18
68
Spécial « Congrès Acoustics 2012 »
Noise source identi cation techniques : simple to advanced applications
The method is rarely used nowadays as it has mainly been
superseded by sound intensity mapping. The main benefit
of sound pressure mapping is the low cost involved. The
limitations are that the method is time consuming, is parti-
cularly susceptible to the influence of background noise and
can only be applied to stationary noise sources.
Since the 1980’s the sound intensity technique based on
a phase-matched pair of microphones, has been used to
measure the acoustic energy flow. This yields not only the
amplitude but also the direction of the sound energy. The
method has been incorporated into a number of interna-
tional standards to determine sound power.
Sound intensity mapping involves the measurement of
sound intensity spectra at a number of discrete points on
a grid close to the object under test. A contour plot is then
produced and superimposed on a photograph to enable
identification and documentation of the noise sources. To
speed up the process a robot is often used to move the
sound intensity probe from position to position.
The main benefit of sound intensity compared to sound
pressure is that it is a vector quantity. The acoustic field
can thus be represented with a magnitude and a direction.
Thus it is possible to determine sound power of a source
even in the presence of background noise. The technique
is now mature so that complete sound intensity mapping
systems are available as a 2 channel sound level meter.
The spatial resolution of a sound intensity map is limited
by the wavelength of sound and the distance between the
measurement points in the mapping grid.
Selective intensity calculates that part of the full measured
sound intensity that is coherent with a specific reference
signal. If, for example, the vibration of a specific compo-
nent is suspected to be the main cause of the radiated
noise, then an accelerometer mounted on that component
can be use to provide a reference signal for the selective
intensity calculation. If the suspicion is correct then the
selective intensity will be close to the full sound intensity
observed. The reference signal may be of any nature:
acoustic, vibration, force, electrical etc whichever provides
the cleanest and least noisy representation of the suspec-
ted cause. The benefit of selective intensity compared to
traditional sound intensity measurement is that it permits
a more precise localization of the sound source.
Nearfield Acoustic Holography
Spatial Transformation of Sound Fields (STSF) was one of
the first commercially available noise source identification
techniques based on nearfield acoustic holography [1]. The
technique requires a regular grid of microphones together
with a number of references transducers. Measurements
of autospectra and cross spectra are made over a plane
which completely covers the test object. Using principal
component decomposition techniques, a model of the
acoustical field is generated from which all acoustical para-
meters (pressure, particle velocity, intensity, power) can
be calculated both closer to and further away from the
test object. The calculations are based on 2D FFT and are
very fast. However there are limitations due to the fact that
unless the test object is measured entirely, artifacts occur
in the resulting mapping. For sources which are not statio-
nary in character, a transient method in the time domain
was developed known as Non-Stationary STSF.
SONAH
The STSF and Non-Stationary STSF methods have been
further developed to overcome some of their practical
limitations. A resultant technique is known as Statistically
Optimised Nearfield Acoustical Holography (SONAH) [2, 3].
The idea consists in fitting a plane wave model to the measu-
rements using a linear decomposition. In case of a double
layer of sensors, it is also possible to differentiate between
sources in front and behind the microphone array.
The main benefits of SONAH, compared to standard NAH,
are the possibility to use a non-uniform grid of micropho-
nes, a low sensitivity to sources outside of the calculation
plane and better performance at low frequencies.
The low sensitivity to sources outside of the calculation
plane makes it possible to perform local measurements;
therefore with SONAH, there is no need to use an array
which entirely covers the object under test, as is requi-
red for traditional NAH. Furthermore, the applicability of
SONAH to an irregular grid of microphones provides more
flexibility. In particular, different microphone array tech-
niques can be combined in the same system. For exam-
ple, combining SONAH with beamforming processing with
the same microphone array system provides an extended
frequency range.
Conformal calculations
Local measurements using SONAH open up the possibility
to perform conformal calculations. This means that quan-
tities such as sound pressure, intensity or velocity can be
calculated on the surface of the test object.
For even more accurate conformal mapping, following the
surface details of the object, it is necessary to acquire a
model (importing or digitizing) of the test object on one
hand, and on the other a precise location of the micro-
phone array.
Fig. 1: Conformal mapping workflow : Import or acquisition of
the geometry, mesh generation, patch measurements,
calculations and displays
Déroulement du travail de cartographie : importation
ou acquisition de la géométrie, génération du maillage,
mesures de correction, calculs et affichage
In order to reduce the preparation time and to simplify
the measurement process, a tracking system based on
infrared sensors is attached to the microphone array
handle, to obtain real-time positions and the orientations
of the array during the measurements. The workflow is
summarized in figure 1. In this process, it is necessary
to assume that the noise source characteristics are
stationary [3].
Using conformal maps reduces the risk of misinterpretation of
the data. Figure 2 shows two examples of applications.
