Design of sound phase diffusers by means of multiobjective optimization approach using ev-MOGA evolutionary algorithm

Abstract

In this paper a new approach to design sound phase diffusers is presented. The acoustic properties of such diffusers are usually increased by using single objective optimization methods. Here we propose the use of a multi- objective (MO) approach to design them in order to take into account several conflicting characteristic simultaneously. Three different MO problems are posed to consider various scenarios where fundamentally the objective is to maximize the normalized diffusion coefficient (following the corre- sponding Audio Engineering Society standard) for the so- called medium frequencies. This single objective could be divided into other several objectives to adjust performances to designer preferences. A multi-objective evolutionary algorithm (called ev-MOGA) is used to characterize the Pareto front in a smart way. ev-MOGA is modified, by using integer codification and tuning some of its genetic operators, to adapt it to the new requirements. Special interest is posed in selecting the diffusers codification properly to eliminate duplicities that would produce a multimodal problem. Precision in the manufacturing process is taking into account in the diffuser codification causing, that the number of different diffusers are quantified. Robust considerations related with the precision manufacturing process are con- sidered in the decision making process. Finally, an optimal diffuser is selected considering designer preferences.