Members

• Nicolò Speciale
  
• Luca De Marchi
• Emanuele Baravelli 
• Alessandro Perelli
  

Project description

In the last years time frequency transformation techniques have acquired a leading role in signal processing. However, such transformations have some restrictive properties which make them unsuitable for some applications. In particular, the possibility of generalizing and adaptively varying the time frequency plane tiling is a major demand. In order to accomplish this task many strategies are possible, such as the application of a preliminary invertible transformation (warping) to reshape the frequency axis in an invertible and flexible way.

A frequency warping map can be chosen according to two different criteria: (1) the distribution of the information along the frequency axis or (2) the way we want to extract information from a signal. The design procedure can start from some analytical warping model or from a set points describing a sampling of a warping function.

We implemented a design procedure and a fast computation algorithm for arbitrary warping maps. Our design strategy produces maps which match the target warping functions tailored to specific applications. In particular, benefits of this approach are being explored in the fields of Structural Health Monitoring (SHM) and biomedical signal processing.

Guided Waves (GWs) can be successfully applied in non-destructive evaluation of structural integrity and material characterization. The use of Lamb waves is especially attractive, but it is complicated by the dispersive nature of the different propagating modes. The Warped Frequency Transform (WFT) is effectively exploited to generate a new time-frequency representation that matches the spectro-temporal structures of the various propagating modes, thus allowing GW characterization. Dispersion compensation through a suitable combination of WFT, Basis Pursuit algorithms and imaging techniques provides defect detection with high localization accuracy. Tests have been performed on both simulations and experimental data recorded from aluminum plates by a scanning laser Doppler vibrometer.  

When long bones are considered as the waveguide, GW propagation analysis can be exploited to extract informations on the integrity as well as the density of the considered cortical bone.  In this context, our time-frequency techniques provide a valuable aid to non-invasive bone assessment including fracture detection and diagnosis of osteoporosis or other bone pathologies.

Contact:

Luca De Marchi       > This e-mail address is being protected from spam bots, you need JavaScript enabled to view it                                                                                                           

Experimental setup for NDT signal analysis

Collaborations

• University of Bologna, Italy - Laboratorio di meccanica computazionale
• Georgia Tech, Atlanta, USA - School of Aerospace Engineering 
• Fudan University, Shanghai, China - Department of Electronics

---

References

• See publications.