EP0819033B1 - Device for early detection of run-out in continuous casting - Google Patents

Description

During continuous casting, the continuous shell can be used during the Growth occurs in the mold places where the Strand shell not or only insufficiently hardened. This Growth defects lead as soon as the strand leaves the mold, to a breakthrough in the strand, through which liquid Steel emerges. The resulting damage to the Casting plant forces a longer downtime and causes high repair costs. So you try Growth defects in the shell before it leaves the mold to recognize. If this succeeds, the exit speed will be so reduced that the potential breakthrough can harden.

Possible breakthrough points are based on the surface temperature profiles found by the in the mold in Temperature sensors attached to the inner wall of the mold be measured. It is known the temperature sensors offset in one or more in the direction of the strand Arrange layers around the strand. When a Fault in the strand shell migrates past the temperature sensors, the measured temperature increases due to the not or only weakly developed strand shell, behind which there is liquid steel, the recorded temperature curves a characteristic in the event of an impending breakthrough Have shape.

To make possible breakthroughs from the recorded temperature profiles to be able to predict, it is known from US-A-4,949,777 the change of each individual temperature sensor recorded temperature with an average of those with all Temperature sensors to compare detected temperature changes and the comparison result thus obtained is exceeded to monitor a predetermined threshold value. If the temporal and spatial distribution of the threshold violations corresponds to a given pattern, is this is a sign of an impending breakthrough.

From T. Tanaka et al: "Trouble Forecasting System by Multi-Neural Network on Continuous Casting Process of Steel Production " in T. Kohonen et al (Ed.): Artificial Neural Networks; Proc. of the 1991 Int. Conf. on Artificial Neural Networks, Espoo, Finland, Elsevier Science Publishers B.V. (North-Holland), 1991, pp. 835 to 840, it is known to Breakthrough early detection as part of a pattern recognition neural networks by the individual temperature sensors save recorded temperature curves and on characteristic Examine patterns.

In a method known from JP-A-4 172 160 the temperatures detected with the temperature sensors neural network, which generates an output signal, if the spatial temperature distribution is one for one impending breakthrough characteristic pattern.

A reasonably reliable prediction of breakthroughs using neural networks requires sufficient training data for the neural network. Here results the problem is that training data from a facility is not can be easily transferred to another system. In addition, the decision criteria according to which breakthroughs are predicted for the plant operator are essentially invisible.

In addition, the known methods for pattern recognition require fully existing temperature patterns, e.g. Temperature curves, which results in a large amount of memory Has. At the same time, the computational effort is very high, because with everyone Changing the temperature pattern, e.g. if the temperature curve is supplemented by a new temperature value and at the same time the oldest temperature value is deleted, one completely new pattern recognition is required.

The invention has for its object a device for early detection of breakthroughs, which indicate only a small amount computational effort a safe and for the plant operator understandable detection of possible breakthroughs guaranteed.

According to the invention the object is achieved by the in the claim 1 specified invention solved.

Advantageous further developments of the device according to the invention are specified in the subclaims.

The early breakthrough detection according to the invention is based on a fuzzy pattern recognition, the rules of which are based on process knowledge be derived. The ones required for pattern recognition There is only information about the temperature profiles from the currently recorded temperatures and one of the representing the current temperature profile and ongoing updated inner state quantity. The pattern recognition can therefore be changed to the previous one for each new temperature value Results of pattern recognition, i.e. the internal state variable, build up so that not a completely new one every time Pattern recognition required due to the temperature curve is. In addition, there is no need to save the temperature profiles, so that overall the pattern recognition by means of the invention Setup faster and more efficiently than procedures which is the pattern recognition based on complete existing patterns.

Figur 1

den prinzipiellen Aufbau einer Stranggießanlage,

Figur 2

eine in der Stranggießanlage verwendete Kokille mit Temperatursensoren in den Kokilleninnenwänden,

Figuren 3 und 4

Beispiele für die mit den Temperatursensoren erfaßten Temperaturverläufe bei unterschiedlichen Wachstumsfehlern in der Strangschale,

Figur 5

ein Beispiel für eine Fuzzy-Mustererkennungseinrichtung zur Bildung eines Vorhersagewertes für die Durchbruch-Wahrscheinlichkeit aufgrund des mit einem Temperatursensor erfaßten Temperaturverlaufs,

Figur 6

ein Beispiel für den beim Auftreten eines bestimmten Wachstumsfehlers erfaßten Temperaturverlauf zusammen mit der in Abhängigkeit davon ermittelten Durchbruch-Wahrscheinlichkeit,

Figur 7

ein Beispiel für die Fuzzy-Zustände der Fuzzy-Mustererkennungseinrichtung,

Figur 8

ein Beispiel für das Fuzzy-Regelwerk der Mustererkennungseinrichtung,

Figur 9

ein verallgemeinertes Ausführungsbeispiel für die Mustererkennungseinrichtung,

Figur 10

ein Beispiel für eine Einrichtung zur Vorhersage der Gesamtwahrscheinlichkeit von Durchbrüchen und

Figur 11

ein Beispiel für die Meßwertaufbereitung der der Mustererkennungseinrichtung zugeführten Signale.

To further explain the invention, reference is made below to the figures of the drawing; show in detail

Figure 1

the basic structure of a continuous caster,

Figure 2

a mold used in the continuous caster with temperature sensors in the mold inner walls,

Figures 3 and 4

Examples of the temperature profiles recorded with the temperature sensors in the case of different growth errors in the strand shell,

Figure 5

an example of a fuzzy pattern recognition device for forming a predictive value for the breakdown probability on the basis of the temperature profile detected with a temperature sensor,

Figure 6

an example of the temperature profile recorded when a certain growth error occurs together with the breakdown probability determined as a function thereof,

Figure 7

an example of the fuzzy states of the fuzzy pattern recognition device,

Figure 8

an example of the fuzzy set of rules of the pattern recognition device,

Figure 9

a generalized embodiment for the pattern recognition device,

Figure 10

an example of a facility for predicting the overall probability of breakthroughs and

Figure 11

an example of the measured value processing of the signals supplied to the pattern recognition device.

Figure 1 shows a schematic representation of a continuous caster. A ladle 1 turns liquid steel 2 into one Distributor 3 cast, which the steel on different strands 4 distributed and also as a buffer and separator for non-metallic Particle serves. From the distributor 3 flows Steel in a mold 5, the inner walls of which are made of copper and water-cooled channels 6 included. Because of the heat dissipation the steel cools down on the mold inner walls and it a solid strand shell 7 is formed. This encloses the liquid steel, so that the strand 4 after leaving the Chill mold 5 transported via rollers 8 and finally in individual Slabs 9 can be cut.

Problems can arise if the strand shell has 7 growth defects having. Then often forms on individual local Issue only a very thin hardened layer that after Leaving the mold 5 can break. In such a case liquid steel escapes, which damages the system, so that a standstill and corresponding repairs become necessary. Around to prevent such breakthroughs in the strand shell 7 the growth defects in the strand shell 7 when they arise located in the mold 5.

As shown in FIG. 2, the mold is in the inner walls 5 temperature sensors 10 in two, offset in the direction of the strand Layers distributed around the strand. It can several levels or only one level can also be provided. Due to changes in the recorded temperature profiles can be concluded on a weak point in the strand shell 7 will. If an error is discovered, the casting speed reduced, so that the cooling time in the mold 5 increases and a sufficiently strong strand shell on the Can form a defect.

By far the most common growth errors, so-called glue, arise from a locally increased friction between the Strand 4 and the inner wall of the mold 5. At the friction point the strand 4 adheres more strongly than in the environment to the Mold inner wall, which is why there is also its speed decreased. This leads to tension in the strand shell 7 so that it breaks open. Liquid steel gets to the Mold inner wall and there leads to a rise in temperature.

In Figure 3 is an example of that with one of the temperature sensors 10 recorded temperature curve shown when a such errors migrate past the relevant temperature sensor 10. While the adhesive on the temperature sensor 10 passes, a significant rise in temperature is measured. If the adhesive has passed the temperature sensor 10, it drops Temperature drops below the temperature level that is normal Casting conditions prevail. This reduction can be attributed on a thickened strand shell behind the glue due to a reduced speed has arisen there.

Another cause of breakthroughs in the strand shell are Air cushions, called cracks, are located between the strand 4 and the mold 5 form.

Figure 4 shows an example of when such occurs Error recorded temperature curve. Because of the low The thermal conductivity of the air is the heat dissipation from line 4 to the mold 5 greatly reduced, so that only one very forms thin strand shell 7. If a crack happens to one of the Temperature sensors 10, it is reflected in the detected Temperature curve as a pronounced slump again. Together makes glue and cracks the cause of over 90% of all Breakthroughs.

The different growth errors in the strand shell 7 thus cause characteristic patterns in the detected Temperature curves. These patterns arise sequentially by new measured values can be added to a temperature curve.

FIG. 5 shows an example of a pattern recognition device 11, which continuously consists of the current temperature values T (i) detected in time steps i by a temperature sensor 10 and the temporal temperature changes ΔT (i) = T (i) -T (i-1) the probability P (i + 1) determines that an adhesive or cracking pattern develops in the recorded temperature profile. Since no pattern recognition can take place solely on the basis of the current values T (i) and ΔT (i), the previously determined breakdown probability P (i) is additionally used as an internal state variable representing the previous temperature profile and together with the current measured values T ( i) and ΔT (i) fed to a fuzzy logic 12, which determines the current breakdown probability P (i + 1). This is temporarily stored in a memory element 13 and fed back to the input of the fuzzy logic 12 in the next time step. By temporarily storing and feeding back the breakthrough probability P (i) determined in the preceding time step, the fuzzy logic 12 is able to recognize the pattern only on the basis of the current temperature T (i) and its change ΔT (i), ie without Knowledge of the temperature profile.

To the operation of the pattern recognition device 11 too illustrate, the temperature profile T is exemplified of an adhesive, as shown in Figure 6, considered:

The temperature T is constant under normal casting conditions and their change over time fluctuates very slightly. The Probability P for a breakthrough is zero here.

At the beginning of an adhesive, the temperature T rises. The probability P is therefore on a small positive Value, for example 0.1, increased.

In the further course of the adhesive, the temperature T, and the change in temperature T over time also increases. Lies now a low probability from the previous step P before, which is synonymous with observing one Is the beginning of the adhesive, then the probability P becomes one medium value, e.g. 0.4, increased. However, is from the previous one Step no low probability P, i.e. of the Beginning of an adhesive, before, the probability P also not changed.

The temperature increase caused by the adhesive is reached now their maximum value, with the temporal Change in temperature T becomes zero. Been up to this Go through the typical temperature curve of an adhesive at this point in time and so far a medium breakthrough probability P is determined, then the probability P becomes a great value, e.g. 0.7, increased.

The adhesive has now passed the temperature sensor 10, and the Temperature T drops to medium in the event of a negative temperature change Values. The probability then follows the scheme above P further, e.g. to 0.9, increased, but below provided that it already has great value.

Due to the thickening of the strand shell at the end of an adhesive The temperature T finally decreases so far that it below the temperature level under normal casting conditions lies. Once this happens and the probability P due to what has happened so far, a very high value , the probability P is at its maximum Value, e.g. 1.0, increased.

FIG. 7 shows the fuzzy state graph of the pattern recognition device 11. The states, i.e. the linguistic values the breakdown probability P (i), form the nodes 14 of the state graph. The probability P (i) can be assume the following linguistic values:

Z = 0, T = very small, S = small, M = medium, B = large, H = very large.

Stand on the transition arrows 15 between the states 14 before the slash the transition conditions, i.e. the Fuzzy rules that cause a change of state; the value after the slash indicates the newly achieved condition on. In the course of pattern recognition, the probability P (i) increases step by step from Z to H only if that Temperature pattern causes the rule sets one after the other R2, R5, R9, R13 and R17 are met. That is with glue or Crack patterns the case. The detected temperature pattern gives way only slightly from these reference patterns, so either keep the current state or the next one in a lower state. If the deviations are larger, depending on the current state reached, one of the rule sets will be activated R3, R8, R12, R16 or R20 active and the probability P (i) becomes Z.

A major influence on the breakthroughs in the strand shell 7 characteristic temperature profiles have changes the casting speed. It therefore makes sense to do this Consider changes Δv (i) additionally in pattern recognition, as shown in dashed lines in Figure 5 is. For example, the casting speed increases, so the dwell time is reduced and with it the cooling time of strand 4 in the mold 5. This means at the same time an increase in the measured temperature. Then kick during a change in the casting speed growth error in the Strand shell 7, so the typical temperature curves for them distorted.

Figure 8 shows an example of a in the fuzzy logic of the Pattern recognition device 11 implemented fuzzy rules, at which in addition to the detected temperature T (i) and the change in temperature ΔT (i) the change in casting speed Δv (i) to determine breakthrough probability P (i) is used. Otherwise, the one shown in Figure 7 Fuzzy state graph and that shown in Figure 8 Fuzzy rules are equivalent to each other. The rules of the rules give the combinations of linguistic values of the Input variables T (i), ΔT (i) and Δv (i) that are met must for the pattern recognition device 11 to be in its state changed or maintained. The temperature will be T (i) assigned the following values:

NB = negative large, NS = negative small, Z = zero, PS = positive small, PM = positive medium, PB = positive large.

The following values are assigned to the temperature change ΔT (i):

NB = negative large, NS = negative small, Z = zero, PS = positive small, PB = positive large.

The following are for changing the casting speed Δv (i) Values provided:

N = negative, Z = zero, PN = positive normal, PE = positive extreme.

The internal state variable, i.e. the temporarily stored probability P (i) takes the following linguistic values on:

Z = zero, T = very small, S = small, M = medium, B = large, H = very large.

For each combination of values of temperature T (i), the temperature change ΔT (i), the change in casting speed Δv (i) and the temporarily stored probability P (i) each results in a specific linguistic value for that predicted by the pattern recognition device 11 Breakthrough probability P (i + 1). The linguistic Predicted breakdown probability values P (i + 1) are coded as follows for the sake of clarity: Z = 1, T = 2, S = 3, M = 4, B = 5, H = 6.

All rules of the fuzzy logic 12 can be read directly from the set of rules. For example, if P (i) = Z and Δv (i) = Z and T = Z and ΔT = Z , then P (i + 1) = 1 (= Z) .

The inference takes place according to the max-min method and the defuzzification according to the focus method.

FIG. 9 shows a generalized exemplary embodiment for the pattern recognition device in which the input variables T (i), ΔT (i) and Δv (i) are combined in an input vector u (i). A first fuzzy logic 16 generates an updated state vector z (i + 1) from the input vector u (i) and a temporarily stored inner state vector z (i), which is temporarily stored in a memory element 17. The temporarily stored state vector z (i) and the input vector u (i) are linked together in a second fuzzy logic 18 to form an output vector y . The pattern recognition device 11 shown in FIG. 5 is a special case of the device shown in FIG. 9 with only one internal state variable z (i) = P (i) , an output variable y (i) = P (i + 1) and with matching transmission behavior of the first fuzzy logic 16 and the second fuzzy logic 18, ie. f = g .

FIG. 10 shows an example of a device for predicting the overall probability of breakthroughs on the basis of the individual temperature profiles detected by the temperature sensors 10. The patterns of certain growth disorders of the strand shell are not only found in a temperature profile, but also due to the expansion of the growth error and the strand movement in adjacent temperature profiles. As FIG. 10 shows, each temperature sensor 10 is followed by its own pattern recognition device 11, which monitors the temperature profile detected in each case for the occurrence of a predetermined pattern. So that the detection of growth errors in the strand shell takes place more reliably, the prediction values P _a and P _b supplied by the pattern recognition devices 11 each of two immediately adjacent temperature sensors 10 are combined in a linking device 19 to form a local breakdown probability P _loc . Erroneous pattern recognitions of an individual pattern recognition device 11 are corrected in that the local breakdown probability P _{loc is} only assigned a large value if both P _a and P _b each have large values. Furthermore, the detection of adhesives or cracks also improves, since increased values for the individual probabilities P _a , P _b can be used to _infer a local breakthrough probability P _loc that is greater than each of the individual probabilities P _a , P _b . The combination of the individual probabilities P _a and P _b to the local breakdown probability P _loc is therefore preferably based on fuzzy inferences.

Since the growth errors in the strand shell move past the individual temperature sensors 10, whereby the direction of movement and the spreading of the growth errors can take place differently, the pattern recognition results P _a and P _b from the pattern recognition devices 11 of two adjacent temperature sensors 10 can have a time offset for the same growth error . However, in order that both pattern recognition results P _a and P _b can be combined in the linking device 19, they must be present at the same time. For this reason, each pattern recognition device 11 is followed by a delay device 20 with which this time offset is compensated. The delay devices 20 each consist of a maximum value holding element, which has the maximum value of each individual probability P (i) at the output of the upstream pattern recognition device 11 P Max (i) = max (P (ik), ..., P (i)) the last k time steps are determined and supplied to the linking device 19.

The maximum value of all local breakthrough probabilities P _loc , which then represents the total probability P _tot for a breakthrough, is ascertained in a logic circuit 21 arranged downstream of all logic devices 19.

The pattern recognition in the pattern recognition devices 11 must be independent of different system and operating conditions be. Therefore, there is 10 between each temperature sensor and the associated pattern recognition device 11 a device 22 arranged for the processing of measured values in which the Input variables of the pattern recognition device 11, that is to say the Temperature T, the temporal change in temperature ΔT and normalized the change in the casting speed Δv over time or transformed that different plant conditions or changing process conditions the detection of glue and crack patterns do not affect or only slightly.

FIG. 11 shows a block diagram of such a device 22 for processing measured values. The temperature values T (i) measured in a time step i are, depending on different system and operating conditions, relatively constant between approx. 100 ° C. and 200 ° C. under normal casting conditions. Adhesives and cracks cause deviations of up to 50 ° C from this constant offset temperature T ₀ . The pattern recognition device 11 can only recognize adhesive and cracking patterns if they start from a constant temperature level. To achieve this, an offset temperature T _{0 is} determined by means of a first-order time-discrete filter 23 and subtracted from the current temperature value T (i) in a subtracting device 24. The temperature thus obtained T A (i) = T (i) -T 0 (i) is optionally smoothed to suppress noise in a filter 25 and then fed to a normalization device 26, in which the temperature deviations caused by typical growth errors from the normal temperature level are limited to a value range between zero and one. The normalized temperature value T _A (i) thus obtained is then fed to the pattern recognition device 11.

The pattern recognition device 11 also receives the temporal change in the temperature ΔT _A (i), which is formed in a device 27 by means of the difference quotient from the output signal of the subtracting device 24 and subsequently normalized to a value range between zero and one in a further normalization device 28.

As already explained above, the change in the casting speed over time can also be an input variable of the pattern recognition device 11. There it changes the rules for pattern recognition in such a way that adhesives and cracks can still be reliably recognized even if their patterns are distorted due to the change in casting speed. The change in the casting speed Δv (i) over time is determined in a device 29 by means of the difference quotient from the casting speed v (i). Often the casting speed v (i) is not increased steadily, but in leaps and bounds. The resulting temperature rise, which arises from the shorter cooling time in the mold 5, however, takes place continuously over a certain period of time. In order to achieve a corresponding change in the rules for pattern recognition during the entire temperature rise, the value Δv (i) must be set to a correspondingly high value during the temperature rise, which simulates a steady increase in the casting speed v (i). This is done with a maximum value holding element 30 which on the output side generates the greatest positive value of Δv (i) from the last k time steps. So the following applies: Δv A (i) = max (Δv (ik), ..., Δv (i) for Δv (i)> 0 and Δv A (i) = Δv (i) for Δv (i) ≤0.

Finally, the value of Δv _A (i) thus obtained is normalized in a normalization device 31 before it is fed to the pattern recognition device 11.

As already mentioned, the influence of changes in the casting speed over time on the temperature profiles can be taken into account by changing the rules for pattern recognition. Another way of reducing the influence of the casting speed changes is to eliminate the temperature changes caused thereby in the recorded temperature profiles before the pattern recognition. This is done by averaging all the temperature values T (i) simultaneously delivered by the temperature sensors 10 on one level in the mold 5 and subtracting the mean value MT (i) thus obtained from the individual temperature values T (i) in a subtractor 32. The temperature difference thus obtained T D (i) = T (i) -MT (i) is independent of temperature changes caused by changes in casting speed and is further supplied to the filter 23 and the subtractor 24. In this case, the adaptation of the pattern recognition by Δv _A (i) can also be omitted, so that the structure of the device for early breakthrough detection becomes simpler.

Alternatively, it can be provided that at constant pouring speed v (i) or small changes in pouring speed v (i) one works without the pouring speed compensation in order not to introduce disturbances into the individual temperature profiles T _A (i) via the mean value MT (i) . For this purpose, the mean value MT (i) of the comparison device 32 is fed via a controllable switching device 33, which switches the mean value MT (i) on to the comparison device 32 only when the change in the casting speed Δv _A (i) exceeds a predetermined threshold value v _S. For this purpose, the values Δv _A (i) and v _{S are} fed to a threshold value detector 34, which controls the controllable switching device 33 on the output side. In order to avoid that the value T _A (i) changes abruptly due to the connection of the mean value MT (i), the value T ₀ (i + 1) of the filter 23 is added via the output of a subtractor 35 T 0 (i + 1) = T (i) -MT (i) -T A (i) set so that the course of T _A (i) continues.

Claims (10)

1. Device for early detection of break-outs during continuous casting with a mould (5) in which temperature sensors (10) are arranged in a manner distributed around the strand (4), each temperature sensor (10) being assigned a pattern recognition device (11) which, from the temperature (T(i)) detected and an internal state variable (P(i)) representing the temperature curve up to that point, updates the internal state variable (P(i)) on the basis of fuzzy conclusions and generates at the output a current predicted value (P(i+1)) for the break-out probability.
2. Device according to Claim 1, characterized in that the predicted value (P(i+1)) is identical with the internal state variable.
3. Device according to Claim 1 or 2, characterized in that each pattern recognition device (11) evaluates the current value (T(i)) and the change (ΔT(i)) of the temperature detected by the respectively associated temperature sensor (10).
4. Device according to one of the preceding claims, characterized in that, to generate the predicted value (P(i+1)) for the break-out probability, the pattern recognition device (11) additionally evaluates the change in the casting rate (Δv(i)).
5. Device according to one of the preceding claims, characterized in that between each temperature sensor (10) and the associated pattern recognition device (11) there is a device (22) for measured-value conditioning in which a time average (T₀(i)) determined on the basis of the temperature curve up to that point is subtracted from the temperature (T(i)) detected.
6. Device according to Claim 5, characterized in that an average (MT(i)) which is formed from the temperature values simultaneously detected by all the temperature sensors (10) distributed around the strand (4) in each case in one and the same plane is additionally subtracted from the detected temperature (T(i)) in the device (22) for measured-value conditioning.
7. Device according to one of the preceding claims, characterized in that the pattern recognition devices (11), which are each assigned to at least two directly adjacent temperature sensors (10), are each connected at the output side to a logic device (19) which links the predicted values (P_a, P_b) supplied by the pattern recognition devices (11) to give a probability value (P_loc) for a local break-out in the region of the adjacent temperature sensors (10).
8. Device according to Claim 7, characterized in that a respective delay device (20) is arranged on the output side of at least those pattern recognition devices (11), the associated temperature sensors (10) of which are arranged above the remaining temperature sensors (10) in the mould (5).
9. Device according to Claim 8, characterized in that, on the output side, the delay device (20) generates the respective maximum value of a predetermined number of the predicted values (P(i+1)) last fed to it.
10. Device according to one of Claims 7 to 9, characterized in that a common logic circuit (21) is arranged on the output side of the logic devices (19) and, from the probabilities (P_loc) for local break-outs, this logic circuit determines a value (P_ges) for the overall probability of a break-out.

Images