
• an explicit online-capable model that calculates setpoints for the flatness control,
• a multi-variable controller for controlling the flatness of the belt
• a prediction of the controlled variables (flatness values), which is included in the dynamic optimization that goes beyond the dead time,
• a feedforward, which takes into account the properties of the incoming strip, the variation of rolling force and thermal crowning and
• an event-triggered scanning system for the consideration of flatness measuring systems with variable sampling time.

The decomposition of the components may advantageously be orthogonal Polynomials, for example with the help of Chebyshev polynomials or Gram polynomials, as described in W.H. Press, S.A. Teukolsky, W.T. Vetterling, B.P. Flannery: Numerical Recipes at C, Cambridge University Press (1992) or A. Ralston, P. Rabinowitz: A first course in numerical analysis, International series in pure applied mathematics, McGraw-Hill (1978).

The flatness of the leaking sheet can be achieved by bending, swiveling and axial displacement of the rolls and by selective multi-zone cooling to be influenced. The individual manipulated variables can help with a multivariable controller from the above-described control difference be determined. In this case, the influence of the rolling force, the incoming Band properties and thermal crowning by a Feedforward control can be compensated.

Fig. 1

ein Schaubild einer Planheitsregelung nach dem Stand der Technik und

Fig. 2

ein Schaubild des erfindungsgemäßen Verfahrens zur modellgestützen prädiktiven Mehrgrößen-Planheitsregelung eines Bandes mit Zerlegung der gemessenen Planheit in orthogonale Komponenten, eine Störgrößenaufschaltung und einer dynamischen Optimierung unter Berücksichtigung von Beschränkungen,

Fig. 3a

Diagramm eines Regelergebnisses bei einem herkömmlichen Regelsystem,

Fig. 3b

Diagramm eines Regelergebnisses bei einem erfindungsgemäßen Regelsystem,

Fig. 4a

Stellgrößenverlaufs-Diagramme bei einem herkömmlichen Regelungssystem,

Fig. 4b

Stellgrößenverlaufs-Diagramm bei einem erfindungsgemäßen Regelungssystem.

The following illustrations show:

Fig. 1

a diagram of a flatness control according to the prior art and

Fig. 2

a diagram of the method according to the invention for model-based predictive multi-size flatness control of a band with decomposition of the measured flatness in orthogonal components, a feedforward control and a dynamic optimization taking into account limitations,

Fig. 3a

Diagram of a control result in a conventional control system,

Fig. 3b

Diagram of a control result in a control system according to the invention,

Fig. 4a

Manipulated variable graphs in a conventional control system,

Fig. 4b

Stellgrößenverlaufs diagram in a control system according to the invention.

As shown in Figure 2, the flatness deviation by means of a measuring system determined and then into orthogonal (independent) components disassembled. The components are compared with values that a online-enabled model of the plant supplies. The resulting difference is called Controlled variable used. This is then used in independent Components decomposed Sollplanheitskurve and the resulting Control difference is a multi-variable controller, consisting of an online-enabled Model and a dynamic optimization involving Manipulated variable restrictions and predicted control variable course, fed. To take account of flatness measuring systems with variable Sampling time is an event-triggered sampling system with one with two Abtasthaltegliedem cooperating event generator provided.

The flatness of the leaking sheet is determined by bending, swiveling and Axial displacement of the rollers as well as through selective multi-zone cooling affected. The individual manipulated variables are calculated using a multivariable controller determined from the above-described control difference. It is the Influence of rolling force, incoming strip properties and thermal Camber compensated by a feedforward control.

Exemplary are the advantages of the new concept compared to the stand The technique is illustrated by simulations in Figures 3a, 3b, 4a and 4b. A model of a Sendzimiergerüstes is used with strong different time constants in the actuators. It will be a flatness error by incorrect counter-displacement of the cone rollers accepted. The new concept regulates the flatness error after about 30 m Band length from (see Figure 3b), while in the current concept a Residual flatness error of 10 lUnits remains (see Figure 3a). This residual error disappears after about 300m tape length. The reason for this is in that with the current concept no dynamic optimization with Consideration of beyond dead time and predicted Controlled variables is used.

On the course of the manipulated variables in Figure 4b can be seen that the inventive multi-size flatness system first bending or Panning responds and then the slow cone rollers nachfährt to correct the flatness errors and thus at any time optimal manipulated variables determined. The current concept (see Figure 4a) creates it does not address the cone rollers at the necessary speed and thus to correct the flatness error.

Claims (10)

1. Method for measuring and/or controlling the flatness of a band during rolling comprising the following steps:

   detecting measured values with a measuring system having time-varied sampling time;

   breaking down of the measured values into independent components and

   control of correcting variables, whereby to take account of the measuring system with time-varied sampling time, an internal model control approach with event-triggered sample-and-hold members is used.
2. Method according to claim 1, characterised by the breaking down of the measured values with the aid of orthogonal polynomials.
3. Method according to one of the claims 1 or 2, characterised in that the independent components are compared with target values from a target value flatness model.
4. Method according to claim 3, characterised by the determination of a deviation value between target values and independent components.
5. Method according to one of the claims 1 to 4, characterised by a disturbance feedforward.
6. Method according to one of the claims 1 to 5, characterised by the use of a multivariable control system.
7. Method according to claim 6, characterised in that the deviation value is passed via a decoupler to the multivariable control system.
8. Method according to one of the claims 1 to 7, characterised in that the correcting variables are determined by means of dynamic optimisation taking account of limitations.
9. Method according to claim 8, characterised in that a prediction of the controlled variables extending beyond the dead time is included in the dynamic optimisation.
10. Method according to claim 1, characterised by the use:

    of an explicit online-capable profile and flatness model which takes account of all the variables involved in the rolling process,

    and of an explicit online-capable model which calculates desired values for the flatness control,

    and of a disturbance feedforward which takes account of the characteristics of the band being fed in, variations in rolling force and thermal bowing.

Images