TWI776618B - Method and system for identifying receiving steel bucket - Google Patents

Method and system for identifying receiving steel bucket Download PDF

Info

Publication number
TWI776618B
TWI776618B TW110128312A TW110128312A TWI776618B TW I776618 B TWI776618 B TW I776618B TW 110128312 A TW110128312 A TW 110128312A TW 110128312 A TW110128312 A TW 110128312A TW I776618 B TWI776618 B TW I776618B
Authority
TW
Taiwan
Prior art keywords
thermal image
temperature
category
colors
tapping hole
Prior art date
Application number
TW110128312A
Other languages
Chinese (zh)
Other versions
TW202307787A (en
Inventor
蘇育德
戴字庭
蕭家科
謝易錚
Original Assignee
中國鋼鐵股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 中國鋼鐵股份有限公司 filed Critical 中國鋼鐵股份有限公司
Priority to TW110128312A priority Critical patent/TWI776618B/en
Application granted granted Critical
Publication of TWI776618B publication Critical patent/TWI776618B/en
Publication of TW202307787A publication Critical patent/TW202307787A/en

Links

Images

Landscapes

  • Image Analysis (AREA)
  • General Factory Administration (AREA)
  • Control And Safety Of Cranes (AREA)
  • Discharge Of Articles From Conveyors (AREA)

Abstract

A system for identifying a receiving steel bucket includes the receiving steel bucket, a thermal sensor and a computer system. The receiving steel bucket includes a mixing brick and an exit hole. The thermal sensor takes a thermal image of the receiving steel bucket. The computer system determines whether the thermal image belongs to a first temperature category or a second temperature category. If the thermal image belongs to the first temperature category, the thermal image is input to a first machine learning model to detect the mixing brick and the exit hole, and determine whether the mixing brick and the exit hole are abnormal. If the thermal image belongs to the second temperature category, the thermal image is input to a second machine learning model to detect the mixing brick and the exit hole, and determine whether the mixing brick and the exit hole are abnormal.

Description

接鋼桶辨識系統與方法Steel drum identification system and method

本揭露是關於接鋼桶辨識系統與方法，特別是用熱影像來判斷接鋼桶中的元件是否異常。The present disclosure relates to a system and method for identifying steel drums, especially using thermal images to determine whether the components in the steel drums are abnormal.

在煉鋼的製程中，轉爐倒出的鋼液會倒進接鋼桶，接鋼桶中有攪拌磚與出鋼孔。攪拌磚用以噴出氣體來攪拌鋼液。出鋼孔在倒入鋼液之前會設置填充砂，填充砂會在出鋼口凝結成塊以阻塞出鋼孔，使鋼液倒入時不會直接洩出。接鋼桶中的鋼液會進行精煉程序，待精煉程序結束到下一個製程(連鑄區)時會拉開底部滑板，利用鋼液壓力衝破凝結塊，完成出鋼。如果攪拌磚沒有正常通氣，則會嚴重影響精煉程序，在習知技術中是以氣壓/氣流計的方式進行判斷或是以目視方式判斷鋼液是否有被攪動。要判斷出鋼孔是否異常，則需要判斷鋼液是否順利流出，如果發生阻塞，則需要由人員進行吹氣將凝結塊熔穿以後才能進行出鋼。因此，上述關於攪拌磚與出鋼孔的辨識方式都仰賴人力且屬於被動方式，必須等到製程有錯誤時才能知道有異常，如何提出一種主動的辨識方式，為此領域技術人員所關心的議題。In the process of steelmaking, the molten steel poured from the converter will be poured into the receiving drum, and there are stirring bricks and tapping holes in the receiving drum. The stirring brick is used to spray gas to stir the molten steel. The tapping hole will be filled with sand before pouring the molten steel. The filling sand will condense into a block at the tapping hole to block the tapping hole, so that the molten steel will not leak out directly. The molten steel in the steel drum will undergo a refining process. When the refining process ends to the next process (continuous casting area), the bottom slide plate will be pulled open, and the liquid steel pressure will be used to break through the coagulation block to complete the tapping. If the stirring brick is not properly ventilated, it will seriously affect the refining process. In the prior art, it is judged by means of air pressure/air flow meter or whether the molten steel is agitated by visual means. To judge whether the tapping hole is abnormal, it is necessary to judge whether the molten steel flows out smoothly. Therefore, the above identification methods for mixing bricks and tapping holes all rely on manpower and are passive methods. It is necessary to wait until there is an error in the process to know that there is an abnormality. How to propose an active identification method is a topic of concern to those skilled in the art.

本發明的實施例提出一種接鋼桶辨識系統，包括接鋼桶、熱感測器與電腦系統。接鋼桶包括攪拌磚與出鋼孔。熱感測器用以拍攝接鋼桶以產生熱影像。電腦系統通訊連接至熱感測器以取得熱影像，判斷熱影像屬於第一溫度類別或第二溫度類別。若熱影像屬於第一溫度類別，將熱影像輸入至第一機器學習模型以偵測攪拌磚與出鋼孔，並判斷攪拌磚與出鋼孔是否異常。若熱影像屬於第二溫度類別，將熱影像輸入至第二機器學習模型以偵測攪拌磚與出鋼孔，並判斷攪拌磚與出鋼孔是否異常。An embodiment of the present invention provides a system for identifying a connected steel drum, which includes a connected steel drum, a thermal sensor and a computer system. The connecting drum includes mixing bricks and tapping holes. The thermal sensor is used to photograph the steel drum to generate thermal images. The computer system is communicatively connected to the thermal sensor to obtain thermal images, and determine whether the thermal images belong to the first temperature category or the second temperature category. If the thermal image belongs to the first temperature category, the thermal image is input to the first machine learning model to detect the stirring brick and the tapping hole, and determine whether the stirring brick and the tapping hole are abnormal. If the thermal image belongs to the second temperature category, the thermal image is input to the second machine learning model to detect the stirring brick and the tapping hole, and determine whether the stirring brick and the tapping hole are abnormal.

在一些實施例中，上述的第一溫度類別為高溫類別，第二溫度類別為低溫類別。In some embodiments, the above-mentioned first temperature category is a high temperature category, and the second temperature category is a low temperature category.

在一些實施例中，電腦系統還用以將熱影像中的每個像素分類為多個顏色的其中之一，這些顏色包括多個高溫顏色與多個低溫顏色。如果具有最多像素的顏色屬於高溫顏色，判斷熱影像屬於高溫類別。如果具有最多像素的顏色屬於低溫顏色，判斷熱影像屬於低溫類別。In some embodiments, the computer system is further configured to classify each pixel in the thermal image as one of a plurality of colors, including a plurality of high temperature colors and a plurality of low temperature colors. If the color with the most pixels belongs to the high temperature color, the thermal image is judged to be in the high temperature category. If the color with the most pixels is a low temperature color, the thermal image is judged to be in the low temperature category.

在一些實施例中，熱影像包括多個像素，每個像素對應至一溫度值，電腦系統還用以計算這些溫度值的平均值。如果平均值大於一臨界值，電腦系統用以判斷熱影像屬於高溫類別。如果平均值小於等於臨界值，電腦系統用以判斷熱影像屬於低溫類別。In some embodiments, the thermal image includes a plurality of pixels, each pixel corresponds to a temperature value, and the computer system is further used to calculate the average value of the temperature values. If the average value is greater than a critical value, the computer system determines that the thermal image belongs to the high temperature category. If the average value is less than or equal to the critical value, the computer system is used to determine that the thermal image belongs to the low temperature category.

在一些實施例中，出鋼孔與攪拌磚位於接鋼桶的底部。In some embodiments, the tapping holes and the stirring bricks are located at the bottom of the connecting drum.

以另一個角度來說，本發明的實施例提出一種接鋼桶辨識方法，由電腦系統執行。接鋼桶包括攪拌磚與出鋼孔，接鋼桶辨識方法包括：透過熱感測器拍攝接鋼桶以產生熱影像；判斷熱影像屬於第一溫度類別或第二溫度類別；若熱影像屬於第一溫度類別，將熱影像輸入至第一機器學習模型以偵測攪拌磚與出鋼孔，並判斷攪拌磚與出鋼孔是否異常；以及若熱影像屬於第二溫度類別，將熱影像輸入至第二機器學習模型以偵測攪拌磚與出鋼孔，並判斷攪拌磚與出鋼孔是否異常。From another perspective, an embodiment of the present invention provides a method for identifying a steel drum, which is executed by a computer system. The connecting steel drum includes stirring bricks and tapping holes. The steel connecting drum identification method includes: photographing the connecting steel drum through a thermal sensor to generate a thermal image; judging whether the thermal image belongs to the first temperature category or the second temperature category; if the thermal image belongs to In the first temperature category, the thermal image is input to the first machine learning model to detect the stirring brick and the tapping hole, and determine whether the stirring brick and the tapping hole are abnormal; and if the thermal image belongs to the second temperature category, the thermal image is input to the second machine learning model to detect the stirring bricks and the tapping holes, and determine whether the stirring bricks and the tapping holes are abnormal.

關於本文中所使用之「第一」、「第二」等，並非特別指次序或順位的意思，其僅為了區別以相同技術用語描述的元件或操作。The terms "first", "second", etc. used in this document do not mean a particular order or order, but are only used to distinguish elements or operations described in the same technical terms.

圖1是根據一實施例繪示接鋼桶辨識系統的示意圖。請參照圖1，接鋼桶辨識系統100包括接鋼桶110、熱感測器120與電腦系統130。接鋼桶110包括攪拌磚111與出鋼孔112，攪拌磚111與出鋼孔112設置在接鋼桶110的底部。熱感測器120例如為紅外線攝影機，用以朝接鋼桶110的內部拍攝熱影像，熱影像包括多個像素，每個像素的顏色反應對應位置的溫度，熱感測器120也可以取得每個像素對應的溫度值。熱感測器120是通訊連接至電腦系統130，在此通訊連接可以用任意有線、無線或用互聯網的方式來達成，電腦系統130可以是工業電腦、雲端伺服器、個人電腦、或任意具有計算能力的電子裝置。熱感測器120所取得的熱影像以及溫度值會傳送至電腦系統130，電腦系統130會根據此熱影像來辨識攪拌磚111與出鋼孔112是否異常。FIG. 1 is a schematic diagram illustrating a steel drum identification system according to an embodiment. Referring to FIG. 1 , the identification system 100 for connecting a steel drum includes a connecting steel drum 110 , a thermal sensor 120 and a computer system 130 . The steel receiving barrel 110 includes a stirring brick 111 and a steel tapping hole 112 , and the stirring brick 111 and the steel tapping hole 112 are arranged at the bottom of the steel receiving barrel 110 . The thermal sensor 120 is, for example, an infrared camera, which is used to shoot a thermal image toward the interior of the steel drum 110 . The thermal image includes a plurality of pixels, and the color of each pixel reflects the temperature of the corresponding position. The temperature value corresponding to each pixel. The thermal sensor 120 is communicatively connected to the computer system 130, where the communication connection can be achieved by any wired, wireless or Internet method. The computer system 130 can be an industrial computer, a cloud server, a personal computer, or any computer with capable electronic device. The thermal image and the temperature value obtained by the thermal sensor 120 will be transmitted to the computer system 130, and the computer system 130 will identify whether the stirring brick 111 and the tapping hole 112 are abnormal according to the thermal image.

具體來說，首先判斷熱影像屬於哪一個溫度類別，在此實施例中設定了兩個溫度類別，分別是高溫類別與低溫類別。由於熱影像中每個像素的顏色都反應一個溫度值，因此可以根據像素的顏色來判斷熱影像屬於高溫類別或是低溫類別。在一些實施例中可以將熱影像中的每個像素分類為多個顏色的其中之一，這些顏色例如為紅色、橘色、黃色、綠色、青色(cyan)與藍色。這些顏色中有些是高溫顏色，有些是低溫顏色，例如紅色、橘色與黃色屬於高溫顏色，而綠色、青色與藍色屬於低溫顏色。在此實施例中熱影像中的每個像素都有三個灰階值，分別是紅色、藍色與綠色，而上述每個顏色都有預設的灰階值範圍，根據這些灰階值範圍可以將像素分類。在把像素分類完以後，如果有最多像素的顏色屬於高溫顏色，則判斷熱影像屬於高溫類別；如果有最多像素的顏色屬於低溫顏色，判斷熱影像屬於低溫類別。Specifically, it is first determined which temperature category the thermal image belongs to. In this embodiment, two temperature categories are set, namely, a high temperature category and a low temperature category. Since the color of each pixel in the thermal image reflects a temperature value, it can be judged whether the thermal image belongs to the high temperature category or the low temperature category according to the color of the pixel. In some embodiments, each pixel in the thermal image may be classified as one of a number of colors, such as red, orange, yellow, green, cyan, and blue. Some of these colors are high temperature colors and some are low temperature colors, such as red, orange and yellow are high temperature colors, while green, cyan and blue are low temperature colors. In this embodiment, each pixel in the thermal image has three grayscale values, namely red, blue and green, and each of the above-mentioned colors has a preset grayscale value range. Classify pixels. After classifying the pixels, if the color with the most pixels belongs to the high temperature color, the thermal image is judged to belong to the high temperature category; if the color with the most pixels belongs to the low temperature color, the thermal image is judged to belong to the low temperature category.

在一些實施例中，也可以根據熱影像對應的溫度值來判斷熱影像屬於哪一個溫度類別。舉例來說，可以將熱影像中所有像素所對應的溫度值平均起來成為一個平均值，如果此平均值大於一臨界值則判斷熱影像屬於高溫類別，反之如果平均值小於等於臨界值，則判斷熱影像屬於低溫類別。本領域具有通常知識者當可將上述內容稍加潤飾，本揭露並不限制如何判斷熱影像屬於哪一個溫度類別。此外，在上述的實施例中共設定兩個溫度類別，但在其他實施例中也可以設定更多溫度類別，本揭露並不在此限。In some embodiments, which temperature category the thermal image belongs to can also be determined according to the temperature value corresponding to the thermal image. For example, the temperature values corresponding to all the pixels in the thermal image can be averaged to form an average value. If the average value is greater than a threshold value, it is determined that the thermal image belongs to the high temperature category. On the contrary, if the average value is less than or equal to the threshold value, it is determined Thermal images fall into the low temperature category. Those with ordinary knowledge in the art can modify the above content slightly, but the present disclosure does not limit how to determine which temperature category the thermal image belongs to. In addition, two temperature categories are set in the above-mentioned embodiments, but more temperature categories can be set in other embodiments, which is not limited in the present disclosure.

如果熱影像屬於高溫類別，則將熱影像輸入至第一機器學習模型以偵測攪拌磚111與出鋼孔112的位置，並判斷攪拌磚111與出鋼孔112是否異常。如果熱影像屬於低溫類別，將熱影像輸入至第二機器學習模型以偵測攪拌磚111與出鋼孔112的位置，並判斷攪拌磚111與出鋼孔112是否異常。上述的第一機器學習模型與第二機器學習模型例如為卷積神經網路、支持向量機等等，本揭露並不在此限。值得注意的是，第一機器學習模型與第二機器學習模型是使用不同的訓練集(training set)來訓練，第一機器學習模型是根據高溫類別的熱影像來訓練，而第二機器學習模型是根據低溫類別的熱影像來訓練，而攪拌磚111與出鋼孔112的位置以及是否異常則是仰賴人工標記。圖2是根據一實施例繪示辨識結果的示意圖。請參照圖2，高溫類別影像210是輸入至第一機器學習模型以辨識出攪拌磚211、出鋼孔212、衝擊區213以及其他區域，而低溫類別影像220是輸入至第二機器學習模型以辨識出攪拌磚221、出鋼孔222、衝擊區223以及其他區域。在其他實施例中，上述的機器學習模型也可以辨識接鋼桶中的其他元件。If the thermal image belongs to the high temperature category, the thermal image is input to the first machine learning model to detect the positions of the stirring block 111 and the tapping hole 112 and determine whether the stirring block 111 and the tapping hole 112 are abnormal. If the thermal image belongs to the low temperature category, the thermal image is input to the second machine learning model to detect the positions of the stirring brick 111 and the tapping hole 112 , and determine whether the stirring brick 111 and the tapping hole 112 are abnormal. The above-mentioned first machine learning model and second machine learning model are, for example, convolutional neural networks, support vector machines, etc., and the disclosure is not limited thereto. It is worth noting that the first machine learning model and the second machine learning model are trained using different training sets. The first machine learning model is trained based on thermal images of high temperature categories, while the second machine learning model is trained using different training sets. The training is based on the thermal images of the low temperature category, and the positions of the stirring bricks 111 and the tapping holes 112 and whether they are abnormal or not are manually marked. FIG. 2 is a schematic diagram illustrating a recognition result according to an embodiment. Referring to FIG. 2 , the high temperature category image 210 is input to the first machine learning model to identify the stirring brick 211 , the tapping hole 212 , the impact zone 213 and other areas, while the low temperature category image 220 is input to the second machine learning model for The stirring block 221, the tapping hole 222, the impact zone 223 and other areas are identified. In other embodiments, the above-mentioned machine learning model can also identify other components in the steel drum.

從圖2可以看出高溫類別影像210與低溫類別影像220的顏色基本上不相同。在習知技術中是以可見光來拍攝接鋼桶110，然而由於接鋼桶110在使用過後桶內會有許多殘鋼，這些殘鋼的分佈、紋理、形狀等特徵每次使用後都不一樣，此外填充砂丟擲的位置也可能不一樣，攪拌磚清洗的清潔度也可能不同，這些都會影響接鋼桶的影像，因此用可見光來辨識攪拌磚與出鋼孔的效果並不夠好。接鋼桶110如果使用完放置許久則溫度會下降，如果連續使用則在拍攝熱影像時溫度較高，在此實施例中對於不同的溫度採用不同的機器學習模型，可以提升準確度。實驗數據如以下表一所示。   單溫模型架構 高低溫雙模型架構 高溫 低溫 訓練準確度 83.84% 83.65% 88.04% 測試準確度 83.79% 83.33% 86.64% 表一 It can be seen from FIG. 2 that the colors of the high temperature category image 210 and the low temperature category image 220 are basically different. In the prior art, visible light is used to photograph the steel receiving drum 110. However, since the steel receiving drum 110 will have a lot of residual steel in the barrel after use, the distribution, texture, shape and other characteristics of these residual steel are different after each use. , In addition, the throwing position of the filling sand may also be different, and the cleaning degree of the mixing bricks may also be different, which will affect the image of the steel drum, so the effect of using visible light to identify the mixing bricks and the tapping holes is not good enough. If the steel bucket 110 is used for a long time, the temperature will drop. If it is used continuously, the temperature will be higher when the thermal image is taken. In this embodiment, different machine learning models are used for different temperatures, which can improve the accuracy. The experimental data are shown in Table 1 below. Single Temperature Model Architecture High and low temperature dual model architecture high temperature low temperature training accuracy 83.84% 83.65% 88.04% test accuracy 83.79% 83.33% 86.64% Table I

從表一可以看出，當採用高低雙模型架構，在高溫時保留了單溫模型架構的準確度(accuracy)，但在低溫時卻大幅度提升了準確度。對照實際生產狀態，低溫狀態的接鋼桶比例為67%，高溫狀態的接鋼桶比例為33%。這樣一來本揭露提出的高低溫雙模型架構對於出現次數較多的低溫接鋼桶可以得到更好的準確度。As can be seen from Table 1, when the high-low dual-model architecture is used, the accuracy of the single-temperature model architecture is retained at high temperature, but the accuracy is greatly improved at low temperature. Compared with the actual production state, the proportion of steel drums in low temperature state is 67%, and the proportion of steel drums in high temperature state is 33%. In this way, the high-low temperature dual-model architecture proposed in the present disclosure can achieve better accuracy for low-temperature steel drums that appear frequently.

圖3是根據一實施例繪示接鋼桶辨識方法的流程圖。請參照圖3，在步驟301，透過熱感測器拍攝接鋼桶以產生熱影像。在步驟302，判斷熱影像的溫度類別。如果熱影像屬於第一溫度類別，在步驟303，將熱影像輸入至第一機器學習模型以偵測攪拌磚與出鋼孔並判斷攪拌磚與出鋼孔是否異常。如果熱影像屬於第二溫度類別，在步驟304，熱影像輸入至第二機器學習模型以偵測攪拌磚與出鋼孔並判斷攪拌磚與出鋼孔是否異常。然而，圖3中各步驟已詳細說明如上，在此便不再贅述。值得注意的是，圖3中各步驟可以實作為多個程式碼或是電路，本發明並不在此限。此外，圖3的方法可以搭配以上實施例使用，也可以單獨使用。換言之，圖3的各步驟之間也可以加入其他的步驟。FIG. 3 is a flow chart illustrating a method for identifying a steel drum according to an embodiment. Referring to FIG. 3 , in step 301 , a thermal image is generated by photographing the steel drum through a thermal sensor. In step 302, the temperature category of the thermal image is determined. If the thermal image belongs to the first temperature category, in step 303, the thermal image is input to the first machine learning model to detect the stirring brick and the tapping hole and determine whether the stirring brick and the tapping hole are abnormal. If the thermal image belongs to the second temperature category, in step 304, the thermal image is input to the second machine learning model to detect the stirring block and the tapping hole and determine whether the stirring block and the tapping hole are abnormal. However, each step in FIG. 3 has been described above in detail, and will not be repeated here. It should be noted that each step in FIG. 3 can be implemented as a plurality of codes or circuits, and the present invention is not limited thereto. In addition, the method of FIG. 3 may be used in conjunction with the above embodiments, or may be used alone. In other words, other steps may be added between the steps in FIG. 3 .

雖然本發明已以實施例揭露如上，然其並非用以限定本發明，任何所屬技術領域中具有通常知識者，在不脫離本發明的精神和範圍內，當可作些許的更動與潤飾，故本發明的保護範圍當視後附的申請專利範圍所界定者為準。Although the present invention has been disclosed above by the embodiments, it is not intended to limit the present invention. Anyone with ordinary knowledge in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. Therefore, The protection scope of the present invention shall be determined by the scope of the appended patent application.

100:接鋼桶辨識系統 110:接鋼桶 111:攪拌磚 112:出鋼孔 120:熱感測器 130:電腦系統 210:高溫類別影像 211,221:攪拌磚 212,222:出鋼孔 213,223:衝擊區 220:低溫類別影像 301~304:步驟 100: Connect the steel drum identification system 110: Connect the steel drum 111: Mixing Bricks 112: tapping hole 120: Thermal sensor 130: Computer Systems 210: High temperature category image 211, 221: Mixing Bricks 212,222: Tap hole 213, 223: Impact Zone 220: Low temperature category image 301~304: Steps

為讓本發明的上述特徵和優點能更明顯易懂，下文特舉實施例，並配合所附圖式作詳細說明如下。 圖1是根據一實施例繪示接鋼桶辨識系統的示意圖。 圖2是根據一實施例繪示辨識結果的示意圖。 圖3是根據一實施例繪示接鋼桶辨識方法的流程圖。 In order to make the above-mentioned features and advantages of the present invention more obvious and easy to understand, the following embodiments are given and described in detail with the accompanying drawings as follows. FIG. 1 is a schematic diagram illustrating a steel drum identification system according to an embodiment. FIG. 2 is a schematic diagram illustrating a recognition result according to an embodiment. FIG. 3 is a flow chart illustrating a method for identifying a steel drum according to an embodiment.

301~304:步驟 301~304: Steps

Claims (8)

一種接鋼桶辨識系統，包括：一接鋼桶，該接鋼桶包括一攪拌磚與一出鋼孔；一熱感測器，用以拍攝該接鋼桶以產生一第一熱影像；以及一電腦系統，通訊連接至該熱感測器以取得該第一熱影像，判斷該第一熱影像屬於一第一溫度類別或一第二溫度類別，其中該第一溫度類別為高溫類別，該第二溫度類別為低溫類別，其中若該第一熱影像屬於該第一溫度類別，將該第一熱影像輸入至一第一機器學習模型以偵測該攪拌磚與該出鋼孔並判斷該攪拌磚與該出鋼孔是否異常，其中該第一機器學習模型是根據一第一訓練集中該高溫類別的熱影像來訓練，在訓練時該攪拌磚與該出鋼孔的位置以及是否異常是由人工標記，其中若該第一熱影像屬於該第二溫度類別，將該第一熱影像輸入至一第二機器學習模型以偵測該攪拌磚與該出鋼孔並判斷該攪拌磚與該出鋼孔是否異常，其中該第二機器學習模型是根據一第二訓練集中該低溫類別的熱影像來訓練，在訓練時該攪拌磚與該出鋼孔的位置以及是否異常是由人工標記。 A connected steel drum identification system, comprising: a connected steel drum, the connected steel drum includes a stirring brick and a steel tapping hole; a thermal sensor for photographing the connected steel drum to generate a first thermal image; and a computer system, connected to the thermal sensor in communication to obtain the first thermal image, and determine that the first thermal image belongs to a first temperature category or a second temperature category, wherein the first temperature category is a high temperature category, and the The second temperature category is a low temperature category, wherein if the first thermal image belongs to the first temperature category, the first thermal image is input into a first machine learning model to detect the stirring brick and the tapping hole and determine the Whether the stirring brick and the tapping hole are abnormal, wherein the first machine learning model is trained according to the thermal image of the high temperature category in a first training set, and the position of the stirring brick and the tapping hole during training and whether the abnormality is By manual marking, wherein if the first thermal image belongs to the second temperature category, the first thermal image is input into a second machine learning model to detect the stirring brick and the tapping hole and determine the stirring brick and the Whether the tapping hole is abnormal, wherein the second machine learning model is trained according to the thermal images of the low temperature category in a second training set, and the positions of the stirring brick and the tapping hole and whether they are abnormal are manually marked during training. 如請求項1所述之接鋼桶辨識系統，其中該電腦系統還用以將該第一熱影像中的每個像素分類為多個 顏色的其中之一，該些顏色包括多個高溫顏色與多個低溫顏色，如果該些顏色中具有最多該些像素的顏色屬於該些高溫顏色的其中之一，該電腦系統用以判斷該第一熱影像屬於該高溫類別，如果該些顏色中具有最多該些像素的顏色屬於該些低溫顏色的其中之一，該電腦系統用以判斷該第一熱影像屬於該低溫類別。 The connected steel drum identification system as claimed in claim 1, wherein the computer system is further configured to classify each pixel in the first thermal image into a plurality of One of the colors, the colors include a plurality of high temperature colors and a plurality of low temperature colors, if the color with the most pixels in the colors belongs to one of the high temperature colors, the computer system is used to determine the first color A thermal image belongs to the high temperature category, and if the color with the most pixels in the colors belongs to one of the low temperature colors, the computer system determines that the first thermal image belongs to the low temperature category. 如請求項1所述之接鋼桶辨識系統，其中該第一熱影像包括多個像素，每一該些像素對應至一溫度值，該電腦系統還用以計算該些溫度值的平均值，如果該平均值大於一臨界值，該電腦系統用以判斷該第一熱影像屬於該高溫類別，如果該平均值小於等於該臨界值，該電腦系統用以判斷該第一熱影像屬於該低溫類別。 The connected steel drum identification system according to claim 1, wherein the first thermal image includes a plurality of pixels, each of the pixels corresponds to a temperature value, and the computer system is further used to calculate the average value of the temperature values, If the average value is greater than a threshold value, the computer system is used to determine that the first thermal image belongs to the high temperature category; if the average value is less than or equal to the threshold value, the computer system determines that the first thermal image belongs to the low temperature category . 如請求項1所述之接鋼桶辨識系統，其中該出鋼孔與該攪拌磚位於該接鋼桶的底部。 The steel drum identification system as claimed in claim 1, wherein the tapping hole and the stirring brick are located at the bottom of the steel drum. 一種接鋼桶辨識方法，由一電腦系統執行，該接鋼桶包括一攪拌磚與一出鋼孔，該接鋼桶辨識方法包括：透過一熱感測器拍攝該接鋼桶以產生一第一熱影像； 判斷該第一熱影像屬於一第一溫度類別或一第二溫度類別，其中該第一溫度類別為高溫類別，該第二溫度類別為低溫類別；若該第一熱影像屬於該第一溫度類別，將該第一熱影像輸入至一第一機器學習模型以偵測該攪拌磚與該出鋼孔並判斷該攪拌磚與該出鋼孔是否異常，其中該第一機器學習模型是根據一第一訓練集中該高溫類別的熱影像來訓練，在訓練時該攪拌磚與該出鋼孔的位置以及是否異常是由人工標記；以及若該第一熱影像屬於該第二溫度類別，將該第一熱影像輸入至一第二機器學習模型以偵測該攪拌磚與該出鋼孔並判斷該攪拌磚與該出鋼孔是否異常，其中該第二機器學習模型是根據一第二訓練集中該低溫類別的熱影像來訓練，在訓練時該攪拌磚與該出鋼孔的位置以及是否異常是由人工標記。 A method for identifying a steel drum is performed by a computer system. The steel drum includes a stirring brick and a tapping hole. The identification method for the steel drum includes: photographing the steel drum through a thermal sensor to generate a first a thermal image; Determine that the first thermal image belongs to a first temperature class or a second temperature class, wherein the first temperature class is a high temperature class, and the second temperature class is a low temperature class; if the first thermal image belongs to the first temperature class , the first thermal image is input into a first machine learning model to detect the stirring brick and the tapping hole and determine whether the stirring brick and the tapping hole are abnormal, wherein the first machine learning model is based on a first A training set of thermal images of the high temperature category is used for training. During training, the positions of the stirring brick and the tapping hole and whether they are abnormal are manually marked; and if the first thermal image belongs to the second temperature category, the first thermal image belongs to the second temperature category. A thermal image is input to a second machine learning model to detect the stirring brick and the tapping hole and determine whether the stirring brick and the tapping hole are abnormal, wherein the second machine learning model is based on a second training set of the The thermal images of the low temperature category are used for training. During training, the positions of the stirring brick and the tapping hole and whether they are abnormal are manually marked. 如請求項5所述之接鋼桶辨識方法，其中判斷該第一熱影像屬於該第一溫度類別或該第二溫度類別的步驟包括：將該第一熱影像中的每個像素分類為多個顏色的其中之一，該些顏色包括多個高溫顏色與多個低溫顏色；如果該些顏色中具有最多該些像素的顏色屬於該些高溫顏色的其中之一，判斷該第一熱影像屬於該高溫類別；以及 如果該些顏色中具有最多該些像素的顏色屬於該些低溫顏色的其中之一，判斷該第一熱影像屬於該低溫類別。 The method for identifying a connected steel drum according to claim 5, wherein the step of judging that the first thermal image belongs to the first temperature category or the second temperature category includes: classifying each pixel in the first thermal image as multiple One of the colors, the colors include a plurality of high temperature colors and a plurality of low temperature colors; if the color with the most pixels in the colors belongs to one of the high temperature colors, it is determined that the first thermal image belongs to the high temperature category; and If the color with the most pixels among the colors belongs to one of the low temperature colors, it is determined that the first thermal image belongs to the low temperature category. 如請求項5所述之接鋼桶辨識方法，其中該第一熱影像包括多個像素，每一該些像素對應至一溫度值，判斷該第一熱影像屬於該第一溫度類別或該第二溫度類別的步驟包括：計算該些溫度值的平均值；如果該平均值大於一臨界值，判斷該第一熱影像屬於該高溫類別；以及如果該平均值小於等於該臨界值，判斷該第一熱影像屬於該低溫類別。 The method for identifying a connected steel drum according to claim 5, wherein the first thermal image includes a plurality of pixels, each of the pixels corresponds to a temperature value, and it is determined that the first thermal image belongs to the first temperature category or the first thermal image. The step of two temperature categories includes: calculating an average value of the temperature values; if the average value is greater than a threshold value, judging that the first thermal image belongs to the high temperature category; and if the average value is less than or equal to the threshold value, judging the first thermal image A thermal image falls into this low temperature category. 如請求項5所述之接鋼桶辨識方法，其中該出鋼孔與該攪拌磚位於該接鋼桶的底部。 The method for identifying a connecting steel drum according to claim 5, wherein the tapping hole and the stirring brick are located at the bottom of the connecting steel drum.
TW110128312A 2021-08-02 2021-08-02 Method and system for identifying receiving steel bucket TWI776618B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW110128312A TWI776618B (en) 2021-08-02 2021-08-02 Method and system for identifying receiving steel bucket

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW110128312A TWI776618B (en) 2021-08-02 2021-08-02 Method and system for identifying receiving steel bucket

Publications (2)

Publication Number Publication Date
TWI776618B true TWI776618B (en) 2022-09-01
TW202307787A TW202307787A (en) 2023-02-16

Family

ID=84957977

Family Applications (1)

Application Number Title Priority Date Filing Date
TW110128312A TWI776618B (en) 2021-08-02 2021-08-02 Method and system for identifying receiving steel bucket

Country Status (1)

Country Link
TW (1) TWI776618B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI575193B (en) * 2014-08-19 2017-03-21 英特爾股份有限公司 System, method, machine-readable storage medium, and device for determining scaling in a boiler
US20200225655A1 (en) * 2016-05-09 2020-07-16 Strong Force Iot Portfolio 2016, Llc Methods, systems, kits and apparatuses for monitoring and managing industrial settings in an industrial internet of things data collection environment
EP3741481A1 (en) * 2019-05-23 2020-11-25 The Boeing Company Additive manufacturing with adjusted cooling responsive to thermal characteristic of workpiece

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI575193B (en) * 2014-08-19 2017-03-21 英特爾股份有限公司 System, method, machine-readable storage medium, and device for determining scaling in a boiler
US20200225655A1 (en) * 2016-05-09 2020-07-16 Strong Force Iot Portfolio 2016, Llc Methods, systems, kits and apparatuses for monitoring and managing industrial settings in an industrial internet of things data collection environment
EP3741481A1 (en) * 2019-05-23 2020-11-25 The Boeing Company Additive manufacturing with adjusted cooling responsive to thermal characteristic of workpiece

Also Published As

Publication number Publication date
TW202307787A (en) 2023-02-16

Similar Documents

Publication Publication Date Title
CN105956618A (en) Converter steelmaking blowing state recognition system and method based on image dynamic and static characteristics
CN110184401B (en) On-line detection method for slag iron state of blast furnace
JP6996413B2 (en) Analytical equipment and analysis program
CN108088654A (en) Projector quality determining method and its electronic equipment
CN110660060B (en) OLED screen color mixing defect detection method, terminal equipment and computer readable medium
TWI776618B (en) Method and system for identifying receiving steel bucket
CN105572080B (en) Contactless float glass tin surface automatic identification equipment and method
CN104637425A (en) Assembly-line LED (light emitting diode) display unit uniformity detection method, system and device
CN114882822A (en) Gamma debugging method, device, equipment and computer readable storage medium
CN115035096A (en) Indicator lamp testing method and related device
CN116086390A (en) Flatness detection device and system for LED display screen
JP2010008159A (en) Visual inspection processing method
CN112458231B (en) Converter slag discharge detection method and system
CN116402813B (en) Neural network-based copper converter converting copper-making period end point judging method
JP7401752B2 (en) Model construction device, prediction device, model construction method, prediction method, and computer program
CN117804595A (en) Intelligent monitoring method, device and medium for LED device based on monitoring technology
JP2009267099A (en) Board inspection method, and inspection result confirmation system of automatic visual inspection
JP4468046B2 (en) Apparatus, method and program for determining rusting state of structure made of steel
CN103728305A (en) Detection method for PCBA (printed circuit board assembly)
CN104505053B (en) Show signal conversion method and device
JP2005252451A (en) Image quality inspection method and image quality inspection apparatus
US10048131B2 (en) Chromaticity test method and chromaticity test apparatus
JP6795046B2 (en) Histogram creation method for slag detection in molten steel flow
CN100386050C (en) Color misregistration reducer
JP2017179598A (en) Furnace bottom tuyere safety evaluation method for bottom blowing converter, furnace bottom tuyere service life extension method and operation method of bottom blowing converter

Legal Events

Date Code Title Description
GD4A Issue of patent certificate for granted invention patent