Abstract
Imaging for damage in plate structures using Lamb waves is a cornerstone of structural health monitoring (SHM). To enhance localization precision, this study introduces a hybrid algorithm that modifies the damage index probability imaging method based on delay-and-sum (DAS) imaging using time-reversed Lamb waves. By incorporating a novel probability distribution function, the researchers improved the accuracy of damage detection. Experimental validation on aluminum plates demonstrates that this hybrid approach significantly outperforms conventional DAS methods in both localization accuracy and image quality.
Introduction: The Challenge of Lamb Wave Imaging
Structural Health Monitoring (SHM) is vital for maintaining the safety of critical infrastructure like aircraft skins and bridges. Lamb waves—ultrasonic waves propagating in thin plates—are frequently employed for their ability to travel long distances. However, their practical application is hindered by dispersion and scattering, which often lead to "ghost images" or artifacts in traditional imaging methods like Delay-and-Sum (DAS), especially when multiple defects are present. The goal is to produce a high-resolution "heat map" that clearly distinguishes real damage from background noise.
Methodology: Fusing Physics and Statistics
The proposed approach integrates three core components to sharpen structural imaging:
- Time Reversal (TR): Utilizing the principle of acoustic reciprocity, signals are recorded, time-reversed, and re-emitted. This process naturally compensates for dispersion, focusing wave energy back onto the scattering source (the damage).
- Delay-and-Sum (DAS) Imaging: This geometric technique identifies damage by finding the intersection of elliptical paths between transmitters and receivers.
- Modified Probability Distribution: The innovation lies in a new probability distribution function used to weight the damage index (DI). Two parameters, $m$ (distribution drop rate) and $\beta$ (shape parameter), allow for "tightening" the probability ellipses, effectively filtering spatial noise.
By combining the physical focusing of TR with these statistical refinements, the algorithm creates a weighted image where peak values pinpoint the most likely defect locations.
Experimental Setup
The study's effectiveness was validated using a controlled laboratory environment.
- Specimen: 1000×1000×1.5 mm aluminum plates.
- Damage Simulation: Circular holes (15 mm diameter) served as artificial defects.
- Configurations: Tests were conducted on plates with both single and double flaws.
- Instrumentation: An 8-sensor network was utilized, controlled by a system including a function generator, oscilloscope, and amplifier.
Key Results and Discussion
The hybrid algorithm demonstrated superior performance over standard DAS across all test scenarios:
- Single Defect: Localization error was reduced to 14.58 mm, compared to 19.04 mm for standard DAS.
- Double Defects: The improvement was most pronounced here. For two holes (D1 and D2), the hybrid method achieved errors of 10.00 mm and 11.18 mm, while standard DAS errors spiked to 29.15 mm and 40.31 mm.
- Parameter Optimization: Experimental tuning identified that $m = 2$ and $\beta = 0.895$ provided the highest imaging fidelity.
The results highlight that the algorithm's strength lies in its ability to resolve multiple defects clearly, a common failure point for simpler geometric methods.
Conclusion and Future Directions
While the hybrid algorithm marks a significant step forward, several avenues for future research remain. The accuracy of the imaging is sensitive to the Lamb wave group velocity; more precise velocity modeling could further minimize errors. Additionally, while circular holes provided a baseline, real-world damage like fatigue cracks presents more complex scattering patterns. Future work will also focus on automating the selection of the $m$ and $\beta$ parameters to make the system more adaptable for field applications.
Summary updated