Wavelet scattering transformation for sub-surface damage detection and localization in PZT sensor

Area of Science:

• Materials Science
• Engineering
• Signal Processing

Background:

• Structural Health Monitoring (SHM) systems rely on piezoelectric sensors, often Lead zirconate titanate (PZT), for structural integrity assessment.
• Subsurface damage in PZT sensors compromises SHM system reliability and accuracy.
• Existing methods may require reference signals from undamaged sensors, limiting practical application.

Purpose of the Study:

• To develop a reference-free method for detecting and localizing subsurface defects in PZT sensors.
• To enhance the reliability and self-diagnostic capabilities of SHM systems.
• To provide an interpretable and non-invasive solution for PZT sensor health monitoring.

Main Methods:

• Experimental measurement of ultrasonic wave interactions in PZT materials using a Coulomb coupling-based excitation scheme.
• Application of Block-Matching 3D (BM3D) filtering to denoise high-resolution but noisy wavefield images.
• Utilizing Wavelet Scattering Transform (WST) for stable, hierarchical, and translation-invariant feature extraction from denoised data.
• Unsupervised defect detection by identifying spatial anomalies that disrupt material symmetry.

Main Results:

• High-resolution, denoised wavefield images were obtained, preserving critical structural details.
• Wavelet Scattering Transform provided robust acoustic pattern representations.
• The method successfully detected and localized subsurface defects without requiring a reference signal.
• Spatial anomalies indicative of defects were identified by deviations from expected material symmetry.

Conclusions:

• The proposed reference-free technique effectively detects and localizes subsurface defects in PZT sensors.
• This approach significantly enhances SHM system reliability through improved sensor self-diagnostics.
• The method offers a non-invasive, interpretable, and practical solution for maintaining sensor integrity.