Improvement of sound source localization by de-convolution method considering a point spread function for model stage

Abstract

Aerodynamic noise generated from Shinkansen train is one of the environmental problems. In order to develop the noise mitigation method effectively, microphone array and de-convolution method is widely used for source localization in a wind tunnel test. On the other hand, in many cases, evaluation point of Shinkansen noise 25 m away from the track along the viaduct section and different from that of the wind tunnel test. This paper describes an evaluation method of sound pressure level (SPL) at an arbitrary observation point based on the sound source localization obtained by de-convolution method. A point spread function (PSF) considering the effect of sound field, sound reflection and diffraction around the model stage, is calculated by numerical analysis (Boundary Element Method) and applied to the de-convolution method. The proposed method is evaluated by a speaker test by comparing with a result obtained by conventional PSF considering a free space. The results show that accuracy of source localization and estimated SPL at the observation point is improved by applying the proposed PSF.