JP6358199B2 - Method and apparatus for determining surface defects of continuous cast slab, and method for producing steel slab using the surface defect determination method - Google Patents

Description

  The present invention relates to continuous casting in which molten steel is discharged into a mold and a slab (slab) is continuously drawn out from the lower end of the mold, and in particular, the presence or absence of surface defects in the slab itself or a product is determined. The present invention relates to a method and apparatus for determining a surface defect of a continuous cast slab, and a method for producing a steel slab using the surface defect determination method.

Conventionally, operation troubles and quality abnormalities are detected by installing a plurality of thermocouples in a continuous casting mold.
For example, in Patent Documents 1 and 2, a three-stage thermocouple is installed near the mold surface, and the molten steel flow in the mold is predicted from the temperature measurement in the mold range where the solidified shell thickness is about 10 mm. A method for determining the presence or absence of inclusions and bubbles that cause surface defects is disclosed.
Patent Documents 3 and 4 similarly disclose a method of detecting a vertical crack of a slab from a temperature measurement value by installing a thermocouple in a mold.
Further, in Patent Document 5, a large number (many types) of operation data is converted so that it can be expressed by a small number (small types) of variables (referred to as feature amounts), and the correspondence between the feature amounts as the representative values and the quality data. A technology for predicting product quality using a performance database of the above is disclosed.

JP 2011-11258 A JP 2012-66278 A JP 2011-206810 A JP 2011-522704 A Japanese Patent No. 5169096

However, the above prior art has the following problems.
In the methods disclosed in Patent Documents 1 and 2, even if the slab temperature is indirectly measured from the thermocouple installed in the mold, it is actually difficult to predict the molten steel internal temperature and further to predict the molten steel flow. . Even if it can be predicted well, actual slab surface defects are not only due to abnormalities in the flow of molten steel, such as bubbles and inclusions, but also, for example, uneven inflow such as flux entrainment (flowing) and runoff, oscillation It is true that many of them are caused by cracks or cooling spots due to abnormal cooling immediately under the mold, and it is difficult to predict these occurrences by measuring the temperature of the slab.

  Patent Document 3 is a method of numerically processing the temperature distribution in the mold width direction and detecting a temperature distribution exceeding a certain threshold value as an abnormal temperature distribution, but this method can detect vertical cracks but does not detect other surface defects. It is difficult to detect the cause of the abnormality. In addition, the temperature distribution in the mold width direction is important information, but in reality, the molten steel flow inside becomes complicated and asymmetric when changing the depth of the immersion nozzle in the mold, especially when electromagnetically controlled. Therefore, it is easy to misdetect temperature distribution abnormality.

  Patent Document 4 relates to a risk of causing a breakout of vertical cracks by calculating an actual temperature value measured by a thermocouple arranged in a mold based on a temperature measurement value obtained without a crack. This is a statistical evaluation. However, in actual manufacturing sites, there are continuous changes such as the difference in molten steel temperature (outgoing steel temperature) (including variations) and the immersion nozzle and casting speed during operation. It is difficult to make a numerical comparison evaluation with the “temperature measurement value obtained without cracks”.

  Patent Document 5 is a product quality prediction technique using a database, but requires equipment and a control system for storing, managing, and reading the database, and there is a concern that the capital investment cost will rise. Moreover, in the Example of patent document 5, although the product quality prediction example using a mold copper plate temperature is presented, the reference regarding the copper plate temperature data number (the number of thermocouples embed | buried in a copper plate) required for quality prediction, The defect occurrence position is not specified.

  The present invention has been made in order to solve the above-mentioned problems, and it is possible to determine the presence or absence of surface defects in a slab itself or product produced by continuous casting and to determine the position of the occurrence of surface defects in a continuous casting slab. It aims at providing the manufacturing method of the steel slab using the method and apparatus and this surface defect determination method.

(1) The method for determining surface defects of a continuous cast slab according to the present invention acquires temperature measurement data of a temperature measuring element embedded in a long side copper plate of a mold, and generates surface defects in the slab based on the temperature measurement data. The determination of the presence or absence, the placement of the temperature measuring element embedded in the long side copper plate, for the casting direction, the position of the temperature measuring element in the uppermost stage is within 200 mm downward from the hot water level control level, The position of the temperature measuring element at the lowest level is 500 mm or more downward from the molten metal surface control level, the interval between adjacent temperature measuring elements is 250 mm or less, the number of stages is 4 or more, and the mold width direction is The position of the temperature measuring element embedded in the position closest to the short-side copper plate of the mold is adjacent to within 250 mm along the direction from the intersection of the long-side copper plate and the short-side copper plate to the center of the mold width. The distance between the temperature measuring elements is 200mm or less, and the number of rows is 8 or more. And the temperature measurement data acquisition step of acquiring the temperature measurement data of the temperature measuring element arranged as described above at a predetermined time interval, and the long side surface copper plate acquired in the temperature measurement data acquisition step A temperature measurement data normalization step of calculating an average value of the temperature measurement data at each time and normalizing the temperature measurement data based on the average value, and the temperature measurement standardized in the temperature measurement data normalization step A principal component analysis step for performing principal component analysis of data, and a residual calculation step for calculating a residual between the temperature measured by the principal component and the principal component score calculated in the principal component analysis step and the temperature measurement data. A surface defect occurrence determination step of determining whether or not a surface defect has occurred in the slab based on the square sum of the residuals calculated in the residual calculation step is provided.

(2) In the device described in (1) above, the position of the temperature measuring element in the lowermost stage is within 900 mm in the casting direction from the molten metal surface control level.

(3) In the above-described (1) or (2), the surface defect occurrence determining step includes generating a surface defect in the slab when the sum of squares of the residual exceeds a predetermined threshold value. It is characterized by determining that there is.

(4) In the device according to any one of (1) to (3), the surface defect occurrence determining step determines for each row of the temperature measuring elements arranged in the casting direction. It is.

(5) The continuous defect slab surface defect determination apparatus according to the present invention acquires temperature measurement data of a temperature measuring element embedded in a long side copper plate of a mold, and surface defect occurrence in the slab based on the temperature measurement data. The determination of the presence or absence, the placement of the temperature measuring element embedded in the long side copper plate, for the casting direction, the position of the temperature measuring element in the uppermost stage is within 200 mm downward from the hot water level control level, The position of the temperature measuring element at the lowest level is 500 mm or more downward from the molten metal surface control level, the interval between adjacent temperature measuring elements is 250 mm or less, the number of stages is 4 or more, and the mold width direction is The position of the temperature measuring element embedded in the position closest to the short-side copper plate of the mold is adjacent to within 250 mm along the direction from the intersection of the long-side copper plate and the short-side copper plate to the center of the mold width. The distance between the temperature measuring elements is 200mm or less, and the number of rows is 8 or more. And temperature measurement data acquisition means for acquiring temperature measurement data of the temperature measuring element arranged as described above at a predetermined time interval, and the temperature measurement data acquisition means acquired for each of the long side copper plates A temperature measurement data normalization unit that calculates an average value of the temperature measurement data at each time , normalizes the temperature measurement data based on the average value, and the temperature measurement data normalized by the temperature measurement data normalization unit Principal component analysis means for performing principal component analysis of data, and residual calculation means for calculating a residual between the temperature measured by the principal component and the principal component score calculated by the principal component analysis means and the temperature measurement data. ,
Surface defect occurrence determination means for determining whether or not surface defects have occurred in the slab based on the sum of squares of the residuals calculated by the residual calculation means is provided.

(6) In the device described in (5) above, the position of the temperature measuring element at the bottom is within 900 mm in the casting direction from the molten metal surface control level.

(7) In the above-described (5) or (6), the surface defect occurrence determination means generates a surface defect in the slab when the sum of squares of the residual exceeds a predetermined threshold value. It is characterized in that it is determined to be present.

(8) In the above (5) to (7), the surface defect occurrence determining means determines for each row of the temperature measuring elements arranged in the casting direction. It is.

(9) A method for manufacturing a steel slab using the surface defect determination method according to the present invention uses the surface defect determination method for a continuously cast slab described in (4) above, and determines the occurrence of the surface defect. The relative position in the mold for identifying the relative position in the mold corresponding to the row of temperature measuring elements determined to have surface defects in the process, and the abnormal flow of molten steel in the mold is estimated based on the relative position in the mold And a molten steel flow control step for controlling the molten steel flow so as to eliminate the estimated molten steel flow abnormality.

In the present invention, a temperature measurement data acquisition step of acquiring temperature measurement data of a temperature measurement element embedded in a relatively wide range in the casting direction and the mold width direction in the long side surface copper plate of the mold, and the acquired temperature measurement An average value of the data is calculated for each long side copper plate of the mold, and the temperature measurement data normalization process for normalizing each temperature measurement data using the average value, and the temperature measurement data normalization process is normalized. A principal component analysis step for performing principal component analysis of the temperature measurement data, and a residual for calculating the residual between the temperature measured by the principal component and the principal component score calculated in the principal component analysis step and the temperature measurement data. By providing a difference calculation step and a surface defect occurrence determination step for determining that a surface defect has occurred when the sum of squares of the residual calculated in the residual calculation step exceeds a predetermined threshold value, Due to abnormalities in molten steel flow for slabs cast under various conditions The presence or absence of surface defects in the slab itself or product, such as the presence of inclusions and bubbles, non-uniform flux flow (powder entrainment, biting, etc.) and cooling spots, and the exact location become able to.
As a result, it is possible to efficiently change the necessity or provision of the surface for the slab after casting and the provisional grade.

It is explanatory drawing of the surface defect determination apparatus of the continuous casting slab which concerns on one embodiment of this invention. It is explanatory drawing of the example of arrangement | positioning of the temperature measuring element which concerns on one embodiment of this invention. It is a flowchart of the surface defect determination method of the continuous casting slab which concerns on one embodiment of this invention. It is a result of the time-sequential change of the deviation degree (the sum of squares of a residual) which concerns on Example 1 of this invention (the 1). It is a result of the time-sequential change of the deviation degree (the sum of the squares of a residual) of the residual which concerns on the Example of this invention (the 2). It is a result of the time-sequential change of the deviation degree (the sum of the squares of a residual) of the residual which concerns on the Example of this invention (the 3). It is a result of the time-sequential change of the deviation degree (the sum of the squares of a residual) of the residual which concerns on the Example of this invention (the 4). It is a flowchart of the manufacturing method of the steel slab which concerns on other embodiment of this invention. It is a figure which shows the example of grouping of the temperature sensing element which concerns on other embodiment of this invention.

[Embodiment 1]
A continuous defect slab surface defect determination apparatus (hereinafter simply referred to as “surface defect determination apparatus 1”) according to an embodiment of the present invention is embedded in a long-side copper plate 3a of a mold 3 as shown in FIG. The thermocouple 5 and the arithmetic device 10 for acquiring the temperature measurement data of the thermocouple 5 and performing principal component analysis to determine the presence or absence of surface defects in the slab (not shown) are provided.
Hereinafter, before describing the arrangement of the thermocouple 5 and the configuration of the arithmetic device 10 according to the first embodiment, first, an outline of principal component analysis of temperature measurement data according to the first embodiment will be described.

<Principal component analysis>
Principal component analysis is a statistical analysis method for synthesizing new variables representing features from many observed variables, and is a method suitable for finding singular points in many temperature measurement data as in the present invention. When principal component analysis is performed on time series information of temperature measurement data, a plurality of bases and basis coefficients are obtained.

Regarding the analysis method of principal component analysis, when the number of thermocouples 5 embedded in the long side copper plate 3a of the mold 3 is 100, that is, when principal component analysis is performed on time series data obtained at 100 measurement points. Will be described as an example.
When acquiring temperature measurement data from 100 thermocouples 5 at intervals of 1 second, time series data X of temperature for t seconds can be expressed as the following equation (1). In formula (1), T represents temperature, the subscript number represents a thermocouple number, and the superscript number represents time.

  In addition, the above time-series data is collectively expressed as a matrix X as in the following equation.

  The principal component y for the matrix X can be obtained as an eigenvector of a covariance matrix obtained by the following equation.

Here, three vectors are selected from the obtained eigenvectors in descending order of eigenvalues. Let these vectors be y ₁ , y ₂ , y ₃ .
The number of elements of the vectors y ₁ , y ₂ and y ₃ is 100, which is equal to the number of thermocouples. In addition, it is assumed that norms of y ₁ , y ₂ , and y ₃ are normalized to 1.

In this way, when, for example, three bases are calculated for the time series data X of temperature by principal component analysis, three types of 100-dimensional vectors y ₁ , y ₂ and y ₃ (bases 1 to 2) shown in the following equation (2) are used. A basis 3) is obtained. These bases 1 to 3 are typical patterns (changes in value with respect to time) for expressing the temperature time-series data X. Each of these three bases is weighted and combined to combine the temperature. The time series data X is to be expressed.

These y ₁ , y ₂ and y ₃ are called “bases”.
In the principal component analysis, basis coefficients a _i ^j are calculated for each time series data (X ¹ , X ^{2 to} X ^t ) of temperature for t seconds with respect to the bases 1 to 3. The basis coefficient a _i ^j is a weighting value for the above-described bases 1 to 3.
By using the basis coefficients a _i ^j for each of the bases 1 to 3, it becomes possible to characterize the time-series data pattern of temperature (a form in which the value changes with time) and extract singular points. become.
The time-series data of the temperature represented by the equation (1) can be approximated by the following equation (3) using the basis 1 to basis 3 (y ₁ , y ₂ , y ₃ ) and the basis coefficient a _i ^j .

  When Expression (3) is described in a matrix, it is expressed as follows.

However, a finite number of bases ( _{three in} the above example, y ₁ , y ₂ and y ₃ ) cannot accurately represent the time series data X of the actually measured temperature, and are approximated by bases and basis coefficients. There is an error between the measured temperature and the actually measured temperature.
Therefore, the present inventors considered to express the temperature time-series data X by further using the residual α as shown in the following equation (4).

If this represents a matrix for a plurality of thermocouple temperature, alpha corresponds to x ^t _res, are expressed as follows.

The number of elements in the vector x ^t _res are thermocouples number 100. Further, the residual can be decomposed into components derived from each row of thermocouples embedded two-dimensionally in the long side copper plate 3a of the mold 3. That is, the degree of deviation from normal fluctuations can be defined for each row of thermocouples arranged in the casting direction.
For example, when the 1,21,41,61,81 thermocouples number of thermocouples A column vector x ^t _res is expressed by the following equation.

Further, the deviation degree Q ^t (A) relating to the A column can be defined by the following equation using the square sum of the residuals. The same applies to the other columns.

The illustrated time-series changes in the deviation of Q ^t is 4-7. 4 to 7, the horizontal axis is min, and the vertical axis is K ² (K: absolute temperature).

The residual α is an error that cannot be approximated only by the base (y ₁ , y ₂ , y ₃ ,...) And the base coefficient a _i ^j , and can be regarded as representing the degree of deviation from the normal state. Here, the normal state is the molten steel flow state in the mold when “a surface defect of the product has not occurred” and occupies most of the casting time.
That is, by monitoring the change in the sum of squares of the residuals, it is possible to detect non-stationary and local temperature behavior, and as a result, it can be determined whether or not surface defects have occurred in the slab.
In the following description, the bases 1 to 3 are referred to as “first principal component” to “third principal component”, and the base coefficient is referred to as “principal component score (principal component score)”.

<Arrangement of thermocouple>
The preferred arrangement of the thermocouple 5 embedded in the casting direction and the mold width direction in the long side surface copper plate 3a of the mold 3 was examined in each of the casting direction and the mold width direction. As a result, the following findings (knowledge i to knowledge vi) were obtained. Hereinafter, each knowledge is demonstrated in order.

≪Arrangement in casting direction≫
The following knowledge was acquired about the range and space | interval which arrange | position in the casting direction of the thermocouple 5 (knowledge i, knowledge ii, knowledge iii). In the following description, “position” represents the position in the casting direction with the molten metal level control level as a base point, and the molten metal surface control level refers to the amount injected when pouring molten steel from the tundish into the mold 3. It is the hot water level that is the target of control.

  In the range from the molten metal level control level to 200 mm in the casting direction, surface defects are likely to occur in the slab, and when the uppermost position of the thermocouple 5 is outside the range of 200 mm from the molten metal level control level, casting at the extreme surface layer of the slab Some cases overlooked the occurrence of defects. Therefore, it is desirable that the uppermost position of the thermocouple 5 is within a range of 200 mm from the molten metal level control level, and more preferably within a range of 180 mm from the molten metal level control level (Knowledge i-1).

  If the lowermost position of the thermocouple 5 is below 500 mm from the molten metal surface control level, the molten steel flow due to the discharge flow from the immersion nozzle can be sufficiently captured, and casting that causes surface defects of the product. Of the defects (inclusions, bubbles), the occurrence of structural defects at a depth of about 8 to 12 mm from the slab surface is not overlooked. Therefore, it is desirable that the upper limit of the lowermost position of the thermocouple 5 is lower than 500 mm from the hot water level control level (Knowledge i-2).

However, it is desirable that the lower limit of the lowermost position of the thermocouple 5 be within a range from the molten metal surface control level to 900 mm (Knowledge i-3). The reason for this is as follows.
The solidified shell has already been sufficiently formed at a position of 900 mm or more from the molten metal level control level, and it is difficult to reflect the mold powder and deoxidation products that cause surface defects at this position even in the slab temperature. In addition, since the mold powder and deoxidation product captured at a position lower than 900 mm from the molten metal level control level are relatively inside the slab, it is considered that surface defects are less likely to occur even when rolled.

  Therefore, the lower limit of the lowest position of the thermocouple 5 is desirably 900 mm. In other words, it is sufficient to embed the thermocouple 5 within the range of 900 mm in the casting direction from the molten metal level control level, and burying the thermocouple 5 below it is effective in detecting cooling abnormality directly under the mold. However, it is not always necessary from the viewpoint of detecting surface defects generated in the slab. Even if the thermocouple 5 is buried below 900 mm, it only increases the thermocouple cost and the data processing load.

  From the above findings (i-1) to (i-3), the range in which the thermocouple 5 is arranged in the casting direction is within the range from the molten metal level control level to 200 mm, and the bottom is between 500 mm and 900 mm. It is desirable to be within the range (Knowledge i).

The interval at which the thermocouples 5 are arranged in the casting direction is sufficient if the interval at which the thermocouples 5 are arranged within the range of 50 mm to 900 mm in the casting direction is 250 mm or less based on the level control level. It was clarified that it can be determined (Knowledge ii).
When the interval between the thermocouples 5 was larger than 250 mm, there was a case where the behavior of the generation of scabs was overlooked.

  The number of stages in which the thermocouple 5 is arranged in the casting direction is preferably four or more from the viewpoint of securing a sufficient number of data to be analyzed for principal component analysis of the temperature measurement data of the thermocouple 5 (knowledge iii). .

≪Arrangement in the mold width direction≫
Similar to the arrangement of the thermocouple 5 in the casting direction, the range, interval, and number of rows in which the thermocouple 5 is arranged in the mold width direction were examined. As a result, the following findings (knowledge iv, knowledge v, and knowledge vi) were obtained and will be described in order.

First, the thermocouple 5 disposed at a position closest to the short side copper plate 3b in the mold width direction of the long side copper plate 3a is 250 mm in the mold width direction from the intersection of the short side copper plate 3b and the long side copper plate 3a. It is desirable to be in the following range, more preferably in the range of 240 mm or less (knowledge iv).
When the thermocouple 5 was not arranged in the range, there was a case where the behavior of casting defect generation in the vicinity of the short side copper plate 3b was overlooked.

In continuous casting, operation is performed by changing the casting width according to the required dimensions of the product. The casting width under the operating conditions investigated this time was approximately 700 mm to 2100 mm. In this case, if the distance between the thermocouples 5 in the mold width direction is 200 mm or less, more preferably 180 mm or less, the presence or absence of surface defects is sufficient. (Knowledge v).
When the interval between the thermocouples 5 in the mold width direction was larger than 200 mm, there was a case in which the behavior of shave generation was overlooked.

  Regarding the number of thermocouples 5 arranged in the mold width direction, if the number is less than 8, the temperature measurement data that is the basis of the analysis target data cannot be secured when performing principal component analysis. It is desirable that the number of rows of thermocouples 5 to be arranged is 8 or more (knowledge vi).

  In FIG. 2, the example of arrangement | positioning of the thermocouple 5 embed | buried under the long side surface copper plate 3a is shown. In FIG. 2, the thermocouple 5 is arranged in the casting direction so that the uppermost position is 50 mm from the molten metal level control level, the lowermost position is 850 mm from the molten metal surface control level, the interval between adjacent thermocouples 5 is 120 mm to 170 mm, and the number of stages is The arrangement of the thermocouples 5 in the mold width direction is 7 stages. The position of the thermocouple 5 closest to the short side copper plate 3b is 250 mm, the interval between the adjacent thermocouples 5 is 133 mm, and the number of rows is A row to P row. Up to 16 columns. Therefore, the arrangement of the thermocouples 5 satisfies the above findings i to vi in both the casting direction and the mold width direction.

  The thermocouple 5 is drilled from the surface that contacts the mold frame (not shown) of the long side copper plate 3a toward the surface that contacts the molten steel, and the hot junction of the thermocouple 5 is in contact with the bottom of the drilled tip. Buried in In the arrangement example shown in FIG. 2, the distance from the bottom of the tip to the surface in contact with the molten steel of the long side copper plate 3a is 15 mm.

  In the first embodiment, the thermocouple 5 is used as the temperature measuring element. However, any type of temperature measuring element can be used as long as it can accurately measure the temperature, such as an optical fiber sensor. There is no.

<Calculation device>
The arithmetic unit 10 is constituted by a computer such as a PC, and a temperature measurement data acquisition unit 11 that acquires temperature measurement data of the thermocouple 5 embedded in the long-side copper plate 3a and normalizes the acquired temperature measurement data. Approximated by the principal component and principal component scores calculated by the principal component analysis means 15, the principal component analysis means 15 that performs principal component analysis of the normalized temperature measurement data, Residual calculation means 17 for calculating a residual between the temperature and the temperature measurement data, and surface defect occurrence for determining the presence or absence of a defect on the slab surface based on the sum of squares of the residual calculated by the residual calculation means 17 Determination means 19.

≪Temperature measurement data acquisition means≫
The temperature measurement data acquisition means 11 is a means for acquiring temperature measurement data from the thermocouple 5 (see FIG. 1) embedded in the long side copper plate 3a at predetermined time intervals.
In the first embodiment, it is desirable that the predetermined time interval for acquiring the temperature measurement data is an interval of 1 second to 30 seconds. The reason is as follows. In order to detect the temperature fluctuation, 1 second or more and 30 seconds or less are sufficient, and when the temperature is acquired at intervals shorter than 1 second, it becomes easy to pick up the influence of disturbance such as mold vibration. In addition, measurement at intervals exceeding 30 seconds increases the risk of overlooking temperature fluctuations caused by abnormalities.

≪Temperature measurement data standardization means≫
The temperature measurement data standardization means 13 calculates the average value for each long side copper plate 3a of the temperature measurement data acquired by the temperature measurement data acquisition means 11 at each time, and subtracts the average value from each temperature measurement data. This standardizes each temperature measurement data.

  In actual manufacturing sites, there are differences in molten steel temperature (steel temperature), continuous operation conditions during casting such as the immersion nozzle depth and casting speed, etc., so that the measurement data acquired by the temperature measurement data acquisition means 11 is used. If the temperature data is simply subjected to principal component analysis by the principal component analysis means 15, it is difficult to make a numerical comparison evaluation with a reference value. Therefore, the temperature measurement data normalization means 13 calculates an average value for each side of the two long side copper plates 3a facing each other of the mold 3 at each time, and normalizes by subtracting the average value from each temperature measurement data. By doing this, the bias of the temperature measurement data is removed, and a relative change in temperature at each thermocouple position is obtained.

≪Principal component analysis means≫
The principal component analysis means 15 is a means for performing principal component analysis of the temperature measurement data normalized by the temperature measurement data normalization means 13 and calculating a principal component and a principal component score. For the principal component analysis means 15, for example, general-purpose statistical analysis software can be used. By performing principal component analysis simultaneously with the acquisition of temperature measurement data by the temperature measurement data acquisition means 11, the principal component and the principal component score are obtained. It is possible to calculate in real time.

≪Residual calculation means≫
The residual calculation means 17 calculates the residual (α in equation (4)) between the temperature approximated from the principal component and the principal component score calculated by the principal component analysis means 15 and the temperature measurement data of the thermocouple 5. To do.
In calculating the residual, it is desirable to approximate the temperature by using at least a second principal component or more, and it is more desirable to approximate the temperature using the third principal component.

≪Surface defect occurrence judging means≫
The surface defect occurrence determination unit 19 determines whether or not a surface defect has occurred in the slab based on the residual calculated by the residual calculation unit 17, and the sum of squares of the residual exceeds a predetermined threshold value. If it is determined that the surface defect is present in the slab.
By setting a threshold value for the sum of squares of the residuals, it is possible to detect and detect with high probability both relatively minor defects that can be relieved by slab care and serious defects that are difficult to establish as a product.
Furthermore, by determining the square sum of the residual α for each column of the thermocouples 5 arranged in the casting direction, it is possible to identify the occurrence position of surface defects and improve the efficiency of slab maintenance in the subsequent process. it can.

≪Storage device, recording / output device≫
The determination result obtained by the surface defect occurrence determination means 19 is stored in a storage device such as a memory and / or output via a recording / output device such as a monitor, display or printer. The slab is inspected according to the output determination result, and, if necessary, treatment such as care is performed, and the slab is conveyed to the next step, for example, a step of performing hot rolling or cold rolling.

Therefore, based on the output of the determination result, it is possible to efficiently change the necessity of the slab after casting and the allocation grade.
Here, examples of the care for the slab include removal of defects on the surface of the slab with a scarf machine, a grinder, or the like. Minor defects are transported to the next process after the care process.
On the other hand, when the surface defect occurrence determination means 19 determines that the occurrence of surface defects is “no” when the sum of squares of the residuals is equal to or less than the threshold value, the slab can be transported to the next process without taking the above care.

In addition, for charges determined to have surface defects, it is possible to prevent defects from occurring in subsequent cast slabs by feedback control that changes the casting conditions. Become.
An example of a specific method regarding this point will be described in a second embodiment described later.

  Since the surface defect determination apparatus 1 as described above does not use a database as in the prior art (for example, Patent Document 5), the capital investment cost does not increase, and a large amount of operation data is unnecessary, so that it is simple. is there.

  Regarding the surface defect determination method of the continuous casting slab using the surface defect determination device 1 configured as described above (hereinafter, sometimes simply referred to as “surface defect determination method”), together with the operation of the surface defect determination device 1, FIG. This will be described based on the flowchart shown in FIG.

<Surface defect judgment method for continuous casting slab>
As shown in FIG. 3, the method for determining surface defects of a continuous cast slab according to Embodiment 1 of the present invention obtains temperature measurement data of the mold 3 measured by the thermocouple 5 embedded in the long side copper plate 3a. A temperature measurement data acquisition step S1, a temperature measurement data normalization step S3 for normalizing the acquired temperature measurement data, a principal component analysis step S5 for performing a principal component analysis of the normalized temperature measurement data, A residual calculation step S7 for calculating a residual between the temperature measured data and the temperature represented by the principal component and the principal component score calculated in the principal component analysis step, and the residual calculated in the residual calculation step S7 And a surface defect occurrence determination step S9 for determining whether or not a surface defect is generated in the slab based on the square sum.
Hereinafter, each step will be described.

≪Temperature measurement data acquisition process≫
The temperature measurement data acquisition step S1 is a step of acquiring temperature measurement data from the thermocouple 5 embedded in the long side copper plate 3a using the temperature measurement data acquisition means 11 at a predetermined time interval.
The time interval for acquiring temperature measurement data is preferably 1 second or more and 30 seconds or less.

≪Temperature measurement data standardization process≫
The temperature measurement data normalization step S3 uses the temperature measurement data normalization means 13 to calculate the average value for each long side surface copper plate 3a at each casting time from the temperature measurement data acquired in the temperature measurement data acquisition step S1, This is a step of normalizing each temperature measurement data by calculating a difference between each temperature measurement data and the average value.

≪Principal component analysis process≫
In the principal component analysis step S5, the principal component analysis of each temperature measurement data normalized in the temperature measurement data normalization step S3 is performed using the principal component analysis means 15, and the principal component and the principal component score of each temperature measurement data are obtained. It is a process of calculating.

≪Residual calculation process≫
Residual calculation step S7 uses residual calculation means 17 to calculate the residual between the temperature measured by the principal component and the principal component score calculated in principal component analysis step S5 and the actual temperature measurement data (formula (4)). This is a step of calculating α).

≪Surface defect occurrence determination process≫
The surface defect occurrence determination step S9 is a step of using the surface defect occurrence determination means 19 to determine whether or not a surface defect has occurred in the slab based on the residual calculated in the residual calculation step S7. Specifically, when the residual sum of squares exceeds a predetermined threshold value, it is determined that surface defects have occurred in the slab, and when the residual sum of squares is less than or equal to the threshold value, It is determined that there is no occurrence.

The determination result of surface defect occurrence is stored in a storage device such as a memory and / or output by a recording / output device such as a monitor, display or printer, and the slab is inspected based on the determination result, and if necessary, maintained Then, it is transported to the next process.
Here, by determining the square sum of the residuals for each column of the thermocouples 5, it is possible to efficiently identify the position where the surface defect has occurred when inspecting the slab.

The cast slab is subjected to a slab treatment in the slab treatment step S11 based on the determination result of the surface defect occurrence, and is then transferred to the rolling step S13.
As the slab treatment, for example, when it is determined that the occurrence of surface defects is “present” in the row in the casting direction of the thermocouple 5 embedded in the long side copper plate 3a, the slab corresponding to the position of the row of the thermocouple 5 The surface of the part may be cared for with a scarf, a grinder, etc., and the slab from which surface defects have been removed by such caring is conveyed to the rolling step S13.
On the other hand, when it is determined that the occurrence of surface defects is “no”, the cast slab is transported to the rolling step S13 without surface maintenance.

  As described above, in the first embodiment, the arrangement of the temperature measuring elements (thermocouples 5) embedded in the long side copper plate 3a of the mold 3 is the same as the position of the uppermost temperature measuring element in the casting direction. Within 200mm from the molten metal surface control level, the position of the bottom temperature sensor is 500mm or more away from the molten metal surface control level, the distance between adjacent temperature sensors is 250mm or smaller, the number of plates is 4 or more, and the mold width As for the direction, the position of the temperature measuring element arranged at the location closest to both short sides is 250 mm along the direction from the intersection of the short side surface and long side surface of the slab width to be measured toward the mold width center. The distance between adjacent temperature measuring elements is 200 mm or less, the number of columns is 8 or more, temperature measurement data acquisition step S1, temperature measurement data normalization step S3, principal component analysis step S5, and residual calculation step Through S7 and the surface defect occurrence determination step S9, the determination result of the surface defect occurrence is stored in the memory. By storing and / or recording / outputting to a printer, display, etc., it is possible to determine the presence or absence of surface defects in the slab without increasing capital investment costs, and appropriate slabs can be determined based on the determination results. By applying the treatment, a slab having excellent surface quality can be efficiently produced.

[Embodiment 2]
A method for manufacturing a steel slab using a surface defect determination method according to another embodiment of the present invention is a method for determining a surface defect of a continuous casting slab during continuous casting of steel described in Embodiment 1 according to the present invention. When the presence or absence of surface defects has been determined, the mold flow abnormality in the mold is estimated from the surface defect occurrence position, and the molten steel flow is performed so as to eliminate the molten steel flow abnormality. As shown in FIG. 8, a temperature measurement data acquisition step S1 for acquiring temperature measurement data of the mold 3 and a temperature measurement for normalizing the acquired temperature measurement data are performed. A data normalization step S3, a principal component analysis step S5 for performing a principal component analysis of the normalized temperature measurement data, a temperature represented by the principal component and the principal component score calculated in the principal component analysis step, and Residual calculation to calculate residual with temperature measurement data Step S7, surface defect occurrence determination step S9 for determining the presence or absence of surface defects in the slab for each column of thermocouples 5 based on the sum of squares of the residuals calculated in the residual calculation step S7, and occurrence of surface defects Based on the in-mold relative position specifying step S15 for specifying the relative position in the mold in the mold width direction corresponding to the row of the thermocouples 5 determined as follows, and the in-mold relative position specified in the in-mold relative position specifying step S15. And a molten steel flow control step S17 for controlling the molten steel flow.

  The temperature measurement data acquisition step S1, the temperature measurement data normalization step S3, the principal component analysis step S5, the residual calculation step S7, and the surface defect occurrence determination step S9 are the same as the contents described in the first embodiment according to the present invention. Therefore, hereinafter, each of the in-mold relative position specifying step S15 and the molten steel flow control step S17 will be described in detail.

<In-mold relative position identification process>
The relative position in the mold specifying step S15 is a process for specifying the relative position in the mold in the mold width direction corresponding to the row of the thermocouples 5 determined as having surface defects in the surface defect occurrence determining step S9.
An example of a method for specifying the relative position in the mold in the mold width direction will be described below.

  First, with respect to the thermocouple 5 embedded in the long side copper plate 3a of the mold 3, two or more rows of thermocouples 5 adjacent to each other in the mold width direction are grouped into one group, and three or more groups are provided for each long side copper plate 3a. Divide into groups.

Next, a relative position in the mold in the mold width direction of each group divided is determined.
Then, based on the relative position in the mold of the group including the row of thermocouples 5 determined to have surface defect occurrence in the surface defect occurrence determination step S9, it corresponds to the row of thermocouples 5 determined to have surface defect occurrence. The relative position in the mold in the mold width direction is specified.

  The relative position in the mold specified in the second embodiment is the end of the mold 3 (near the short side surface of the mold 3) or near the center.

  The group whose relative position in the mold is the end of the mold 3 is one or two groups closest to the short side surface of the mold 3 (two or four groups with respect to both short side surfaces of the mold 3). .

  On the other hand, when the group in which the relative position in the mold is near the center of the mold 3 is grouped into an odd number of groups, the center is one group in the mold width direction in the case of three groups, and the center in the group of five or more groups. Or three groups including both adjacent groups of the central group, and when grouped into an even number of groups, 2 or 4 on both sides of the center line in the mold width direction of the long side copper plate 3a Group.

If the row of thermocouples 5 determined to have surface defects in the surface defect occurrence determination step S9 belongs to a group that is an end of the mold 3, the relative position in the mold where the surface defects have occurred is in the mold width direction. Identifies the end.
Further, when the row of thermocouples 5 determined to have surface defect occurrence in the surface defect occurrence determination step S9 belongs to a group near the center of the mold 3, the relative position in the mold where the surface defect has occurred is in the mold width direction. Identify near the center.

The method for specifying the relative position in the mold will be specifically described in the case where the rows of the thermocouples 5 are grouped into 6 groups as shown in FIG.
As shown in FIG. 9, the thermocouples 5 embedded in the long side surface copper plate 3a of the casting mold 3 from the A row to the P row are connected to the AB row (group 1), the CE row (group 2), When the group is grouped into six groups of FH column (group 3), IK column (group 4), LN column (group 5), and OP column (group 6), mold 3 The relative position in the mold of the group 1 and the group 6 closest to the short side surface of the mold is the end of the mold 3. Further, the relative positions in the mold of the group 3 and the group 4 on both sides of the center line C of the long side copper plate 3 a are set near the center of the mold 3.

When it is determined in the surface defect occurrence determination step S9 that the surface defect has occurred in the B row, the B row belongs to the group 1 whose relative position in the mold is the end of the mold 3, and thus the mold in which the surface defect has occurred. The inner relative position is specified as the end of the mold 3.
Further, when it is determined in the surface defect occurrence determination step S9 that the surface defect has occurred in the F row, the F row belongs to the group 3 in which the relative position in the mold is near the center of the mold 3, and thus the mold in which the surface defect has occurred. The inner relative position is specified to be near the center of the mold 3.

  In the example shown in FIG. 9, the group 2 and the group 5 do not correspond to either the end portion or the central portion of the mold 3.

<Molten steel flow control process>
The molten steel flow control step S17 estimates the molten steel flow abnormality in the mold 3 from the relative position in the mold identified as having surface defects in the in-mold relative position specifying step S15, and eliminates the estimated molten steel flow abnormality. This is a process for controlling the flow of molten steel.

The molten steel flow abnormality estimated in the molten steel flow control step S17 includes insufficient molten steel flow or excessive molten steel flow.
In the case of insufficient molten steel flow, the bubbles of inert gas blown from the immersion nozzle (not shown) in the mold 3 are taken into the solidified shell of the slab and become surface defects. It tends to occur at the end in the mold width direction.
On the other hand, in the case of excessive molten steel flow, powder floating on the molten metal surface in the mold 3 is caught in the slab and becomes a surface defect. The surface defect caused by excessive molten steel flow is near the center of the mold 3 in the mold width direction. It is easy to occur.

  Therefore, when it is determined that a surface defect has occurred at the end of the mold 3 in the in-mold relative position specifying step S15, in the molten steel flow control step S17, it is estimated that the molten steel flow is insufficient, and the insufficient molten steel flow is eliminated. To control the molten steel flow. Specifically, the molten steel flow can be promoted by increasing the amount of molten steel blown from an immersion nozzle (not shown).

  On the other hand, when it is determined in the relative position specifying step S15 in the mold that a surface defect has occurred near the center of the mold 3, the molten steel flow control step S17 estimates that the molten steel flow is excessive and eliminates the molten steel flow excess. To control the flow of molten steel. Specifically, the molten steel flow can be suppressed by reducing the amount of molten steel blown from an immersion nozzle (not shown).

  As described above, by adjusting the control condition of the molten steel flow so as to eliminate the molten steel flow abnormality in the molten steel flow control step S17, a surface defect is generated in the slab manufactured after it is determined that the surface defect is generated. Can be prevented.

  From the above, when it is determined by the surface defect determination method according to the present invention that surface defect has occurred, the relative position in the mold where the surface defect has occurred is determined based on the column of the thermocouple 5 that has acquired the determined temperature measurement data. The countermeasure is followed by identifying and estimating a molten steel flow abnormality in the mold 3 from the identified relative position in the mold and controlling the molten steel flow so as to eliminate the estimated molten steel flow abnormality. Therefore, the occurrence of surface defects in the same charge can be reduced.

  In the above-mentioned mold relative position specifying step S15, the thermocouples 5 are grouped in units of columns, and the relative position in the mold where the surface defect has occurred is specified by the position of the group in the mold width direction. However, the relative position (end portion or center portion) in the mold may be specified based on the position in the mold width direction of the rows of the thermocouples 5 without being grouped in units of rows of the thermocouples 5.

  The molten steel flow control step S17 performs control such as promoting or suppressing molten steel flow in the mold by the amount of molten steel blown from the immersion nozzle. In the case of a casting machine capable of electromagnetic flow control, The flow of molten steel in the mold 3 can be controlled by electromagnetic flow. Therefore, in this Embodiment 2, molten steel flow control process S17 may perform electromagnetic flow control. In addition, control of molten steel flow can be performed online.

  Furthermore, although the thermocouple 5 is used as the temperature measuring element in the second embodiment, as in the first embodiment, any method can be used as long as the temperature can be accurately measured, such as an optical fiber sensor. There is no problem even with the temperature measuring element.

A specific experiment for confirming the effect of the arrangement of the thermocouple 5 in the method and apparatus for determining surface defects of the continuous cast slab of the present invention was conducted, and the results will be described below.
In the experiment, the surface defect determination apparatus 1 shown in FIG. 1 is used to determine the surface defect by changing the arrangement of the thermocouple 5 embedded in the long-side copper plate 3a of the mold 3, and the actual surface defect probability is evaluated. It is to do.
As shown in Table 1, the thermocouples 5 are arranged in the range satisfying the findings (i) to (vi) described in the first embodiment, and Examples 1 to 5 of the present invention and Comparative Examples. Comparative Examples 1 to 5 that are not included in the scope of the present invention were examined.

  The thermocouple 5 is a JIS-T type thermocouple, which is drilled from the surface contacting the mold frame of the long side copper plate 3a toward the surface contacting the molten steel, and the hot junction of the thermocouple 5 is provided at the bottom of the drilled tip. Buried in contact. The distance from the bottom of the tip to the surface of the long side copper plate 3a contacting the molten steel was 15 mm.

Using the vertical bending type continuous casting machine in which the thermocouple 5 was embedded in this way, aluminum killed molten steel was continuously cast. The casting conditions were a casting thickness of 220 to 300 mm, a casting width of 1900 to 2200 mm, and a molten steel throughput of 3.0 to 7.5 ton / min.
Molten steel is supplied from a tundish (not shown) into the mold 3 by a dipping nozzle. The molten steel discharge angle of the molten steel discharge hole in the dipping nozzle is downward 15 ° to 45 °, and the dipping depth is controlled by the molten metal surface. The distance from the level to the upper end of the molten steel discharge hole was in the range of 180 mm to 300 mm. Ar gas was used as the inert gas blown from the immersion nozzle. The molten steel in the mold 3 is applied with a moving magnetic field in the opposite direction along the opposing long side copper plate 3a from the magnetic field generator, so that the molten steel in the mold 3 is horizontally aligned along the solidified shell interface. A swirling flow was imparted.

  The cast slab is transported to the process of rolling without maintenance with the scarf or grinder etc. (hereinafter referred to as “uncleaned”), and subjected to hot rolling, cold rolling, etc. Later, surface defects were continuously measured with an on-line surface defect meter. And the comparison with the determination result of the surface defect generation | occurrence | production obtained by the surface defect determination apparatus 1 was performed, and the surface defect middle rate was evaluated.

Principal component analysis was performed on all temperature measurement data in the same charge, and the principal component analysis was performed once on the temperature measurement data for each charge, and the first to third principal components and the principal component score were calculated. . Then, it calculates a residual between the actual temperature measurement data with temperature which is approximated by the principal component and principal component score, if the square sum of residuals exceeds 100K ^2, the surface defects of the slab and "present" Judged.

  In Example 1, the time interval for acquiring the temperature measurement data was 5 seconds, and the computer language MATLAB (registered trademark) for technical calculation manufactured by Mathworks was used for the analysis of the principal component analysis.

4 to 7 show examples of changes over time of the square sum of the residual α calculated for each column from the A column to the P column of the thermocouple 5. In E column, G column, H column, M column, N column, and O column, unsteady behavior in which the sum of squares of the residual α exceeds 100K ² is seen, and it is determined that surface defects are present. It was.

In this example, the number of slabs determined to have surface defects “existence” based on the square sum of residual α (= X), the number of slabs in which surface defects were detected after actual rolling (= Y) The detection rate (target) rate Z = Y / X × 100 [%] was determined.
Furthermore, the number of slabs (= Q) when surface defects actually occurred even though it was not determined that surface defects were found to be “present” was the total number of slabs (= R) where surface defects were actually generated by rolling. What was divided was evaluated as a non-detection rate (missing rate) P = Q / R × 100 [%].
Each of Invention Examples 1 to 5 and Comparative Examples 1 to 5 was evaluated for a casting amount of about 300 charges (about 300 tons per charge).

The results of the surface defect detection rate and non-detection rate are shown in Table 1 above.
In the arrangement of the thermocouple 5 satisfying the present invention (Invention Example 1 to Invention Example 5), the defect detection rate is a high value exceeding 81%, indicating that the occurrence of surface defects on the surface of the slab can be determined well. The non-detection rate was 41% to 55%.

In Comparative Example 1, the number of stages of the thermocouple 5 is three, which is less than four, the detection rate is as low as 47%, and the non-detection rate is as high as 69%.
In Comparative Example 2, the number of rows of the thermocouple 5 was 7 rows smaller than 8 rows, the detection rate was as low as 58%, and the non-detection rate was as high as 80%.
In Comparative Example 3, the interval in the mold width direction of the thermocouple 5 was 257 mm, which was larger than 250 mm, the detection rate was 57%, and the non-detection rate was 65%, which was a lower result than the present invention example. .
In Comparative Example 4, the uppermost position in the casting direction of the thermocouple 5 was 210 mm larger than 200 mm from the molten metal surface control level, the detection rate was as low as 47%, and the non-detection rate was as high as 85%.
In Comparative Example 5, the lowermost position in the casting direction of the thermocouple 5 is 480 mm outside the range of 500 mm to 900 mm from the molten metal surface control level, and the detection rate is 67%, which is a lower result than the example of the present invention. .
From the above, it was shown that when the thermocouple 5 embedded in the long-side copper plate 3a of the mold 3 is not disposed within the scope of the present invention, it is not possible to satisfactorily determine the occurrence of surface defects in the slab.

In Example 2, productivity was compared depending on whether or not the surface defect determination apparatus 1 shown in FIG. 1 was applied.
In Example 2, the arrangement of the thermocouple 5 embedded in the long-side copper plate 3a of the mold 3 is the same as in the case of Invention Example 4 shown in Table 1, and the same casting conditions as in Example 1 are used. did.

As a result of comparing the required days from the start of casting of the slab to the shipment of the product through subsequent processing such as rolling, when the number of days when the surface defect determination device 1 is not applied is normalized to 100, the present invention is applied. When the surface defect determination apparatus 1 is applied, the required number of days is 88, and the lead time from production to shipment is improved by 12% (= (100-88) / 100 × 100).
This is because the introduction of the surface defect determination apparatus 1 makes it possible to efficiently perform the necessity of surface maintenance for the slab after casting and change of the assigned grade.

  As described above, the continuous casting slab surface defect determination method and apparatus according to the present invention is applied to perform continuous casting, thereby detecting the occurrence of defects on the surface of the slab before the rolling process, and surface care for the slab after casting. It has been proved that productivity can be improved by efficiently changing the grades of the necessity and provision.

  Next, since the experiment which compares the defect incidence rate of the slab surface at the time of manufacturing a steel slab by the manufacturing method of the steel slab using the surface defect determination method concerning this invention is demonstrated, the result is demonstrated.

  The experiment is the same as in Example 1 described above, and when the flow of molten steel is controlled based on the determination result obtained by using the surface defect occurrence determination method according to the present invention for two different charges, The rate at which slab defects actually occurred (defect generation rate) was obtained for two separate charges when the molten steel flow was not controlled, and the two were compared.

  When the steel slab manufacturing method according to the present invention is applied, the presence or absence of surface defects is determined for each column of temperature measurement data obtained by a thermocouple. The relative position in the mold where the surface defect occurred was specified, and the molten steel flow was controlled based on the relative position in the mold.

  As shown in FIG. 9, the arrangement and grouping of the thermocouples embedded in the mold 3 are 16 rows and 6 groups in the mold width direction, and the groups 1 (AB row) and 6 (OP row) are arranged. The relative position in the mold is the end of the mold 3, the relative position in the mold of the group 3 (FH row) and the group 4 (I-K row) is near the center of the mold 3, and surface defects are found in the continuously cast slab. The relative position in the mold in the mold width direction of the group to which the thermocouple row determined to be generated belongs was specified.

  In the case where the molten steel flow in the mold 3 is not controlled using the determination result obtained by the surface defect determination method according to the present invention, the surface defect according to the present invention is assumed to be 100 when the ratio of slabs in which defects that need to be generated are generated. When the flow of molten steel in the mold 3 was controlled using the determination result of surface defect generation by the determination method, the ratio of slabs in which defects that needed to be generated occurred was 60, and the defect generation rate was greatly improved.

  As described above, using the determination result of the surface defect generation obtained by the surface defect determination method according to the present invention, the relative position in the mold where the surface defect has occurred is specified, and the molten steel in the mold 3 is based on the relative position in the mold. It has been demonstrated that by controlling the flow, defects generated on the surface of the slab after the control of the molten steel flow can be reduced.

DESCRIPTION OF SYMBOLS 1 Surface defect determination apparatus 3 Mold 3a Long side copper plate 3b Short side copper plate 5 Thermocouple 10 Arithmetic device 11 Temperature measurement data acquisition means 13 Temperature measurement data normalization means 15 Principal component analysis means 17 Residual calculation means 19 Surface defect generation Judgment means

Claims (9)

Obtaining temperature measurement data of the temperature measuring element embedded in the long side surface copper plate of the mold, and based on the temperature measurement data is a surface defect determination method of the continuous casting slab for determining the presence or absence of surface defects in the slab,
The arrangement of the temperature measuring element embedded in the long side copper plate,
Regarding the casting direction, the position of the temperature measuring element on the uppermost stage is within 200 mm below the molten metal level control level, and the position of the temperature measuring element on the lowermost stage is located at a position 500 mm or more below the molten metal surface control level, adjacent to each other. The interval between the temperature measuring elements is 250 mm or less, the number of steps is 4 or more,
Regarding the mold width direction, the position of the temperature measuring element embedded in the position closest to the short side copper plate of the mold is set in a direction from the intersection of the long side copper plate and the short side copper plate toward the center of the mold width. Within 250mm, the interval between the temperature measuring elements adjacent to each other is 200mm or less, the number of rows is 8 or more,
A temperature measurement data acquisition step of acquiring temperature measurement data of the temperature measurement element arranged as described above at a predetermined time interval ;
Normalization of temperature measurement data that calculates an average value of the temperature measurement data acquired for each of the long side surface copper plates at each time in the temperature measurement data acquisition step and normalizes the temperature measurement data based on the average value Process,
A principal component analysis step for performing a principal component analysis of the temperature measurement data normalized in the temperature measurement data normalization step;
A residual calculation step of calculating a residual between the temperature measured data and the temperature represented by the principal component and the principal component score calculated in the principal component analysis step;
A surface defect determination method for a continuous cast slab, comprising a surface defect occurrence determination step for determining whether or not a surface defect has occurred in the slab based on a sum of squares of the residuals calculated in the residual calculation step.   The method for determining a surface defect in a continuous cast slab according to claim 1, wherein the position of the temperature measuring element at the lowest level is within 900 mm in the casting direction from the level control level.   The said surface defect generation | occurrence | production determination process determines with the generation | occurrence | production of a surface defect in the said slab, when the square sum of the said residual exceeds the predetermined threshold value determined beforehand, The surface defect generation | occurrence | production is characterized by the above-mentioned. Method for determining surface defects in continuous cast slabs.   4. The surface defect determination method for a continuous cast slab according to claim 1, wherein the surface defect occurrence determination step is performed for each row of the temperature measuring elements arranged in a casting direction. 5. . It is a surface defect determination device for a continuous casting slab that acquires temperature measurement data of a temperature measuring element embedded in a long side surface copper plate of a mold, and determines the presence or absence of surface defect occurrence in the slab based on the temperature measurement data,
The arrangement of the temperature measuring element embedded in the long side copper plate,
Regarding the casting direction, the position of the temperature measuring element on the uppermost stage is within 200 mm below the molten metal level control level, and the position of the temperature measuring element on the lowermost stage is located at a position 500 mm or more below the molten metal surface control level, adjacent to each other. The interval between the temperature measuring elements is 250 mm or less, the number of steps is 4 or more,
Regarding the mold width direction, the position of the temperature measuring element embedded in the position closest to the short side copper plate of the mold is set in a direction from the intersection of the long side copper plate and the short side copper plate toward the center of the mold width. Within 250mm, the interval between the temperature measuring elements adjacent to each other is 200mm or less, the number of rows is 8 or more,
Temperature measurement data acquisition means for acquiring temperature measurement data of the temperature measurement element arranged as described above at a predetermined time interval ;
Normalization of temperature measurement data for calculating an average value of the temperature measurement data acquired for each of the long side surface copper plates by the temperature measurement data acquisition means at each time , and normalizing the temperature measurement data based on the average value Means,
Principal component analysis means for performing principal component analysis of the temperature measurement data normalized by the temperature measurement data normalization means;
Residual calculation means for calculating a residual between the temperature measured data and the temperature represented by the principal component and the principal component score calculated by the principal component analysis means;
A surface defect determination device for a continuous casting slab, comprising surface defect occurrence determination means for determining whether or not a surface defect has occurred in the slab based on a sum of squares of the residual calculated by the residual calculation means.   6. The apparatus for determining surface defects of a continuous cast slab according to claim 5, wherein the position of the temperature measuring element at the lowest stage is within 900 mm in the casting direction from the level control level.   The surface defect occurrence determining means determines that a surface defect has occurred in the slab when the square sum of the residuals exceeds a predetermined threshold value. The surface defect determination apparatus of the continuous casting slab described in 1.   The surface of the continuously cast slab according to any one of claims 5 to 7, wherein the surface defect occurrence determining means determines for each row of the temperature measuring elements arranged in the casting direction. Defect determination device. A method for producing a steel slab using the method for determining surface defects of a continuously cast slab according to claim 4,
In-mold relative position identifying step for identifying a relative position in the mold corresponding to the row of temperature measuring elements determined to have surface defect occurrence in the surface defect occurrence determining step;
A steel casting comprising: a molten steel flow control step for estimating a molten steel flow abnormality in the mold based on the relative position in the mold and controlling the molten steel flow so as to eliminate the estimated molten steel flow abnormality. A manufacturing method of a piece.