Design of finite-size acoustic metasurfaces with a combination of Machine Learning and Reduced Order Models

This doctoral project is part of a larger, multidisciplinary and international project VAMOR: “Vibro-Acoustic Model Order Reduction” (Grant agreement no. 101119903) funded under the Marie-Skłodowska-Curie Actions Doctoral Networks within the Horizon Europe Program of the European Union.

The aim of this PhD research is to optimize acoustic metasurfaces of finite size using machine learning algorithms. While simple mass-spring systems or C-shaped resonators are well suited to the study of basic principles, more complex unit cells are required for broadband and low-frequency sound and vibration attenuation/absorption.
Although the periodicity of the structure implies infinite structures at least in one planar dimension, this is not the case for practical applications, as edges induce a considerable effect. To effectively manage the increased computational effort associated with calculating the response of the finite-dimensional structure many times over, the DC will undertake MOR approaches in combination with machine learning strategies to result in high-performance numerical models that can be used in an optimization scheme. The aim of this research is to incorporate MOR techniques into machine learning-based optimization for the simultaneous design of a host structure and a customized metasurface for automotive applications. Experimental validation (absorption coefficient in an impedance tube – attenuation in an alpha cabin) of the designed structure (poroelastic core with optimized internal resonator) will be carried out to validate the approach.

More informations on the link below.