Methods and devices for diagnosing abnormalities in continuous casting machines

TWI945723BActive Publication Date: 2026-11-11JFE STEEL CORP
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
TW114108422
Authority / Receiving Office
TW · TW
Patent Type
Patents
Current Assignee / Owner
Priority Date
2024-03-13
Filing Date
2025-03-07
Publication Date
2026-11-11
Estimated Expiration
2045-03-06

AI Technical Summary

Technical Problem

Existing methods for diagnosing abnormalities in continuous casting machines, such as roller spacing deviations, are prone to inaccuracies due to positional deviations of measurement instruments and environmental biases, leading to false detections and delays in identifying actual issues.

Method used

A method and device that utilize a roller spacing meter on a dummy bar to measure roller spacing at multiple positions, comparing current and past data for two roller pairs of the same type, using principal component analysis to calculate a Q-value for accurate anomaly detection, reducing false detections and predicting anomalies.

Benefits of technology

Enhances the accuracy of roller spacing anomaly detection, reduces false alarms, and allows for earlier identification of issues, minimizing production downtime by automating the monitoring process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure TWG2TB001920475_001
    Figure TWG2TB001920475_001
  • Figure TWG2TB001920475_002
    Figure TWG2TB001920475_002
  • Figure TWG2TB001920475_003
    Figure TWG2TB001920475_003
Patent Text Reader

Abstract

This invention provides an anomaly diagnosis method and device for a continuous casting machine that can accurately detect anomalies in roll spacing. An anomaly diagnosis method for a continuous casting machine (1) detects anomalies in the roll spacing of opposing roll pairs (3a) in the thickness direction of the casting plate, comprising a group of support rolls (2). The multiple support rolls (2) clamp the casting plate (S) within the continuous casting machine (1) and guide it towards the casting direction. In the anomaly diagnosis method for the continuous casting machine (1), a roll spacing meter (11) installed on the dummy bar (9) measures the roll spacing of multiple roll pairs (3a) at multiple positions in the width direction of the casting plate. The dummy bar (9) moves in conjunction with the pulling movement of the casting plate (S). For two roll pairs (3a) of the same type, the roll spacing at the same position in the width direction of the casting plate is used to detect anomalies in the roll spacing. The support rolls (3) have multiple segmented rolls and roll bearing seats, and two roll pairs (3a) of the same type have at least the same bearing position.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to a method and device for diagnosing abnormalities in a continuous casting machine. Prior Technology

[0002] The guide rollers (support rollers) arranged on the exit side of the mold in a continuous casting machine support and guide the casting slab by clamping and pulling the slab relative to the thickness direction. In this guide roller group, to achieve stable continuous casting and avoid bulging or internal cracking that would degrade the quality of the casting slab, the roller spacing of the roller pairs is strictly managed.

[0003] Typically, the gap between the guide rollers is measured during each casting by a measuring device mounted on the dummy bar. The dummy bar is a device that guides the initial portion of the cast slab poured from the mold to the pinch rollers located downstream of the continuous casting machine.

[0004] As a prior art related to this point, Patent Document 1 discloses a method that compares the peak value of the roller spacing of each roller obtained by using a roller spacing meter with a preset roller spacing to determine whether there is an abnormality in the roller spacing.

[0005] Moreover, Patent Document 2 discloses the following method: measuring the roller gap of a pair of rollers at at least two locations, comparing the difference between the two measured values ​​with the previous value, and determining that the roller gap is abnormal when it exceeds the threshold value. [Existing technical documents] [Patent Literature]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 2012-245531 [Patent Document 2] Japanese Patent Application Publication No. 2006-247687 Summary of the Invention

[0007] [The problem that the invention aims to solve] Furthermore, in the aforementioned method, sometimes the roller gap measurement cannot be performed normally and is identified as abnormal due to reasons such as the positional deviation of the measuring instrument caused by the serpentine movement of the guide rod during the pulling movement, or the guide rod crossing over the roll chock.

[0008] Furthermore, when the roller spacing deviation exceeds the threshold value due to bias and unevenness of the measured values, which depend on the characteristics of the sensor or the measurement environment, it may sometimes be judged as an abnormal roller spacing.

[0009] Moreover, in cases where something is deemed abnormal, there are often instances where actual inspections of the support rollers of the continuous casting equipment reveal no abnormalities.

[0010] Therefore, the present invention was made with regard to the aforementioned problem, and aims to provide an anomaly diagnosis method and anomaly diagnosis device for a continuous casting machine that can accurately detect anomalies in the roll spacing. [Methods for solving problems]

[0011] (1) By means of one aspect of the present invention, an abnormality diagnosis method for a continuous casting machine is provided, which detects the abnormality of the roller spacing of a pair of rollers facing each other in the thickness direction of the casting plate, comprising a group of multiple support rollers, wherein the multiple support rollers clamp the casting plate inside the continuous casting machine and guide it in the casting direction, wherein the abnormality diagnosis method for the continuous casting machine is provided by means of a roller spacing meter provided on the dummy bar, wherein the roller spacing of the multiple roller pairs is measured at multiple positions in the width direction of the casting plate, the dummy bar moves in conjunction with the pulling movement of the casting plate, and for two roller pairs of the same type, the roller spacing at the same position in the width direction of the casting plate is used to detect the abnormality of the roller spacing, wherein the support roller has multiple segmented rollers and roller bearing seats, and the two roller pairs of the same type have at least the same bearing position.

[0012] (2) In the abnormal diagnosis method of the continuous casting machine in (1), the two roller pairs of the same type are roller pairs in the same section.

[0013] (3) In the abnormality diagnosis method of the continuous casting machine in (1) or (2), when detecting the abnormality of the roll interval, the current comparison difference data is used to detect the abnormality of the roll interval, wherein the current comparison difference data is the comparison difference data of the roll interval measured at the same position of two roll pairs of the same type as the element.

[0014] (4) In the abnormal diagnosis method of the continuous casting machine in (3), the measurement of the roll interval is repeatedly performed at the time of starting the continuous casting of the continuous casting machine, and the abnormality of the roll interval is detected by using the current comparison data and the past comparison data, wherein the past comparison data is the comparison data measured in the past multiple continuous castings.

[0015] (5) In the abnormal diagnosis method of the continuous casting machine in (4), the past poor data is the most recent consecutive poor data.

[0016] (6) In the abnormal diagnosis method of the continuous casting machine in (5), based on the data of multiple points (n1 points) in the current comparative poor data and the data of multiple points (n1×n2 points) in the comparative poor data measured in the most recent (n2) continuous casting, a vector data (n1×n3 points, where n3=n2+1) is calculated as an element, and the deviation of the vector data from the normal state is statistically evaluated to detect abnormalities.

[0017] (7) According to one aspect of the present invention, an abnormality diagnosis device for a continuous casting machine is provided, which detects abnormalities in the roller spacing of roller pairs facing each other in the thickness direction of the casting plate, comprising a group of multiple support rollers, wherein the multiple support rollers clamp the casting plate inside the continuous casting machine and guide it in the casting direction, the abnormality diagnosis device for the continuous casting machine comprising: a roller spacing meter, disposed on a dummy bar that moves in conjunction with the pulling movement of the casting plate, measuring the roller spacing of the multiple roller pairs at multiple positions in the width direction of the casting plate; and a diagnosis unit that, for two roller pairs of the same type, detects abnormalities in the roller spacing using the roller spacing at the same position in the width direction of the casting plate, wherein the support rollers have multiple segmented rollers and roller bearing seats, and the two roller pairs of the same type have at least the same bearing position. [The effects of the invention]

[0018] According to one aspect of the present invention, an anomaly diagnosis method and anomaly diagnosis device for a continuous casting machine are provided, which can detect anomalies in the roll spacing with good accuracy. Simple Explanation of the Diagram

[0019] Figure 1 is a schematic diagram showing a continuous casting machine according to one embodiment of the present invention. Figure 2 is a schematic diagram showing the structure of the support roller. Figure 3 is a side view showing the method of measuring roll gap using a roll gap meter. Figure 4 is a schematic diagram illustrating the concept of Q value. Figure 5 is a graph showing the trend of the deviation of the roll gap and the Q value when a failure occurs. Figure 6 is a graph showing the trend of the deviation of the roller spacing and the Q value when a false detection occurs. Figure 7 is a graph showing the trend of the deviation of the roll spacing when a false detection occurs. Figure 7(A) shows the roll pair that was falsely detected, and Figure 7(B) shows the roll pair that is the same type as the support roll 3 in Figure 7(A) and is close to each other in the same section. Figure 8 is a graph showing the number of false detections in the embodiment and the comparative example. Figure 9 is a graph showing the anomaly detection rate and detection details in the embodiments and comparative examples. Implementation

[0020] In the following detailed description, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same or similar parts are indicated by the same or similar symbols, and repeated descriptions are omitted. The drawings are illustrative and may include variations from reality. Furthermore, the embodiments shown below illustrate apparatus or methods for embodying the technical concept of the present invention; the technical concept of the present invention does not limit the materials, structures, and arrangements of the constituent parts to those described below. Various modifications can be made to the technical concept of the present invention within the scope of the claims as defined in the patent claims.

[0021] <Structure of a Continuous Casting Machine> Hereinafter, the present invention will be further described in detail using drawings. FIG1 is a schematic diagram showing a continuous casting machine 1 according to an embodiment of the present invention.

[0022] The continuous casting machine 1 includes a mold for continuous casting 2, multiple support rollers 3, multiple sections 4, a tundish 5, a sliding nozzle 6, an immersion nozzle 7, a pinch roller 8, a dummy bar 9, and an abnormality diagnosis device 10.

[0023] Multiple support rollers 3 constitute a support roller group, which encloses the casting plate S within the continuous casting machine 1 and guides it towards the casting direction. The casting direction is the direction of the long side of the casting plate S in Figure 1, and is also the direction of movement of the casting plate S containing the solidified billet shell within the continuous casting machine 1. Furthermore, within the continuous casting machine 1, the width direction (the front-back direction in Figure 1) of the casting plate S is referred to as the casting plate width direction, and the thickness direction of the casting plate S is referred to as the casting plate thickness direction. Furthermore, two support rollers 3 facing each other in the casting plate thickness direction are also referred to as roller pair 3a.

[0024] Furthermore, the support roller 3 has multiple segmented rollers 31 and at least one roller bearing seat (bearing) 32. Figure 2 shows an example of the structure of the support roller 3 in a section 4. In the example shown in Figure 2, five roller pairs 3a are provided in the section 4. Figure 2 only shows the five support rollers 3 on one side of the roller pair 3a in the thickness direction of the cast plate. Furthermore, the five support rollers 3 on the other side of the thickness direction of the cast plate are also designed with the same shape as in Figure 2. As shown in Figure 2, support rollers 3A to 3E each have three segmented rollers 31 and multiple roller bearing seats 32 provided at the ends of the segmented rollers 31. Furthermore, the outer diameter or material of each segmented roller 31 is all the same. In section 4 of Figure 2, support rollers 3A to 3E are classified into two types. Here, the type of support roller 3 refers to those classified at least by the position of the roller bearing seat 32, i.e., the bearing position. In the example of Figure 2, support rollers 3A, 3C, and 3E are roller pairs 3a of the same type with the same bearing position, and support rollers 3B and 3D are roller pairs 3a of the same type with the same bearing position.

[0025] Section 4 is a frame with multiple roll pairs 3a. The continuous casting machine 1 forms a support roll group by incorporating multiple sections 4. Furthermore, a cooling unit (not shown) for spraying cooling water is provided in section 4.

[0026] The feed trough 5 is an intermediate container for holding the molten steel (molten metal) M transferred from the self-pouring vessel (not shown). A sliding nozzle 6 is positioned at the bottom of the feed trough 5, and its opening is adjusted to regulate the flow rate of the molten steel M. An immersion nozzle 7 is positioned on the lower surface of the sliding nozzle 6. The molten steel M held by the feed trough 5 is injected into the mold 2 via the sliding nozzle 6 and the immersion nozzle 7.

[0027] The pinch roller 8 is positioned downstream of the continuous casting machine 1 in the casting direction, and pulls the casting plate S in the casting direction by rotating. The dummy bar 9 is used in the initial stage of continuous casting, and is a device that pulls the initial casting portion of the casting plate S cast by the mold 2 to the pinch roller 8, and moves in conjunction with the pulling movement of the casting plate S.

[0028] Furthermore, a roller spacing gauge 11 is provided on the dummy bar 9. The roller spacing gauge 11 measures the spacing between the roller pair 3a in the thickness direction of the cast plate, i.e., the roller spacing. The roller spacing gauge 11 is a measuring component that uses a known measuring component such as a rotary encoder or a transmission transformer to measure the roller spacing. Moreover, multiple roller spacing gauges 11 are provided along the width direction of the cast plate, measuring the roller spacing at multiple positions of the roller pair 3a in the width direction of the cast plate. For example, three roller spacing gauges 11 are provided on the dummy bar 9, measuring the roller spacing at three locations in the width direction of the cast plate. As shown in Figure 3, the roller spacing gauge 11 contacts the roller pair 3a when the dummy bar 9 passes through the support roller group, thereby measuring the roller spacing of the roller pair 3a.

[0029] The anomaly diagnosis device 10 is a component that detects anomalies in the roll spacing based on the measurement results of the roll spacing gauge 11, and may include, for example, a computer. The anomaly diagnosis device 10 includes an acquisition unit 100, a memory unit 101, and a diagnosis unit 102. Furthermore, when the anomaly diagnosis device 10 is a computer, a program is executed by a computing device such as a central processing unit (CPU), thereby functioning as the acquisition unit 100 and the diagnosis unit 102. The acquisition unit 100 acquires the measurement results of the roll spacing gauge 11 (roller spacing data) and, as needed, obtains operating conditions from a host computer. The memory unit 101 stores the roll spacing data acquired by the acquisition unit 100, and also stores past normal roll spacing data, or principal component analysis results based on the normal roll spacing data (described later), and other data required for anomaly determination. The memory unit 101 can be a primary memory device such as Random Access Memory (RAM), or a secondary memory device capable of storing data such as a Hard Disk Drive (HDD) or Solid State Drive (SSD). The diagnostic unit 102 diagnoses roll gap anomalies based on roll gap data. Details regarding the diagnostic method used by the diagnostic unit 102 to diagnose roll gap anomalies will be described later.

[0030] Furthermore, in this embodiment, the continuous casting machine 1 is a vertically curved type. In this type of continuous casting machine 1, with a portion of the head of the dummy bar 9 inserted into the lower part of the mold 2, molten steel M is injected into the mold 2, thereby initiating continuous casting. Subsequently, the molten steel M is cooled in the mold 2 to form a solidified shell, becoming a semi-solidified casting plate S. Then, the dummy bar 9 is pulled in the casting direction, thereby pulling the semi-solidified casting plate S in the casting direction. Subsequently, the casting plate S passes through the support roller group and is completely solidified by secondary cooling by the cooling components. Then, when the front end of the casting plate S reaches the pinch roller 8 on the downstream side in the casting direction, it separates from the dummy bar 9 and is then cut to a predetermined length, thereby becoming a slab or similar casting plate.

[0031] <Abnormal Diagnostic Methods> Next, the abnormality diagnosis method of the continuous casting machine 1 in this embodiment will be described. In this embodiment, firstly, the roll gap meter 11 measures the roll gap of each roll pair 3a at the initial stage of casting when the dummy bar 9 passes through the support roll group (measurement step). In the measurement step, the roll gap meter 11 measures the roll gap at multiple positions in the casting width direction for each roll pair 3a. In this embodiment, for example, the roll gap meter 11 measures the roll gap of each roll pair 3a at three locations in the casting width direction. The measurement results are sent to the acquisition unit 100 and stored in the memory unit 101. The measurement step ends by measuring the roll gap of all roll pairs 3a for which abnormality diagnosis of roll gap is to be performed. Furthermore, as an example, the abnormality diagnosis of the roll gap of the support roll 3 included in a section 4 shown in FIG2 will be described below. In the example shown in FIG2, the roll gap is measured at three positions Pa, Pb, and Pc in the casting width direction shown by the dashed lines. That is, for example, for support roller 3A, the roller interval is measured at the intersection of the support roller 3A and the other support roller 3 (not shown) forming roller pair 3a with the three positions Pa, Pb, and Pc shown by the dashed lines, namely position Pa1, position Pb1, and position Pc1.

[0032] After the measurement step is completed, the diagnostic unit 102 uses the roller spacing measured in the measurement step at the same position in the casting width direction for two roller pairs 3a of the same type within the same section 4 to detect the presence or absence of roller spacing abnormalities (diagnostic step). There is no particular limitation as long as the two roller pairs 3a of the same type are within the same section 4. To improve the accuracy of abnormality detection, it is preferable to use roller pairs 3a that are close to each other. For example, in the case of Figure 2, as two roller pairs 3a of the same type, they could also be roller pairs 3a of support roller 3A and support roller 3E, but it is preferable to use roller pairs 3a of support roller 3A and support roller 3C (or support roller 3C and support roller 3E).

[0033] In the diagnostic step, firstly, for two roll pairs 3a of the same type, the difference in roll spacing at the same position in the width direction of the cast plate is calculated. The difference in roll spacing at the same position in the width direction of the cast plate is calculated separately at multiple positions in the width direction measured in the measurement step, preferably all positions. Furthermore, the data using this difference in roll spacing as an element is also called the comparison difference data. For example, in the case shown in Figure 2, roll pair 3a (3a1) including support roll 3A and roll pair 3a (3a3) including support roll 3C are roll pairs 3a of the same type. Furthermore, in the measurement step, for roll pair 3a1 and roll pair 3a3, the roll spacing at positions Pa1, Pb1, and Pc1 (e.g., da1, db1, dc1) and the roll spacing at positions Pa3, Pb3, and Pc3 (e.g., da3, db3, dc3) are measured respectively. Next, the difference in roll spacing at the same position along the width direction of the cast plate is calculated (for example, the difference between the roll spacing at position Pa1 and the roll spacing at position Pa3: (da1-da3)). Furthermore, this difference in roll spacing is calculated separately at positions Pa, Pb, and Pc (in the example given, for instance, Δda=da1-da3, Δdb=db1-db3, Δdc=dc1-dc3). Here, the comparison difference data calculated in the diagnostic step is also referred to as the current comparison difference data.

[0034] Subsequently, feature quantities are extracted from the calculated current comparison difference data and the most recent comparison difference data. The most recent comparison difference data is the comparison difference data that uses the difference in roll interval measured in the most recent consecutive castings as a feature, and it is the same support roll 3 and the comparison difference data of the casting plate width direction position as the object of anomaly diagnosis. In the most recent comparison difference data, similar to the current comparison difference data, it includes the difference in roll interval at multiple positions in the casting plate width direction. Moreover, the most recent comparison difference data uses the comparison difference data collected in the most recent consecutive castings, but preferably it uses the comparison difference data collected in at least the most recent two consecutive castings, combined with the current comparison difference data to use three comparison difference data. As for the past comparison difference data, it is sufficient to have data from at most five recent times in total. If we take Δda as an example, in the case of a measurement at most two times before the measurement time of the current comparison difference data, which is considered as the most recent past, the index representing the current comparison difference data is set as k. Furthermore, if we set the index of the most recent worst comparison data as k-1 and the index of the most recent two-last comparison data as k-2, then each comparison data can be recorded as Δda(k), Δda(k-1), and Δda(k-2).

[0035] In the continuous casting machine 1, casting is performed using the dummy bar 9 whenever continuous casting begins. Therefore, the interval between each roll is measured and the difference data is calculated at the start of each continuous casting operation. Furthermore, in the so-called continuous casting operation where multiple ladles are used to continuously perform continuous casting, the measurement is only performed when the dummy bar is inserted at the very beginning of the continuous casting. Moreover, the past difference data is stored in the memory unit 101.

[0036] For feature extraction, dimensional compression based on principal component analysis is preferred. Specifically, a data vector is calculated by combining multiple points (n1 points) of the current comparative data with multiple points of the comparative data obtained from multiple consecutive castings in the past (n2 times). The number of measurements is calculated by combining the current measurement with the most recent multiple (n2 times) measurements to form n3 (=n2+1). Since the number of points in the comparative data is n1, the data vector becomes a vector data with data (n1×n3 points) as features.

[0037] Regarding the vector of n1×n3 data points as elements, in the example described above, both n1 and n3 are 3. That is, for the comparative data (Δda, Δdb, Δdc) at the three locations, the index of the current comparative data is set to k, and the indices of the two most recent comparative data are set to k-1 and k-2 respectively, forming a vector containing data from three data acquisition points. This vector is denoted by ΔD, which in this example includes the following elements. ΔD=(Δda(k) Δdb(k) Δdc(k) Δda(k-1) Δdb(k-1) Δdc(k-1) Δda(k-2) Δdb(k-2) Δdc(k-2))T (T represents transpose)

[0038] Furthermore, for ΔD, principal component analysis is performed in advance using vector data from the normal state where the roll spacing was not abnormal during past operations. From the principal component vectors (n1×n3) obtained as the analysis result, one or more principal component vectors (in the case of multiple vectors, fewer than n1×n3) that represent the characteristics of the normal state are determined as feature vectors (hereinafter, also simply referred to as feature vectors). This method represents the characteristics with fewer than n1×n3 vectors, hence it is called dimensional compression. Moreover, the hyperplane used to calculate the Q statistic (also called the Q value) is calculated using the previously determined principal component vectors that represent the characteristics of the normal state, i.e., feature vectors. The coupling coefficients of the principal component vectors used to calculate this hyperplane are also calculated in advance. The data of these analysis results are stored in advance in the memory unit 101. Furthermore, principal component analysis and the calculation of the Q statistic are well-known statistical methods; for details, please refer to the literature on statistical analysis, etc.

[0039] Furthermore, for the detected difference data vector ΔD and characteristic quantity of the roll gap, the presence or absence of roll gap abnormality is diagnosed by evaluating the deviation from the normal state where the roll gap is not abnormal.

[0040] The evaluation of the deviation from the normal state, i.e., the diagnosis of abnormal roll spacing, can be performed as follows. As shown in Figure 4, the distance between the hyperplane set by the feature quantities (principal component vectors) extracted in the past under normal conditions and the vector data (ΔD) containing the comparative difference data measured at multiple locations this time and the corresponding comparative difference data of the most recent past is set as the Q value. Furthermore, this Q value is compared with a threshold value. If the Q value exceeds the threshold value, an abnormal roll spacing is diagnosed. Moreover, at least one of the two roll pairs 3a is diagnosed as having an abnormality. The threshold value can be appropriately set according to the desired detection accuracy of the abnormality. For example, by setting the threshold value low, the abnormality of the roll spacing can be predicted. Furthermore, the calculated Q value can be standardized using the upper 80% of all Q values, thereby setting the threshold value to the standardized value.

[0041] The abnormal diagnosis of the roller spacing is preferably performed on all two roller pairs 3a of the same type that are close to each other within the same section 4. For example, in the case of Figure 2, it is preferable to perform abnormal diagnosis on the combination of support roller 3A and support roller 3C, the combination of support roller 3B and support roller 3D, and the combination of support roller 3C and support roller 3E respectively.

[0042] As in this embodiment, in the case of a vertically curved continuous casting machine 1, as shown in FIG1, the support roller 3 is divided into a vertical section, a curved section, and a horizontal section. Therefore, in the measurement of the roller spacing using the roller spacing meter 11, the measurement environment changes depending on the position of the support roller 3 being measured, thus the measurement value will be biased based on the position of the support roller 3. However, when the support roller 3 is normal, the measurement values ​​of support rollers 3 that are close in position, i.e., have the same measurement environment, will shift synchronously. Conversely, if the support roller 3 is abnormal, this synchronization will break down. In contrast, in this embodiment, the measurement values ​​of multiple roller spacings of the same support roller 3 that are close in position are used to capture the synchronous shift between multiple measurement values ​​as a feature, and the roller spacing data of the support roller 3 to be judged is compared with the roller spacing data of the support roller in a normal state. In this way, it can be determined whether the support roller 3 is normal or abnormal.

[0043] Furthermore, in this embodiment, the roller spacing of two roller pairs 3a that are close to each other in the same segment 4 and have the same type of bearing position is processed into a data set, and the difference data between them is used to detect anomalies. In this structure, the difference between the roller spacing data of two roller pairs 3a that are close to each other in the same segment 4 and have the same type of bearing position is used. In this way, measurement anomalies caused by the positional deviation of the roller spacing meter 11 or the physical shape of the cross-section roller 31 on the roller bearing seat 32 can be eliminated, thereby improving the accuracy of anomaly detection. Furthermore, in this embodiment, by using the current difference data and data from the most recent past times, the accuracy of anomaly detection can be further improved. Moreover, according to this embodiment, false detections are less frequent, thereby reducing the frequency of operators entering the equipment when anomalies occur.

[0044] Furthermore, in previous methods such as Patent Document 1 and Patent Document 2, it is necessary to monitor three items separately: comparison with the preceding and following roller pairs, comparison of time series data of roller pair measurement values, and comparison of measurement results at multiple locations of a roller pair. This places a heavy burden on the monitor. However, according to this embodiment, since all these items can be monitored automatically, the burden on the monitor can be reduced.

[0045] Furthermore, in previous methods such as Patent Documents 1 and 2, when a fault is detected, the support rolls must be replaced abruptly. This results in significant time loss for continuous casting machine production due to the time available for emergency personnel arrangements. However, according to this embodiment, by performing a diagnostic using the Q-value, roll interval anomalies can be predicted in advance compared to the current method of setting a threshold value for roll interval deviation. This allows for more leeway in personnel arrangements, thereby reducing production time loss.

[0046] <Variation Example> The present invention has been described above with reference to specific embodiments, but it is not intended to limit the invention by these descriptions. By referring to the description of the present invention, it will be clear to those skilled in the art that other embodiments of the present invention, including various modifications, are also apparent in conjunction with the disclosed embodiments. Therefore, it should be understood that the embodiments of the invention described in the claims also include embodiments that individually or in combination include the modifications described in this specification.

[0047] For example, in the described embodiment, the continuous casting machine 1 is a vertically curved type, but the present invention is not limited to this example. For example, the continuous casting machine 1 may also be a curved type or other types.

[0048] Furthermore, in the described embodiment, the measurement step is performed during the initial stage of casting when the dummy bar 9 passes through the support roller group, but the present invention is not limited to this example. For example, if the roller interval can be measured during the stage of inserting the dummy bar 9 into the continuous casting machine 1 before continuous casting, the measurement can also be performed before continuous casting.

[0049] Furthermore, in the aforementioned embodiment, it is assumed that two types of roller pairs 3a exist in one segment 4, but the present invention is not limited to this example. Multiple types of roller pairs 3a can exist in one segment 4, as long as multiple types exist respectively. For example, three types of roller pairs 3a can also exist in one segment 4. [Example]

[0050] The embodiments described below are examples performed by the inventors. In these embodiments, in a continuous casting machine 1 equipped with 158 roll pairs 3a containing support rolls 3 with diameters of 320 mm to 345 mm, the accuracy of anomaly detection resulting from the applicability of the anomaly diagnosis method of the present embodiment was verified based on roll interval measurement data from eight months of previous continuous casting. In previous measurements, the roll interval of the roll pairs 3a was measured at three different locations in the casting width direction. Furthermore, in this embodiment, for two roll pairs 3a of the same type and closest distance in the same segment 4, the Q statistic was calculated using a data vector created by combining the current difference data with the most recent past data, similar to the present embodiment. For the past difference data, the data from the previous two measurements were used. Moreover, as a comparative example, the same previous anomaly diagnosis method as in Patent Document 1 was used to diagnose the anomaly. Specifically, in the comparative example, the difference between the measured roll interval and the reference value was calculated, and the presence or absence of an anomaly was determined by judging whether the difference exceeded a threshold value.

[0051] Figure 5 shows the deviation and Q value of the roller spacing in the faulty support roller 3. Furthermore, in Figure 5 and Figure 6 (described later), the deviation is the difference between the previously measured value of the same roller spacing (the measured value in one roller pair 3a) and the reference value. It has been confirmed that in the faulty support roller, the Q value increases at the time the fault is detected, thus allowing the detection of roller spacing anomalies by setting a threshold value. Moreover, Figure 6 shows the deviation and Q value when a roller anomaly is falsely detected. As shown in Figure 6, it has been confirmed that the deviation becomes excessively large at the time of false detection, but the Q value does not change at the same time. Therefore, it has been confirmed that the anomaly diagnosis method according to the described embodiment can detect roller spacing anomalies with good accuracy.

[0052] Figure 7 shows the trend of roller spacing deviation when false detection occurs. Furthermore, Figure 7(A) shows the trend in support roller 3 when false detection occurs, and Figure 7(B) shows the trend of roller spacing deviation in support roller 3 of the same type and close distance within the same segment 4 as in Figure 7(A), within the same period as in Figure 7(A). In Figures 7(A) and 7(B), the time enclosed by a circle represents the time when false detection occurs. Moreover, Figures 7(A) and 7(B) depict three data points at three locations relative to the width of the casting plate. As shown in Figure 7(A), at the time enclosed by a circle, the deviation exceeds the threshold value. However, it is evident that the deviation tends to increase with the passage of the measurement date and time, and the same trend is observed in the deviation of the same type of support roller 3 shown in Figure 7(B). That is, it can be confirmed that by considering the time variation of the measured roller spacing value and the measured value of the same type of support roller as described in the embodiment, false detection can be reduced.

[0053] Figures 8 and 9 show the results of applying the anomaly diagnosis method of the described embodiment to all roll interval anomalies and false detections occurring over eight months. Furthermore, Figure 9 shows the following scenario: in the anomaly prediction result, anomalies were detected more than two days earlier than in the comparative example. As shown in Figure 8, it was confirmed that by using Q-value diagnosis, the number of false detections was significantly reduced in the embodiment compared to the comparative example. Moreover, as shown in Figure 9, it was confirmed that in the embodiment, the anomaly detection rate was improved to 100% compared to the comparative example, and roll interval anomalies could be detected earlier than in the comparative example. That is, the anomaly diagnosis method according to the described embodiment can achieve a significant reduction in false detections and can detect anomalies that could not be detected in previous methods, thus improving detection accuracy. In addition, roll interval anomalies can be detected earlier than in previous methods.

[0054] 1: Continuous casting machine 2: Casting mold 3. 3A~3E: Support rollers 3a: Roller pair 4: Section 5: Feeding trough 6: Sliding nozzle 7: Dipping nozzle 8: Pinch rollers 9: Die-starting rod 10: Abnormal diagnostic device 11: Roller gap meter 31: Dividing roller 32: Roller bearing housing (bearing) 100: Acquisition Department 101: Memory Department 102: Diagnostic Department M: Molten steel (molten steel) Pa, Pa1, Pa3, Pb, Pb1, Pb3, Pc, Pc1, Pc3: Position S: Casting plate

Claims

1. A method for diagnosing anomalies in a continuous casting machine, comprising detecting anomalies in the roller spacing of opposing roller pairs in the thickness direction of a casting plate, wherein the multiple support rollers clamp the casting plate within the continuous casting machine and guide it toward the casting direction, wherein the method measures the roller spacing of the multiple roller pairs at multiple positions in the width direction of the casting plate using a roller spacing meter installed on a dummy bar, the dummy bar moving in conjunction with the pulling movement of the casting plate, and for two roller pairs of the same type, detecting anomalies in the roller spacing using the roller spacing at the same position in the width direction of the casting plate, wherein the support rollers have multiple segmented rollers and roller bearing seats, and the two roller pairs of the same type have at least the same bearing position.

2. The method for diagnosing abnormalities in a continuous casting machine as described in claim 1, wherein, The two roller pairs of the same type are roller pairs within the same section.

3. The method for diagnosing anomalies in a continuous casting machine as described in claim 1 or claim 2, wherein, When detecting anomalies in the roll spacing, the current comparison difference data is used to detect the anomalies in the roll spacing, which is the comparison difference data of the roll spacing measured at the same position for two roll pairs of the same type as the element.

4. The method for diagnosing abnormalities in a continuous casting machine as described in claim 3, wherein, The roll gap measurement is performed repeatedly at the start of continuous casting on the continuous casting machine. The current differential data and past differential data are used to detect abnormalities in the roll gap. The past differential data are differential data measured in previous continuous casting operations.

5. The method for diagnosing abnormalities in a continuous casting machine as described in claim 4, wherein, The past poor data refers to the most recent consecutive poor data.

6. The method for diagnosing abnormalities in a continuous casting machine as described in claim 5, wherein, Based on the multi-point (n1-point) data in the current comparative poor data and the multi-point (n1×n2-point) data in the comparative poor data measured in the most recent (n2) consecutive castings, a vector data (n1×n3-point, where n3=n2+1) is calculated as an element. The deviation of the vector data from the normal state is statistically evaluated to detect anomalies.

7. An anomaly diagnosis device for a continuous casting machine, for detecting anomalies in the roller spacing of opposing roller pairs in the thickness direction of a group of support rollers comprising multiple support rollers, wherein the multiple support rollers clamp and guide the casting plate in the casting direction within the continuous casting machine, the anomaly diagnosis device for the continuous casting machine comprising: A roll gap meter is installed on a dummy bar that moves in conjunction with the pulling movement of the casting plate, and measures the roll gap of multiple roll pairs at multiple positions in the width direction of the casting plate; and a diagnostic unit detects abnormalities in the roll gap using the roll gap at the same position in the width direction of the casting plate for two roll pairs of the same type, wherein the support roller has multiple segmented rolls and roll bearing seats, and the two roll pairs of the same type have at least the same bearing position.

Citation Information

Patent Citations

  • Abnormality detection method for continuous casting facility

    JP2019155379A

  • Slab Pulling Method and Device for Horizontal Continuous Casting Device

    TW184204B