Un modelo de caja gris para la acústica de recintos industriales

MARTÍN SEQUEIRA; VÍCTOR H. CORTÍNEZ

Rosario

Congreso; Primeras Jornadas Regionales de Acústica AdAA 2009; 2009

Asociación de Acústicos Argentinos

Noise pollution constitutes a serious problem within industrial rooms. Because of the high costinvolved in noise control systems, their design is an important aspect that must be considered. There are several computer models for this purpose, from simple analytical formulations to more complex methods based on geometrical acoustics. An alternative approach to model the different relationships involved in such situations is the use of artificial neural networks (ANN). One difficulty associated with this type of technique is the limited predictive extrapolation properties, which leads to erroneous results. In this connection, this paper deals with an approach that combines the theoretical structure of a Sabine type acoustic model with an ANN to correct the aforementioned model, in situations where it presents a known considerable inaccuracy. This grey model approach may improve the accuracy with respect to the ANN or the Sabine model used separately.