JP7617403B2 - Blast furnace tuyere appearance evaluation method, tuyere appearance evaluation device, and tuyere appearance evaluation program - Google Patents

Description

The present invention relates to a tuyere landscape evaluation method, a tuyere landscape evaluation device, and a tuyere landscape evaluation program for evaluating the landscape of a tuyere based on a tuyere image.

Tyueres are arranged at equal intervals around the circumference of the lower part of the blast furnace, and high-temperature hot air, oxygen, and pulverized coal fuel are blown in through these tuyere. A raceway is formed in front of the tuyere, and pulverized coal and coke are burned in this raceway. The heat and reducing gas generated in the raceway reduce the sintered ore to produce molten iron, so the condition of the raceway has a significant impact on the operation of the blast furnace.

In recent years, in order to reduce production costs during blast furnace operation, a method of injecting large amounts of pulverized coal into the tuyere has been adopted. However, if the combustion state of the pulverized coal in the raceway deteriorates for some reason, the unburned pulverized coal will not serve as a heat source in the blast furnace and will accumulate inside the furnace, undesirably impairing the gas permeability inside the furnace and causing operation to become unstable or resulting in an increase in the reducing agent ratio. For these reasons, various technologies have been proposed to monitor the combustion state of the raceway.

Patent Document 1 discloses an imaging device (CCD camera) that captures thermal images of the raceway at two different wavelengths through an observation window in the blast furnace tuyere, a digital conversion device that converts the image signal output by this imaging device into a digital image, and a method of detecting the temperature inside the furnace from the shape of the histogram by calculating the temperature from the two-color brightness of the pixels of the digital image converted by the digital conversion device.

JP 2002-309307 A

However, with the detection method of Patent Document 1, if the brightness of the image captured by the CCD camera is excessively high, the temperature distribution cannot be accurately represented, and therefore the appearance of the tuyere cannot be accurately grasped.

In view of this, the present invention aims to provide a blast furnace tuyere landscape evaluation method, a tuyere landscape evaluation device, and a tuyere landscape evaluation program that can accurately grasp the tuyere landscape.

In order to solve the above problems, the blast furnace tuyere landscape evaluation method of the present invention includes: (1) a tuyere image acquisition process for continuously acquiring tuyere images using a camera equipped with an image sensor; a raceway identification process for identifying an image portion corresponding to a raceway among each tuyere image acquired in the tuyere image acquisition process; a luminance distribution generation process for creating a luminance distribution of the image portion corresponding to the raceway identified in the raceway identification process for each of the three primary colors of light; a parameter calculation process for calculating skewness and/or kurtosis from a luminance distribution created in the luminance distribution generation process in which the average luminance value is less than a luminance threshold value; an individual tuyere landscape judgment process for judging the individual tuyere landscape for each tuyere based on the magnitude of the skewness and/or kurtosis calculated in the parameter calculation process; and a comprehensive tuyere landscape evaluation process for judging at least the normality or abnormality of the comprehensive tuyere landscape of the entire blast furnace based on an evaluation value calculated from the tabulation result of the judgments in the individual tuyere landscape judgment process.

(2) The method for evaluating the tuyere landscape of a blast furnace described in (1), characterized in that in the parameter calculation process, both skewness and kurtosis are calculated, and in the individual tuyere landscape judgment process, the quality of the individual tuyere landscape is judged for each tuyere based on the magnitude of the skewness and kurtosis calculated in the parameter calculation process, and the cause of the defect in the individual tuyere landscape is estimated, and in the overall tuyere landscape evaluation process, if the overall tuyere landscape is judged to be abnormal, the cause of the abnormality in the overall tuyere landscape is estimated based on the cause of the defect in the individual tuyere landscape for each tuyere estimated in the individual tuyere landscape judgment process.

(3) The method for evaluating the appearance of a blast furnace tuyere according to (1) or (2), characterized in that, if the average value of red brightness is less than the brightness threshold, the parameter calculation process is performed based on the red brightness distribution, if the average value of red brightness is equal to or greater than the brightness threshold and the average value of green brightness is less than the brightness threshold, the parameter calculation process is performed based on the green brightness distribution, and if the average values of red and green brightness are equal to or greater than the brightness threshold, the parameter calculation process is performed based on the blue brightness distribution.

(4) The method for evaluating the appearance of a blast furnace tuyere according to any one of (1) to (3), characterized in that the brightness threshold is 200.

In order to solve the above problem, the blast furnace tuyere landscape evaluation device according to the present invention includes (5) a raceway identification unit that identifies an image portion corresponding to a raceway among each tuyere image continuously acquired by a camera equipped with an image sensor; a luminance distribution generation unit that creates a luminance distribution of the image portion corresponding to the raceway identified by the raceway identification unit for each of the three primary colors of light; a parameter calculation unit that calculates skewness and/or kurtosis from a luminance distribution created by the luminance distribution generation unit in which the average luminance value is less than a luminance threshold; an individual tuyere landscape judgment unit that judges the individual tuyere landscape for each tuyere based on the magnitude of the skewness and/or kurtosis calculated by the parameter calculation unit; and a comprehensive tuyere landscape evaluation unit that judges at least whether the comprehensive tuyere landscape of the entire blast furnace is normal or abnormal based on an evaluation value calculated from the tabulation result of the judgments made by the individual tuyere landscape judgment unit.

Furthermore, in order to solve the above problem, the blast furnace tuyere landscape evaluation program of the present invention causes a process computer to execute (6) a raceway identification step of identifying an image portion corresponding to a raceway among each tuyere image continuously acquired by a camera equipped with an image sensor, a luminance distribution generation step of creating a luminance distribution of the image portion corresponding to the raceway identified by the raceway identification step for each of the three primary colors of light, a parameter calculation step of calculating skewness and/or kurtosis from the luminance distribution created in the luminance distribution generation step in which the average luminance value is less than a luminance threshold value, an individual tuyere landscape judgment step of judging the individual tuyere landscape for each tuyere based on the magnitude of the skewness and/or kurtosis calculated in the parameter calculation step, and a comprehensive tuyere landscape evaluation step of judging at least whether the comprehensive tuyere landscape of the entire blast furnace is normal or abnormal based on an evaluation value calculated from the aggregation result of the judgments in the individual tuyere landscape judgment step.

According to the present invention, the appearance of the tuyere can be accurately grasped by using the brightness distribution of RGB (the three primary colors of light) where the average brightness value is below a certain value.

FIG. 2 is a schematic enlarged view of a tuyere portion in this embodiment. FIG. 2 is a functional block diagram of a controller. 1 is a schematic diagram of three luminance distributions. 4A to 4C are schematic diagrams of three representative luminance distributions different from those in FIG. 3 . This is a skewness distribution created by the skewness distribution/kurtosis distribution generating unit 122E. This is a kurtosis distribution created by the skewness distribution/kurtosis distribution generating unit 122E. This is a two-dimensional coordinate consisting of skewness and kurtosis, which indicates the area occupied by at least one of the normal ranges D1 and D2. Figure 7A shows a plot of actual data. This is a two-dimensional coordinate of skewness and kurtosis showing the area where both normal ranges D1 and D2 overlap. This is a two-dimensional coordinate system consisting of skewness and kurtosis showing the normal range D1. This is a two-dimensional coordinate system consisting of skewness and kurtosis showing the normal range D2. 7B is a flowchart showing a process from the imaging process to the calculation of normal ranges D1 and D2 in FIG. 7A. 13 is a flowchart for performing evaluation processing of the overall tuyere landscape.

Figure 1 is a schematic enlarged view of the periphery of the tuyere of a blast furnace to which the blast furnace tuyere landscape evaluation method of this embodiment is applied, and also shows a functional block diagram of a calculation processing device. Several dozen tuyere 4 are provided in the circumferential direction of the furnace at the bottom of the blast furnace body 3. The tuyere 4 is provided with an injection lance 5 for injecting pulverized coal and a hot air supply pipe 6 that communicates with a hot air stove (not shown), and high-temperature gas heated in the hot air stove is injected into the furnace together with the pulverized coal.

The wind pressure of the gas blown into the furnace creates a cavity (i.e., a raceway 1) inside the furnace. Inside this raceway 1, a group of coke particles is burned while swirling. At the interface of the raceway 1, a high-temperature combustion reaction occurs in which the coke and pulverized coal burn and produce carbon monoxide. An observation window 7 is provided at the rear end of each tuyere 4 on the outside of the furnace. The observation window 7 may be provided in all tuyere 4 or only in some of the tuyere 4. It is possible to determine which tuyere 4 has the observation window 7 installed.

The observation window 7 is provided to monitor the inside of the furnace, and has been used to visually observe the brightness of the tuyere, the swirling state of the coke, the dripping of unreduced ore, etc. By observing these, it is possible to estimate and predict the reaction state and fluctuations in the raceway and the lower part of the cohesive zone. However, visual observation has problems such as poor continuity and objectivity of the observation, and the inability to perform post-mortem analysis because the data is not stored. Therefore, in this embodiment, a camera 10 for optically observing the raceway 1 is installed behind the observation window 7. The camera 10 can be installed in all tuyere 4 or only in some tuyere 4 depending on the observation window 7. However, the more tuyere 4 equipped with the camera 10, the more accurately the overall state of the raceway 1 in the furnace circumferential direction can be grasped, and the more accurately the tuyere appearance can be evaluated.

<About the camera>
The camera 10 is equipped with an imaging element 10A, and can continuously capture images of the tuyere including the raceway 1 through the observation window 7 at a predetermined frame rate. That is, the "tuyere image" is an image of the state of the raceway captured through the observation window 7 using the camera 10 installed near the tuyere 4. The tuyere image may also include the injection lance 5 located in front of the tuyere and the pulverized coal injected from the injection lance 5. A CCD sensor can be used for the imaging element 10A. The CCD sensor is equipped with a color filter of R, G, or B for each pixel, and the light component transmitted through this color filter is received by the pixel (photodiode). The pixel outputs an image signal according to the intensity of the received light to the arithmetic processing device 12. In this embodiment, a CCD sensor is used as the imaging element, but the present invention is not limited to this, and a CMOS sensor can also be used. In addition, the camera 10 is not limited to a camera that captures still images, but also includes a camera that captures videos.

<Regarding the Processing Unit>
The arithmetic processing device 12 includes a storage unit 121 and a controller 122. The controller 122 performs imaging control for controlling the imaging operation of the camera 10, and performs analysis and evaluation processing for analyzing tuyere images acquired by the imaging operation and evaluating the appearance of the tuyere. The storage unit 121 sequentially stores the tuyere images and the like acquired from the camera 10. The processing performed by the controller 122 can be realized by a program, and this program can be stored in the storage unit 121.

The controller 122 may be a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). It is also possible to have at least a portion of the processing performed by the CPU or MPU executed as a circuit using an ASIC (Application Specific Integrated Circuit). The memory unit 121 may be a server or the like.

(Regarding imaging control)
The controller 122 outputs a control signal (trigger signal) to the camera 10 to start the imaging operation. As a result, the camera 10 continuously captures tuyere images at a predetermined frame rate. The controller 122 can also control the time during which the camera 10 performs the imaging operation, that is, the imaging time. The imaging time is not particularly limited, but if the imaging time is too short, the obtained brightness distribution may be insufficient, and the accuracy of detecting abnormalities in the tuyere scenery may decrease. If the imaging time is too long, the obtained brightness distribution becomes enormous, and data processing becomes complicated. Therefore, the imaging time can be appropriately set, for example, in the range of 1 to 60 seconds. The frame rate of the camera 10 is not particularly limited, but can be set to, for example, 30 (frames/second). The acquired tuyere image data is stored sequentially in the storage unit 121.

The controller 122 calculates the brightness distribution of the raceway included in the tuyere image for each R component, G component, and B component, selects a brightness distribution with a relatively low brightness value from among these, judges the individual tuyere scenery for each tuyere based on the selected brightness distribution, and uses the judgment result to evaluate the overall tuyere scenery. Here, "individual tuyere scenery" refers to the state of the raceway and the vicinity of the raceway in each tuyere 4 in the furnace. The individual tuyere scenery includes, for example, the state of the pulverized coal injected from the injection lance 5, the swirling state of the coke in the raceway, and the raw ore falling into the tuyere 4. Note that structures such as the injection lance 5 are regarded as parts other than the raceway (background) in the tuyere image in the raceway identification unit 122A (described later), and are therefore not included in the individual tuyere scenery. In addition, the "overall tuyere scenery" refers to the tuyere scenery viewed comprehensively from the individual tuyere scenery of the entire blast furnace.

Figure 2 is a functional block diagram of the controller 122 corresponding to this analysis and evaluation process. Referring to the figure, the controller 122 includes a raceway identification unit 122A, a luminance distribution generation unit 122B, an RGB selection unit 122C, a parameter calculation unit 122D, a skewness distribution/kurtosis distribution generation unit 122E, a threshold calculation unit 122F, an individual tuyere landscape determination unit 122G, and a comprehensive tuyere landscape evaluation unit 122H. Each unit will be described in detail below.

(Regarding the raceway identifying portion 122A)
The raceway identification unit 122A compares any two of the consecutively acquired tuyere images, and regards the part corresponding to the pixel group with a large brightness difference as raceway 1, and the part corresponding to the pixel group with a small brightness difference as the part other than the raceway (i.e., the background), thereby identifying the image part in which raceway 1 is present.

Because the state of the raceway 1 changes from moment to moment, when comparing two tuyere images taken at different times, the brightness difference in the image portion corresponding to the raceway 1 is relatively large. On the other hand, because the blowing lance 5 and the like are structures, the brightness difference in the image portion corresponding to the raceway 1 is relatively small when comparing two tuyere images taken at different times. Therefore, by quantitatively analyzing this brightness difference, it is possible to accurately identify the image portion of each tuyere image that corresponds to the raceway 1.

The brightness difference for evaluating the background can be set appropriately based on operational records. For example, a portion corresponding to a pixel group with a brightness difference of 5 or less can be regarded as the background. For example, the current tuyere image and the tuyere image from the previous frame can be used as the tuyere images to be compared.

In this embodiment, the image portion corresponding to the raceway 1 is identified based on the brightness difference between the two tuyere images, but the present invention is not limited to this, and the image portion corresponding to the raceway 1 may be identified based on the absolute brightness value. In other words, since structures such as the injection lance 5 appear relatively dark while the raceway 1 appears relatively bright, the image portion corresponding to the raceway 1 can also be identified based on the absolute brightness value. The absolute brightness value that separates the background from the raceway 1 can be set appropriately based on operational performance. For example, the portion corresponding to a pixel group with an absolute brightness value of 10 or less can be regarded as the background.

After dividing the tuyere image into background and raceway 1 using the above-mentioned method, a binarization process can be performed to obtain a mask image in which the brightness of the part corresponding to the background is 0 (black) and the brightness of the part corresponding to raceway 1 is 256 (white). After obtaining the mask image, a brightness distribution, which will be described later, can be created based on the brightness of the part corresponding to raceway 1 (brightness measured before binarization process).

(Regarding the brightness distribution generating unit 122B)
The luminance distribution generating unit 122B creates a luminance distribution based on the electrical signals output from each pixel corresponding to the raceway 1 identified by the raceway identifying unit 122A. The luminance distribution is created for each of the three primary colors of light, R, G, and B.

The positions of structures such as the blowing lance 5 may be slightly displaced due to vibrations during blast furnace operation, and may be significantly displaced due to maintenance of the blowing lance 5 during blast furnace shutdowns. Therefore, when creating a luminance distribution using the luminance distribution generating unit 122B, it is desirable for the raceway identifying unit 122A to identify the image portion corresponding to the raceway 1 for each tuyere image. This makes it possible to create a more accurate luminance distribution of the image portion corresponding to the raceway 1.

However, if the positions of the injection lance 5 and the like hardly change during blast furnace operation, a pixel group corresponding to the raceway 1 may be identified in advance, and a brightness distribution may be created based on the electrical signals output from these pixels. This method simplifies the processing, as it is no longer necessary to identify the image portion corresponding to the raceway 1 for each tuyere image. The method for identifying the pixel group corresponding to the raceway 1 has been described above, so the explanation will not be repeated.

(Regarding the RGB selection unit 122C)
The RGB selection unit 122C calculates the average value of the luminance values (hereinafter also referred to as the luminance average value) for each of the R, G, and B luminance distributions created by the luminance distribution generation unit 122B, and selects the luminance distribution in which the luminance average value is less than a threshold value as the luminance distribution to be used by the parameter calculation unit 122D described later. This threshold value (threshold value for the luminance average value of the image) is referred to as the luminance threshold value. Whether the luminance average value is less than the luminance threshold value is determined in the order of R, G, and B. That is, the luminance distribution is selected in the order of priority of R, G, and B. When comparing the luminance at the same pixel with the three primary colors of light (R, G, and B), the luminance is higher in the order of R, G, and B, so that the judgment of the individual tuyere landscape for each tuyere and the evaluation of the overall tuyere landscape can be performed more accurately in this order. However, even if the luminance distribution of B is selected, it goes without saying that the judgment of the individual tuyere landscape for each tuyere and the evaluation of the overall tuyere landscape can be performed sufficiently accurately based on this. If the average brightness value is excessively high, the image will be excessively bright, making it impossible to accurately determine abnormalities in the individual tuyere appearances from the skewness or kurtosis of the brightness distribution.
The brightness threshold is determined by past operational experience, but is preferably 200.

Here, as another method (comparative example), when the average brightness value of R is 200 or more, a method can be considered in which the aperture value of the camera 10 is reset to reduce the average brightness value of R to less than 200, and then the brightness distribution is acquired again. However, this method requires a process of resetting the aperture value, which makes the process cumbersome. Furthermore, in order to accurately create a brightness distribution, it is necessary to continuously capture tuyere images again with the camera 10 after resetting the aperture value, which makes the work cumbersome.

(Regarding the parameter calculation unit 122D)
The parameter calculation unit 122D calculates the variance and standard deviation from the luminance distribution selected by the RGB selection unit 122C and the average luminance value of the selected luminance distribution (calculated by the RGB selection unit 122C), and calculates the skewness and/or kurtosis based on the calculation results. The variance, standard deviation, skewness, and kurtosis are calculated by the following formulas.

The calculation of skewness and/or kurtosis is performed for all tuyere images acquired within the imaging time. However, the present invention is not limited to this, and a method of calculating skewness and kurtosis for any number of acquired tuyere images may also be used. The calculation results of skewness and kurtosis are stored in the memory unit 121 for each tuyere 4.

Figure 3 shows three luminance distributions, with the vertical and horizontal axes corresponding to the number of pixels and luminance value, respectively. Luminance distribution II is approximately normal, and is approximately evenly divided into bright and dark regions with the average luminance as the boundary. In other words, luminance distribution II is evaluated to have a skewness of approximately 0. Luminance distribution I is skewed toward the smaller luminance value side compared to luminance distribution II, and shows a state in which there are relatively many dark regions. In other words, luminance distribution I is evaluated to have a relatively large skewness (skewness > 0) compared to luminance distribution II. Luminance distribution III is skewed toward the larger luminance value side compared to luminance distribution II, and shows a state in which there are relatively many bright regions. In other words, luminance distribution III is evaluated to have a relatively small skewness (skewness < 0) compared to luminance distribution II.

Figure 4 shows three schematic luminance distributions different from those in Figure 3. Luminance distribution V is approximately normal, and is evaluated to have a kurtosis of approximately 0. Luminance distribution IV is more peaked than luminance distribution V, and shows a state in which areas of specific luminance are relatively numerous. That is, luminance distribution IV is evaluated to have a relatively large kurtosis (kurtosis > 0) compared to luminance distribution V. Luminance distribution VI is less peaked than luminance distribution V, and shows a state in which bright and dark areas are relatively evenly distributed. That is, luminance distribution IV is evaluated to have a relatively small kurtosis (kurtosis < 0) compared to luminance distribution II.

(Regarding the skewness distribution/kurtosis distribution generating unit 122E)
The skewness distribution and kurtosis distribution generating unit 122E creates a skewness distribution and/or kurtosis distribution based on the skewness and/or kurtosis calculated by the parameter calculation unit 122D. As described above, the skewness and/or kurtosis of each tuyere image calculated by the parameter calculation unit 122D is stored in the storage unit 121 for each tuyere 4. Of the skewness and/or kurtosis of each tuyere image stored in the storage unit 121, the skewness and/or kurtosis of each tuyere image of the tuyere 4 in a state where the individual tuyere landscape is normal is read from the storage unit 121 to create a skewness distribution and/or kurtosis distribution. FIG. 5 is a schematic diagram showing the skewness distribution of a certain tuyere 4 in a state where the individual tuyere landscape is normal, created by the skewness distribution and kurtosis distribution generating unit 122E. FIG. 6 is a schematic diagram showing the kurtosis distribution of a certain tuyere 4 in a state where the individual tuyere landscape is normal, created by the skewness distribution and kurtosis distribution generating unit 122E. The horizontal axis of Fig. 5 represents skewness, and the vertical axis represents the logarithmic relative frequency of skewness, whereas the horizontal axis of Fig. 6 represents kurtosis, and the vertical axis represents the logarithmic relative frequency of kurtosis.

(Regarding the threshold calculation unit 122F)
The threshold calculation unit 122F calculates thresholds to be used in the individual tuyere appearance judgment unit 122G described later based on the skewness distribution and/or kurtosis distribution in a state where the individual tuyere appearance is normal, which are created by the skewness distribution/kurtosis distribution generation unit 122E. The average value P of the skewness distribution in FIG. 5 is calculated, and the value obtained by adding 2σ ₁ to the average value P is set as the upper limit threshold T1 of the skewness, and the value obtained by subtracting 2σ ₁ from the average value P is set as the lower limit threshold T2 of the skewness (σ ₁ indicates the standard deviation in the skewness distribution). In addition, the average value Q of the kurtosis distribution in FIG. 6 is calculated, and the value obtained by adding 2σ ₂ to the average value Q is set as the upper limit threshold T3 of the kurtosis, and the value obtained by subtracting 2σ ₂ from the average value Q is set as the lower limit threshold T4 of the kurtosis (σ ₂ indicates the standard deviation in the kurtosis distribution). Each threshold value calculated here is stored in the storage unit 121.

(Regarding the individual tuyere appearance determination unit 122G)
The individual tuyere appearance determination unit 122G draws normal ranges D1 and D2 on a two-dimensional coordinate system consisting of skewness and kurtosis based on the threshold values T1 to T4 calculated by the threshold calculation unit 122F (see FIG. 7A). The normal ranges D1 and D2 can also be notified to the operator by a method different from the above-mentioned method (for example, a method of displaying numerical values in an area outside the two-dimensional coordinate system on the display screen).
The individual tuyere landscape determination unit 122G performs the above-mentioned processes (the processes performed by the raceway identification unit 122A, the luminance distribution generation unit 122B, the RGB selection unit 122C, and the parameter calculation unit 122D) on the newly acquired tuyere image to calculate the skewness and kurtosis, and plots these calculated skewness and kurtosis in two-dimensional coordinates to create a scatter diagram (see FIG. 7B). It goes without saying that these two-dimensional coordinates and scatter diagrams must be created for each tuyere 4.

Here, the individual tuyere landscape determination unit 122G can determine the individual tuyere landscape for each tuyere 4 based on whether a predetermined percentage or more of the plotted data is included in the normal range D1 and/or D2 (the definition of the individual tuyere landscape has been described above, so the description will not be repeated). That is, the individual tuyere landscape can be determined based on whether a predetermined percentage or more of the plotted data is included in either of the normal ranges D1 and D2 (see Figures 7A and B), the individual tuyere landscape can be determined based on whether a predetermined percentage or more of the plotted data is included in both the normal ranges D1 and D2 (see Figure 7C), the individual tuyere landscape can be determined based on whether a predetermined percentage or more of the plotted data is included in the normal range D1 (see Figure 7D), or the individual tuyere landscape can be determined based on whether a predetermined percentage or more of the plotted data is included in the normal range D2 (see Figure 7E). However, it is preferable to determine the individual tuyere landscape based on both the normal ranges D1 and D2, as this improves accuracy. The specified percentage can be set appropriately based on operational performance, but it is desirable to set it at least 70% or more.

In addition, by using both normal ranges D1 and D2 as the basis for discrimination, the cause of the poor individual tuyere appearance can also be estimated. For example, referring to Figures 7A and B, if 70% of the total number of plots are included in the upper right region Y, which is separated by the upper threshold value T1 of skewness and the upper threshold value T3 of kurtosis, it can be estimated that the individual tuyere appearance is poor due to ore fall. In addition, if 70% of the total number of plots are included in the upper left region Z, which is separated by the lower threshold value T2 of skewness and the upper threshold value T3 of kurtosis, it can be estimated that the individual tuyere appearance is poor due to a decrease in the amount of pulverized coal injection caused by clogging of the tuyere 4 or injection lance 5.

In the above embodiment, the normal ranges D1 and D2 are set to 2σ ₁ from the average value of the skewness distribution and 2σ ₂ from the average value of the kurtosis distribution, but the method of setting the normal ranges is not limited to this. They may be set to mσ ₁ (m is a natural number other than 2) from the average value of the skewness distribution and nσ ₂ (n is a natural number other than 2) from the average value of the kurtosis distribution according to the operational performance of the blast furnace.

(About the Comprehensive Tuyere Landscape Evaluation Section 122H)
The overall tuyere landscape evaluation unit 122H judges whether the overall tuyere landscape of the entire blast furnace is normal or not based on an evaluation value calculated from the aggregation result of the judgments made by the individual tuyere landscape judgment unit 122G. Furthermore, if the overall tuyere landscape evaluation unit 122H judges that the overall tuyere landscape is abnormal, it can estimate the cause of the abnormality of the overall tuyere landscape based on the cause of the poor individual tuyere landscape for each tuyere estimated by the individual tuyere landscape judgment unit 122G. Note that the definition of the overall tuyere landscape has been described above, so explanation will be omitted.

Whether the overall tuyere landscape is normal or not can be determined, for example, by taking the ratio of the number of tuyere landscapes determined to be good (hereinafter referred to as the good tuyere ratio) to the total number of tuyere landscapes determined by the individual tuyere landscape determination unit 122G as an evaluation value, and determining whether the evaluation value exceeds a threshold value. In this embodiment, the total number of tuyere landscapes determined by the individual tuyere landscape determination unit 122G is set to 40, and the threshold value is set to 0.75 based on past operating performance, but is not limited to this.

Table 1 shows the results of collating the judgments of individual tuyere appearances of blast furnaces operating under certain operating conditions. Table 2 shows the results of collating the judgments of individual tuyere appearances of blast furnaces operating under different operating conditions. In Tables 1 and 2, tuyere judged to satisfy both normal ranges D1 and D2 was evaluated as "A". Tuyere with 70% or more of the total number of plots plotted in area Y of Figure 7A was evaluated as "B1". Tuyere with 70% or more of the total number of plots plotted in area Z of Figure 7A was evaluated as "B2".

Referring to Table 1, the number of tuyere rated as "A" (individual tuyere appearance is good) was 9, and the good tuyere ratio was calculated to be 0.225. Therefore, since the good tuyere ratio was below the threshold value (0.75), it was rated as "the overall tuyere appearance is abnormal."

Referring to Table 2, the number of tuyere rated as "A" (individual tuyere appearance is good) was 39, and the good tuyere ratio was calculated to be 0.975. Therefore, since the good tuyere ratio exceeded the threshold value (0.75), it was rated as "overall tuyere appearance is normal."

If the overall tuyere appearance is determined to be abnormal, the blast furnace operator can take action to improve it. Improvement actions include increasing the amount of coke charged from the top of the furnace (increasing the amount of coke) and removing blockages in the tuyere and supply pipes (cleaning the equipment).

When the individual tuyere landscape is judged using both skewness and kurtosis in the individual tuyere landscape judgment unit 122G as in this embodiment, the cause of the defect in the individual tuyere landscape can be estimated as described above. When the overall tuyere landscape evaluation unit 122H judges that the overall tuyere landscape is abnormal, it can estimate the cause of the abnormality in the overall tuyere landscape based on this estimated cause of the defect. When reporting the judgment result of the overall tuyere landscape, the estimated result (the cause of the abnormality in the overall tuyere landscape) is also reported, making it easier for the operator to select improvement actions.

For example, in Table 1, since the evaluations are concentrated in the "B1" range (poor individual tuyere appearance due to raw ore falling), it is presumed that the cause of the abnormality in the overall tuyere appearance is "insufficient reduction inside the furnace." Therefore, by reporting the evaluation result of the overall tuyere appearance as "abnormality in the overall tuyere appearance due to insufficient reduction," the operator can easily select the remedial action "increase the skeleton." Also, if the evaluations are concentrated in the "B2" range, it is presumed that the cause of the abnormality in the overall tuyere appearance is "clogging of the tuyere 4 or injection lance 5." Therefore, by reporting the evaluation result of the overall tuyere appearance as "abnormality in the overall tuyere appearance due to clogging of pulverized coal," the operator can easily select the remedial action "clean the equipment."

The evaluation results of the overall tuyere landscape in the overall tuyere landscape evaluation unit 122H can be displayed by the controller 122 on an output device such as a display provided in the arithmetic processing device 12 or an output device such as a display provided outside the arithmetic processing device 12. However, this is not limited to this, and notification may be given by other methods such as audio output.

The processes performed by the above-mentioned components will be described with reference to the flowcharts of FIGS.
Fig. 8 is a flowchart showing the process from the imaging process to the calculation of the normal ranges D1 and D2 in Fig. 7A. The process shown in the flowchart in Fig. 8 is performed when the individual tuyere landscape is normal. Whether or not the individual tuyere landscape is normal can be determined based on, for example, the coke ratio. Fig. 9 is a flowchart for performing an evaluation process of the overall tuyere landscape.

Referring to FIG. 8, in step S101 (corresponding to the tuyere image acquisition step in claim 1), the controller 122 operates the camera 10 according to a predetermined frame rate to continuously capture tuyere images. The captured tuyere image data is sequentially recorded in the memory unit 121. When the predetermined image capture time has elapsed (Yes in step S102), the process proceeds to step S103.

In step S103 (corresponding to the raceway identification process in claim 1), the raceway identification unit 122A performs a process of comparing any two of the consecutively acquired tuyere images, and regarding the portion corresponding to the pixel group with a large brightness difference as the raceway, and regarding the portion corresponding to the pixel group with a small brightness difference as the background. It goes without saying that the process shown in step S103 is performed for each acquired tuyere image.

In step S104 (corresponding to the luminance distribution generation step in claim 1), the luminance distribution generation unit 122B creates a luminance distribution for the raceway-corresponding portion in each tuyere image identified in step S103. The luminance distribution is created for each of the three primary colors of light: R, G, and B.

The RGB selection unit 122C calculates the luminance average value based on the luminance distribution of R created in step S104, and determines whether the luminance average value of R is less than 200 (luminance threshold) (step S105). If the luminance average value of R is less than 200 (step S105 Yes), the luminance distribution of R is selected as the luminance information to be used by the parameter calculation unit 122D (step S106). If the luminance average value of R is 200 or more (step S105 No), the RGB selection unit 122C calculates the luminance average value based on the luminance distribution of G created by the luminance distribution generation unit 122B, and determines whether the luminance average value of G is less than 200 (luminance threshold) (step S107). If the luminance average value of G is less than 200 (step S107 Yes), the luminance distribution of G is selected as the luminance information to be used by the parameter calculation unit 122D (step S108). If the average brightness value of G is 200 or more (step S107 No), the brightness distribution of B is selected as the brightness information to be used by the parameter calculation unit 122D (step S109).

The parameter calculation unit 122D calculates the variance and standard deviation based on any one of the RGB luminance distributions selected in step S106, step S108, or step S109 and the average luminance value of this selected luminance distribution (calculated in step S105 or S107), and calculates the skewness and kurtosis based on this calculation result (step S110). Step S110 corresponds to the parameter calculation step in claim 1. The calculation results of the skewness and kurtosis are stored in the memory unit 121 for each tuyere. Note that the explanation of the calculation method of the skewness and kurtosis will not be repeated.

In step S111, the skewness distribution/kurtosis distribution generation unit 122E creates a skewness distribution and a kurtosis distribution based on the skewness and kurtosis calculated in step S110.

The threshold calculation unit 122F calculates the threshold used in the individual tuyere landscape determination unit 122G based on the skewness distribution and kurtosis distribution created in step S111 (step S112). The method for calculating the threshold will not be described again.

Referring to FIG. 9, the individual tuyere landscape determination unit 122G performs the processes shown in S101 to S110 in FIG. 8 to calculate the skewness and kurtosis of each newly acquired tuyere image for each tuyere, and determines the individual tuyere landscape for each tuyere based on the calculated skewness and kurtosis using the threshold calculated in step S112 in FIG. 8 (step S114). Step S114 corresponds to the individual tuyere landscape determination process in claim 1. In this flowchart, since both skewness and kurtosis are parameters, in addition to determining whether the individual tuyere landscape is good or bad, it is also possible to estimate the cause of the poor individual tuyere landscape. The method of determining the individual tuyere landscape has been described above, so a description is omitted. The determination of the individual tuyere landscape is performed until the number of tuyere reaches a predetermined number (in this embodiment, all tuyere) (step S115 No), and when the predetermined number is reached (step S115 Yes), the determination of the individual tuyere landscape is terminated and the process proceeds to step S116.

The overall tuyere landscape evaluation unit 122H determines whether the overall tuyere landscape is normal or not based on the individual tuyere landscapes for each tuyere determined in step S114 (step S116). Step S116 corresponds to the overall tuyere landscape evaluation process in claim 1. Whether the overall tuyere landscape is normal or not can be determined by whether the proportion of good tuyeres exceeds a threshold value.

If the overall tuyere landscape evaluation unit 122H determines that the overall tuyere landscape is normal (Yes in step S116), it reports that the overall tuyere landscape is normal (step S117). On the other hand, if the overall tuyere landscape evaluation unit 122H determines that the overall tuyere landscape is abnormal, it estimates the cause of the abnormality in the overall tuyere landscape based on the cause of the poor individual tuyere landscapes for each tuyere estimated in step S114 (step S118), and reports the estimated result (the cause of the abnormality in the overall tuyere landscape) together with the fact that the overall tuyere landscape is abnormal (step S119). The method for estimating the cause of the abnormality in the overall tuyere landscape will not be repeated.

The operator can take corrective action depending on the cause of the abnormality, but we will not repeat the explanation on this point.

REFERENCE SIGNS LIST 10 Camera 10A Image sensor 12 Processing device 121 Storage unit 122 Controller 122A Raceway identification unit 122B Brightness distribution generation unit 122C RGB selection unit 122D Parameter calculation unit
122E Skewness distribution/kurtosis distribution generating unit 122F Threshold calculation unit
122G Individual Tuyere Landscape Judgment Section 122H Overall Tuyere Landscape Evaluation Section

Claims (6)

A raceway condition evaluation method for determining whether a raceway of a blast furnace is normal or abnormal , comprising:
A tuyere image acquisition process of continuously acquiring tuyere images by a camera equipped with an image sensor;
a raceway identification step of identifying an image portion corresponding to a raceway among each of the tuyere images acquired in the tuyere image acquisition step;
a luminance distribution generating step of generating a luminance distribution of an image portion corresponding to the raceway identified by the raceway identifying step for each of the three primary colors of light;
a parameter calculation step of calculating skewness and/or kurtosis from a luminance distribution in which an average luminance value is less than a luminance threshold value among the luminance distributions generated in the luminance distribution generation step;
An individual raceway state determination step of determining the raceway state for each tuyere based on the magnitude of the skewness and/or kurtosis calculated in the parameter calculation step;
A comprehensive raceway state evaluation step of comparing an evaluation value calculated from a result of tallying up the judgments in the individual raceway state judgment step with a predetermined threshold value to judge at least the raceway state of the entire blast furnace;
A method for evaluating the state of a raceway in a blast furnace, comprising: In the parameter calculation step, both skewness and kurtosis are calculated,
In the individual raceway state determination step, the raceway state is determined for each tuyere based on the magnitude of the skewness and kurtosis calculated in the parameter calculation step, and the cause of the poor raceway state is estimated.
The blast furnace raceway condition evaluation method according to claim 1, characterized in that, when it is determined that the raceway condition of the entire blast furnace is abnormal, the comprehensive raceway condition evaluation step estimates a cause of the abnormality in the raceway condition of the entire blast furnace based on the cause of the poor raceway condition of each tuyere estimated in the individual raceway condition judgment step. When the average value of the red luminance is less than the luminance threshold, the parameter calculation step is performed based on the red luminance distribution;
When the average value of red luminance is equal to or greater than the luminance threshold value and the average value of green luminance is less than the luminance threshold value, the parameter calculation step is performed based on the luminance distribution of green;
The method for evaluating the state of a blast furnace raceway according to claim 1 or 2, characterized in that, when the average value of the brightness of red and green is equal to or greater than the brightness threshold value, the parameter calculation step is carried out based on the brightness distribution of blue. The method for evaluating a state of a blast furnace raceway according to any one of claims 1 to 3, characterized in that the brightness threshold value is 200. A raceway condition evaluation device for determining whether a raceway of a blast furnace is normal or abnormal ,
A raceway identification unit that identifies an image portion corresponding to a raceway among each tuyere image continuously acquired by a camera equipped with an image sensor;
a luminance distribution generating unit that generates a luminance distribution of an image portion corresponding to the raceway identified by the raceway identifying unit for each of the three primary colors of light;
a parameter calculation unit that calculates skewness and/or kurtosis from a luminance distribution in which an average luminance value is less than a luminance threshold value among the luminance distributions generated by the luminance distribution generation unit;
An individual raceway state determination unit that determines the raceway state for each tuyere based on the magnitude of the skewness and/or kurtosis calculated by the parameter calculation unit;
A comprehensive raceway state evaluation unit that compares an evaluation value calculated from a result of tallying up the judgments in the individual raceway state judgment unit with a predetermined threshold value to judge at least the raceway state of the entire blast furnace;
A blast furnace raceway condition evaluation device comprising: A raceway condition evaluation program for determining whether a raceway of a blast furnace is normal or abnormal ,
A raceway identification step of identifying an image portion corresponding to a raceway among each of the tuyere images continuously acquired by a camera equipped with an image sensor;
a luminance distribution generating step of generating a luminance distribution of an image portion corresponding to the raceway identified by the raceway identifying step, for each of the three primary colors of light;
a parameter calculation step of calculating skewness and/or kurtosis from a luminance distribution in which an average luminance value is less than a luminance threshold value among the luminance distributions generated in the luminance distribution generation step;
An individual raceway state determination step of determining the raceway state for each tuyere based on the magnitude of the skewness and/or kurtosis calculated in the parameter calculation step;
A comprehensive raceway state evaluation step for determining at least the raceway state of the entire blast furnace by comparing an evaluation value calculated from the aggregation result of the judgment in the individual raceway state judgment step with a predetermined threshold value;
A raceway condition evaluation program for causing a process computer to execute the above.