Time domain nearfield acoustical holography with three-dimensional linear deconvolution.

Area of Science:

• Acoustics
• Signal Processing
• Computational Physics

Background:

• Nearfield acoustical holography (NAH) is crucial for acoustic field analysis.
• Traditional NAH methods often rely on circular convolution, limiting accuracy for non-stationary sources.
• Reconstructing time-domain acoustic fields requires robust deconvolution techniques.

Purpose of the Study:

• To propose and validate a novel formulation for time-domain reconstruction of non-stationary acoustic fields using NAH.
• To enhance the accuracy and applicability of NAH for transient acoustic phenomena.
• To compare the performance of the proposed method against standard NAH techniques.

Main Methods:

• Development of a three-dimensional (3D) linear deconvolution formulation.
• Utilizing a Green's function sampled in time and spatial domains.
• Application of a 3D patch extrapolation algorithm to mitigate truncation errors.
• Comparison of Tikhonov regularization parameter prediction methods: L-curve, generalized cross-validation (GCV), and empirical Bayesian.

Main Results:

• The proposed linear deconvolution method significantly outperforms circular convolution-based NAH, reducing error by up to a factor of three.
• Accurate reconstructions are achieved even over large back-propagation distances.
• The 3D patch extrapolation algorithm effectively reduces truncation errors.
• Generalized cross-validation demonstrated the best prediction of noise levels and back-propagation distances for the transient baffled piston model.

Conclusions:

• The proposed linear deconvolution formulation offers superior performance for non-stationary acoustic field reconstruction in NAH.
• The method provides accurate and reliable results for transient sources and extended propagation distances.
• Generalized cross-validation is recommended for optimal regularization parameter selection in this context.