Computational reduction techniques for numerical vibro-acoustic analysis of hearing aids

Ester Creixell Mediante

Research output: Book/ReportPh.D. thesisResearch

738 Downloads (Pure)

Abstract

Numerical modelling is a key point for vibro-acoustic analysis and optimization of hearing aids. The great number of small components constituting the devices, and the strong structure-acoustic coupling of the system make it a challenge to obtain accurate and computationally efficient models. In this thesis, several challenges encountered in the process of modelling and optimizing hearing aids are addressed. Firstly, a strategy for modelling the contacts between plastic parts for harmonic analysis is developed. Irregularities in the contact surfaces, inherent to the manufacturing process of the parts, introduce variations on the final contact points in practice, making the contact properties unknown. The suggested technique aims at characterising the contact in terms of distributed stiffness values, which are identified by means of a model updating method that matches simulation to experimental data. Secondly, the applicability of Model Order Reduction (MOR) techniques to lower the computational complexity of hearing aid vibro-acoustic models is studied. For fine frequency response calculation and optimization, which require solving the numerical model repeatedly, a computational challenge is encountered due to the large number of Degrees of Freedom (DOFs) needed to represent the complexity of the hearing aid system
accurately. In this context, several MOR techniques are discussed, and an adaptive reduction method for vibro-acoustic optimization problems is developed as a main contribution. Lastly, topology optimization techniques for structure acoustic interaction problems are investigated with the aim of evaluating their applicability to the design of hearing aid parts. The strong fluid-structure interaction between the air and some of the thin, soft parts makes it necessary to include the effects of the interface variations in the optimization, which poses a challenge due to the need of interpolating between solid and fluid elements. Two techniques are compared in this context for a 2D hearing aid suspension design problem.
Original languageEnglish
PublisherTechnical University of Denmark, Department of Electrical Engineering
Number of pages176
Publication statusPublished - 2018

Fingerprint

Dive into the research topics of 'Computational reduction techniques for numerical vibro-acoustic analysis of hearing aids'. Together they form a unique fingerprint.

Cite this