Session: 04-03: SMRs, Advanced Reactors and Fusion

Paper Number: 131784

131784 - Multiple Knowledge Exploration Functions for Advanced Reactor Design and Safety in Knowledge Management System Implemented in the Arkadia 

Abstract: 

Japan Atomic Energy Agency (JAEA) is developing ARKADIA (Advanced Reactor Knowledge- and AI-aided Design Integration Approach through the whole plant lifecycle) to meet the expectations for safe, economical and sustainable carbon-free energy. ARKADIA is an integrated numerical analysis system that utilizes AI to provide the best possible solution to any issue that may arise in the design process, safety assessment and operation of power plants. From the platform of ARKADIA, users have access to several systems with capabilities for finding solutions to problems that may arise in the design and operation of nuclear power plants. These systems are the Virtual Plant Life System (VLS) as a numerical simulator, the Enhanced and AI-aided optimization System (EAS) as a controller for the numerical analyses and the design optimization, and the Knowledge Management System (KMS) as a knowledge base. This paper presents an overview of KMS, and the knowledge exploration functions to be developed in KMS as well as information provisions to assist the EAS and VLS. KMS preserves and manages the knowledge including know-how accumulated in the development project of the prototype fast breeder reactor "MONJU" and related research and development (R&D). Since the knowledge is diverse and vast, it has been difficult for users to find the required knowledge in the database by manual. For this reason, multiple exploration functions are being developed in the KMS to match the characteristics of the information in user’s requirements. First, a large number of technical documents are sorted and categorized systematically. Then indexes are added to the documents to quickly find specific data in KMS through the portal site. In addition, a search function has been enhanced to enable cross-search of all information registered in KMS. This function also improves the efficiency of search by utilizing AI. For instance, it analyzes syntax and meaning to find documents closely related in meaning to the search text. Additionally, it presents correct keyword candidates when a search is conducted using the wrong keywords. For information regarding facilities, an enhancement of the search function for CAD drawings and spatial data has been introduced. With this function, confirmations that used to require on-site visits can now be handled remotely. Furthermore, the possibility of information retrieval using knowledge graphs is being developed to enhance efficiency of information retrieval, along with the evaluation flows in ARKADIA. This function can integrate rules, standards, interpretations, guidelines, design information, and related feasibility check information into a knowledge graph. Finally, milestones in KMS development, integration with the other two systems, and further enhancements are presented.

Presenting Author: Masanori Yoshikawa Japan Atomic Energy Agency

Presenting Author Biography: Masanori Yoshikawa studied quantum theory at Kyoto University. At the university he studied on broadcasting of entanglement states in generalized probability theories, a set of physical theories including classical and quantum theories as special cases. At present he belongs to a group of developing the knowledge management system of ARKADIA. His primary research areas include AI technology for controlling nuclear power plants.

Authors:

Akiyuki Seki Japan Atomic Energy Agency
Yuki Kondo Japan Atomic Energy Agency
Ryuta Hashidate Japan Atomic Energy Agency
Masanori Yoshikawa Japan Atomic Energy Agency
Kenji Yokoyama Japan Atomic Energy Agency
Shigeru Takaya Japan Atomic Energy Agency
Yasuhiro Enuma Japan Atomic Energy Agency
Taira Hazama Japan Atomic Energy Agency
Takashi Wakai Japan Atomic Energy Agency
Tai Asayama Japan Atomic Energy Agency