Data-driven matched field processing for Lamb wave structural health monitoring

Area of Science:

• Engineering
• Acoustics
• Materials Science

Background:

• Matched field processing (MFP) is a model-based localization technique.
• MFP is widely used in underwater acoustics for complex environments.
• MFP is attractive for structural health monitoring (SHM) but faces implementation challenges due to model accuracy and computational cost.

Purpose of the Study:

• Introduce data-driven matched field processing (DD-MFP) as a novel framework.
• Develop a method to build propagation models directly from measured data.
• Enhance damage localization in SHM by overcoming MFP limitations.

Main Methods:

• Developed a data-driven framework for building propagation models from experimental data.
• Analyzed the DD-MFP framework's performance under unmodeled multipath interference.
• Applied DD-MFP to experimental measurements of an aluminum plate to localize two nearby scatterers.

Main Results:

• The DD-MFP framework successfully built models of multimodal propagation environments directly from measured data.
• DD-MFP demonstrated robust localization performance, distinguishing two nearby scatterers.
• Significantly smaller localization errors and finer resolutions were achieved compared to traditional delay-based models.

Conclusions:

• Data-driven matched field processing offers a viable and effective solution for damage localization in SHM.
• DD-MFP overcomes the limitations of traditional MFP by utilizing data-driven models.
• This framework provides a promising advancement for accurate and efficient structural health monitoring.