Boundary layer noise subtraction in hydrodynamic tunnel using robust principal component analysis.

Area of Science:

• Acoustics
• Hydrodynamics
• Signal Processing

Background:

• Acoustic studies of propellers in hydrodynamic tunnels are crucial for design.
• Boundary layer noise (BLN) significantly complicates acoustic signal recovery.
• Poor signal-to-noise ratios necessitate advanced denoising techniques.

Purpose of the Study:

• To develop and validate a novel denoising method for acoustic signals affected by BLN.
• To improve the signal-to-noise ratio in hydrodynamic tunnel measurements.
• To assess the effectiveness of the proposed method for acoustic source localization.

Main Methods:

• Decomposition of the wall-pressure cross-spectral matrix (CSM) using Robust Principal Component Analysis (RPCA).
• Leveraging the low-rank property of acoustic CSM and the sparse property of BLN CSM.
• Incorporating a prewhitening strategy to handle spatially correlated background noise.

Main Results:

• The RPCA algorithm successfully recovers acoustic signals, even with negative signal-to-noise ratios when BLN is decorrelated.
• A prewhitening strategy enhances BLN reduction in realistic scenarios with partially correlated noise.
• Demonstrated improvement in BLN reduction within a large hydrodynamic tunnel.

Conclusions:

• The proposed RPCA-based denoising method is effective for reducing BLN in hydrodynamic tunnel acoustics.
• The technique shows promise for improving acoustic source localization accuracy.
• This method offers a significant advancement in analyzing propeller acoustics under challenging noise conditions.