RESUMEN
Sonic crystals have been demonstrated to be good candidates to substitute for conventional diffusers in order to overcome the need for extremely thick structures when low frequencies have to be scattered, however, their performance is limited to a narrow band. In this work, multiobjective evolutionary algorithms are used to extend the bandwidth to the whole low frequency range. The results show that diffusion can be significantly increased. Several cost functions are considered in the paper, on the one hand to illustrate the flexibility of the optimization and on the other hand to demonstrate the problems associated with the use of certain cost functions. A study of the robustness of the optimized diffusers is also presented, introducing a parameter that can help to choose among the best candidates. Finally, the advantages of the use of multiobjective optimization in comparison with conventional optimizations are discussed.
RESUMEN
An exhaustive study has been made into the potential improvement in attenuation and focusing of phononic crystal arrays resulting from the deliberate creation of vacancies. Use is made of a stochastic search algorithm based on evolutionary algorithms called the epsilon variable multi-objective genetic algorithm which, in conjunction with the application of multiple scattering theory, enables the design of devices for effectively controlling sound waves. Several parameters are analyzed, including the symmetries used in the distribution of holes and the optimum number of holes. The validity and utility of the general rules obtained have been confirmed experimentally.
