Naked pouring detection of ladle based on multi-sensor fusion
YAO Yu1, LIU Man-xian2, ZHANG Zhao-jie3, DING Ding1, WANG Lin-hao4, LIU Yang5
1. Research and Development Center, Nanjing Tianxiang Intelligent Equipment Technology Co., Ltd., Nanjing 211300, Jiangsu, China; 2. Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China; 3. Storage System Automation Division, Tiandi Science and Technology Co., Ltd., Beijing 100013, China; 4. Jizhong Engineering Technician College, Jizhong Energy Xingtai Mining Group Company, Xingtai 054000, Hebei, China; 5. Design and Research Institute Co., Ltd., University of Science and Technology Beijing, Beijing 100083, China
Abstract：Protective casing will be very likely to break and fall off when ladle is poured. That will cause large-scale splashing of molten steel, which will seriously affect the production quality of steel. In order to effectively monitor the ladle pouring state, a method of detecting the ladle pouring state based on infrared thermometry is studied. It is verified through experiments that there is a large temperature difference between the normal pouring state and the naked pouring state of the ladle, and the effective classification of the ladle pouring state can be achieved by setting the temperature threshold. The ladle pouring state classification method based on deep learning is also studied. The ladle pouring image is preprocessed by median filter and average gray grade-gradient 2D Otsu segmentation method, and the model is trained through an improved target detection algorithm which is based on YOLO v3. The experimental result shows that the accuracy of naked pouring of the ladle is 99.2%, and it has better generalization performance. Based on the above research, a ladle naked pouring detection equipment has been developed, which realizes accurate alarm of the pouring state and has a good field application effect.
姚煜, 刘漫贤, 张兆杰, 丁丁, 王林浩, 刘洋. 基于多传感器融合的钢包裸浇检测[J]. 中国冶金, 2021, 31(5): 104-110.
YAO Yu, LIU Man-xian, ZHANG Zhao-jie, DING Ding, WANG Lin-hao, LIU Yang. Naked pouring detection of ladle based on multi-sensor fusion[J]. China Metallurgy, 2021, 31(5): 104-110.
Redmon J, Farhadi A. YOLO v3: An incremental improvement[J]. ArXiv Preprint ArXiv: 2018,1804:02767.
[14]
HE K, ZHANG X, REN S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 37(9):1904.
[15]
ZHANG X, ZHU X. An efficient and scene-adaptive algorithm for vehicle detection in aerial images using an improved YOLO v3 framework[J]. International Journal of Geo-Information, 2019, 8(11):483.
2021, Vol. 31 
