Denoising of cross-spectral matrices using canonical coherence

Area of Science:

• Acoustics
• Aerodynamics
• Signal Processing

Background:

• Microphone arrays in wind tunnels are susceptible to flow noise, complicating aerodynamic noise-source mapping.
• Existing noise suppression methods for cross-spectral matrices (CSM) have limitations, including remaining off-diagonal noise or high computational costs.

Purpose of the Study:

• To introduce a novel, computationally efficient method for aerodynamic noise-source mapping.
• To address the limitations of existing denoising techniques by offering automated parameter selection.

Main Methods:

• Development of a unique canonical-coherence-based method for noise suppression.
• Application of the method to averaged cross-spectral matrices (CSM) for aerodynamic noise-source mapping.
• Validation through simulated and real wind tunnel measurements.

Main Results:

• The proposed canonical-coherence-based method demonstrates high computational speed and automated parameter selection.
• The technique effectively suppresses flow noise, improving the accuracy of aerodynamic noise-source mapping.
• Performance is validated in both simulated and real-world measurement scenarios.

Conclusions:

• The canonical-coherence-based method provides a significant advancement in aerodynamic noise-source mapping.
• This approach overcomes the limitations of previous methods, offering a fast and reliable solution for wind tunnel acoustics.