-
- Takaya Shigeru
- Japan Atomic Energy Agency
-
- Seki Akiyuki
- Japan Atomic Energy Agency
-
- Yoshikawa Masanori
- Japan Atomic Energy Agency
-
- Yan Xing L.
- Japan Atomic Energy Agency
説明
<p>Enhancement of the ability to manage abnormal situations is important for improvement of the safety of nuclear power plants. It is needed to investigate potential risks thoroughly in advance, and to prepare countermeasures against the identified risks. In addition, in case of occurrence of an abnormal situation, plant operators are required to recognize the plant situation promptly and select a suitable countermeasure. However, the human ability to perform it is limited because the number of such abnormal situations in actual nuclear power plants is indefinite. Due to the advent of AI, it becomes possible to compensate for such limitation, by learning abnormal situations and assessing the effectiveness of prepared countermeasures virtually. The present study aims to develop such AI-based system to support plant operators to deal with abnormal situations steadily. Although many previous studies about detection of anomalies have been conducted, few studies consider countermeasures, especially against unexperienced abnormal situations. In this study, a novel plant operator support system that can estimate anomalies in a plant and propose countermeasures adaptively is proposed by using several AI technologies such as deep neural network and reinforcement learning. A plant simulator is used to prepare training data for AI. The combination of the proposed AI-based system and the plant simulator makes it possible to identify abnormal situations unknown to operators and propose countermeasures. The design and performance of the proposed system is illustrated using High Temperature engineering Test Reactor (HTTR) in Japan Atomic Energy Agency as an example.</p>
収録刊行物
-
- Proceedings of the ... International Conference on Nuclear Engineering. Book of abstracts : ICONE
-
Proceedings of the ... International Conference on Nuclear Engineering. Book of abstracts : ICONE 2023.30 (0), 1344-, 2023
一般社団法人 日本機械学会
- Tweet
キーワード
詳細情報 詳細情報について
-
- CRID
- 1390016803389790720
-
- ISSN
- 24242934
-
- 本文言語コード
- en
-
- データソース種別
-
- JaLC
- Crossref
-
- 抄録ライセンスフラグ
- 使用不可