Robust sparse representation for adaptive sensing of turbulent phenomena

Ion Candel (1, 2) , Cornel Ioana (1) , Bertrand Reeb (2)

1 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
2 DTG - EDF - Division Technique Générale

In this study, the authors propose a method for turbulence characterisation by using sparse representation of a channel's impulse response. They consider the case of moving vortices created naturally or artificially that do not conserve their physical properties when observed at two distinct positions in space. The existing amplitude-based techniques fail to provide an accurate representation when the physical properties of the dynamic turbulence are altered. A two stages approach is proposed in this study. The first one deals with the design of robust waveforms for sensing of turbulent phenomena. The second stage consists of sparsely representing the decomposition of the turbulence's impulse response, based on a physically driven decomposition basis. The tests conducted in a reduced scale experimental facility show, on real data, the efficiency of the turbulence tracking. They compare several types of signals and show that the wideband signals are best suited for the application, achieving a high resolution combined with excellent results in terms of robustness.

Mots clés: adaptive sensing; wide band signals; turbulence; sparse representation

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Ion Candel, Cornel Ioana, Bertrand Reeb. Robust sparse representation for adaptive sensing of turbulent phenomena. IET Signal Processing, 2013, 28 (3), pp.285-290. ⟨10.1049/iet-spr.2013.0353⟩. ⟨hal-00985500⟩