Single-Station Coda Wave Interferometry: A Feasibility Study Using Machine Learning

Area of Science:

• Geophysics
• Materials Science
• Artificial Intelligence

Background:

• Traditional coda wave interferometry relies on dual-station setups.
• A single-station approach offers potential advantages in deployment and cost-effectiveness.
• Assessing subsurface changes in highly scattering media remains a challenge.

Purpose of the Study:

• To evaluate the feasibility of a single-station coda wave interferometry technique.
• To investigate the use of ultrasonic measurements for detecting velocity changes.
• To apply machine learning for analyzing reflected coda wave signals.

Main Methods:

• Finite-difference simulations of wave propagation were employed.
• Ultrasonic measurements were simulated to detect velocity variations up to 1.6m depth.
• 1D convolutional neural networks were utilized for signal prediction and analysis.

Main Results:

• The single-station method demonstrated feasibility in identifying damage zones.
• The approach proved robust against variations in crack density, crack length, and attenuation.
• The influence of noise and sensor width on detection was analyzed.

Conclusions:

• The proposed single-station coda wave interferometry is a viable method for damage detection.
• The workflow integrates machine learning for enhanced analysis.
• The technique is transferable for defect detection in concrete structures.