Abstract
We proposed a laser heating test system called the selective laser thermoregulation system to confirm the
reliability of SiC/SiC ceramic matrix composites, which are expected to improve the efficiency of aircraft
engines. We fabricated the proposed system and confirmed that it can heat a SiC ceramic sample to
over 1400 °C and safely maintain this temperature. Furthermore, we developed AI for estimating the laser
power for automatic determination of the parameters which is necessary to realize the required temperature
distribution. We created datasets for machine learning by numerical simulation of laser heating
and compared three types of fully connected neural networks. The AI learned the relationship between
the laser power and the temperature distribution and then estimated the laser power from an untrained
temperature distribution. We found that three or four layers of fully connected neural network are sufficient
for realizing AI which can estimate laser power from a temperature distribution.
