[2K04-06] Operation Management, Inspection and Maintenance

[2K04] Development of Evaluation Method of the Number of SG Tubes Broken using Gaussian Process Regression

*seitaro sakurai¹, Akira Nakamura¹ (1. Institute of Nuclear Safty System, Incorporated)

Keywords:machine-learning, pressurized water reactor, steam generator tube rupture, numerical simulation

Using Gaussian process regression, we developed a machine learning method to automatically evaluate the number of broken tubes in the event of a steam generator tube rupture in a pressurized water reactor. We extracted feature values from changes in reactor coolant pressure. We focused on the feature values and predicted number of broken tubes in a numerical simulation by machine learning.