Time/frequency multiple signal classification beamforming based on principle component analysis for reducing platform and flow noise and identifying continuous and impulsive ground targets on UAV.
Ramon A. Silva, Yong-Joe Kim · The Journal of the Acoustical Society of America · 2011
When a microphone array is mounted on an UAV, most existing beamforming methods cannot be used to adequately identify continuous and impulsive targets on the ground due to high-level platform and flow noise. Here, we propose to develop a time/frequency beamforming method based on a principle component analysis and multiple signal classification (MUSIC) algorithm. This method can reduce the effects of platform and flow noise, e.g., engine and boundary layer noise, by removing noise-associated principle components from measured signals. It can also pinpoint the exact target locations while most existing methods can only detect target directions. In order to validate the proposed method, a cross-shaped microphone array is installed on the bottom surface of an UAV. The UAV is then placed in a wind tunnel operating at Mach 0.05–0.1 and its engine is turned on to simulate flight cruising conditions. Two loudspeakers are used to simulate continuous and impulsive ground targets. The target locations estimated from the proposed method are compared to the actual loudspeaker locations. Through the wind tunnel experiment, it is shown that the proposed beamforming method can be used to effectively suppress platform and flow noise and successfully identify the transient targets.