JP5644910B1 - Abnormality detection method and blast furnace operation method - Google Patents

Abstract

An abnormality detection method and a blast furnace operation method capable of automatically detecting a deviation in luminance information in the blast furnace circumferential direction are provided. An anomaly detection device includes a tuyere camera and an image processing device installed near a plurality of tuyere of a blast furnace. The representative luminance vector collection unit 771 of the image processing device 7 determines the representative luminance based on the luminance value of each pixel for each tuyere image captured in advance by the tuyere camera, and collects the representative luminance vector in time series. To do. The index extraction unit 773 performs principal component analysis of representative luminance vectors collected in time series, and extracts principal component vectors. In addition, the representative luminance vector collection unit collects representative luminance vectors from tuyere images taken simultaneously by the tuyere camera during operation. The abnormality detection processing unit 775 calculates the length of a perpendicular line drawn from the collected representative luminance vector in the principal component vector direction as an evaluation value, and detects an abnormality in the blast furnace by comparing the evaluation value with a predetermined threshold value. [Selection] Figure 2

Description

  The present invention relates to an abnormality detection method and a blast furnace operation method for detecting an abnormality of a blast furnace from a tuyere image taken by a camera installed in the vicinity of a plurality of tuyere of a blast furnace.

  One of the criteria for realizing stable blast furnace operation is luminance information of the raceway section in the blast furnace observed through the tuyere of the blast furnace. This luminance information includes information important for the operation of the blast furnace, such as the level of furnace heat, the burnup of pulverized coal, and the fall information of unmolten ore. Observation of luminance information through the tuyere is carried out by a sensory test performed by an operator looking into the viewing window several times a day. In recent years, there is an increasing number of cases where a camera is installed near the tuyere and an image (feather image) photographed by this camera is displayed on a monitor in a monitoring room for centralized monitoring (see, for example, Patent Document 1). ).

  Here, a plurality of tuyere are provided in the circumferential direction of the blast furnace. In both cases of looking directly into the sight glass and centralized monitoring in the monitoring room, the operator pays attention to the variation in luminance information at each tuyere in the blast furnace circumferential direction (luminance information bias in the blast furnace circumferential direction). By doing so, the occurrence of an abnormality in the vicinity of each tuyere is intuitively detected.

  On the other hand, in Patent Document 2, a camera and a luminance meter are attached to three or more of the plurality of tuyere in the blast furnace circumferential direction, and a drop detected by calculating the number of unmelted ore drops from above in each place A technique for adjusting the ratio of ore and coke introduced from the top of the furnace based on the frequency is disclosed.

Japanese Patent Application Laid-Open No. 2004-183956 Japanese Patent Laid-Open No. 5-186811

  However, the technique of Patent Document 2 described above uses a tuyere image obtained by photographing the vicinity of a plurality of tuyere for the purpose of detecting the number of unmelted ore drops throughout the blast furnace. It does not detect the deviation of luminance information in the blast furnace circumferential direction that an operator is gazing at to detect an abnormality.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide an abnormality detection method and a blast furnace operation method capable of automatically detecting a deviation in luminance information in the blast furnace circumferential direction. And

  In order to solve the above-described problems and achieve the object, the abnormality detection method according to the present invention is an abnormality detection for detecting an abnormality of the blast furnace from a tuyere image taken by a camera installed in the vicinity of a plurality of tuyere of a blast furnace. A method of determining representative luminance based on a luminance value of each pixel for each tuyere image simultaneously photographed by the camera in advance, and collecting representative luminance vectors determined by the representative luminance in time series; , Performing a principal component analysis of the representative luminance vector collected in the time series, extracting the principal component vector, and collecting the representative luminance vector from the tuyere image photographed simultaneously by the camera during operation, A calculation step of calculating a length of a perpendicular line from the representative luminance vector in the principal component vector direction as an evaluation value, and comparing the evaluation value with a predetermined threshold value Characterized in that it comprises a, an abnormality detecting step of detecting an abnormality of the blast furnace by the.

  According to the present invention, it is possible to automatically detect a deviation in luminance information in the blast furnace circumferential direction.

FIG. 1 is a schematic diagram illustrating a schematic configuration example of a blast furnace to which an abnormality detection device is applied. FIG. 2 is a schematic diagram illustrating a configuration example of the abnormality detection device. FIG. 3 is a schematic diagram illustrating an example of a tuyere image. FIG. 4 is a diagram illustrating an example of a time-series change in luminance value of a tuyere image. FIG. 5 is a diagram illustrating a distribution example of time-series representative luminance vectors in an eight-dimensional space. FIG. 6 is a diagram illustrating the principal component direction and the departure direction from the principal component. FIG. 7 is an explanatory diagram illustrating calculation of an evaluation value. FIG. 8 is a diagram for explaining abnormality determination using an evaluation value. FIG. 9 is a flowchart illustrating a processing procedure of index extraction processing. FIG. 10 is a flowchart showing the processing procedure of the abnormality detection processing.

  Hereinafter, an embodiment for carrying out an abnormality detection method and a blast furnace operation method of the present invention will be described with reference to the drawings. Note that the present invention is not limited to the embodiments. Moreover, in description of drawing, the same code | symbol is attached | subjected and shown to the same part.

  FIG. 1 is a schematic diagram illustrating a schematic configuration example of a blast furnace 1 to which the abnormality detection apparatus 10 (see FIG. 2) of the present embodiment is applied. As shown in FIG. 1, a blast furnace 1 is charged with iron ore 21 and coke 23 from the top of the furnace, separates pig iron (hot metal) 25 obtained at the bottom of the furnace from slag 27 and discharges it outside the furnace. Then, hot air is blown from the tuyere 11 provided in the lower part of the furnace, and the iron ore 21 is reduced and melted using the coke 23 as a heat source to obtain molten iron 25. Since the specific gravity of the slag 27 is smaller than that of the hot metal 25, the slag 27 is separated into the upper layer of the hot metal 25.

  In the blast furnace 1, one end of a blower pipe 13 for blowing hot air is connected to the tuyere 11. A lance 15 is installed in the middle of the blast pipe 13 through the blast pipe 13, and pulverized coal is introduced into the hot air by the lance 15 (arrow Y11). Hot air (arrow Y13) blown through the blow pipe 13 is introduced into the blast furnace 1 from the tuyere 11 together with pulverized coal, and burns mainly in a combustion space called a raceway 17 ahead of the tuyere 11 in the hot air blowing direction. Contribute to.

  A tuyere observation unit 3 is installed on the other end side of the blower tube 13 facing the tuyere 11. This tuyere observation unit 3 is a tuyere for photographing the state of the blast furnace 1 in operation, specifically, the state inside the blower tube 13 and the state inside the blast furnace 1 (furnace state) via the tuyere 11. A camera (camera) 31, a viewing window 33 for visually observing the state of the blast furnace 1, and a half for branching the optical path shown by the alternate long and short dash line in FIG. 1 to the tuyere camera 31 side and the viewing window 33 side The mirror 35 is installed and unitized.

  Here, a plurality of tuyere 11 are arranged in the circumferential direction of the blast furnace 1, and the tuyere observation unit 3 is provided on the other end side of the blower tube 13 having one end connected to the plurality of tuyere 11. The tuyere cameras 31 are installed in whole or in part, and each anomaly detection device 10 is configured.

  FIG. 2 is a schematic diagram illustrating a configuration example of the abnormality detection apparatus 10 according to the present embodiment, and tuyere 11 arranged in the circumferential direction of the blast furnace 1 and tuyere camera 31 installed as tuyere observation unit 3. The positional relationship is also shown. In the present embodiment, as shown in FIG. 2, tuyere cameras 31 (31-1 to 31-8) are installed in eight tuyere 11 selected at substantially equal intervals along the circumferential direction of the blast furnace 1. ing. Each tuyere camera 31 outputs image data of tuyere images obtained by photographing the situation of the blast furnace 1 to an image acquisition device 5 to be described later. The tuyere cameras 31 need only be installed in at least a plurality of tuyere 11, and the tuyere 11 to be installed and the number thereof may be set as appropriate.

  The abnormality detection device 10 includes a plurality of tuyere cameras 31, the image acquisition device 5, and the image processing device 7 installed in a plurality (for example, eight) of tuyere 11 as described above. The image acquisition device 5 and the image processing device 7 are realized using a general-purpose computer such as a workstation or a personal computer.

  The image acquisition device 5 captures image data of tuyere images (moving images) continuously photographed by the tuyere cameras 31 during operation of the blast furnace 1 and transfers them to the image processing device 7 as needed.

  The image processing apparatus 7 includes an input unit 71, a display unit 73, a recording unit 75, and a processing unit 77 as main functional units.

  The input unit 71 is for inputting information necessary for detecting an abnormality of the blast furnace 1 and outputs an input signal corresponding to the operation input to the processing unit 77. The input unit 71 is realized by an input device such as a keyboard, a mouse, a touch panel, and various switches. The display unit 73 is for performing monitor display of tuyere images taken by each tuyere camera 31, for example, abnormality notification of the blast furnace 1, and the like, and various types based on display signals input from the processing unit 77. Display the screen. The display unit 73 is realized by a display device such as an LCD, an EL display, or a CRT display.

  The recording unit 75 is realized by an update recordable flash memory, a built-in hard disk connected by a data communication terminal, an information recording medium such as a memory card, and a read / write device thereof, and appropriately adopts a recording device according to the application. Can be used. In the recording unit 75, a program for operating the image processing device 7 and realizing various functions provided in the image processing device 7, data used during the execution of the program, and the like are recorded in advance. Alternatively, it is temporarily recorded for each processing.

  The processing unit 77 is realized by a CPU or the like, and based on an input signal input from the input unit 71, a program or data recorded in the recording unit 75, instructions to each unit constituting the image processing apparatus 7 and data The operation of the image processing apparatus 7 is controlled by performing transfer or the like. The processing unit 77 includes a representative luminance vector collection unit 771, an index extraction unit 773, and an abnormality detection processing unit 775, monitors the status of the blast furnace 1 by image processing the tuyere image, and detects abnormalities in the blast furnace 1. Perform processing to detect.

  Next, the principle of abnormality detection of the blast furnace 1 performed by the abnormality detection apparatus 10 will be described. FIG. 3 is a schematic diagram illustrating an example of a tuyere image. As shown in FIG. 3, in the tuyere image, the lance 15 in front of the tuyere 11 and the pulverized coal 29 thrown from the lance 15 are photographed in the form of smoke together with the state in the blast furnace 1 in the back of the tuyere 11. Is done. The luminance information of such tuyere images includes information important for the operation of the blast furnace 1 such as the level of furnace heat, the burnup degree of pulverized coal, and the fall information of unmolten ore as described above. When the above situation is normal, the same tuyere image is taken by each tuyere camera 31-1 to 31-8, but a situation different from the normal state (referred to as “abnormal” in this specification). When this occurs, the luminance value decreases or increases in the tuyere image in which the abnormality that has occurred is captured.

  FIG. 4 is a diagram illustrating an example of time-series changes in luminance values obtained from tuyere images of a plurality of tuyere 11. FIG. 4 shows luminance value changes L11, L13, L15, and L17 obtained from four tuyere images.

  The luminance value change of the tuyere image at normal time shows regularity according to a normal furnace heat change (normal furnace state change) that occurs with time. For example, as shown in a period T11 surrounded by a broken line in FIG. 4, the luminance value of each tuyere image changes while drawing a similar locus, although the value is slightly larger or smaller. That is, in the period T111 in which the temperature in the blast furnace 1 tends to increase, the luminance value gradually increases in all tuyere images, and in the period T113 in which the temperature in the blast furnace 1 is maintained, the luminance value change becomes flat, and the blast furnace 1 In the period T115 in which the temperature tends to decrease, the luminance value gradually decreases.

  On the other hand, when a furnace state change (abnormal furnace state change) occurs due to the occurrence of an abnormality in a part of the blast furnace 1, as shown in a period T13 surrounded by a one-dot chain line in FIG. The direction of the brightness value change L15 of the tuyere image partially deviates from the direction of normal furnace state changes drawn by the brightness value changes L11, L13, and L17 of the other tuyere images. Furthermore, when an abnormality occurs in the entire blast furnace 1 and an abnormal furnace state change expands during a period T15 surrounded by a two-dot chain line in FIG. 4, the luminance value changes L11, L13, L15 of each tuyere image are expanded. , L17 deviate from the direction of normal furnace condition change as a whole, and each shows a separate change tendency.

  The types of abnormalities appearing in the time-series changes in luminance values as described above include those caused by changes in furnace heat in the specific tuyere 11 or all tuyere 11 and abnormalities in the pulverized coal burn-up rate, falling of unmolten ore. However, various things can be considered. For example, the case where the lance 15 is broken in the blower pipe 13 is given (when the lance is broken). If the lance 15 breaks and falls out of the field of view of the tuyere camera 31, the state behind the tuyere 11 that has been blocked by the lance 15 until then is photographed. The high brightness area in the image increases. Further, as another abnormality, the melt level in the blast furnace 1 (that is, the upper surface level of the slag 27) may rise beyond the liquid level that is safe for operation (when noro springs). The rise in the melt level causes operational troubles such as melting the tuyere 11. When the melt level reaches the field of view of the tuyere camera 31, a state in which the slag 27 rises in the tuyere image is photographed, and the low-luminance region in the image increases. As another abnormality, the scattering direction of the pulverized coal supplied from the lance 15 may change (when the PCI flow direction is abnormal). Even in such a case, the brightness value in the image fluctuates before and after the change.

In the present embodiment, the index is extracted in advance by paying attention to the above-described time-series change of the luminance value obtained from each tuyere image (index extraction process). For this purpose, first, the representative luminance determined for each of the eight tuyere images photographed during operation is set as the representative luminance of the tuyere cameras 31-1 to 31-8 that photographed the corresponding tuyere images, and the following formula ( Collected as a set of vector information (representative luminance vector) V (t) shown in 1).

  Here, as representative luminance, for example, the maximum value (maximum luminance), minimum value (minimum luminance), average value (average luminance), and intermediate value (intermediate luminance) of the luminance value of each pixel in the corresponding tuyere image Etc. can be used. Which value should be used as the representative luminance may be selected according to the type of abnormality to be detected. For example, if it is desired to detect when the spring is swelled, the average luminance that reflects the increase in the low luminance region is used.

  The representative luminance vector V (t) collected as described above can be represented by one point in the N-dimensional space (N is the number of tuyere cameras 31; in this embodiment, the 8-dimensional space). FIG. 5 is a diagram showing a distribution example of the time-series representative luminance vector V (t) in the 8-dimensional space. When a sufficient number of representative luminance vectors V (t) are collected, a point P1 in the 8-dimensional space representing the representative luminance vector V (t) collected from the normal tuyere image is represented by the representative luminance vector V (t). It is distributed in the shape of an ellipsoid distributed in the direction of the main principal component, that is, the direction of change in the normal furnace state of the luminance value.

FIG. 6 is a diagram for explaining the principal component direction A and the degree of deviation from the principal component. Of the points representing the representative luminance vector V (t), the normal point P21 is an ellipse dispersed in the principal component direction (long axis direction of the ellipsoid) A of the representative luminance vector V (t ₁ ) as described above. Distributed body. For example, when the luminance change as shown in FIG. 5 occurs in the normal operation range, the point P23 representing the representative luminance vector V (t ₂ ) is also present in the ellipsoidal distribution. On the other hand, at the time of abnormality, one or more of the representative luminances deviate from the direction of normal furnace condition change and change to an unexpected direction, and therefore a component deviating from the main component orthogonal to the main component direction A (hereinafter, “ This is called “outlier component”). Therefore, the point P25 representing the representative luminance vector V (t ₃ ) when the abnormal furnace state change occurs in a part of the blast furnace 1 as in the period T13 in FIG. 4 is an unexpected change from the ellipsoidal distribution. (Arrow Y23).

  Therefore, in the index extraction process, principal component analysis is performed on a sufficient number of representative luminance vectors V (t) collected as described above, and the degree of deviation from the principal component direction A is extracted as an index.

  At the time of abnormality detection, a representative luminance vector V (t) is collected from tuyere images taken during operation in the same manner, and an abnormality in the blast furnace 1 is detected by thresholding the evaluation value (abnormality detection). processing). FIG. 7 is an explanatory diagram for explaining calculation of an evaluation value, and FIG. 8 is a diagram for explaining abnormality determination using the evaluation value.

For example, as shown in FIG. 7, it is assumed that a representative luminance vector V (t ₄ ) represented by a point P3 is newly collected during operation. The evaluation value is √ [{V (t ₄ ) using the inner product A ^t · V (t ₄ ) of the collected representative luminance vector V (t ₄ ) and the principal component vector A (unit vector of length 1). } ² − {A ^t · V (t ₄ )} ² ]. This evaluation value is the length of a perpendicular line drawn from the representative luminance vector V (t ₄ ) in the principal component vector A direction, and represents the degree of deviation from the principal component direction of the representative luminance vector V (t ₄ ). As the value increases (the greater the change in the unexpected direction), the evaluation value increases.

In the abnormality determination using the evaluation value, the ratio of the evaluation value, that is, the component deviating from the main component, falls within a normal range that is predetermined as a threshold range of −α or more and α or less as shown by hatching in FIG. Abnormality is determined depending on whether it belongs to an abnormal range that is outside the threshold range. For example, when the evaluation value is within the threshold range as in the representative luminance vector V (t ₅ ) represented by the point P41 in the 8-dimensional space, the situation of the blast furnace 1 is determined to be normal. On the other hand, when the evaluation value is out of the threshold range as in the representative luminance vector V (t ₆ ) represented by the point P43, the situation of the blast furnace 1 is determined to be abnormal.

  Next, a specific processing procedure performed by the abnormality detection apparatus 10 will be described. FIG. 9 is a flowchart illustrating a procedure of index extraction processing performed by the image processing apparatus 7. FIG. 10 is a flowchart showing a processing procedure of abnormality detection processing performed by the image processing device 7. The abnormality detection device 10 performs the abnormality detection method and the image processing device 7 by performing the abnormality detection processing according to the processing procedure of FIG. 10 after the image processing device 7 performs the index extraction processing according to the processing procedure of FIG. 9 in advance. Implement blast furnace operation method.

  That is, in the index extraction process, as shown in FIG. 9, the representative luminance vector collection unit 771 first starts from eight tuyere images transferred from the image acquisition device 5 in advance at a timing such as every predetermined time interval. A representative luminance vector V (t) is collected (step S11; collection step). Specifically, the representative luminance vector collection unit 771 determines and determines the representative luminance based on the luminance value of each pixel for each of the eight tuyere images photographed by each tuyere camera 31 at the corresponding timing. A representative luminance vector V (t) shown in the above formula (1) determined by the representative luminance for every eight tuyere images is obtained. The representative luminance vector collection unit 771 collects the representative luminance vector V (t) in a time series by collecting the representative luminance vector V (t) for a predetermined period.

  In subsequent step S13, the index extraction unit 773 performs principal component analysis on the series of representative luminance vectors V (t) collected during the past operation as described above, and extracts the principal component vector A (extraction step). . Then, the index extraction unit 773 stores the extracted principal component vector A in the recording unit 75 (step S15).

In the operation after the index extraction process is performed as described above, the abnormality detection process shown in FIG. 10 is executed at predetermined time intervals, for example. In this abnormality detection process, first, the process of step S21-step S23 is performed as a calculation step. That is, the representative luminance vector collection unit 771 first collects the representative luminance vector V (t) from the eight tuyere images transferred from the image acquisition device 5 at the corresponding timing by the same procedure as Step S11 in FIG. (Step S21). Thereafter, the abnormality detection processing unit 775 reads the principal component vector A from the recording section 75, the representative luminance vector V (t) collected in step S21, the inner product of the read principal component vectors A A ^t · V ^(t) √ [{V (t)} ² − {A ^t · V (t)} ² ] is calculated as an evaluation value (step S23).

Then, the process of step S25-step S27 is performed as an abnormality detection step. That is, the abnormality detection processing unit 775 performs threshold processing on the evaluation value calculated in step S23 (step S25), and determines whether or not the evaluation value is outside the predetermined threshold range (step S27).

  If the evaluation value is outside the threshold range (step S27: Yes), the abnormality detection processing unit 775 performs a process of determining that an abnormality has occurred in the blast furnace 1 and displaying a warning to that effect on the display unit 73. (Step S29). In addition, the process here should just be able to alert | report to the operator that abnormality had occurred at least in the blast furnace 1, and it is good also as a structure which alert | reports generation | occurrence | production of abnormality by outputting a warning sound with output devices, such as a speaker. In addition, the operation condition of the blast furnace 1 is controlled according to whether or not it is determined that an abnormality has occurred in the blast furnace 1 in step S29, and the generated abnormality is dealt with.

  As described above, in the present embodiment, the representative luminance vector V (t) corresponding to a set of eight representative luminances for each tuyere image photographed simultaneously by each tuyere camera 31-1 to 31-8. Principal component analysis was performed and the principal component vector A was extracted as an index. Then, the degree of deviation from the principal component calculated by the inner product of the representative luminance vector V (t) collected during operation and the principal component vector A is calculated as an evaluation value, and the occurrence of abnormality is detected.

  According to this, in some or all of the tuyere images taken by the tuyere cameras 31-1 to 31-8 in the vicinity of the plurality of tuyere 11 in the blast furnace 1 circumferential direction, the luminance value is the direction of normal furnace condition change. Can be detected as a situation (abnormality) different from the situation of the normal blast furnace 1 in the blast furnace 1 during operation. Therefore, conventionally, an operator visually detects the vicinity of each tuyere 11 by looking through the viewing window 33, or by intuitively monitoring the tuyere images near each tuyere 11 in the monitoring room. It is possible to automatically detect the deviation of the luminance information in the one round direction of the blast furnace.

DESCRIPTION OF SYMBOLS 1 Blast furnace 11 tuyere 13 Blower pipe 15 Lance 17 Raceway 21 Iron ore 23 Coke 25 Hot metal 27 Slag 10 Abnormality detector 3 Tuyere observation unit 31 (31-1 to 31-8) Tuyere camera 5 Image acquisition device 7 Image Processing device 71 Input unit 73 Display unit 75 Recording unit 77 Processing unit 771 Representative luminance vector collection unit 773 Index extraction unit 775 Abnormality detection processing unit

Claims (5)

An abnormality detection method for detecting an abnormality of the blast furnace from a tuyere image taken by a camera installed in the vicinity of a plurality of tuyeres of the blast furnace,
A collecting step of determining representative luminance based on a luminance value of each pixel for each tuyere image simultaneously captured by the camera in advance, and collecting representative luminance vectors determined by the representative luminance in time series;
An extraction step of performing principal component analysis of representative luminance vectors collected in time series and extracting the principal component vectors;
A calculation step of collecting the representative luminance vector from a tuyere image photographed simultaneously by the camera during operation, and calculating a length of a perpendicular drawn from the representative luminance vector in the principal component vector direction as an evaluation value;
An abnormality detection step of detecting an abnormality of the blast furnace by comparing the evaluation value with a predetermined threshold value;
An abnormality detection method comprising:   The abnormality detection method according to claim 1, wherein the representative luminance vector is collected using the maximum luminance value in the tuyere image as the representative luminance.   The abnormality detection method according to claim 1, wherein the representative luminance vector is collected using an average value of luminance values in the tuyere image as the representative luminance.   The abnormality detection method according to claim 1, wherein the representative luminance vector is collected using the minimum luminance value in the tuyere image as the representative luminance.   An abnormality of the blast furnace is detected using the abnormality detection method according to any one of claims 1 to 4, and the operating condition of the blast furnace is controlled according to whether or not an abnormality of the blast furnace is detected. Blast furnace operation method.

Images