DEVELOPMENT OF A MACHINE LEARNING-ASSISTED CORE SIMULATION FOR BOILING WATER REACTOR OPERATIONS

The research focuses on improving core simulation procedures in Boiling Water Reactors (BWRs) by leveraging machine learning techniques. Aimed at better fuel planning and enhanced safety, a machine learning model has been developed to predict errors in existing low-fidelity, diffusion-based core simulators. The machine learning models have demonstrated the capability to accurately and efficiently predict errors in core eigenvalue and power distribution in BWR Operations. This results in a significant improvement over conventional simulation methods in nuclear reactors without increasing computational complexity.