A converter furnace image anomaly detection system based on incremental learning has been proposed to simplify the research and development process for industrial software and enhance the accuracy and efficiency of image anomaly detection. This system employs machine vision technology to capture images of the converter furnace. It introduces a deep residual network to form an image anomaly detection model, using the captured images to train this model. The system is implemented through low-code development methods and optimizes the iterative updates of the model by incorporating incremental learning algorithms. The recognition accuracy of the converter furnace image anomaly detection model is compared across various neural network architectures. Additionally, the disparities in model accuracy and time consumption between incremental learning and full learning are assessed on a low-code platform. Experimental results demonstrate that this system exhibits good precision and stability in image anomaly detection. On the low⁃code development platform, the system software based on incremental learning performs excellently in handling large-scale data and real-time scenarios, providing an efficient and low⁃cost solution for converter furnace image anomaly detection.
WANGS L, WANGY K, YANGB, et al. Fog manufacturing: New paradigm of industrial Internet manufacturing based on hierarchical digital twin[J]. Computer Integrated Manufacturing Systems, 2019, 25(12): 3070-3080. DOI: 10.13196/j.cims.2019.12.009(Ch ).
XIAOX W, WANGF, WANGX H, et al. Bionic structure and cyber-physical system for intelligent power plant oriented to the industrial Internet[J]. Transactions of China Electrotechnical Society, 2020, 35(23): 4898-4911. DOI: 10.19595/j.cnki.1000-6753.tces.191593(Ch ).
YANGS. Industry 4.0 and the Industrial Internet: Comparison, Enlightenment and Countermeasures[J]. Contemporary Finance and Economics,2015(8):99-107.DOI:10.13676/j.cnki.cn36-1030/f.2015.08.010(Ch ).
[9]
ROJKOA. Industry 4.0 concept: Background and overview[J]. International Journal of Interactive Mobile Technologies (IJIM), 2017, 11(5): 77. DOI: 10.3991/ijim.v11i5.7072 .
TANGT, TENGL, WUJ, et al. Realizing digitalization in an all-round way is the only way to intelligent manufacturing—Interpretation of The Road to Intelligent Manufacturing: Digital Factory[J]. China Mechanical Engineering, 2018, 29(3): 366-377. DOI: 10.3969/j.issn.1004-132X.2018.03.018(Ch ).
WANGG H. Analysis on the process and end point control of medium and high carbon steel smelting in converter[J]. Metallurgy and Materials, 2023, 43(2): 105-107. DOI: 10.3969/j.issn.1674-5183.2023.02.037(Ch ).
[16]
MASANAM, LIUX, TWARDOWSKIB, et al. Class-incremental learning: survey and performance evaluation on image classification[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(5): 5513-5533. DOI: 10.1109/TPAMI.2022.3213473 .
[17]
DEL M, ALJUNDIR, MASANAM, et al. A continual learning survey: Defying forgetting in classification tasks[J]. IEEE Transactions On Pattern Analysis And Machine Intelligence, 2022, 44(7): 3366-3385. DOI: 10.1109/TPAMI.2021.3057446 .
[18]
SARFRAZF, ARANIE, ZONOOZB. Sparse coding in a dual memory system for lifelong learning[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(8): 9714-9722. DOI: 10.1609/aaai.v37i8.26161 .
[19]
KIRKPATRICKJ, PASCANUR, RABINOWITZN, et al. Overcoming catastrophic forgetting in neural networks[J]. Proceedings of the National Academy of Sciences of the United States of America, 2017, 114(13): 3521-3526. DOI: 10.1073/pnas.1611835114 .
GAOB, HEJ, GUOY F. Research on development and application of low code cost management and control system for oil downhole operation[J]. China CIO News, 2023(4): 28-30. DOI: 10.3969/j.issn.1001-2362.2023.04.011(Ch ).
LIUQ M. Design of hospital COVID-19 prevention and control system based on low-code platform[J]. China Digital Medicine, 2022, 17(12): 49-53. DOI: 10.3969/j.issn.1673-7571.2022.12.010(Ch ).
[29]
杨丹. 低代码平台在金融行业的创新研究及实践[J]. 金融电子化, 2022(3): 18-20.
[30]
YANGD. Innovative research and practice of low code platform in financial industry[J]. Financial Computerizing, 2022(3): 18-20 (Ch).
ZHANGM. The Design and Implementation of an SAP Low⁃Code Development Platform[D]. Jinan: Shandong University, 2020.DOI: 10.27272/d.cnki.gshdu.2020.001067(Ch ).
WANGH Y, QIANQ, WUX. Incremental learning of material absorption coefficient regression based on parameter penalty and experience replay[J]. Chinese Journal of Engineering, 2023, 45(7): 1225-1231. DOI: 10.13374/j.issn2095-9389.2022.05.03.006(Ch ).
[37]
SIMONVANK, ZISSERMANA. Very deep convolutional networks for large-scale image recognition[EB/OL]. [2023-05-29].