The is the least “littered” with other noises.

The
model of the sound environment with the use of controlled samples differs from the
usual ones, including the fact that it can not use independent models for each noise,
due to the complexity of isolating “clean” samples containing the sounds
of only their sources of noise. The model of a particular
noise source, as a rule, depends on the models of other noise:

 ,

Where

 –
the model under consideration,

 

 –
noise parameters,

 

 –
parameters of the noise source,

 

 –
a lot of models of other noises.

This problem is solved by choosing the most independent
samples and taking into account the remaining dependencies when choosing the management
of their parameters. By
listening and analyzing records, the developer finds areas where the noise of this
source is the least “littered” with other noises. The
area should be long enough. Next,
it is necessary to correlate the selected noise with the parameters of its source
operation. This can be done using the indicator information
in the cabin, which has got into the video image of the recording, and in its absence
– using spectral analysis and knowledge of operating modes of the aircraft’s mechanisms
in different flight modes. For example, the noise
spectrum of the screw usually has characteristic peaks representing the main rotational
speed of the screw and its harmonics multiplied by the number of blades:

 ,

Where

 –
frequency of the i- th harmonic,

 

 –
the frequency of the screw,

 

 –
number of propeller blades,

 

 –
Harmonic number.

 Recalculating the frequencies
of the harmonics into the frequency of the screw and comparing it with the nominal
operating conditions of the power plant, it is possible to isolate the corresponding
characteristic sections of the recording (see the example in Fig.2).

 

Fig. 2.
Analysis of noise spectrum of a turboprop aircraft

 

The influence of models of individual noise on each
other can be taken into account on the basis of knowledge of the mutual dependencies
of the corresponding mechanisms in different flight regimes. For
example, the noise pattern of a propeller of a turboprop aircraft in takeoff and
landing and cruising modes, as a rule, can not be effectively separated from turbine
noise. At the same time, a linear change in the loudness
and frequency bands of such a sample leads to a “failure” of the frequency
components of the turbine at low revolutions of the screw, which “by ear”
is manifested as the absence of turbine noise. Therefore,
a turbine sample must be extracted from another part of the record, for example,
at the time when the screw starts spinning, when its noise is insignificant, and
the control dependence must take into account the noise amplification of the turbine
in the noise pattern of the screw by reducing the loudness characteristic (Fig.3)

 

Fig. 3.
Illustration of the volume correction of image playback to take into account their
interdependence

 

The problem of the lack of data on the absolute levels
of noise in the cabin can only be solved using expert judgment.

 First you need to get the
recording of the sounds of the aircraft. The simplest method is to record the time
of flight.

The resulting recording provides a panoramic view
of the sound at different stages of the flight and serves as part of the specimen.
Part of the recording oscilloscope, from the stage of engine start up to the transition
to cruise flight mode, is shown in Fig. 5. The main problem when working with such
records may be the lack of identification of sources of noise and their binding
to the parameters of the mechanisms. The subsequent identification of the sources
in this case should be carried out expertly, by listening to the characteristic
parts of the record and their reconciliation with the operational documentation
of the aircraft. In this way, it will be possible to identify the main stages of
the flight (Fig. 5) and identify the main sources of noise (starters, engines, screws).
The binding of these noise to the parameters of their sources is carried out using
spectral analysis. To do this, in the spectrum was the most powerful and characteristic
component of the work of the aircraft (for example, the revolutions of the propeller).
Further, as images were selected fragments with the most pronounced noises of one
source and the smallest components of others.

Fig.5. Oscillogram
of the flight sound fragment of the aircraft.