Crack characterization from ultrasonic array data via multi-angular-range scattering matrix reconstruction and...

Area of Science:

• Materials Science
• Non-Destructive Testing
• Acoustics

Background:

• Ultrasonic scattering matrices are crucial for characterizing small defects.
• Accurate crack characterization is essential for structural integrity assessment.
• Existing methods face challenges with noise and limited angular data.

Purpose of the Study:

• To propose a novel framework for scattering matrix denoising, extrapolation, and crack parameter regression.
• To enhance the accuracy and robustness of crack size and orientation estimation.
• To validate the framework's effectiveness through simulations and experimental studies.

Main Methods:

• Development of a multi-angular-range scattering-matrix denoising and extrapolation network (MARSM-DENet).
• Implementation of an adaptive optimal angular-range scattering-matrix regression network (AOAR-SMRNet).
• Utilizing simulation and experimental data for crack-like slots.

Main Results:

• MARSM-DENet successfully denoises and extrapolates scattering matrices across wider angular ranges.
• AOAR-SMRNet achieves highly accurate crack size estimation (MAE: 0.043λ-0.048λ, R² > 0.98).
• Experimental validation shows RMSEs of 0.084λ for size and 3.644° for orientation.

Conclusions:

• The proposed framework significantly improves ultrasonic crack characterization.
• Wider angular range reconstructions retain critical defect information and enhance noise robustness.
• The method demonstrates practical robustness and effectiveness for non-destructive testing applications.