Overflying measurements using microphone arrays: an investigation of algorithms for sound source localization of complex moving objects
Examensarbete för masterexamen
Sound and vibration (MPSOV), MSc
It is common to find airports in the vicinity of greater cities where aircraft causes health inflicting noise pollution. Deutsches Zentrum für Luft und Raumfahrt (German Aerospace Center) have been a contributor in reducing sound emissions from aircraft by proposing design changes backed by their research and experiments. To be able to evaluate various design attempts in a real life scenario, sound sources must be localized and quantified in an overflying measurement. At their private airfield they conduct experiments using a microphone phased array. With post processing techniques and the deconvolution algorithm, DAMAS, they have managed to localize and quantify sound sources of aircraft. DAMAS has been used by the German Aerospace Center since 2006. In recent time they have been eager to investigate more novel algorithms for sound source localization, and one in particular, called CLEANT. The goal of this thesis is to investigate how CLEANT works and implement it in python code in order to evaluate the results. This would be among the first attempts to use CLEANT on overflying measurements at this scale. Both CLEANT and DAMAS requires beamforming to be performed in order to enable further deconvolution of the results. The implementation of CLEANT was successful and interesting results were obtained. The main advantage of CLEANT is that it operates exclusively in the time domain whereas DAMAS is a frequency domain algorithm. This means that CLEANT can take the movement of the object into account whereas frequency domain algorithms are better suited for stationary sources. Although CLEANT can cause a masking effect leading to undetected sources who are very close to each other, CLEANT can operate in lower frequencies than DAMAS. It also scales better for greater amount of focus points in the grid, meaning that for higher resolution or larger area of analysis, CLEANT is faster.
sound source localization, aircraft, beamforming, deconvolution, microphone array