Data-driven and calibration-free Lamb wave source localization with sparse sensor arrays.

Area of Science:

• Acoustics
• Signal Processing
• Materials Science

Background:

• Lamb wave localization typically requires accurate velocity characteristics, which are often difficult to obtain in practice.
• Complexity of Lamb waves and sparse sensor arrays present challenges for traditional localization methods.
• A need exists for Lamb wave localization techniques that minimize dependence on prior velocity information.

Purpose of the Study:

• To develop and evaluate two novel methods for acoustic source localization using sparse sensor arrays.
• To address the challenge of unreliable or unavailable Lamb wave velocity information.
• To improve the accuracy and reduce systematic errors in Lamb wave localization.

Main Methods:

• Developed two methods exploiting the sparse structure of Lamb wave frequency-wavenumber representations.
• Method 1: Utilized sparse recovery techniques for velocity extraction from calibration data.
• Method 2: Employed kurtosis and support earth mover's distance to measure sparseness for source localization without prior calibration.

Main Results:

• Both novel methods demonstrated the ability to localize acoustic sources with less than 1 cm error.
• Experimental validation was performed on an isotropic aluminum plate using acoustic emission data.
• The proposed methods exhibited reduced systematic error compared to traditional time-of-arrival localization techniques.

Conclusions:

• The developed Lamb wave localization methods effectively reduce reliance on a priori velocity information.
• These techniques offer a promising alternative for accurate acoustic source identification in complex scenarios.
• The findings contribute to advancements in non-destructive testing and structural health monitoring.