Regularization for improving the deconvolution in real-time near-field acoustic holography

Area of Science:

• Acoustics
• Signal Processing
• Computational Physics

Background:

• Near-field acoustic holography (NAH) locates sound sources using microphone arrays.
• Real-time NAH (RT-NAH) extends this to nonstationary sources via time-wavenumber domain analysis.
• Incorporating evanescent waves in RT-NAH improves spatial resolution but creates ill-posed deconvolution problems.

Purpose of the Study:

• To address the ill-posed deconvolution problem in RT-NAH.
• To compare singular value decomposition with Tikhonov regularization against optimum Wiener filtering for deconvolution.
• To evaluate the accuracy of reconstructed sound fields for nonstationary sources.

Main Methods:

• Formulation of time-dependent sound pressure propagation using convolution in the time-wavenumber domain.
• Backward propagation of the acoustic pressure field via deconvolution.
• Comparison of two deconvolution techniques: SVD/Tikhonov regularization and Wiener filtering.
• Simulation using monopoles driven by nonstationary signals.

Main Results:

• Both regularization and Wiener filtering improve deconvolution accuracy.
• Regularization methods, particularly Tikhonov, demonstrate superior performance in reconstructing sound fields.
• Enhanced accuracy is observed even in the presence of simulated measurement noise.

Conclusions:

• Regularization techniques are crucial for solving the ill-posed deconvolution problem in RT-NAH.
• The chosen regularization methods offer significant advantages for accurate acoustic field reconstruction.
• RT-NAH with regularization provides a robust solution for analyzing nonstationary sound sources.