Microstructural characterization of bimodal composite metal foams under compression with machine learning

Cellular materials are gaining popularity in today's major sectors. The aim is to develop high-performance materials to meet customer and application demands. This study revolves around the beginning of the failure; computed tomography and statistical image analysis assisted with machine learning were employed to quantitatively characterize the occurring processes within the structure at the beginning of the well-known plateau effect. Simple techniques (e.g., easily obtainable shape parameters, forest decision, 2D image slices) were employed to sort the structural elements into seven classes based on their visual appearances. The presented method can provide additional information about the fracture mode of the different foams, possibly applying/extending them for other types of closed-cell (composite or not) foams.