Learning the propagation properties of rectangular metal plates for Lamb wave-based mapping.

Area of Science:

• Robotics
• Non-Destructive Testing
• Acoustic Sensing

Background:

• Inspecting large metal structures like storage tanks and ship hulls requires efficient, reliable methods.
• Lamb waves offer potential for long-range non-destructive testing, but integrated robotic systems are lacking.
• Current methods often require prior knowledge of material properties, which is not always available.

Purpose of the Study:

• To develop a Lamb-wave-based robotic system for inferring plate geometry without prior material property knowledge.
• To enable autonomous inspection of large metal structures in challenging environments.
• To advance ultrasonic Simultaneous Localization and Mapping (SLAM) for structural health monitoring.

Main Methods:

• Combined focalization to adjust Lamb wave propagation model parameters.
• Employed beamforming to determine plate boundaries using acoustic measurements from a mobile unit.
• Utilized simulated annealing optimization to recover the optimal space-domain beamformer.
• Assessed beamformer focusing ability on filtered maps for robust geometry estimation.

Main Results:

• Accurate plate geometry inference was achieved without prior knowledge of propagation models.
• The proposed method demonstrated robustness across diverse experimental conditions.
• The optimal beamformer outperformed models based on predetermined propagation properties, especially in non-nominal scenarios.

Conclusions:

• The developed method enables accurate plate geometry inference for large structures using Lamb waves, even without prior material data.
• This robotic approach enhances the efficiency and reliability of non-destructive testing for industrial applications.
• The findings pave the way for more autonomous and adaptable structural monitoring systems.