Verification strategy for artificial intelligence components in nuclear plant instrumentation and control systems

The application of artificial intelligence (AI) systems in nuclear power plants has the potential to facilitate significant advancements in the industry. However, nuclear system engineers and regulators have expressed concerns regarding the reliability of AI components owing to a lack of documentati...

Full description

Saved in:
Bibliographic Details
Published inProgress in nuclear energy (New series) Vol. 164; p. 104842
Main Authors Park, JaeKwan, Kim, TaekKyu, Koo, SeoRyong
Format Journal Article
LanguageEnglish
Published Elsevier Ltd 01.10.2023
Subjects
Artificial intelligence
Deep learning model
Software development process
Software testing
Verification and validation
Software testing
Deep learning model
Artificial intelligence
Verification and validation
Software development process
Online AccessGet full text

Cover

More Information
Summary:The application of artificial intelligence (AI) systems in nuclear power plants has the potential to facilitate significant advancements in the industry. However, nuclear system engineers and regulators have expressed concerns regarding the reliability of AI components owing to a lack of documentation and appropriate verification methods that differ from those used for traditional software. Accordingly, this study proposes a systematic approach for the verification and validation of AI components based on nuclear software regulatory guidelines. The proposed strategy includes process verification and functional testing activities specifically tailored to AI components for instrumentation and control systems. This approach serves as a foundation for fostering understanding and agreement among system engineers, operating licensees, and regulatory bodies on the use of AI components in nuclear power plants.
ISSN:0149-1970
DOI:10.1016/j.pnucene.2023.104842