A time-frequency energy segmentation reconstruction method for multimodal ultrasonic guided waves.

Area of Science:

• Materials Science
• Acoustics
• Signal Processing

Background:

• Multimodal ultrasonic guided wave (UGW) signal reconstruction is crucial for accurate material nondestructive testing.
• Existing methods rely heavily on precise time-frequency (TF) ridge extraction, which is often incomplete or imprecise, especially with overlapping modes.
• This limitation hinders comprehensive analysis and accurate defect detection.

Purpose of the Study:

• To develop a novel TF energy segmentation reconstruction method for UGW signals.
• To overcome the dependency on complete TF ridge extraction for accurate mode separation.
• To improve the precision of UGW signal analysis in material nondestructive testing.

Main Methods:

• Proposed a TF energy segmentation reconstruction method that does not require complete TF ridge extraction.
• Introduced an adaptive noise variance estimation Bayesian filter for robust TF ridge extraction, even in noisy or overlapping regions.
• Employed a region growing algorithm to retrieve mode energy from both time and frequency domains for reconstruction.

Main Results:

• The proposed method achieved high reconstruction accuracies of 96.9% for simulated signals and 92.5% for experimental signals.
• Successfully separated and reconstructed UGW modes even with incomplete TF ridge information.
• Demonstrated the ability to recover modes with rapidly changing instantaneous frequency or group delay, outperforming traditional single-domain methods.

Conclusions:

• The TF energy segmentation reconstruction method offers a robust and accurate approach for UGW signal analysis.
• This technique enhances material nondestructive testing by providing more precise information through improved mode separation.
• The method's effectiveness is validated through numerical simulations and experimental data, showing significant improvements over existing techniques.