Separating Propeller Cavitation Noise and Cavitation Tunnel System Noise using Blind Source Separation(BSS)
Invited paper
Indonesia Hydrodynamic Lab
Tuesday 2 june, 2015, 14:40 - 15:00
0.8 Rome (118)
Abstract:
A passive sonar detects a surface ship or a submarine around it by using
vessel noises originating from the vessel to be detected. The vessel noise
consists of three components, namely propeller, flow and machinery noise.
Propeller noise consists of two components, i.e. a cavitation and a non-
cavitation noise. To investigate a propeller cavitation noise, we have
conducted an experiment done by using a cavitation tunnel. The K16B type
cavitation tunnel belonging to the Indonesian Hydrodynamics Laboratory (IHL)
Surabaya, Indonesia is used to generate the cavitation of a B-series four-
blades propeller model. This propeller is placed in its measurement section
and then rotates with different rotation speeds, variation of water flow,
and set in three different pressure conditions inside the tunnel. With these
three kinds of variations, the propeller cavitation can be generated.
Acoustic noise recorded in the cavitation tunnel, caused by the propeller
cavitation, is mixed with the noise originated from the tunnel components
such as impellers (which serves to flow the water in the tunnel) and the fan
at the top of the measurement section (which perform to manage the
temperature of a dynamometer in the tunnel low). To separate these two noise
components, i.e. cavitasi propeller and cavitation tunnel system noise, we
proposed to use blind source separation (BSS) methods. These techniques are
applied by assuming that the two noise components are convolutive mixed.
Several BSS techniques that have been used to separate two speech signals
that convolutive mixed, are applied to separate these two noise components
result from this experiment. Since frequency domain BSS can give a good
result of separating these two noise components, then so does a multistage
BSS in a frequency-domain and a time-domain. The application of the second
technique has been done in separating the signals originating from the
reverberant condition.