Avtonomni sistemi za klasifikacijo virov hrupa na osnovi prostorskega filtriranja : doktorsko delo

slovenian

doktorska dela

2021

automation   dissertations   localization   microphone array   noise measurements   self-organizing maps   source classification

Univerza v Ljubljani, Fakulteta za strojništvo
(Obvezni izvod spletne publikacije)

This doctoral thesis presents the results of research that enabled the development of a advanced method for measuring noise. This method will be the basis for an improved scientific research in the understanding of human sound perception and the development of applied methods for measuring and monitoring noise, with the aim of improving the environmental habitat and development of more acceptable machinery and equipment. The direct result of new scientific findings is a four-channel microphone antenna with optimized geometry and new associated algorithms with which we achieved signal-to-noise ratios that exceed all commercial systems for detecting the direction of sound sources. As we wanted to mimic human actions and track trends of environmental noise measurements, we used psychoacoustic features because they directly describe human perception of sound. We introduced the concept of immission directivity. We also developed a more efficient algorithm to detect the direction of sound arrival and introduced a modified growing selforganizing map for classification of different noise sources. The prototype system was tested in a real environment of noise measurements.

15.05.2023