Cutting-edge of application of AI technology to PRA (2). Development of probabilistic risk assessment methodology using artificial intelligence technology. Part 2. Automatic fault detection method for building reliability database
- 1. AdvanceSoft Corporation, Tokyo (Japan)
- 2. Japan Atomic Energy Agency, Oarai Research and Development Institute, Fast Reactor Cycle System Research and Development Center, Oarai, Ibaraki (Japan)
Description
To streamline and standardize the probabilistic risk assessment (PRA) of nuclear power plants, a three-year project titled 'Development of Probabilistic Risk Assessment Methodology Using Artificial Intelligence Technology' is underway. This paper introduces the 2022 prototype development and the outcomes and challenges (including expert evaluations) of the subproject focused on 'Automatic Fault Detection Method for Building Reliability Database.' Using natural language processing (NLP) frameworks and pre-trained large-scale NLP models, an AI tool methodology was designed and prototyped to construct a reliability database from NUCIA. This approach enables the automatic extraction of essential information for reliability database construction, such as fault locations (systems and components) and failure causes. (J.P.N.)
Availability note (English)
Available from DOI: https://doi.org/10.3327/jaesjb.66.11_560Abstract (Japanese)
原子力発電所の確率論的リスク評価の省力化・等質化を目指して,「AI技術を活用した確率論的リスク評価手法の高度化研究」を3ヵ年計画で実施中である。本稿では,この計画のうち「信頼性データベース構築のための自動故障判定手法の開発」の2022年度の試作内容と得られた成果・課題(専門家による評価)について紹介する。NUCIAから信頼性データベースを作成するAIツールの方法論を,自然言語処理フレームワークや事前学習済み大規模自然言語処理モデル等を活用して構築し,試作した。これにより,信頼性データベース構築に必要な情報である故障発生個所(系統・機器)や故障原因等を自動抽出可能となった。(著者)Additional details
Additional titles
- Original title (Japanese)
- 確率論的リスク評価手法へのAI技術活用の最前線.2.AI技術を活用した確率論的リスク評価手法の高度化研究その2.信頼性データベース構築のための自動故障判定手法の開発
Identifiers
Publishing Information
- Journal Title
- Nippon Genshiryoku Gakkai-Shi (Atomos)
- Journal Volume
- 66
- Journal Issue
- 11
- Series
- 雑誌名:日本原子力学会誌
- Journal Page Range
- p. 560-564
- ISSN
- 1882-2606
INIS
- Country of Publication
- Japan
- Country of Input or Organization
- Japan
- INIS RN
- 56005200
- Subject category
- S21: SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS;
- Descriptors DEI
- ARTIFICIAL INTELLIGENCE; DATA BASE MANAGEMENT; DECISION MAKING; ENERGY CONSERVATION; NUCLEAR POWER PLANTS; PROBABILISTIC ESTIMATION; PROGRAMMING LANGUAGES; RISK ASSESSMENT
- Descriptors DEC
- CALCULATION METHODS; MANAGEMENT; NUCLEAR FACILITIES; POWER PLANTS; THERMAL POWER PLANTS
Optional Information
- Notes
- 11 refs., 1 fig., 1 tab.; Abstract translated from Japanese in support of AI