Multi-fidelity reduced order modelling approach for the inverse identification of acoustic material properties

Abstract: Vibro-acoustic materials are crucial for various industries from automotive and aerospace to consumer electronics and architecture. Automotive applications include sound-absorbing foams in car doors, floors, and engine bays. To use acoustic materials in industry, accurate prediction of the vibro-acoustic behaviour of complex structures is required which is possible by accurate material characterization. The acoustic-material characterization is classically done by direct measurement methods like traction tests, ultrasonic testing, Impedance Tube Measurements etc. The direct approach for material parameter identification assumes that all boundary and initial conditions of the mechanical system (i.e. specimen, experimental device and set-up) are well known and that the only unknowns are the material parameters. Since the direct measurements rely on material samples, results are highly dependent on the inhomogeneity and anisotropy of the samples and boundary conditions. Therefore, this research aims to investigate the possibility of identifying acoustic material properties by inverse techniques. For instance, inverse approaches to estimate the elastic properties of foam from bending beam vibrations.