Machine learning model using transition learning to estimate mechanical properties of aluminum foam from X-ray CT images

Aluminum foam is a material that contains many pores and has excellent impact energy absorption and thermal insulation properties. The foaming process affects the mechanical properties of the aluminum foam due to variations in the pore structures. In this study, we attempted to estimate the plateau stress of Al-Mg-Si aluminum alloy A6061 aluminum foam from its X-ray CT images using machine learning. Using a machine learning model that has already been trained on industrial pure aluminum A1050 aluminum foam, a model was created to estimate the plateau stress of A6061 aluminum foam by transition learning. It was found that the use of transfer learning reduced the training time by 26% while achieving the same estimation accuracy as the model created by learning from the beginning without using transfer learning.