Nonnegative unmixing methodology applied on Brillouin Optical Fiber Sensor

Edouard Buchoud (1) , Valeriu Vrabie (2) , Jerome I. Mars (1) , Alexandre Girard (3) , Guy d'Urso (3) , Sylvain Blairon (3) , Jean-Marie Henault (3)

1 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique
2 CRESTIC - Centre de Recherche en Sciences et Technologies de l'Information et de la Communication - EA 3804
3 EDF - EDF

As a complement to conventional sensors, Distributed Optical Fiber Sensors (DOFS) have gradually played a prominent role in Structural Health Monitoring (SHM) for the last decade. The distributed Brillouin sensor enables to measure strain along kilometers of cable with a spatial resolution of 1 meter. The challenge is to have a centimeter spatial resolution to improve structure defaults detection and localization. A numerical model, based on the sensor physic, is first proposed in order to study Brillouin spectra distortion depending on strain distribution within spatial resolution. Then, based on nonnegative least squares (NNLS) problem, Brillouin spectra are decomposed into several elementary spectra. The estimated central frequencies and maxima permit to estimate a centimeter frequency distribution within the spatial resolution. It has been verified with numerical and experimental data: that method enables to enhance the accuracy and spatial resolution of the sensor from meter to centimeter.

Mots clés: NonNegative Matrix Factorization; NonNegative Least Square (NNLS) problem; spatial resolution enhancement; B-OTDA; Optical Backscatter Reflectometer; Structural Health Monitoring

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

Edouard Buchoud, Valeriu Vrabie, Jerome I. Mars, Alexandre Girard, Guy d'Urso, et al.. Nonnegative unmixing methodology applied on Brillouin Optical Fiber Sensor. EUSIPCO 2013 - 21th European Signal Processing Conference, Sep 2013, Marrakech, Morocco. pp.1569739197. ⟨hal-00986252⟩