21
68
Spécial « Congrès Acoustics 2012 »
Noise source identi cation techniques : simple to advanced applications
To increase the resolution for a selected radius, a hard
(closed) sphere is preferable. In the context of Spherical
Harmonics Beamforming, a hard sphere makes the proces-
sing more stable. The dynamic range, related to the highest
sidelobe level, is determined by the number of micropho-
nes and their distribution on the sphere. Examples are
presented in figure 7.
Fig. 7 : Pressure contribution based on Spherical Harmonics
Beamforming measurements in a car cabin. Air
conditioning noise Top: Engine at 3000 RPM, alignment
based on pictures. Bottom: idle condition, conformal
Spherical Beamforming based on cabin model
Contribution de la pression à partir de mesures de
faisceaux sphériques harmoniques dans l’habitable
d’un véhicule. Bruit de l’air conditonné. En haut :
Moteur à 3 000 tours/min, alignement sur la base des
photos. En bas : Conditions de repos, méthode des
faisceaux conforme à partir du modèle de l’habitacle
Sound Quality Metrics Mapping
Mapping sound intensity permits the localization of noise
sources. However, to understand the effect of the noise
source on a person, it is usually more appropriate to map
sound quality metrics such as loudness, sharpness or
impulsiveness [10]. The arrays required for such mapping
are the same as for SONAH or beamforming, only extra
processing needs to be added.
Selection of an NSI technique
Beamforming techniques assume that the sound field can
be modeled by typically plane waves or point sources,
whereas NAH using an array and sound mapping using
a single probe make no such assumptions. Therefore,
broadly speaking, at measurement distances far from the
source (several wavelengths), beamforming techniques
can be used, whilst close to the source (fractions of a
wavelength) NAH techniques are applicable.
The lower and upper limiting frequency of interest gives an
indication of the techniques which can be used (Fig.8).
Roughly speaking, the lower the frequency of interest,
the larger the array must be. The upper frequency limit
is set by the average spacing between the microphones.
However, the upper limit for Beamforming and Spherical
Beamforming can be increased by using different algo-
rithms and by an optimization of the microphone positions
to adjust the Maximum Side Lobe pattern.
Fig. 8 : Lower & upper frequency limits of arrays
Fréquences limites supérieures et inférieures des réseaux
The resolution obtainable for NAH/SONAH is typically
equal to the average spacing between the microphones.
For Beamforming, the resolution depends on the wave-
length, the nature of the sound source (coherent or non-
coherent), on the algorithm employed and also on the
size of the array (Fig.9). For the classical delay and sum
algorithm, resolution is about one wavelength. This can
however be improved by employing a deconvolution tech-
nique such as refined beamforming [7].
Another factor to be considered is the measurement area
covered by the array (Fig. 10). For NAH, the array must be
large enough so that virtually all acoustical energy passes
through the array area. In practice this means that the
array must often be of greater dimensions than the object
under test. For SONAH, even though the array only covers
part of the source, good representation of the sound field
can be obtained inside the covered area. For spherical
beamforming with omnidirectional directivity, provided that
the measurement distance is more than about twice the
radius of the spherical array (to avoid near field effects),
there is no limitation to the area covered.
Fig. 9 : Resolution of various acoustical arrays
Résolution de divers réseaux acoustiques
Fig. 10: Measurement area & distance
Zone de mesure et distance
