The Use of Convolutional Neural Network for Segmenting Signals of Electrically-Actuated Valves

6/15/2021 2021 - #02 Modelling processes at nuclear facilities

Kotsoev K.I. Trykov E.L. Trykova I.V.

https://doi.org/10.26583/npe.2021.2.14

Electrically"actuated valves (EAV) represent one of the most numerous classes of equipment at nuclear power plants. The main problem of diagnosing EAV failures is the lack of operational (online) automated control of the technical condition of the EAV when the power unit is operating at full capacity.

In this regard, an important task is diagnosing the EAV according to the current and voltage signals consumed during the ‘opening’ and ‘closing’ operations of the EAV. The current and voltage signals are time series measured at regular intervals. The current (and voltage) signals can be received online and contain all the necessary information for the online diagnostics of the EAV condition.

The essence of the approach is to be able to calculate active power signals from the current and voltage signals, and then extract characteristics (‘diagnostic signs’) from certain sections (segments) of the active power signals, according to the values of which the state of the EAV can be diagnosed.

The paper is focused on the problem of automating the division of active power signals into segments (segmentation). In order to transfer the segmentation process to automatic mode, the authors have developed an algorithm based on the use of a convolutional neural network.

The developed convolutional neural network makes it possible to perform online segmentation of active power signals of the EAV. The network has shown good results, which will allow automated monitoring of the technical condition of the EAV when the reactor is operating at full capacity. As a result, the quality of the EAV operation is improved while the failure rate is reduced.

References

1. Abonyi J., Szeifert F., Babuska R. Modified Gath-Geva Fuzzy Clustering for Identification of Takagi-Sugeno Fuzzy Models. IEEE Systems, Man and Cybernetics, Part B. 2002, pp. 612-621.
2. Matveev A.V., Zhidkov S.V., Adamenkov A. K.., Galivets E.Yu., Usanov D.A. An Integrated Approach to Diagnosing Еlectrodrived Valves to applied to the Tasks of Resource Management. Armaturostroenie. 2009, no. 2 (59), pp. 53-59 (in Russian).
3. MT 1.2.3.02.999.0085-2010. Methodology «Diagnosting of Pipeline Еlectrodrived Valves». Moscow. Rosenergoatom Publ., 2010, 239 p. (in Russian).
4. Ronneberger O., Fischer P., Brox T. U-Net: Convolutional Networks for Biomedical Image Segmentation. arXiv:1505.04597v1 [cs.CV]. 18 May 2015, 8 p.
5. Long J., Shelhamer E., Darrell T. Fully Convolutional Networks for Semantic Segmentation. arXiv:1411.4038v2 [cs.CV]. 8 Mar 2015, 10 p.
6. Xiaoya Li. Dice Loss for Data-imbalanced NLP Tasks. arXiv:1911.02855v3 [cs.CL]. 29 Aug 2020, 12 p.
7. Sorensen Th. A Method of Establishing Groups of Equal Amplitude in Plant Sociology Based on Similarity of Species Content and its Application to Analyses of the Vegetation on Danish Cc of the Vegetation on Danish Commons. Biologiske Skrifter. 1948, v. 5, pp. 1-34.
8. Milletari F. V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation. arXiv:1606.04797v1 [cs.CV]. 15 Jun 2016, 11 p.
9. Drozdzal M., Vorontsov E., Chartrand G., Kadoury S., Pal Ch. The Importance of Skip Connections in Biomedical Image Segmentation. arXiv:1608.04117v2 [cs.CV]. 22 Sep 2016, 9 p.
10. Kingma D. P., Ba J. Adam: A Method for Stochastic Optimization. arXiv:1412.6980v9 [cs.LG]. 30 Jan 2017, 15 p.

convolutional neural network time series segmentation electrically-actuated valves automated system

Link for citing the article: Kotsoev K.I., Trykov E.L., Trykova I.V. The Use of Convolutional Neural Network for Segmenting Signals of Electrically-Actuated Valves. Izvestiya vuzov. Yadernaya Energetika. 2021, no. 2, pp. 158-167; DOI: https://doi.org/10.26583/npe.2021.2.14 (in Russian).