Detection of Coolant Leakage in Fusion Reactor by Machine Learning of Optical Fiber Scattering Signal

Abstract

The test blanket modules for the ITER fusion reactor have been designed. As for Japan, the TBM has a complex structure for breeding tritium while cooling with water and is exposed to high temperatures, radiation, and magnetic fields during operations. One potential accident is the blanket module's Loss of Coolant Accident (LOCA). Currently, leakage detection is planned through differential pressure. Identifying leakage locations and developing rapid detection methods are necessary to improve safety. The authors propose using optical fibers wrapped around a mock blanket submodule and applying machine learning to frequency modulation of scattered light for leakage detection. This study aims to develop a method to detect internal leakage and external heating using linear optical fiber arrangements.