Um sistema de identificação de fontes de ruído utilizando a técnica de beamforming

Dissertação (mestrado) - Universidade Federal de Santa Catarina, Centro Tecnológico. Programa de Pós-Graduação em Engenharia Mecânica.This work describes the development and the application of an acoustical measurement system for identification of noise sources. Currently known techniques are able to map the noise radiation from a specific region in space using the sound field measured by a fixed array of microphones. The developed system uses a 32-microphones planar array processed by a classical frequency domain Beamforming algorithm. Different array geometries were studied using computational simulation and finally a multi-arm spiral array was designed based on the results from the simulations. To overcome some limitations of the available commercial systems, which are typically expensive and present a lot of operational constraints, low cost microphones and a flexible and portable signal acquisition platform were used. The microphones were designed and built using a commercially available electret cartridge coupled with a simple circuitry for the signal conditioning, making the easily integration with the acquisition system. Experiments on the microphones response characteristics allowed to select the most adequate samples, considering an expected operational range of 15 kHz. To validate the system, experiments were carried out in an anechoic chamber and it showed good agreement between the theoretical and experimental results. Finally, two different cases from the automotive industry were used to test the application of the system for the purpose of noise source identification. First, the sound transmission through a complex automotive structure with acoustic treatment was evaluated, showing clearly the regions of sound leakage, particularly at high frequencies. In a second experiment the radiated noise out of the engine compartment from a vehicle operating at a dynamometer was mapped. In this case the system was capable of presenting good results using the cross spectrum matrix diagonal removal technique, enabling the identification of the major noise sources