Optimal design of a sound reflector by particle swarm optimization

Abstract

For room acoustic design, the shape of the room and the materials need to be well considered. Designing such rooms requires extensive experience. Numerical analyses can minimize designers’ burdens by the automatic generation of many room candidates and the selection of an optimized one on the basis of objective measurements. Research on both numerical analysis methods and optimal design methods is required, but few studies are done regarding the latter. This paper proposes an optimal method by combining geometrical acoustic simulation with an optimization technique known as “particle swarm optimization.” Although the final aim is to achieve a global acoustic design, this paper is limited to local shape design, which is still important, considering that local shapes affect the characteristics of the global sound field. The proposed method is used for designing sound reflectors, whose aim is to scatter first reflected sound waves across the entire field. Hence, the optimization variables are sound reflectors shapes and materials, and the objective function is to evaluate the uniformity of a sound field in terms of room acoustic indices (T20, C80, and Ts). Compared with the initial reflectors, the optimized sound reflectors exhibited a higher uniformity in the room acoustic indices.