Refractory abnormality detection system, display terminal device, refractory abnormality detection method, and refractory abnormality detection program

Abstract

To provide an improved refractory abnormality detection technology.SOLUTION: A refractory abnormality detection system includes an acquisition unit for acquiring a thermal image of a container lined with the refractory, imaged by an infrared thermography, and a control unit for determining abnormality of the thermal image when a feature amount related to the temperature of the thermal image exceeds a threshold set in advance. The threshold is a value calculated in relation to the thermal images serving as multiple references, and calculated on the basis of predetermined statistical processing performed with respect to the feature amount related to the respective temperatures.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a refractory anomaly detection system, a display terminal device, a refractory anomaly detection method, and a refractory anomaly detection program used for refractory linings of refining vessels and the like. [Background technology]

[0002] Refractories used in molten iron vessels, molten steel vessels, etc. are divided into inner linings that come into contact with high-temperature molten materials and outer linings such as the shell that are the housing side. The inner lining refractories come into contact with molten iron, molten steel, and slag. Hereinafter, molten metals mainly containing Fe, such as molten iron and molten steel, are collectively referred to as molten iron. The inner lining refractories are damaged by chemical dissolution due to contact with molten iron and slag, cracks caused by thermal load, and subsequent peeling. Therefore, the inner lining refractories must be replaced with new ones before they are completely worn out. In addition to visually monitoring the lining, the damage status is monitored by thermocouples embedded in the refractory lining, infrared thermography to monitor the shell temperature, and a 3D measuring device to measure the remaining thickness.

[0003] For example, Patent Document 1 proposes a refractory anomaly detection system including a first addition unit that calculates data which is a collection of temperature difference data by subtracting temperature data of corresponding coordinates of current thermal imaging data, which is made up of temperature data photographed to capture the temperature of the refractory at the current time, from reference thermal imaging data, which is made up of temperature data photographed to capture the temperature of the refractory in a reference state of the refractory, a mode calculation unit that calculates the most frequent value of such data, and an offset calculation unit that calculates an offset value based on the most frequent value.

[0004] Furthermore, Patent Document 2 proposes an inspection device for a molten metal storage vessel in which refractory material is applied to the inner side of a shell that forms an outer shell, the inspection device comprising: a plurality of imaging means for capturing thermal images of the entire shell; and an inspection means for calculating a maximum value of a temperature distribution function of the entire shell from the thermal images of the entire shell captured by the plurality of imaging means, and determining that there is a possibility that the refractory material has melted if the calculated maximum value is equal to or greater than a predetermined value, and outputting predetermined information. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2018-205235 A [Patent Document 2] JP 2016-56398 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, the above conventional techniques have the following problems. The technique described in Patent Document 1 requires the acquisition of temperature data at corresponding coordinates. When the field of view is shifted between the reference thermal image and the measurement thermal image, image registration is required. Various useful techniques have been proposed, such as SIFT (Scale-invariant feature transform). For example, when applied to a torpedo whose shooting position shifts slightly each time, a method is considered in which the ridgeline is detected as a feature, the feature is matched, and the image is converted. However, feature extraction is not easy because slag adheres to the iron shell of the torpedo. A method using maximum pooling is also considered to make the image robust against slight position shifts, but the number of pixels is reduced, making it difficult to determine abnormalities. In addition, when photographing with a portable infrared thermography, the position of the subject is different each time, so the same problem occurs. Furthermore, in Patent Document 1, the most frequent value of the temperature difference between the reference thermal image and the measured thermal image is used to calculate the offset. When such temperature difference shows a normal distribution, the method of Patent Document 1 is effective. However, the amount of damage to refractories varies depending on the location, and in many cases, multiple types of refractories with different thermal conductivities are used. Furthermore, refractories are often repaired by partially replacing only the deteriorated parts during intermediate repairs. Therefore, the temperature difference does not necessarily follow a normal distribution. For example, in a case where a distribution shows two maximum values, if the offset is made at the smaller maximum value (the maximum value with a large absolute value of the difference), the offset amount becomes large and abnormality detection becomes impossible. On the other hand, if the offset is made at the larger maximum value (the maximum value with a small absolute value of the difference), the offset amount becomes small, and overdetection occurs in which a non-abnormal container is judged to be abnormal. In this way, when Patent Document 1 is used, it is necessary to classify the offset value according to various conditions such as the repair location of the refractory, the thermal conductivity, and the remaining thickness value of the refractory. Therefore, it is difficult to practically operate Patent Document 1. It is possible to classify refractories into ranges at the boundaries of materials or thicknesses and use the method of Patent Document 1. However, since refractories are subject to the heat of the surrounding refractories, they can only be applied to places that are a certain distance away from the boundary, making this an extremely limited method of use and unrealistic. In addition, the mode is a value with a class width, and even in an ideal vessel with a normal distribution, the resolution is regulated by the class width. If the class width is set small, there is a risk that the overall trend cannot be grasped. For example, if the number of classes is calculated for data of 20,000 pixels according to Sturges's formula, the number of classes will be 15 (calculated based on 1 + log2 20,000). If the number of classes is 15 when the temperature range is 0 degrees to 400 degrees, the class width will be 25 degrees, which will lead to a decrease in accuracy. As such, the technology of Patent Document 1 could not be used in actual operation of a steelworks.

[0007] The technology described in Patent Document 2 judges anomalies using multiple high temperature points. Therefore, in the case of a container with a low operating rate and an overall low temperature, if the threshold for anomaly detection is set small, there will be many false positives, and if the threshold is set large, anomalies cannot be detected. As such, there is room for improvement in refractory anomaly detection technology.

[0008] In view of the above circumstances, an object of the present disclosure is to improve refractory anomaly detection technology. [Means for solving the problem]

[0009] (1) A refractory anomaly detection system according to an embodiment of the present disclosure, an acquisition unit for acquiring a thermal image of the refractory-lined vessel taken by an infrared thermograph; a control unit that determines that the thermal image is abnormal when a feature amount related to temperature of the thermal image exceeds a preset threshold value, The threshold value is a value calculated based on a predetermined statistical process executed on the temperature-related feature amounts calculated for a plurality of reference thermal images.

[0010] (2) A refractory anomaly detection system according to an embodiment of the present disclosure, in (1), The feature amount related to temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value, where α is 0.95 or more.

[0011] (3) A refractory anomaly detection system according to an embodiment of the present disclosure, in (2), The difference values ​​are calculated by classifying the average temperatures of the plurality of reference thermal images.

[0012] (4) A refractory anomaly detection system according to an embodiment of the present disclosure, in any one of (1) to (3), The control unit performs a pre-processing to extract a vessel shell from the thermal image acquired by the acquisition unit.

[0013] (5) A refractory anomaly detection system according to an embodiment of the present disclosure, in (2) or (3), The statistical processing is a process of calculating a temperature obtained by adding A times the standard deviation to the average value of the difference values ​​calculated for each of the reference thermal images as a threshold value, where A is 2 or more.

[0014] (6) A refractory anomaly detection system according to an embodiment of the present disclosure, in (2) or (3), The statistical processing is a process of calculating a threshold temperature by adding B times the interquartile range to the third quartile of the difference values ​​calculated for each of the plurality of reference thermal images, where B is 1 or more.

[0015] (7) A display terminal device according to an embodiment of the present disclosure, a temperature characteristic quantity of a thermal image of a refractory-lined container taken by an infrared thermograph is compared with a preset threshold value to determine whether the thermal image is abnormal or not, and the result is displayed; The threshold value is a value calculated based on a predetermined statistical process executed on the temperature-related feature amounts calculated for a plurality of reference thermal images.

[0016] (8) A display terminal device according to an embodiment of the present disclosure, in (7), The feature amount related to temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value, where α is 0.95 or more.

[0017] (9) A method for detecting an anomaly in a refractory material according to an embodiment of the present disclosure, Obtaining thermal images of the refractory-lined vessel taken with an infrared thermograph; If a feature amount relating to temperature of the thermal image exceeds a preset threshold, the thermal image is determined to be abnormal; The threshold value is a value calculated based on a predetermined statistical process executed on the temperature-related feature amounts calculated for each of a plurality of reference thermal images.

[0018] (10) A refractory anomaly detection method according to an embodiment of the present disclosure, in (9), The feature amount related to temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value, where α is 0.95 or more.

[0019] (11) A refractory anomaly detection method according to an embodiment of the present disclosure, in (10), The difference values ​​are calculated by classifying the average temperatures of the plurality of reference thermal images.

[0020] (12) A refractory anomaly detection method according to an embodiment of the present disclosure, in any one of (9) to (11), A pre-processing step is performed to extract the vessel shell from the acquired thermal image.

[0021] (13) A refractory anomaly detection method according to an embodiment of the present disclosure, in (10) or (11), further comprising: The statistical processing is a process of calculating a temperature obtained by adding A times the standard deviation to the average value of the difference values ​​calculated for each of the reference thermal images as a threshold value, where A is 2 or more.

[0022] (14) A refractory anomaly detection method according to an embodiment of the present disclosure, in (10) or (11), further comprising: The statistical processing is a process of calculating a threshold temperature by adding B times the interquartile range to the third quartile of the difference values ​​calculated for each of the plurality of reference thermal images, where B is 1 or more.

[0023] (15) A refractory anomaly detection program according to an embodiment of the present disclosure, On the computer, obtaining a thermal image of the refractory-lined vessel taken by an infrared thermograph; determining that the thermal image is abnormal when a feature amount related to temperature of the thermal image exceeds a preset threshold value; Run the command, The threshold value is a value calculated based on a predetermined statistical process executed on the temperature-related feature amounts calculated for each of a plurality of reference thermal images.

[0024] (16) A refractory anomaly detection program according to an embodiment of the present disclosure, in (15), The feature amount related to temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value, where α is 0.95 or more. Effect of the Invention

[0025] According to an embodiment of the present disclosure, refractory anomaly detection technology can be improved. [Brief description of the drawings]

[0026] [Figure 1] 1 is a block diagram showing a schematic configuration of a refractory anomaly detection system according to an embodiment of the present disclosure. FIG. [Diagram 2] FIG. 1 is a diagram showing an overview of infrared thermography according to an embodiment of the present disclosure. [Diagram 3] FIG. 13 is a diagram showing a modified example of an outline of infrared thermography according to an embodiment of the present disclosure. [Figure 4] 10 is a flowchart showing an operation of setting a threshold value in the refractory anomaly detection system according to an embodiment of the present disclosure. [Diagram 5] 4 is a flowchart showing a determination operation performed by the refractory anomaly detection system according to the embodiment of the present disclosure. [Figure 6] FIG. 13 is a diagram showing a first pre-processing. [Figure 7] FIG. 13 is a diagram showing a second pre-processing. [Figure 8] FIG. 13 is a diagram showing a third pre-processing. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0027] Hereinafter, a refractory anomaly detection technique according to an embodiment of the present disclosure will be described with reference to the drawings.

[0028] In each drawing, the same or corresponding parts are denoted by the same reference numerals. In the description of this embodiment, the description of the same or corresponding parts will be omitted or simplified as appropriate.

[0029] An overview and configuration of a refractory anomaly detection technology according to this embodiment will be described with reference to Fig. 1. The refractory anomaly detection technology according to this embodiment is executed by a refractory anomaly detection system 1. This technology is used to detect anomalies in refractories of refining vessels and the like (hereinafter also simply referred to as vessels). Refining vessels and the like are used in the steelmaking industry when transporting, holding, and performing various types of refining of molten steel or molten pig iron.

[0030] First, an outline of this embodiment will be described, and details will be described later. In general, refining vessels such as molten iron vessels and molten steel vessels have a structure in which an outer lining refractory and an inner lining refractory are lined inside an exterior shell. The outer lining refractory is constructed so that it can be reused even if the inner lining refractory is damaged and destroyed. The inner lining refractory is damaged by chemical melting and abrasion due to molten metal and slag held in the furnace, as well as peeling due to thermal load. Here, damage to the outer lining refractory can be prevented by dismantling and repairing the inner lining refractory before it is damaged and destroyed. In other words, damage to the outer lining refractory can be prevented by determining an abnormality in the refractory and dismantling and repairing the inner lining refractory when an abnormality in the refractory occurs. By preventing damage to the outer lining refractory in this way, troubles such as molten metal leaking out of the vessel can be avoided and waste can be minimized. Here, an abnormality in the refractory refers to a state in which the inner lining of one ordinary brick, which is a standard brick as defined in JIS R2101-1983, has disappeared and the outer lining refractory has been exposed.

[0031] The refractory anomaly detection system 1 according to the present embodiment aims to accurately determine the anomaly of the refractory. Specifically, the refractory anomaly detection system 1 according to the present embodiment includes an acquisition unit that acquires a thermal image of a container lined with a refractory taken by an infrared thermograph (infrared camera), and a control unit that determines the thermal image as abnormal when a feature value related to temperature of the thermal image exceeds a preset threshold value. The threshold value is a value calculated based on a predetermined statistical process executed on each feature value related to temperature calculated for a plurality of reference thermal images. Here, the more reference thermal images there are, the higher the accuracy of determining whether the thermal image of the container is abnormal or not. The reference thermal images are preferably 30 or more, and more preferably 100 or more. In addition, a thermal image selected in advance may be used as the reference thermal image. For example, if the temperature data at a position corresponding to the container lined with a refractory contains a lot of extremely low temperature data (for example, less than 150°C), it is highly likely that the temperature other than the container shell is being measured, so the reference thermal image is not used.

[0032] In this way, the refractory anomaly detection system 1 according to this embodiment uses a threshold value that is statistically processed using a plurality of reference thermal images. More specifically, a feature amount related to temperature is calculated for each of the plurality of reference thermal images, and a predetermined statistical process is performed on the feature amount to calculate a threshold value. In this way, it is possible to detect an anomaly in a refractory with high accuracy, and the refractory anomaly detection technology can be improved.

[0033] Next, each component of the refractory anomaly detection system 1 will be described in detail.

[0034] The refractory anomaly detection system 1 includes an information processing device 10 and a display terminal device 11. The information processing device 10 and the display terminal device 11 are any devices used by a user (for example, a manager or person in charge of a steelworks). For example, a general-purpose electronic device or a dedicated electronic device can be adopted as the information processing device 10 and the display terminal device 11. Note that the refractory anomaly detection system 1 in this embodiment will be described as including the information processing device 10 and the display terminal device 11. However, the configuration of the refractory anomaly detection system 1 is not limited to this. For example, the information processing device 10 may have the function of the display terminal device 11, and in this case, the information processing device 10 may function as the refractory anomaly detection system 1.

[0035] As shown in FIG. 1, the information processing device 10 includes an acquisition unit 101, a control unit 102, a storage unit 103, a communication unit 104, and an output unit 105.

[0036] The acquisition unit 101 acquires data used for the operation of the information processing device 10. Specifically, the acquisition unit 101 acquires a thermal image of a container lined with a refractory material, captured by the infrared thermographer 2. In other words, the acquisition unit 101 functions as an interface for connection with the infrared thermographer 2. As a method for connecting with the infrared thermographer 2, any method such as USB, HDMI (registered trademark), or Bluetooth (registered trademark) can be used.

[0037] An overview of photographing by the infrared thermograph 2 is shown with reference to Figures 2 and 3. Figure 2 shows a state in which the refractory material 3 is photographed by multiple fixed infrared thermographs 21, 22. Figure 3 shows a state in which the refractory material 3 is photographed by moving a single portable infrared thermograph 23. As shown in Figures 2 and 3, in this embodiment, the infrared thermograph 2 photographs the refractory material 3 from multiple different directions. In this way, the infrared thermograph 2 can capture a wide range of the refractory material 3 in its field of view.

[0038] The control unit 102 includes at least one processor, at least one dedicated circuit, or a combination of these. The processor is a general-purpose processor such as a central processing unit (CPU) or a graphics processing unit (GPU), or a dedicated processor specialized for a specific process. The dedicated circuit is, for example, a field-programmable gate array (FPGA) or an application specific integrated circuit (ASIC). The control unit 102 executes processes related to the operation of the information processing device 10 while controlling each unit of the information processing device 10.

[0039] The storage unit 103 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The semiconductor memory is, for example, a random access memory (RAM) or a read only memory (ROM). The RAM is, for example, a static random access memory (SRAM) or a dynamic random access memory (DRAM). The ROM is, for example, an electrically erasable programmable read only memory (EEPROM). The storage unit 103 functions as, for example, a main storage device, an auxiliary storage device, or a cache memory. The storage unit 103 stores data used in the operation of the information processing device 10 and data obtained by the operation of the information processing device 10. For example, the storage unit 103 stores a thermal image acquired by the acquisition unit 101. The thermal image may be stored in an external device different from the information processing device 10. In this case, the information processing device 10 uses the thermal image by accessing the external device that stores the thermal image via the communication unit 104.

[0040] The communication unit 104 includes at least one external communication interface. The communication interface may be either a wired communication interface or a wireless communication interface. In the case of wired communication, the communication interface is, for example, a LAN (Local Area Network) interface or a USB (Universal Serial Bus). In the case of wireless communication, the communication interface is, for example, an interface compatible with mobile communication standards such as LTE (Long Term Evolution), 4G (4th generation), or 5G (5th generation), or an interface compatible with short-range wireless communication such as Bluetooth (registered trademark). The communication unit 104 receives data used in the operation of the information processing device 10, and transmits data obtained by the operation of the information processing device 10.

[0041] The output unit 105 includes at least one output interface. The output interface is an interface for connecting to the display terminal device 11 in a predetermined manner. As the connection method, any method such as USB, HDMI (registered trademark), or Bluetooth (registered trademark) can be used.

[0042] The display terminal device 11 includes, for example, an LCD (liquid crystal display) or an organic EL (electro luminescence) display. The display terminal device 11 displays and outputs data obtained by the operation of the information processing device 10.

[0043] The functions of the information processing device 10 are realized by executing a program according to this embodiment on a processor equivalent to the information processing device 10. That is, the functions of the information processing device 10 are realized by software. The program causes a computer to execute the operations of the information processing device 10, thereby causing the computer to function as the information processing device 10. That is, the computer functions as the information processing device 10 by executing the operations of the information processing device 10 in accordance with the program.

[0044] In this embodiment, the program can be recorded on a computer-readable recording medium. The computer-readable recording medium includes a non-transitory computer-readable medium, such as a magnetic recording device, an optical disc, a magneto-optical recording medium, or a semiconductor memory. The program is distributed, for example, by selling, transferring, or lending a portable recording medium such as a DVD (digital versatile disc) or a CD-ROM (compact disc read only memory) on which the program is recorded. The program may also be distributed by storing the program in the storage of an external server and transmitting the program from the external server to another computer. The program may also be provided as a program product.

[0045] A part or all of the functions of the information processing device 10 may be realized by a dedicated circuit equivalent to the control unit 102. In other words, a part or all of the functions of the information processing device 10 may be realized by hardware.

[0046] (Setting the threshold) Next, an operation of the information processing device 10 according to the present embodiment will be described. Fig. 4 is a flowchart showing an operation of setting a threshold value in the refractory anomaly detection system 1 according to an embodiment of the present disclosure.

[0047] Step S100: The acquisition unit 101 of the information processing device 10 acquires a plurality of reference thermal images captured by the infrared thermograph 2.

[0048] Step S110: The control unit 102 performs pre-processing on the multiple reference thermal images acquired by the acquisition unit 101. The pre-processing is a process for extracting the container shell from the thermal images acquired by the acquisition unit 101. In other words, the pre-processing is an image process for extracting the part of the container to be judged from the thermal images.

[0049] Step S120: The control unit 102 calculates the average temperature of each of the pre-processed reference thermal images.

[0050] Step S130: The control unit 102 calculates a feature quantity related to temperature for each of the multiple reference thermal images. In this embodiment, the feature quantity related to temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value. α is 0.95 or more. If α is 0.95 or more, a certain degree of accuracy can be ensured. Note that, in terms of accuracy, it is preferable that α is 0.995 or more.

[0051] Step S140: The control unit 102 sets a threshold based on statistical processing of the feature amount. For example, the statistical processing is a process of obtaining the average value and standard deviation of the feature amount of all reference images from the feature amount of each reference image. The control unit 102 sets a threshold based on the average value and standard deviation of the feature amount. Specifically, the control unit 102 sets a value obtained by adding A times the standard deviation to the average value of the feature amount as the threshold. A is, for example, a value of 2 or more. Note that, when the population of data follows a normal distribution, the probability that a certain piece of data will take a value exceeding the average value + 2σ and the probability that a certain piece of data will take a value exceeding the average value + 3σ are 2.28% and 0.13%, respectively, according to the standard normal distribution table. Therefore, by setting A to a value of 2 or more, abnormal data exceeding the threshold can be effectively determined.

[0052] The statistical processing may be, for example, a process of determining the third quartile and interquartile range of the feature of all the reference images from the feature of each reference image. The control unit 102 sets a threshold value based on the third quartile and interquartile range of the feature. Specifically, the control unit 102 sets a value obtained by adding B times the interquartile range to the third quartile of the feature as the threshold value. B is, for example, a value of 1 or more. The interquartile range is about 1.34σ, and the value obtained by adding 1 times the interquartile range to the third quartile is about 2σ. The value obtained by adding 1.5 times the interquartile range to the third quartile is about 2.68σ. According to the standard normal distribution table, the probability of exceeding the value obtained by adding 1 times the interquartile range to the third quartile and the probability of exceeding the value obtained by adding 1.5 times the interquartile range to the third quartile are 2.28% and 0.37%, respectively. Therefore, by setting B to a value of 1 or more, abnormal data exceeding the threshold can be effectively determined.

[0053] (Determining Abnormality) Next, a determination operation performed by the refractory anomaly detection system according to an embodiment of the present disclosure will be described with reference to the flowchart of FIG.

[0054] Step S200: The acquisition unit 101 of the information processing device 10 acquires a thermal image of the target object captured by the infrared thermograph 2.

[0055] Step S210: The control unit 102 performs preprocessing on the thermal image of the object to be determined acquired by the acquisition unit 101. This preprocessing is similar to the preprocessing of step S110 in Fig. 4. That is, the preprocessing is a process of extracting the container shell from the thermal image of the object to be determined acquired by the acquisition unit 101. This preprocessing includes the above-mentioned first preprocessing, second preprocessing, and third preprocessing.

[0056] Step S220: The control unit 102 calculates the average temperature of the thermal image to be determined.

[0057] Step S230: The control unit 102 calculates a feature quantity related to the temperature of the thermal image to be determined. In this embodiment, the feature quantity related to the temperature is a difference value between the average temperature value of the thermal image and the 100α percentile value. α is 0.95 or more, and the same value as α used when setting the threshold value is used.

[0058] Step S240: The control unit 102 determines whether the feature amount of the thermal image to be determined exceeds the threshold value set in step S140. If the feature amount of the thermal image to be determined exceeds the threshold value, the process proceeds to step S250. If the feature amount of the thermal image to be determined does not exceed the threshold value, the process proceeds to step S260.

[0059] Step S250: The control unit 102 determines that the thermal image is abnormal. The control unit 102 outputs the determination result from the output unit 15. The output unit 15 displays and outputs the determination result on the display terminal device 11. For the containers determined to be abnormal, an offline refractory inspection is performed.

[0060] Step S260: The control unit 102 determines that the thermal image is normal. The control unit 102 outputs the determination result from the output unit 15. The output unit 15 displays and outputs the determination result on the display terminal device 11. The target container determined to be normal can continue to be used as is.

[0061] As described above, according to the refractory anomaly detection system 1 of this embodiment, by using a threshold value that is statistically processed using a plurality of reference thermal images, it is possible to detect an anomaly in a refractory with high accuracy, and the refractory anomaly detection technology can be improved. More specifically, according to the refractory anomaly detection system 1 of this embodiment, a feature amount related to temperature is calculated for each of a plurality of reference thermal images, and a threshold value is calculated by performing a predetermined statistical process on the feature amount, so that an anomaly in a refractory can be detected with high accuracy.

[0062] (Example) An example of detecting a refractory anomaly using the refractory anomaly detection system 1 according to this embodiment will be described below. In this example, the shell temperature of a torpedo, a molten iron transport vessel roughly measuring 20 m in total length, 4.3 m in width, and 4.3 m in height, was measured by infrared thermography. In this example, after the measurement, the presence or absence of a refractory anomaly in the furnace was actually inspected visually. The example will be described below in detail.

[0063] In this example, a torpedo receiving molten iron in a blast furnace was placed in front of an infrared thermograph, and a thermal image was taken.

[0064] The pre-processing according to this embodiment includes a first pre-processing, a second pre-processing, and a third pre-processing. The first pre-processing is a process for removing high temperature areas unrelated to refractory abnormalities, and more specifically, is an image process for removing the effects of the sun or the like captured in the thermal image.

[0065] With reference to FIG. 6, the first pre-processing will be described. First, the control unit 102 binarizes the original data 501 using a threshold value related to temperature (hereinafter, also referred to as the temperature threshold value) to generate binarized data 502. The temperature threshold value is, for example, 100° C. In this process, the control unit 102 may perform noise removal processing. After noise removal processing, expansion processing and contraction processing may be performed. When performing these processes, the control unit 102 uses a maximum value filter or a minimum value filter. This makes it possible to clarify the outline of the pixels. Pixel 505 indicates a pixel to be excluded. Next, the control unit 102 performs labeling processing on the binarized data 502 according to the pixel position to generate labeling data 503. The labeling processing is a process of labeling adjacent pixels and dividing them into elements. By this process, the same labeling is performed on adjacent pixels. The control unit 102 generates processed data 504 from which outlying areas have been removed based on the labeling data 503. Specifically, among the labeled pixels, a group (here, pixel 505) that has an area smaller than the main body of the container is determined to be an outlying area by using a threshold value. Such outlying areas are caused by the influence of the sun, and are unnecessary pixels. This process removes pixel 505 that is caused by the influence of the sun.

[0066] Next, the second pre-processing will be outlined with reference to FIG. 7. The second pre-processing is a process for removing high temperature areas unrelated to refractory abnormalities, specifically, image processing for removing the effects of railroad tracks. The position of the railroad tracks is fixed within the image for each shooting location. Therefore, the effects of the railroad tracks are removed by removing data outside a set range for each shooting location.

[0067] First, the control unit 102 detects the height and horizontal positions (hereinafter also referred to as top, bottom, left and right positions) of the container of the object to be determined using data 601 after the first pre-processing. The data 602 includes auxiliary lines. The height position is indicated by auxiliary line 604. The horizontal position is indicated by auxiliary line 605 and auxiliary line 606. The control unit 102 estimates the shooting location based on the detected top, bottom, left and right positions and determines the position of the track. The shooting position of the container varies greatly in the horizontal direction, but varies little in the height direction. The control unit 102 estimates the shooting location using the characteristics of the shooting position. In addition, the position of the track can be specified based on the shooting location and the relative relationship between the position of the container on the thermal image and the position of the track. The control unit 102 generates processed data 603 excluding the track position 607 based on the specified track position.

[0068] Next, an overview of the third pre-processing will be described with reference to FIG. 8. The third pre-processing is a process for improving the accuracy of refractory abnormality determination, and specifically, is image processing for removing the influence of the cast steel of the vessel. The torpedo uses thick cast steel around the upper opening of the vessel. This is to prevent the iron shell from being damaged by slag that scatters and flows out from the opening. Therefore, if the cast steel is included as a target, the temperature will be dulled even if there is a refractory abnormality, making it difficult to detect the abnormality. Therefore, in this embodiment, the cast steel is excluded from the determination target.

[0069] First, the control unit 102 generates data 701 by removing low-temperature parts (below 100°C) from the processed data after the first pre-processing and the second pre-processing have been performed. The data 701 includes cast steel hardware 704. Next, the control unit 102 detects the height and horizontal ends of the container of the object to be determined. The data 702 includes auxiliary lines indicating such ends. The height ends are indicated by an upper end auxiliary line 705 and a lower end auxiliary line 706. The horizontal position is indicated by a left end auxiliary line 707 and a right end auxiliary line 708.

[0070] Next, the control unit 102 detects the straight body part of the container. Specifically, the control unit 102 counts the number of pixels in the horizontal direction of the container from the bottom end auxiliary line 706 toward the top end auxiliary line 705. The control unit 102 identifies a position (hereinafter also referred to as a horizontal line) where the number of pixels corresponding to the container in the horizontal direction exceeds a value obtained by multiplying the number of pixels from the top end to the bottom end of the container by 5 / 6. The control unit 102 detects the straight body part based on the intersection point between the horizontal line and the pixel corresponding to the container. The data 703 is data including an auxiliary line indicating the range of the straight body part determined based on the intersection point. An auxiliary line 709 and an auxiliary line 710 indicate the range of the detected straight body part.

[0071] Next, the control unit 102 sets a reference line 715. The reference line 715 is a line that passes through the center of the straight body portion. The data 711 is data that includes the reference line 715.

[0072] Next, the control unit 102 performs filtering using a maximum value filter and a minimum value filter to set a reference point 716. Specifically, the control unit 102 detects the point on the reference line 715 where the temperature is the lowest, and sets this point as the reference point 716. Data 712 is data that indicates the reference point 716.

[0073] Here, the bottom end of the cast steel hardware 704 is a place on the reference line 715 where the temperature changes suddenly. Therefore, the control unit 102 detects the point where the difference between the average temperature values ​​in three pixels above and below the reference point 716 toward the bottom end is maximum. The control unit 102 sets a straight line perpendicular to the reference line 715 from the maximum point as the bottom end auxiliary line 717 of the cast steel hardware 704. Data 713 is data that indicates the bottom end auxiliary line 717 of the cast steel hardware 704.

[0074] Next, the control unit 102 detects the horizontal end of the cast steel hardware 704. Specifically, the control unit 102 detects the point where the difference between the average values ​​of the temperatures in three pixels on the left and right of the lower end auxiliary line 717 of the straight body part of the cast steel hardware 704 is maximum. The control unit 102 sets a straight line that passes through the point and is parallel to the height direction as the horizontal end auxiliary lines 718, 719 of the cast steel hardware 704. The data 714 is data indicating the horizontal end auxiliary lines 718, 719 of the cast steel hardware 704. By setting the lower end auxiliary line 717 and the horizontal end auxiliary lines 718, 719 of the cast steel hardware 704 in this way, the range of the cast steel hardware 704 in the thermal image is specified, and the control unit 102 can exclude the influence of the cast steel hardware 704 by excluding the range.

[0075] In this example, the difference between the average temperature and the 100α percentile temperature was calculated from the data of 10,000 normal torpedoes obtained in the process of creating the reference thermal image. In the following Examples 1-9, only α was changed, with α = 0.9995, 0.999, 0.9985, 0.998, 0.997, 0.995, 0.99, 0.97, and 0.95, respectively. Table 1 shows the statistical data of the calculated difference values ​​(average, first quartile, median, third quartile, interquartile range, and standard deviation).

[0076] [Table 1]

[0077] In this embodiment, the temperature obtained by adding A times the standard deviation (A=3 in this embodiment) to the average value was set as threshold 1. The temperature obtained by adding B times the interquartile range (B=1.5 in this embodiment) to the third quartile was set as threshold 2. Table 2 shows thresholds 1 and 2 in each of the embodiments 1-9.

[0078] [Table 2]

[0079] Next, the difference between the average temperature and the 100α percentile temperature was calculated for 14 thermal image data (hereinafter referred to as measurements 1-14) that were actually measured. In addition, the presence or absence of abnormalities in the refractory, which was precisely determined by visual inspection after the operation was stopped, is shown in Table 3.

[0080] [Table 3]

[0081] Table 4 shows the judgment results based on threshold 1 for measurement 1-7, which had a refractory abnormality, and measurement 8-14, which had a normal refractory. Similarly, Table 5 shows the judgment results based on threshold 2 for measurement 1-7, which had a refractory abnormality, and measurement 8-14, which had a normal refractory. Note that in each table, "Abnormal judgment results" show the judgment results for measurement 1-7, and "Normal judgment results" show the judgment results for measurement 8-14.

[0082] [Table 4]

[0083] [Table 5]

[0084] As shown in Tables 4 and 5, according to the refractory anomaly detection method of this embodiment, whether threshold value 1 or threshold value 2 is set, the refractory anomaly judgment was successful in 14% or more. Moreover, whether threshold value 1 or threshold value 2 is set, normal refractory was not erroneously judged as abnormal. Furthermore, in Examples 1-6, the result of abnormality judgment was 100%. From this, it is considered that α≧0.995 is preferable. Even if threshold value 1 is used for Example 9, it is useful because it is possible to improve the durability of the outer lining refractory and avoid refractory troubles by detecting abnormalities within the durability range. For example, if the durability is 20 charges, when an event in which an abnormality can be detected with a probability of 14% is repeatedly tried 20 times, the probability that an abnormality can be detected within 20 charges is 95%.

[0085] Alternatively, the reference thermal image may be divided into classes according to the average temperature, and thresholds (hereinafter also referred to as class-specific thresholds) may be set based on the classified population. In this case, a corresponding class-specific threshold is used according to the class of the average temperature of the target thermal image.

[0086] The class thresholds will be described with reference to Table 6. The class thresholds shown in Table 6 are thresholds corresponding to three classes of average temperatures: less than 160°C, 160°C or more and less than 180°C, and 180°C or more (hereinafter also referred to as the first class threshold, the second class threshold, and the third class threshold, respectively). The first class threshold is a threshold used when the average temperature of the thermal image of the target is less than 160°C. The second class threshold is a threshold used when the average temperature of the thermal image of the target is 160°C or more and less than 180°C. The third class threshold is a threshold used when the average temperature of the thermal image of the target is 180°C or more. The first class threshold is a threshold set based on the images (hereinafter also referred to as the first class reference images) that have an average temperature less than 160°C among all the reference thermal images. Specifically, first, for all the first class reference images, the difference value between the average temperature value of the thermal image and the 100α percentile value is calculated as the feature value. Next, the third quartile and the interquartile range of the feature value of the first class reference image are calculated. Next, a value obtained by adding B times the interquartile range to the third quartile of the feature is set as a threshold. The second class threshold is a threshold set based on images (hereinafter also referred to as second class reference images) of all reference thermal images with an average temperature of 160°C or more and less than 180°C. Specifically, for all reference images of the second class, a difference value between the average temperature value of the thermal image and the 100α percentile value is calculated as a feature. Next, the third quartile and the interquartile range of the feature of the reference images of the second class are calculated. Next, a value obtained by adding B times the interquartile range to the third quartile of the feature is set as a threshold. The third class threshold is a threshold set based on images (hereinafter also referred to as third class reference images) of all reference thermal images with an average temperature of 180°C or more. Specifically, for all reference images of the third class, a difference value between the average temperature value of the thermal image and the 100α percentile value is calculated as a feature. Next, the third quartile and interquartile range of the feature of the reference image of the third class are calculated. Then, the third quartile of the feature is added to B times the interquartile range and set as the threshold. In this embodiment, the above B is set to 1.5. Note that "Overall" in Table 6 is a reference value, and indicates the case where a threshold is set without classifying. The threshold for "Overall" is the same as threshold 2 in Table 2.

[0087] [Table 6]

[0088] Table 7 shows the excess temperatures relative to the thresholds set in Examples 1-7 for data showing refractory abnormalities, both with class-specific thresholds and with a constant threshold. When the threshold is constant (no class-specific thresholds are used), the excess temperatures at 160-180°C are relatively small. On the other hand, when the class-specific thresholds are used, the excess temperatures are relatively large regardless of the average temperature. In this way, by using class-specific thresholds that are classified according to the average temperature of the reference thermal image, anomalies can be detected more accurately.

[0089] [Table 7]

[0090] In Table 6 above, the first, second, and third threshold levels are used, but the method of dividing the levels is not limited to this. The upper and lower limits of the levels and the number of levels may be set appropriately depending on the variance of the data.

[0091] In the present embodiment, the feature amount is the difference between the average temperature value of the thermal image and the 100α percentile value, but is not limited to this. Any other feature amount may be used as long as it can be an index for determining an abnormality. For example, the feature amount may be a value calculated using the median temperature or the like.

[0092] Although the refractory anomaly detection system 1 according to the present embodiment performs an anomaly determination using a feature amount, the present invention is not limited to this. The refractory anomaly detection system 1 may perform an anomaly determination using an absolute value of the temperature of a thermal image as a threshold value.

[0093] In the present embodiment, the information processing device 10 includes the output unit 105, and the result of calculation by the control unit 102 of the information processing device 10 is displayed and output from the display terminal device 11 by the output unit 105. However, the present embodiment is not limited to this. For example, the technology according to the present embodiment may be executed in a Software as a Service format (SaaS format). In this case, the information processing device 10 functions as a cloud server. In this case, when receiving a request from the display terminal device 11, the information processing device 10 causes the display terminal device 11 to output the judgment result. In this case, the display terminal device 11 includes a communication unit, receives the judgment result from the information processing device 10 via the communication unit, and displays information related to the judgment result. In other words, the display terminal device 11 displays the result of judging whether the thermal image is abnormal or not by comparing the feature amount related to the temperature of the thermal image of the container lined with the refractory material, which is taken by the infrared thermography, with a preset threshold value. The request for the judgment result to the information processing device 10 can be made by any method. For example, the display terminal device 11 may include an input unit that accepts input from a user, and may transmit a request for a determination result to the information processing device 10 based on the content of the input accepted by the input unit.

[0094] Although the present disclosure has been described based on the drawings and embodiments, it should be noted that those skilled in the art can easily make various modifications and corrections based on the present disclosure. Therefore, it should be noted that these modifications and corrections are included in the scope of the present disclosure. For example, the functions included in each means or step can be rearranged so as not to be logically inconsistent, and multiple means or steps can be combined or divided into one. [Explanation of symbols]

[0095] 1 Refractory Abnormality Detection System 10. Information processing device 101 Acquisition Department 102 Control section 103 Storage section 104 Communications Department 105 Output section 11 Display terminal device 2, 21, 22, 23 Infrared Thermography 501 Original Data 502 Binarized Data 503 Labeling Data 504 Processed Data 505 pixels 601, 602, 603 Data 604, 605, 606 auxiliary lines 607 Track position 701,702,703,711,712,713,714 Data 704 Cast steel hardware 705 Top auxiliary line 706 Bottom auxiliary line 707 Left end auxiliary line 708 Right end auxiliary line 709, 710 auxiliary lines 715 Reference Line 716 Reference point 717 Bottom auxiliary line 718, 719 Horizontal auxiliary line

Claims

1. an acquisition unit for acquiring a thermal image of the refractory-lined vessel taken by an infrared thermograph; a control unit that determines that the thermal image is abnormal when a feature amount related to temperature of the thermal image exceeds a preset threshold value, A refractory anomaly detection system, wherein the threshold value is a value calculated based on a predetermined statistical processing performed on each of the temperature-related feature quantities calculated for a plurality of reference thermal images.

2. 2. The refractory anomaly detection system according to claim 1, wherein the feature amount related to temperature is a difference value between an average temperature value of the thermal image and a 100α percentile value, where α is 0.95 or more.

3. The refractory anomaly detection system according to claim 2 , wherein the difference values ​​are calculated by classifying the difference values ​​into classes according to average temperatures of the plurality of reference thermal images.

4. The refractory anomaly detection system according to claim 2 or 3, wherein the control unit performs a pre-processing step of extracting a vessel shell from the thermal image acquired by the acquisition unit.

5. 4. The refractory anomaly detection system according to claim 2 or 3, wherein the statistical processing is a process of calculating a temperature that is a threshold value obtained by adding A times the standard deviation to an average value of the difference values ​​calculated for each of the plurality of reference thermal images, where A is 2 or more.

6. 4. The refractory anomaly detection system according to claim 2 or 3, wherein the statistical processing is a process of calculating a threshold temperature that is a third quartile of the difference values ​​calculated for each of the plurality of reference thermal images plus B times an interquartile range, where B is 1 or more.

7. a temperature characteristic quantity of a thermal image of a refractory-lined container taken by an infrared thermograph is compared with a preset threshold value to determine whether the thermal image is abnormal or not, and the result is displayed; The threshold value is a value calculated based on a predetermined statistical process executed on each of the temperature-related feature amounts calculated for a plurality of reference thermal images.

8. 8. The display terminal device according to claim 7, wherein the feature amount relating to temperature is a difference value between an average temperature value of the thermal image and a 100α percentile value, where α is 0.95 or more.

9. Obtaining thermal images of the refractory-lined vessel taken with an infrared thermograph; If a feature amount relating to temperature of the thermal image exceeds a preset threshold, the thermal image is determined to be abnormal; the threshold value is a value calculated based on a predetermined statistical process executed on each of the temperature-related feature quantities calculated for a plurality of reference thermal images; Refractory anomaly detection method.

10. 10. The refractory anomaly detection method according to claim 9, wherein the feature amount related to temperature is a difference value between an average temperature value of the thermal image and a 100α percentile value, where α is 0.95 or more.

11. The refractory anomaly detection method according to claim 10, wherein the difference values ​​are calculated by classifying the difference values ​​into classes according to average temperatures of the plurality of reference thermal images.

12. The refractory anomaly detection method according to claim 10 or 11, further comprising the step of carrying out a pre-processing step of extracting a vessel shell from the acquired thermal image.

13. 12. The refractory anomaly detection method according to claim 10, wherein the statistical processing is a process of calculating a temperature that is a threshold value obtained by adding A times the standard deviation to an average value of the difference values ​​calculated for each of the plurality of reference thermal images, where A is 2 or more.

14. 12. The refractory anomaly detection method according to claim 10, wherein the statistical processing is a process of calculating a threshold temperature that is a third quartile of the difference values ​​calculated for each of the plurality of reference thermal images plus B times an interquartile range, where B is 1 or more.

15. On the computer, obtaining a thermal image of the refractory-lined vessel taken by an infrared thermograph; determining that the thermal image is abnormal when a feature amount related to temperature of the thermal image exceeds a preset threshold value; Run the command, The threshold value is a value calculated based on a predetermined statistical process executed on each of the temperature-related feature quantities calculated for a plurality of reference thermal images. Refractory anomaly detection program.

16. 16. The refractory anomaly detection program according to claim 15, wherein the feature amount relating to temperature is a difference value between an average temperature value of the thermal image and a 100α percentile value, where α is 0.95 or more.

Images