TY - GEN
T1 - Automation Trustworthiness and Transparency in Nuclear Power Plants: Definitions, Methodologies, and Case Studies
T2 - 19th International Conference on Probabilistic Safety Assessment and Analysis, PSA 2025
AU - Khalid, Muhammad Hammad
AU - Bui, Ha
AU - Al Rashdan, Ahmad
AU - Reihani, Seyed
AU - Mohaghegh, Zahra
N1 - Publisher Copyright:
© Proceedings of the 19th International Conference on Probabilistic Safety Assessment and Analysis, PSA 2025. All rights reserved.
PY - 2025/6/15
Y1 - 2025/6/15
N2 - Nuclear power plants (NPPs) are integrating automation to improve efficiency and safety, but challenges remain in architecture design, trustworthiness, transparency, and licensing. This paper reports on some key results of a project, sponsored by the U.S. Department of Energy (DOE), that focuses on developing risk-informed methodologies for automation trustworthiness and transparency. Four key contributions of the project are: (i) Conducting a comprehensive literature review on automation trustworthiness and transparency to establish their definitions and evaluate strengths and limitations of existing methodologies; (ii) Developing a risk-informed methodology for evaluating and improving automation trustworthiness; (iii) Developing a methodology for quantifying automation transparency and (iv) Demonstrating the feasibility and practicality of the proposed methodologies for two case studies: (a) an AI-based fire classifier, and (b) an AI-based anomaly detection system. This paper focuses on Contributions (i), (iii), and (iv-b).
AB - Nuclear power plants (NPPs) are integrating automation to improve efficiency and safety, but challenges remain in architecture design, trustworthiness, transparency, and licensing. This paper reports on some key results of a project, sponsored by the U.S. Department of Energy (DOE), that focuses on developing risk-informed methodologies for automation trustworthiness and transparency. Four key contributions of the project are: (i) Conducting a comprehensive literature review on automation trustworthiness and transparency to establish their definitions and evaluate strengths and limitations of existing methodologies; (ii) Developing a risk-informed methodology for evaluating and improving automation trustworthiness; (iii) Developing a methodology for quantifying automation transparency and (iv) Demonstrating the feasibility and practicality of the proposed methodologies for two case studies: (a) an AI-based fire classifier, and (b) an AI-based anomaly detection system. This paper focuses on Contributions (i), (iii), and (iv-b).
KW - AI-based anomaly detection system
KW - automation transparency
KW - automation trustworthiness
KW - nuclear power plant
KW - uncertainty analysis
UR - https://www.scopus.com/pages/publications/105021941225
UR - https://www.ans.org/pubs/proceedings/article-58392/
U2 - 10.13182/PSA2025-46899
DO - 10.13182/PSA2025-46899
M3 - Conference contribution
AN - SCOPUS:105021941225
T3 - Proceedings of the 19th International Conference on Probabilistic Safety Assessment and Analysis, PSA 2025
SP - 643
EP - 652
BT - Proceedings of the 19th International Conference on Probabilistic Safety Assessment and Analysis, PSA 2025
PB - American Nuclear Society
Y2 - 15 June 2025 through 18 June 2025
ER -