Diagnostics of failures of technological equipment of chemical industries using artificial intelligence
- Authors: Zubov D.V.1, Lebedev D.A.1
-
Affiliations:
- Issue: No 2 (2024)
- Pages: 30-40
- Section: Articles
- URL: https://ogarev-online.ru/2454-0714/article/view/359414
- DOI: https://doi.org/10.7256/2454-0714.2024.2.70729
- EDN: https://elibrary.ru/XBIJYK
- ID: 359414
Cite item
Full Text
Abstract
The paper considers the problem of automated recognition of single emergencies in chemical and oil refining industries. Modern chemical and technological production facilities are maintained and managed by a small number of personnel, which increases the burden on each operator. To reduce the number of operator errors, their training is regularly conducted on simulators equipped with a set of both standard situations (routine start-up, shutdown, normal process management, switching from one mode to another) and emergency scenarios (column depressurization, pump failure, failure of the power supply system). Nevertheless, it is impossible to foresee all possible failures during operator training, and even a trained operator may not notice the first signs of an accident, and therefore it is necessary to create a decision support system that helps the operator to recognize failures of technological equipment in a timely manner. To recognize failures, it is proposed to use a neural network trained on an array of simulated accident data. An industrial simulator based on the RTsim platform was used to simulate typical accidents. The novelty of the research lies in the use of artificial intelligence methods to diagnose the property of the technological process according to the SCADA system and the use of data for training a neural network not from a real object (which will always be insufficient), but from a model that exactly corresponds to a specific technological site. The number of simulated scenarios used to train a neural network can be quite large, which reduces the proportion of erroneous system responses. The developed system confidently copes with the recognition of individual equipment failures. The results obtained can be used to help process operators and to improve emergency protection systems. The analysis of the time required by the system to recognize an emergency situation can be used to design new production facilities, modify the control and management system.
About the authors
Dmitrii Vladimirovich Zubov
Email: dvzubov@gmail.com
ORCID iD: 0000-0002-0703-1577
Danila Aleksandrovich Lebedev
Email: lebedev.d.a@muctr.ru
ORCID iD: 0009-0007-2873-2341
References
Цифровые тренажеры технологических процессов РТСИМ. Карьера для обеспечения безопасности в нефтехимическом комплексе / Р. Р. Габдрахманов, Э. В. Гарифуллина, В. В. Бронская [и др.] // Наука, технологии, общество: Экологический инжиниринг в интересах устойчивого развития территорий : Сборник научных статей, Красноярск, 09–10 ноября 2023 года. – Красноярск: Красноярский краевой Дом науки и техники Российского Союза научных и инженерных общественных объединений, 2023. – С. 131-135. – EDN BLETRM. Титова, Л. М. Цифровые двойники при подготовке инженерных кадров химических производств в высшей школе / Л. М. Титова, Н. П. Мемедейкина // Международный форум KAZAN DIGITAL WEEK-2023 : Сборник материалов, Казань, 20–22 сентября 2023 года / Сост. Р.Ш. Ахмадиева, Р.Н. Минниханов. Под общей редакцией Р.Н. Минниханова. – Казань: Научный центр безопасности жизнедеятельности, 2023. – С. 954-960. – EDN RNNMLO. Имитация процессов управления агрегатом производства метанола на тренажере / В. Л. Перов [и др.]. – Текст : непосредственный // Тр. Моск. хим. – технол. ин-та им. Д.И.Менделеева. – 1988. – Вып. 152. – С. 12-18. Siguo Bi, Cong Wang, Bochun Wu, Shuyan Hu, Wutao Huang, Wei Ni, Yi Gong, Xin Wang. A comprehensive survey on applications of AI technologies to failure analysis of industrial systems// Engineering Failure Analysis, Volume 148, 2023, 107172, ISSN 1350-6307. URL: https://doi.org/10.1016/j.engfailanal.2023.107172. Pallavi Kumari, Bhavana Bhadriraju, Qingsheng Wang, Joseph Sang-Il Kwon. A modified Bayesian network to handle cyclic loops in root cause diagnosis of process faults in the chemical process industry // Journal of Process Control. Volume 110, 2022, Pages 84-98. URL: https://doi.org/10.1016/j.jprocont.2021.12.011 Chunbo Pang, Dawei Duan, Zhiying Zhou, Shangbo Han, Longchao Yao, Chenghang Zheng, Jian Yang, Xiang Gao, An integrated LSTM-AM and SPRT method for fault early detection of forced-oxidation system in wet flue gas desulfurization // Process Safety and Environmental Protection. Volume 160, 2022, Pages 242-254. URL: https://doi.org/10.1016/j.psep.2022.01.062 Hao Yin, He Xu, Weiwang Fan, Feng Sun. Fault diagnosis of pressure relief valve based on improved deep Residual Shrinking Network // Measurement. Volume 224, 2024, 113752. URL: https://doi.org/10.1016/j.measurement.2023.113752 Имитация аварий оборудования в системах моделирования химико-технологических процессов / Д. В. Зубов, А. В. Лобанов, Е. Б. Филиппова, Д. А. Лебедев // Математические методы в технологиях и технике. – 2022. – № 6. – С. 31-35. – doi: 10.52348/2712-8873_MMTT_2022_6_31. – EDN GSNFHJ. Разработка программных модулей цифрового двойника химико-технологического производства / Е. Б. Филиппова, Д. В. Зубов, Р. С. Крашенинников, А. В. Лобанов // Автоматизация в промышленности. – 2021. – № 7. – С. 16-20. – doi: 10.25728/avtprom.2021.07.04. – EDN CJBIPX
Supplementary files

