EP2452233A1

EP2452233A1 - Control method for an influencing device for a rolled product

Info

Publication number: EP2452233A1
Application number: EP10729838A
Authority: EP
Inventors: Klaus Weinzierl; Hans-Ulrich LÖFFLER
Original assignee: Siemens AG
Current assignee: Primetals Technologies Germany GmbH
Priority date: 2009-07-08
Filing date: 2010-06-28
Publication date: 2012-05-16
Also published as: EP2280323A1; CN102473003B; WO2011003765A1; CN102473003A

Abstract

The invention relates to a control device (3) actuating an influencing device (1), so that said influencing device correspondingly influences the rolled product (2). The control device (3) captures at least one measurement variable (F, M, T) in connection with the influencing of the rolled product (2) in real time, said variable depending on a first property of the rolled product (2) at the time of capturing the measurement variable (F, M, T). The control device derives a corresponding model parameter (F', M', T') by means of a model (11) of the rolled product (2) describing the processes in the influencing device (1). The model (11) comprises a first partial model (12) based on mathematical and physical equations, by means of which the value of the first property can be derived. The model comprises a second partial model (13) based on mathematical and physical equations, by means of which the time progression of a second property of the rolled product (2) can be derived during the influencing. The value of the first property of the rolled product (2) at the time of capturing the measurement variable (F, M, T) depends on the time curve of the second property, or on the value of the second property of the rolled product at the time of capturing the measurement variable (F, M, T) and/or the measurement variable (F, M, T) also depends on the value of the second property of the rolled product (2) at the point in time of capturing the measurement variable (F, M, T) in addition to depending on the value of the first property. The control device (3) derives the time curve of the second property of the rolled product (2) and the value of the first property of the rolled product (2) at the time of capturing the measurement variable (F, M, T) by means of the model (11), and derives the model parameter (F', M', T') using the derived time curve of the second property of the rolled product (2) and the value of the first property of the rolled product (2) at the time of capturing the measurement variable (F, M, T). The control device adapts the second partial model (13) using the deviation of the captured measurement variable (F, M, T) from the model parameter (F', M', T'), but not the first partial model, or adapts both the first partial model (12) and the second partial model (13) by optimizing a cost function (K1), wherein, however, a penalty term (K2) depending on the adaption of the first partial model (12) as such is included in the cost function (K) in addition to the deviation of the measurement variable (F, M, T) from the model parameter (F', M', T').

Description

description

Control method for an influencing device for a rolling stock

The present invention relates to a control method for an influencing device for a rolling stock,

wherein a control device controls the influencing device with control commands, so that the influencing device influences the rolling stock in accordance with the control commands,

wherein the control device, in conjunction with the influencing of the rolling stock by the influencing device, detects in real time at least one measured variable that depends on the value of a first property of the rolling stock at the time of the detection of the measured variable,

wherein the control device determines a model variable corresponding to the detected measured variable by means of a model of the rolling stock which describes the processes in the influencing device,

- wherein the model comprises a first sub-model based on mathematical-physical equations, by means of which the value of the first property of the rolling stock can be determined at the time of the acquisition of the measured variable, - the model comprising a second submodel based on mathematical-physical equations whose temporal evolution of a second property of the rolling stock can be determined during influencing,

- Where the value of the first property of the rolling stock to

Time of detection of the measured variable from the time course of the second property or the value of the second property of the rolling stock at the time of detecting the measured variable depends and / or the measured variable in addition to the dependence of the value of the first property and the value of the second property of the rolling stock at the time Depending on the measured variable,

- The control device by means of the model, the time course of the second property of the rolling stock, the value the first property of the rolling stock at the time of detecting the measured variable and using the determined time characteristic of the second characteristic of the rolling stock and the value of the first characteristic of the rolling stock at the time of detecting the measured variable determines the model size.

The present invention further relates to a control program which comprises machine code which can be processed directly by a control device for an influencing device for a rolling stock and whose execution by the control device causes the control device to carry out such a control method. Furthermore, the present invention relates to a control device for an influencing device for a rolling stock, wherein the control device is designed such that it executes such a control method during operation. Finally, the present invention relates to an influencing device for a rolling stock, which is controlled by such a control device.

In the production of rolling stock, for example steel in a hot strip mill, models are used to predict in particular material properties such as microstructure, phase components and mechanical properties. The models can be used, for example, to determine the activation commands for the hot strip line online or offline. Alternatively, a pure prediction of the properties is possible.

The models should be as accurate as possible. It is therefore necessary to check and adapt the models after installation. The adaptation takes place on the basis of measured variables, which are compared with corresponding model sizes. For some models, the appropriate measures can be easily determined. For other models it is hard and partial not known at all in the prior art to determine the corresponding measured variables online.

A control method of the aforementioned type is known, for example, from DE 10 2007 025 447 A1. In the known method, the first submodel is a heat transfer model, the second submodel is a phase transformation model. The behavior of a steel volume in a cooling section is modeled. There is a coupled determination of possible adaptation factors for the first submodel and the second one

Partial model. Within the scope of the known method, possible value pairs for the adaptation factors are stored. If another rolling stock is later influenced, possible value pairs are determined again. By comparing the value pairs determined in the two influencing processes, the amount of permissible value pairs can be reduced. Ideally, a single value pair can be determined in this way, so that therefore the adaptation factors for the first and the second partial model can be determined unambiguously. In many cases, however, there remains an uncertainty as to what the physically correct values of the adaptation factors actually are. In particular, in many cases there remains an uncertainty in that an error of the first partial model is compensated by an oppositely directed, but equally large error of the second partial model.

The object of the present invention is to provide possibilities by means of which the adaptation factor for the second partial model can be correctly determined.

The object is achieved by a control method having the features of claim 1. Advantageous embodiments of the control method according to the invention are the subject matter of dependent claims 2 to 10. According to the invention, a control method of the type mentioned by the fact that the control device

either the second partial model, but not the first partial model, is adapted on the basis of the deviation of the detected measured variable from the model size,

or by optimizing a cost function, both the first submodel and the second submodel are adapted, but in the cost function, in addition to the deviation of the measured variable from the model size, a constraint dependent on the adaptation of the first submodel as such is received.

The procedure according to the invention only apparently leads to a simple shifting of the uncertainty in the adaptation of the first partial model. Because in many cases, due to other circumstances, so based on additional information on the influencing device continuous rolling, which are different from the second property to decide whether the second property of the rolling stock when detecting the at least one measured variable has a predetermined value. It is therefore possible that in this case the control device exclusively adapts the first, but not the second, submodel based on the deviation of the measured variable from the model size and adapts the second submodel only if it can not be decided whether the second property of the rolling stock when detecting the at least one measured variable has the predetermined value.

It is possible to carry out the alternative adaptation of the first and the second submodel as described last, depending on the position of the individual case ("as it is"), but it is preferred that the adaptation of the first submodel be carried out in advance can alternatively be done online or offline.

Regardless of whether the first partial model is adapted or not, it is preferred that the adaptation of the second partial model takes place online. The control method according to the invention can be used in various embodiments. For example, it is possible for the first submodel to comprise a heat transfer model by means of which the heat transfer of the rolling stock to its surroundings is modeled and which as the first property supplies the temperature of the rolling stock. Alternatively or additionally, the first part model may comprise a rolling model, by means of which rolling forces and / or rolling moments occurring during rolling of the rolling stock are modeled and which as a first property delivers the rolling forces and / or rolling moments occurring during rolling of the rolling stock. As a further output variable, the rolling model can deliver, for example, a forming energy which is introduced into the rolling stock during the respective rolling process. The second submodel may, for example, comprise a phase transformation model by means of which the phase transformation of the rolling stock is modeled and which supplies the phase components of the rolling stock as a second property. Particularly good results have been provided by phase transformation models in which the phase transformation is modeled using Gibbs free enthalpy. Alternatively or additionally, the second partial model may comprise a structural model, by means of which the recrystallization of the rolling stock is modeled and which supplies the microstructure of the rolling stock as a second property.

The model size and, correspondingly, the measured variable can be selected as required. It is at least different from the second property. It may also be different from the first property. In many cases, however, it will correspond to the first property. This will be discussed later.

Due to the fact that the detected measured variable is available in real time, it is possible to adapt the second partial model online. This procedure is preferred. In particular (but not exclusively) in the case of an adaptation of the second partial model is online preferably that the control device, an initial state of the first rolling stock and a desired final state of the first rolling stock are predetermined and that the control device determines the drive commands using the model.

The object is further achieved by a control program of the type mentioned, whose execution by the control device causes the control device executes an inventive control method.

The control program is usually stored on a data carrier in machine-readable form. The data carrier can be part of the control device. The object is further achieved by a control device of the type mentioned, which is designed such that it executes an inventive control method during operation. Finally, the object is achieved by an influencing device for a rolling stock, which is controlled by a control device according to the invention.

Further advantages and details emerge from the following description of exemplary embodiments in conjunction with the drawings. In a schematic representation:

1 shows schematically an influencing device for a rolling stock,

2 and 3 possible embodiments of the influencing device of FIG. 1,

4 shows a flow chart,

5 shows a control block diagram and FIGS. 6 to 9 are flowcharts.

According to FIG. 1, an influencing device 1 for a rolling stock 2 is controlled by a control device 3. The influencing device 1 is passed through by the rolling stock 2. In Connection with the passage of the rolling stock 2 through the influencing device 1 influences the influencing device 1 the rolling stock 2. For example, the influencing device 1 according to FIG. 2 can be designed as a rolling train. In this case, the influencing device 1 has at least one rolling stand 4, in which the rolling stock 2 is rolled. In most cases, according to the representation of FIG. 2, a plurality of rolling stands 4 are present, in which the rolling stock 2 is successively rolled.

Alternatively, the influencing device 1 according to FIG. 3 can be designed as a cooling section for the rolling stock 2. In this case, the rolling stock 2 is cooled in the cooling section. Most is adjusted by appropriate control valves 5 quantitatively, with what amount of a coolant 6 (for example, water), the rolling stock 2 is acted upon.

The control device 3 is designed such that it executes a control method for the influencing device 1 during operation. In general, the control device 3 is programmed for this purpose with a control program 7, which is stored in a data carrier 8, which is part of the control device 3. For example, the data carrier 8 may be designed as a hard disk of the control device 3. The storage in the disk 8 is of course in machine-readable form.

According to FIG. 1, the control program 7 comprises machine code 9, which can be processed directly by the control device 3. The execution of the machine code 9 by the controller 3 causes the controller 3 to execute the control process. The control program 7 may have been supplied to the control device 3 in any desired manner. By way of example only, a mobile data carrier 10 is shown in FIG 1, on which the control program 7 is stored in machine-readable form. The mobile data carrier 10 is formed according to the illustration of FIG 1 as a USB memory stick. However, any other configurations are also possible, for example as a CD-ROM or SD memory card.

The control method executed by the control device 3 will be explained in more detail below in conjunction with FIG. Partially in this context, reference is also made to FIGS. 1 to 3 and to FIG. 5.

According to FIG. 4, the control device 3 controls the influencing device with control commands A in a step S1. It is thereby achieved that the influencing device 1 influences the rolling stock 2 in accordance with the control commands A.

In the case of the formation of the influencing device 1 as a rolling train according to FIG. 2, the control commands A cause, for example, that the individual rolling stands 4 of the rolling train apply the rolling stock 2 with corresponding rolling forces F and rolling moments M. In the case of the formation of the influencing device 1 as a cooling section according to FIG. 3, the control device 3 can control, for example, the individual control valves 5 of the cooling section such that they apply a defined time and local quantity of coolant flow to the rolling stock 2. A combined approach is also possible. This is possible in particular when the coolant 6 is applied - at least partially - between the rolling stands 4 to the rolling stock 2. The corresponding procedure is known as such in the prior art.

By way of example, reference is made to US Pat. No. 7,310,981 B2. It is not as such subject of the present invention.

In connection with the influence of the rolling stock 2 by the influencing device 1, the control device 3 detects in a step S2 in real time at least one measured variable F, M, T. The measured variables F, M, T will be discussed in more detail later. By way of example, as possible measuring large F, M, T but for the rolling mill, the rolling forces F, the rolling moment M and the temperature T mentioned, for the cooling section, the temperature T of the rolling stock 2 when leaving the cooling section.

The control device 3 has according to FIG 5 internally a model 11 of the rolling stock 2. The model 11 describes the processes in the influencing device 1. The processes may include internal processes in the rolling stock 2 and / or interactions of the rolling stock 2 with the influencing device 1. In particular, the model 11 according to FIG. 5 comprises a first and a second submodel 12, 13. The submodels 12, 13 are based on mathematical-physical equations, in particular on algebraic and / or differential equations.

Using the first partial model 12, the control device 3 can determine the value of a first property of the rolling stock 2 at the time of the acquisition of the measured variable F, M, T. The term "using" is to be understood in a broad sense: it means that the first submodel 12 is at least used inter alia for determining the value of the first property, but it should not be ruled out that the first property can be determined In particular, it should not be ruled out that the second submodel 13 is used to determine the corresponding value in addition to the first submodel 12. Furthermore, it may be possible to use the first submodel 12 (the Term "under use" has just been defined) a temporal course of the first property is determined. Various examples:

It is possible for the first partial model 12 to model the heat conduction in the rolling stock 2, including the heat transfer to the surroundings. In this case, the first partial model 12 is a

Heat transfer model. The first property is the temperature T of the rolling stock 2. In the case of the embodiment of the first part model 12 as a heat transfer model, the first is based Submodel 12 on the Fourier equation of heat conduction and another equation describing the heat transfer to the environment. In order to determine the temperature T of the rolling stock 2 which is expected according to the first partial model 12, the heat equation and the heat transfer equation must be solved iteratively in small time steps. In this case, it is therefore necessary to determine the time course of the temperature T, even if later only a single temperature T is needed, namely the time of detection of the measured variable F, M, T.

If the temperature T of the rolling stock 2 at the time of detection of the measured variable F, M, T falls below the transition temperature of the rolling stock 2, phase transitions and the energy released thereby must be taken into account in the correct time when determining the temperature T. It may therefore be necessary, in addition to the first submodel 12, to take into account at least one further submodel 13, namely a phase conversion model by means of which the phase transformations of the rolling stock 2 are modeled. Such a partial model 13 supplies as output variables the phase components of the rolling stock 2, which are needed to determine the energy released in the phase transformation. Alternatively, the phase transformation model may determine the changes in phase contributions. It may be a second submodel 13 in the sense of the present invention. This will be discussed later.

Alternatively, the first partial model 12 may be, for example, a rolling model, by means of which rolling forces F and / or rolling moments M occurring during rolling of the rolled stock 2 are modeled and the rolling stock forces F and / or rolling moments M occurring during rolling of the rolled stock 2 are the first characteristic. In this case, no determination of a time course of the first property is required. However, the rolling model 12 may be coupled to a heat transfer model, a phase transformation model and / or a structural model. The heat transfer model and the phase model were already explained. A microstructure model is a model by means of which the recrystallization of the rolling stock 2 is modeled and which as a second property provides the microstructure of the rolled stock 2 and a concomitant material hardening. The heat transfer model, the phase transformation model and / or the structural model may be second partial models 13 in the sense of the present invention. This will also be discussed later. The temporal development of a second property of the rolling stock 2 during the influencing can be determined by means of the second partial model 13. The term "temporal development" means that, starting from the respective instantaneous state of the rolling stock 2, the temporal course of the second property of the rolling stock 2 is ascertained in a small amount, the second submodel 13 being based, at least in part, on differential equations. For example, if the first submodel 12 is a rolling model, the second submodel 13 may also be a heat transfer model.

According to FIG. 4, the control device 3 determines in one

Step S3 by means of the model 11 in the result

the time profile of the second property of the rolling stock 2,

an expected value of the first property of the rolling stock 2 at the time of detecting the measured variable F, M, T and

- Using the determined time course of the second property of the rolling stock 2 and the expected value of the first property of the rolling stock 2 at the time of detecting the measured variable F, M, T, a model size F ', M', T '.

As already mentioned, it may be possible for the individual investigations to take place successively but separately from each other. In many cases, however, the investigations coupled. For example, when phase transformations occur, a heat transfer model (first submodel 12) and a phase transformation model (second submodel 13) are coupled together such that the temporal histories of temperature T and phase transformation are only detectable coupled. In other cases, it may be possible to first isolate the time history of the second property and then determine the value of the first property and

Finally, on the basis of the determined time course of the second property and the value of the first property to determine the model size F ', M', T '.

The model size F ', M', T 'may correspond to the first property. For example, the first sub-model 12 may be a rolling model and the model size F ', M' may correspond to the expected rolling force F 'or the expected rolling torque M' of a particular rolling operation. Likewise, the first part model 12 may be a heat transfer model and the model size T 'correspond to the expected temperature T' of the rolled stock 2 when leaving the influencing device 1. Regardless of whether the model size F ', M', T 'determined using the model 11 corresponds to the first property or not, the model size F', M ', T' corresponds to the measured quantity F, M, T. The model size F ', M', T 'is the quantity that is expected as the measured variable F, M, T on the basis of the model-based determination.

The measured variable F, M, T and the corresponding model size F ', M', T 'and the first and the second property can in principle be chosen arbitrarily, provided that the following conditions are met:

- The measured variable F, M, T must depend on the value of the first property of the rolling stock 2 at the time of detection of the measured variable F, M, T. If the model size F ', M', T 'corresponds to the first property, this is trivially the case. But there are also other cases possible. For example, the rolling force F and the rolling moment M (= possible measured variables) temperature-dependent (= possible first property), but not identical with the temperature T. - The measured variable F, M, T must be in addition to the dependency of

Value of the first property also depend on the value of the second property of the rolling stock 2. Thus, for example, the flow curve (= core property of a rolling model) is not only temperature-dependent (= possible first property), but also phase- and assembly-dependent (= possible second properties).

The value of the first property at the time of the acquisition of the measurand F, M, T must depend on the value of the second property or on its time course. Thus, for example, temperature development and phase transformation mutually influence one another so that the two time courses can only be determined together, but not independently of one another. Also, the rolling force F (possible first property) depends on the recrystallization of the rolling stock 2 (possible second property). The first of the three conditions mentioned above must always be met. The second and third of said conditions may be met alternatively or cumulatively.

According to FIG. 4, the control device 3 determines in one

Step S4 the deviation of the detected measured variable F, M, T from the corresponding model size F ', M', T '. Furthermore, in a step S5, the control device 3 uses the deviation to determine an adaptation factor k2 for the second submodel 13. It therefore adapts the second submodel 13. The adaptation factor k2 is referred to below as the second adaptation factor because the second submodel 13 is adapted therewith.

Starting from step S5, the control device 3 returns to step S1. In the later, renewed execution of step S3, the just made adaptation of the second partial model 13 is taken into account. The first partial model 12 is not adapted at all by the control device 3 as part of the step S5. For this procedure to lead to correct results, it must be ensured that the first submodel 12 is correct in terms of model accuracy. Below is shown in conjunction with FIG 6, how this can be ensured.

FIG. 6 is based on the procedure of FIG. 4. In addition to steps S 1 to S 5 of FIG. 4, steps S 6 to S 8 are present.

In step S6, additional information I is evaluated via the rolling stock 2 passing through the influencing device 1. The additional information I is different at least from the second property of the rolling stock 2. They may also be different from the first property of the rolling stock 2. Alternatively, you can match the first property. On the basis of the additional information I it is decided whether the second property of the rolling stock 2 when detecting the measured variable F, M, T has a predetermined value. Depending on whether this condition is met or not, the logical value OK (or TRUE or

FALSE).

In step S7, the value of the logical variable OK is checked. Depending on the result of the check, either step S5 or step S8 is executed. The step S5 has already been explained. In step S8, the control device 3 exclusively adapts the first partial model 12, but not the second partial model 13. Thus, in step S8 it determines an adaptation factor k1 (hereinafter referred to as the first adaptation factor k1) analogously to step S5 first partial model 12 is adapted. It is possible that, as shown in FIG. 6, the steps S6 and S7 are executed automatically by the control device 3. Alternatively, it is possible that the controller 3 according to FIG 1 by an operator 14 a corresponding decision E is specified.

The procedure of FIG. 6, that is to say the alternative adaptation of the first partial model 12 or of the second partial model 13, will be explained in more detail below on the basis of examples.

First example:

The rolling stock 2 should be steel. It is rolled in a rolling train (compare purely by way of example FIG. 2). The first sub-model 12 is a rolling model, the second sub-model 13 is a phase transformation model. The temperature of the rolling stock 2 when entering the first rolling stand 4 of the rolling train is known. The temperature behind the last roll stand 4 of the rolling train is detected and is therefore also known. For example, due to the detected temperatures, it may be known that the entire rolling process takes place in the austenitic region. It can therefore be assumed "in good conscience" that no phase transformations take place during the entire rolling process, so that the rolling stock 2 is always completely present in the "austenite" phase. Therefore, in step S8, based on the detected rolling forces F and / or rolling moments M (= simultaneously first properties and measured variables) and the corresponding model sizes F ', M', the rolling model (= first submodel 12) can be adapted, since within the currently considered rolling stock 2 the phase transformation model (= second partial model 13) does not come into play. Otherwise, that is, when the temperature of the rolling stock 2 is below the transformation temperature of steel, the phase conversion model is adapted based on the deviation in step S5.

Second example: As before, with the difference that the rolling model as such is correct, but additionally comprises as first partial model 12 a structural model. In this case, the texture model corresponds to the first sub-model 12. The first property, that is, the microstructure, is different in this case from the model size F ', M'. When the rolled stock 2 is rolled in the austenitic region, the texture model can be adapted. When ferritic rolling occurs, the phase transformation model is adapted.

In the event that the rolling model, the phase transformation model and the structural model are present, it is even possible to adapt all three models (rolling model, structural model and phase transformation model) using the same measured variable F, M. Because the recrystallization of the rolling stock 2 is a function of the phase components, the temperature and the time between two rolling operations. If the rolled material 2 is rolled in the austenitic region and slowly, it can be assumed "with good conscience" that the rolled material 2 is in the austenite phase and the recrystallization is complete For a slow rolling in the ferritic range, the phase transformation model (= one of the second submodels 13) can be adapted Partial model 13) are adapted, for example, in the case of rapid rolling in the ferritic region, no adaptation takes place.

Third example: The rolling stock 2 - for example steel - is rolled in a rolling train - for example, the rolling mill of FIG 2 -. The first partial model 12 is a rolling model, the second partial model 13 is a structural model. The temperature T of the rolling stock 2 when entering the first rolling stand 4 is known. The approximate cooling rate of the rolled stock 2 is known empirically. Furthermore, the distances of the rolling stands 4 from each other and the mass flow are known. The result is thus the time that elapses between the individual rolling passes. If the mass flow is low enough and the temperature T of the rolling stock 2 is high enough, it can be assumed on the basis of empirical data that the recrystallization of the rolled stock 2 has already "ceased" long when the rolled stock 2 enters a specific rolling stand 4 is. Thus, it can be assumed "with good conscience" that the rolling stock 2 enters the respective rolling stand 4 in a completely recrystallized manner. Therefore, in step S8, based on the detected rolling forces F and / or rolling moments M (= simultaneous measured quantities and first properties) Walzmodell (= first part model 12) are adapted, as in the context of the rolling pass under consideration the structural model (= second sub-model 13) does not come to fruition. Conversely, if a rapid rolling of the rolled material 2 takes place, the structural model can be adapted.

Fourth example:

The rolling stock 2 should be steel again. It is cooled in a cooling path - for example, the cooling section of FIG 3. The first submodel 12 is a heat transfer model, the second submodel 13 is a phase change model. The temperature of the rolling stock 2 when entering the cooling section is known or is detected. The temperature T of the rolling stock 2 when the rolled stock 2 leaves the cooling section is detected. It is usually below the transformation temperature of the rolling stock 2. On the basis of the alloy properties of the rolling stock 2 and the residence time of the rolling stock 2 in the cooling section can be decided whether the rolling stock 2 is completely converted in the course of cooling in the cooling section (Austentitanteil = zero ). If this is the case, the heat transfer model (= first partial model 12) can be adapted in step S8 since the phase conversion model (= second sub-model 13) does not come into play in the context of the considered rolling stock 2. Otherwise, the phase transformation model can be adapted. Combinations of the above procedures are also possible. For example, the model 11 may comprise all four of said sub-models 12, 13, ie a rolling model, a structural model, a phase transformation model and a heat transfer model. The measured variables F, M, T are, in particular, the rolling forces F and the rolling moments M of the rolling stands 4 and the temperature T of the rolling stock 2 on the output side of the rolling train in question. If, in such a case, the rolling stock 2 is rolled slowly in the austenitic region, it is possible to adapt the rolling force model and the heat transfer model based on the resulting rolling forces F and / or rolling moments M and the resulting temperatures T. These two models are in this case first partial models 12 in the sense of the present invention. Because of the slow rolling, the structural model (one of the second partial models 13) does not come into play, due to austenitic rolling, not the phase transformation model (further second partial model 13). If, on the other hand, a faster rolling takes place in the austenitic region, the structural model can be adapted on the basis of the resulting changed rolling forces F and / or rolling moments M. In a slow ferritic rolling, the phase transformation model can be adapted on the basis of the resulting rolling forces F, rolling moments M and temperatures T. In the case of fast ferritic rolling, the phase transformation model and the structural model can be adapted, the former preferably based on the detected temperature, the latter preferably on the basis of the detected rolling forces F and rolling moments M.

It can be seen from the above explanations that it is advantageous according to the representation of FIG. 7 if the adaptation of the first partial model 12 takes place in advance. The adaptation of the first partial model 12, since it is done in advance, alternatively be done online or offline. The second submodel 13, on the other hand, should, as shown in FIG 7 preferably be adapted online. The term "online" means that the evaluation of the measured values F, M, T recorded in real time takes place in parallel to the current control of the influencing device 1 by the control device 3. In the case of an offline evaluation, however, the evaluation is decoupled from the current operation.

As far explained so far, only the second partial model 13 is adapted when adapting the second partial model 13. The first submodel 12 is used in conjunction with the

Adapting the second partial model 13 not adapted at all. However, it is also possible to make a slight adaptation of the first partial model 12. In this case, step S5 of FIGS. 4 and 6 is replaced by steps S9 and S10, see FIG. 8. In step S9, control device 3 determines a cost function K. In the cost function K, the deviation of the measured variable F, M, T of the model size F ', M', T '. In addition, the cost function K receives a penalty, which depends on the adaptation of the first partial model 12 as such.

The cost function K thus consists of two sub-functions Kl, K2, where - the subfunction Kl depends on the amount of deviation of the measured variable F, M, T from the corresponding model size F ', M', T 'and

- The second sub-function K2 depends on the amount of change of the first adaptation factor kl as such.

As a rule, the subfunctions K1, K2 are the squares or a higher power of the corresponding values. In step S10, the control device 3 determines those adaptation factors k1, k2 at which the cost function K assumes its minimum value. These values are adopted as corresponding adaptation factors kl, k2. For the utilization of the first and second properties of the rolling stock 2 determined by means of the model 11, various case designs are possible. For example, it is possible to determine the drive commands A independently of the model 11. In this case, the model 11 is trained only for eventual later control intervention and / or used to predict the properties of the rolling stock 2. It is better, however, if the model 11 is involved in the determination of the drive commands A. In this case, the corresponding first and second properties of the rolling stock 2 are determined by means of the model 11 for a plurality of possible control commands A. Those control commands A for which the resulting properties of the rolling stock 2 correspond as well as possible to desired desired properties of the rolling stock 2 are used to control the influencing device 1. FIG. 9 shows in simplified form a possible procedure.

According to FIG. 9, the control device 3 assumes an initial state Z of the rolling stock 2, for example in a step S11. The initial state Z describes the rolling stock 2 before influencing by the influencing device 1. In a step S 12, the control device 3 accepts a desired final state Z * of the rolling stock 2. The final state Z * refers to the state that the rolling stock 2 should have after influencing the influencing device 1.

The final state Z * includes, at least among other properties of the rolling stock 2, which can be determined by means of the rolling model 11 and / or the additional model 12. In a step S13, the control device 3 sets the control commands A-even if only for the time being. In a step S 14, the control device 3 determines, using the rolling model 11 and the additional model 12, expected properties of the rolling stock 2 after influencing the influencing device 1.

In a step S15, the controller 3 determines the logical variable OK. The logical variable OK takes the value "TRUE" then and only then, if the expected properties determined in step S14 - within permissible tolerances - coincide with the desired properties accepted in step S12 and determined by the end state Z *, otherwise the logical variable OK takes the value Value "FALSE".

In a step S16, the value of the logical variable OK is checked. If the logical variable OK is "FALSE", the controller 3 proceeds to a step S17, in step S17 the controller 3 varies the drive commands A. Then, it returns to the step S14, whereas if the logical variable OK is " TRUE ", the influencing device 1 is actuated with the now final drive commands A. The corresponding

Step 9 is deliberately provided with the reference symbol S1 in FIG. 9, since it also corresponds to the step S1 of FIGS. 4 and 6. The present invention has many advantages. In particular, it is comparable in complexity to the prior art approaches but shows superior results. The above description is only for explanation of the present invention. The scope of the present invention, however, is intended to be determined solely by the appended claims.

Claims

claims

1. Control method for an influencing device (1) for a rolling stock (2),

- wherein a control device (3) controls the influencing device (1) with control commands (A), so that the influencing device (1) influences the rolling stock (2) in accordance with the control commands (A),

- wherein the control device (3) in conjunction with the influencing of the rolling stock (2) by the influencing device (1) in real time at least one measured variable (F, M,

T), which depends on the value of a first property of the rolling stock (2) at the time of detecting the measured variable (F, M, T),

- wherein the control device (3) by means of a model of the rolling stock (2) describing the processes in the influencing device (1) has a model size (F ', M', T ') which corresponds to the detected measured variable (F, M, T) ),

wherein the model (11) comprises a first submodel (12) based on mathematical-physical equations, by the use of which the value of the first property of the rolling stock (2) can be determined at the time of the acquisition of the measured variable (F, M, T),

- wherein the model (11) comprises a mathematical-physical equations based second sub-model (13) by means of which the temporal evolution of a second property of the rolling stock (2) during the influencing (1) can be determined,

- The value of the first property of the rolling stock (2) at the time of detecting the measured variable (F, M, T) from the time course of the second property or the value of the second property of the rolling stock (2) at the time of detecting the measured variable (F , M, T) and / or the measured variable (F, M, T) in addition to the dependence of the value of the first property on the value of the second property of the rolling stock (2) at the time of detecting the measured variable (F, M, T) depends - The control device (3) by means of the model (11) the time course of the second property of the rolling stock

(2), the value of the first property of the rolling stock (2) at the time of detecting the measured quantity (F, M, T) and using the determined time characteristic of the second property of the rolling stock (2) and the value of the first property of the rolling stock (2) determines the model size (F ', M', T ') at the time the measured variable (F, M, T) is detected,

- wherein the control device (3)

- Based on the deviation of the detected measured variable (F, M, T) of the model size (F ', M', T '), although the second sub-model (13), but not the first sub-model

(12) adapted,

or by optimizing a cost function (K), both the first partial model (12) and the second partial model

(13) adapts, in the cost function (K), in addition to the deviation of the measurand (F, M, T) from the model size (F ', M', T '), one of the adaptation of the first submodel (12) as such Strafterm (K2) enters.

2. Control method according to claim 1,

characterized,

in that on the basis of additional information (I) about the rolling device (2) passing through the influencing device (1), which is different from the second characteristic, it is decided whether the second characteristic of the rolling stock (2) is involved in detecting the at least one measured quantity ( F, M, T) has a predetermined value,

- That the control device (3) in the case of

Deviation of the measured variable (F, M, T) of the model size (F ', M', T ') only the first, but not the second partial model (12, 13) adapted and

- That the control device (3) makes the adaptation of the second partial model (13) only in the negative.

3. Control method according to claim 2,

characterized in that the adaptation of the first partial model (12) takes place in advance.

4. Control method according to claim 1, 2 or 3,

characterized in that the adaptation of the second partial model (13) takes place online.

5. Control method according to one of claims 1 to 4,

characterized in that the first sub-model (12) comprises a heat transfer model, by means of which the heat transfer of the rolling stock (2) is modeled to its environment and as the first property, the temperature (T ') of the rolling stock (2) supplies.

6. Control method according to one of claims 1 to 4,

characterized in that the first part model (12) comprises a rolling model, by means of which during rolling of the rolling stock

(2) occurring rolling forces (F ') and / or rolling moments (M') are modeled and the first property of the rolling of the rolling stock (2) occurring rolling forces (F ') and / or rolling moments (M') supplies.

7. Control method according to one of claims 1 to 6,

characterized in that the second partial model (13) a

Phase transformation model comprises, by means of which the phase transformation of the rolling stock (2) is modeled and which provides as a second property, the phase components of the rolling stock (2).

8. Control method according to one of claims 1 to 6,

characterized in that the second partial model (13) comprises a structural model, by means of which the recrystallization of the rolling stock (2) is modeled and which supplies the microstructure of the rolling stock (2) as a second property.

9. Control method according to one of the above claims,

characterized in that the results obtained from the model (11) average model size (F ', M', T ') corresponds to the first property.

10. Control method according to one of the above claims,

characterized in that the control device (3) an initial state (Z) of the first rolling stock (2) and a desired final state (Z *) of the first rolling stock (2) are predetermined and that the control device (3) the drive commands (A) using of the model (11).

11. Control program comprising machine code (9) which can be processed directly by a control device (3) for an influencing device (1) for a rolling stock (2) and whose execution by the control device (3) causes the control device (3) a control method according to one of the above claims performs.

12. Control program according to claim 11,

characterized in that it is stored on a data carrier (8, 10) in machine-readable form.

13. Control program according to claim 12,

characterized in that the data carrier (8) is part of the control device.

14. Control device for an influencing device (1) for a rolling stock (2), characterized in that the control device is designed such that it carries out a control method according to one of claims 1 to 10 during operation.

15. influencing device for a rolling stock (2),

characterized in that it is controlled by a control device (3) according to claim 14.