Search


Information
ICMST-Tohoku 2018
Oct. 23 - 26, 2018
Sendai, Japan
ICMST-Shenzhen 2016
Nov 1 - 4, 2016
Shenzhen, China
(EXIT THIS PAGE)
ICMST-Kobe 2014
Nov 2(Sun) - 5(Wed), 2014
Kobe, Japan
Nuclear Regulation Authority Outline of the New Safety Standards for Light Water Reactors for Electric Power Generation
For Public Comment
Outline of New Safety Standard (Design Basis)
For Public Comment
New Safety Standards (SA) Outline (Draft)
For Public Comment
Outline of New Safety Standard(Earthquake and Tsunami)(DRAFT)
Issues
 

Vol.10 No.2(Aug)
Vol.10 No.1(May)
Vol.9 No.4(Feb)
Vol.9 No.3(Nov)
< Other Issues

 

Occasional Topics
OTjapan Measures for Tsunami Striking Nuclear Power Station in Japan
Special Article: The Great Tohoku Earthquake (1)
OTjapan The Tragedy of “To Be” Principle in the Japanese Nuclear Industry
EJAMOT_CN3_Figure1_The_outside_view_of_CEFR OTChinaPlanning and Consideration on SFR R&D Activities in China
< All Occasional Topics

Featured Articles
EJAM7-3NT72 A New Mechanical Condition-based Maintenance Technology Using Instrumented Indentation Technique
EJAM7-3NT73 Survey robots for Fukushima Daiichi Nuclear Power Plant

JSM
Contacts
(EJAM): ejam@jsm.or.jp
(JSM): secretariat@jsm.or.jp
HP: http://www.jsm.or.jp
(in English)

 

Vol.16 No.2previousAA182
Academic Articles
Regular Paper Vol.16 No.2 (2025) p.12 - p.20
 

Performance Evaluation of Multivariate Time Series Anomaly Detection by Two-Stage Autoencoder using Power Plant Data

 

Yasunori TAGUCHI1,*,Susumu NAITO1,Yuichi KATO1,Kouta NAKATA1,Shinya TOMINAGA2 ,Naoyuki TAKADO2,Ryota MIYAKE2,Yusuke TERAKADO2,Toshio AOKI2 ,Yukio TAKAMORI3and Eiichi OOKUMA4

 
1 Corporate Research & Development Center, Toshiba Corporation, 1 Komukai-Toshiba-cho, Saiwai-ku, Kawasaki-shi, Kanagawa, 212-8582, Japan
2 Isogo Nuclear Engineering Center, Toshiba Energy Systems & Solutions Corporation, 8 Shinsugita-cho, Isogo-ku, Yokohama-shi, Kanagawa, 235-8523, Japan
3 Sigma Power Ariake Corporation, 72-34 Horikawa-cho, Saiwai-ku Kawasaki-shi, Kanagawa, 212-8585, Japan
4 Power Systems Div., Toshiba Energy Systems & Solutions Corporation, 2-4 Suehiro-cho, Tsurumi-ku Yokohama-shi, Kanagawa, 230-0045, Japan
 
Abstract
Power plants are one of the most important social infrastructure facilities for supporting people's lives, and therefore require highly reliable operation and maintenance. This is achieved through the installation of numerous sensors, and in large plants, the number of process variables can be in the thousands. Multivariate time series data are obtained from these process variables and displayed on multiple screens and monitored by operators. If an operator discovers an anomaly, the presence or absence of failure or deterioration is checked on-site. However, because the number of operators is limited, the number of process variables that can be visually observed is also limited. If the number of process variables is too large compared to the number of operators, there may be delays in the detection of anomalies. To overcome this limitation, we previously proposed a two-stage autoencoder to provide automatic and accurate early detection of anomalies from multivariate time series data. The latest version was developed to suppress false positives caused by false correlations in training data, and its effectiveness has been demonstrated using simulation data that mimics the behavior of a nuclear power plant. In this paper, we demonstrate the anomaly detection performance using operational data from an operating thermal power plant. The results confirm that this method is effective for suppressing false positives and that repair work that was overlooked by the comparison methods was correctly detected as abnormalities.
 
Keywords
plant monitoring, anomaly detection, multivariate time series data, autoencoder, two-stage autoencoder
 
Full Paper: PDF
Article Information
Article history:
Received 24 September 2024
Accepted 21 January 2025