EDN: TLOWQN
Authors & Affiliations
Arkadov G.V.1, Kotsoev K.I.2,3, Trykova I.V.2
1 Limited liability company “Kvant program”, Moscow, Russia
2 Scientific and Technical Center “Diaprom”, Obninsk, Russia
3 Bauman Moscow State Technical University, Moscow, Russia
Trykova I.V.2 – programmer. Contacts: e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it..
Kotsoev K.I.2,3 – Head of Data Science Department.
Arkadov G.V.1 – President of the Russian Institute for Systems Engineering, Сand. Sci. (Tech.).
Abstract
In reactor installations with a water–water power reactor, the appearance of loose, loosely anchored and outsider objects in the main circulation circuit is not excluded. These objects, moving in the coolant flow, can collide with the inner walls of the elements of the main circulation circuit, which can lead to damage to the equipment of the main circulation circuit. The experience of operating reactor installations shows that early detection of such events can minimize equipment damage and increase the safety level of NPP operation. Detection of loose, loosely anchored and outsider objects is carried out by bursts of acoustic noise from the impacts they cause in the coolant flow on the surface of equipment and pipelines. The main difficulty in detecting such events is that during the operation of the reactor plant, a variety of regular noises occur, which should not lead to false alarms of the safety system. We propose an effective mechanism based on artificial intelligence technologies that allows us to determine whether a specific acoustic signal is generated by a “regular” mode of operation or by the usual operation of equipment, or whether it is noise from a collision with a loose (loosely anchored) object that requires an immediate response from the personnel operating the NPP.
Keywords
nuclear power plant, artificial neural networks, variational autoencoder, loose and loose anchored objects, operational safety, acoustic anomalies, early detection
Article Text (PDF, in Russian)
References
1. Online monitoring for improving performance of nuclear power plants. Part 2: Process and component condition monitoring and diagnostics. Vienna: IAEA, 2008. 82 p.
2. Maksimov I.V., Perevezentsev V.V. A localization method of loose part for VVER NPPs. Izvestiya vuzov. Yadernaya Energetika, 2019, no. 4, pp. 28–38. DOI: 10.26583/npe.2019.4.02.
3. Regulatory Guide 1.133. Loose Part Detection Program for the Primary System of Light Water Cooled Reactors: tech. rep. U.S. Nuclear Regulatory Commission, 1981. 6 p.
4. Operation and Maintenance of Nuclear Power Plants. Part 12 Loose Part Monitoring. American Society of Mechanical Engineers (ASME), 2015. 523 p.
5. IEC 60988. Nuclear power plants. Instrumentation important to safety. Acoustic monitoring systems for detection of loose parts: Characteristics, design criteria and operational procedures, 2009, 75 p.
6. Maksimov I.V., Pavelko V.I., Perevezentsev V.V., Trykov E.L. Metod vydeleniya poleznogo signala dlya sistemy obnaruzheniya svobodnykh, slabozakreplennykh i postoronnikh predmetov v glavnom tsirkulyatsionnom konture reaktornoy ustanovki s vodo-vodyanym energeticheskim reaktorom. [Valid signal isolation method for loose parts monitoring system in the main circulation circuit of WWER reactor]. Vestnik MGTU im. N.E. Baumana. Seriya: Priborostroyeniye – Herald of the Bauman Moscow State Technical University, Series Instrument Engineering , 2018, no. 1, pp. 4–15. DOI: 10.18698/0236-3933-2018-1-4-15.
7. Arkadov G.V., Pavelko V.I., Finkel’ B.M. Sistemy diagnostirovaniya VVER [VVER diagnostic systems]. Moscow, Nauka Publ., 2019. 391 p.
8. Joao Pedro Cardoso Pereira. Unsupervised Anomaly Detection in Time Series Data using Deep Learning. Dr. elec. and comp. eng. Diss. Thesis. 2018. 94 р.
9. Kniaz V.V., Vygolov O.V., Fedorenko V.V., Sevrykov V.D. Glubokiye svertochnyye avtoenkodery: stereootozhdestvleniye dlya vosstanovleniya trekhmernykh modeley slabo teksturirovannykh ob"yektov [Deep convolutional autoencoders: Stereo matching for three-dimensional model reconstruction of low-textured objects]. Mashinnoye obucheniye i analiz dannykh – Machine Learning and Data Analysis, 2017, vol. 3, issue 2, pp. 119–134. DOI: 10.21469/22233792.3.2.03.
10. Analiz audiodannykh s pomoshch'yu glubokogo obucheniya i Python (chast' 1) [Analysis of audio data with help of deep learning and Python (part 1)]. 28.02.2020. Available at: https://nuancesprog.ru/p/6713/ (accessed 08.08.2023).
11. Spectrogram. From Wikipedia, the free encyclopedia. Available at: https://en.wikipedia.org/wiki/Spectrogram (accessed 08.08.2023).
12. Window function. From Wikipedia, the free encyclopedia. Available at: https://en.wikipedia.org/wiki/Window_function#Hann_and_Hamming_windows (accessed 08.08.2023).
UDC 621.039.4
Problems of Atomic Science and Technology. Series: Nuclear and Reactor Constants, 2023, issue 3, 3:18
