My research activity focuses on the investigation of the mechanical properties of complex lattice structures manufactured through additive manufacturing technologies. These structures belong to the broader class of metamaterials, i.e., engineered materials whose effective properties primarily arise from their internal architecture rather than solely from their base material composition, enabling mechanical performances unattainable with conventional materials.

Despite their potential, additive manufacturing processes inherently introduce geometric imperfections and deviations from the nominal CAD design. Such defects may significantly degrade the mechanical performance of the component, affecting stiffness, strength, and fatigue resistance. Within this framework, my research aims at the qualification and monitoring of defects and imperfections in lattice metamaterials using non-destructive testing techniques based on acoustic resonance analysis.

The adopted methodology relies on the extraction of the acoustic spectrum of the component and its comparison with reference configurations. The presence of damage or geometric irregularities alters the effective mechanical properties of the structure, which in turn modifies its natural frequencies. Since a direct correlation exists between the natural frequencies of a system and its mechanical parameters—such as stiffness and mass distribution—variations in the frequency spectrum can be exploited to detect and characterize structural degradation.

In this context, a neural network-based framework has been developed to identify fatigue-induced defects, determine their spatial location, and estimate their extent. Dedicated neural network architectures were designed for each specific task. The models were trained and validated on numerically generated datasets obtained through finite element simulations and subsequently tested on experimental data, demonstrating satisfactory generalization capability from the numerical to the physical domain.

My current research activity is focused on the development of machine learning models capable of directly inferring the static mechanical properties of a component from its acoustic spectrum, namely from its natural frequencies. The ultimate goal is to establish a reliable, rapid, and fully non-destructive methodology for the mechanical characterization and structural health assessment of additively manufactured lattice metamaterials.

Click here for more information and here to view my posters.