Real-Time Safety Decision Support for Nuclear Power Plant Operators
Overview
This study focuses on improving the safety of next-generation nuclear microreactors by developing a real-time decision support system. Unlike traditional nuclear reactors, microreactors operate under new and evolving conditions, making it difficult for operators to rely solely on past experiences and existing guidelines.
Challenges in Nuclear Safety
- Existing safety data is primarily based on Light Water Reactors (LWRs), which may not fully apply to microreactors.
- Limited cases of abnormal operation increase uncertainty in decision-making.
- Traditional regulatory guidelines do not always account for modern operational methods and emerging safety concerns.
Proposed Approach
To address these challenges, this research introduces a real-time decision support system that helps operators assess risks and verify operational guidelines. The system combines knowledge graphs and retrieval-augmented generation (RAG) with large language models (LLMs) to analyze current operational conditions and match them with relevant past incidents and safety regulations.
Expected Benefits
- Provides real-time hazard identification to support safer reactor operations.
- Enhances regulatory compliance by linking operational scenarios to existing safety guidelines.
- Reduces uncertainty in high-risk situations, helping operators make informed decisions.
Future Directions
While the current system is designed around past accident cases, future research will focus on identifying new and emerging operational risks. By analyzing patterns in decision-making failures, the goal is to develop a more adaptive and predictive safety management model.
