Artificial Intelligence for Estimating Multiple Irradiation Conditions from Temperature Distribution

An accelerated heating test method named Selective Laser Thermoregulation (SLT) is capable of dynamically controlling heat input by scanning a laser. Therefore, the irradiation conditions of the SLT system are highly flexible. The development of AI, which predicts the irradiation conditions that can reproduce the required temperature distribution, is required to determine the irradiation conditions automatically. In this paper, we report the development of AI that suggests irradiation conditions that can reproduce the required temperature distribution when irradiating a laser at a fixed point. The AI well predicted the irradiation coordinates, but it predicted the laser power, the beam radius, and the irradiation time different from the original values. However, the temperature distributions calculated using the irradiation conditions predicted by AI and the required temperature distributions were similar. We evaluated the similarity of the temperature distribution; 94.6% of data of MAPE are less than 5%.