PhD Defence - Joannès Chambon

Equivalent source methods for 3D acoustic imaging in complex environments

The PhD defence will be given in English, hosted by a jury composed of Jérôme Antoni, Simon Bouley (MicrodB), Mirjam Snellen (TU Delft), Jean-Hugh Thomas (LAUM), Vincent Valeau (Institut P Prime) and Paolo Chiariotti (Politecnico di Milano).

Abstract

With the growing interest for acoustic source identification in three dimensions, a set of new issues with respect to two-dimensional acoustic imaging with microphone arrays are to be tackled. One is the model chosen to describe the acoustic propagation in the environment where the measurements are performed. This may feature first order phenomena such as scattering effects, convection or inherent directivity patterns related to the physical origin of the sources. In practical, this model takes the form of a transfer function between sources and microphones, discretized into a matrix then featured to an inversion algorithm. In this thesis, the equivalent source method is surveyed as a simple and flexible simulation tool to build physically accurate transfer functions able to account for the above-cited phenomena. In-depth approaches to tune equivalent sources for acoustic imaging purposes are proposed and assessed with respect to various test cases including both analytical models, actual wind tunnel measurements and simulations. Finally, a more prospective algorithm is designed with a view to fully involve equivalent sources in the imaging process and bypass formalism of transfer functions computed independently from measurements: equivalent sources are tailored to concomitantly match with microphone pressures and a given boundary condition, then propagated toward regions of interest to seize the directivity of the emitted sound field. Performance of this method is benchmarked side by side with wide-spread acoustic imaging algorithms on a mock-up fitted with flush-mounted sources.