Machine learning for turning optical Fiber Specklegram Sensor into a spatially-resolved sensing system. Proof of concept

Fiber Specklegram Sensors (FSSs) are highly sensitive to external perturbations, however, trying to locate perturbation's position remains as a barely addressed study. In this work, a system able to classify perturbations according to the place they have been caused along a multimode optical fi...

Descripción completa

Detalles Bibliográficos
Autores: Rodríguez Cuevas, Alberto, Fontana, Marco, Rodríguez Cobo, Luis|||0000-0002-2068-2956, Lomer Barboza, Mauro Matías|||0000-0002-4721-4247, López Higuera, José Miguel|||0000-0002-8615-8487
Tipo de recurso: artículo
Fecha de publicación:2018
País:España
Institución:Universidad de Cantabria (UC)
Repositorio:UCrea Repositorio Abierto de la Universidad de Cantabria
Idioma:inglés
OAI Identifier:oai:repositorio.unican.es:10902/16224
Acceso en línea:http://hdl.handle.net/10902/16224
Access Level:acceso abierto
Palabra clave:Fiber optic sensors
Multimode waveguides
Neural networks
Pattern recognition
Speckle
Speckle interferometry
Descripción
Sumario:Fiber Specklegram Sensors (FSSs) are highly sensitive to external perturbations, however, trying to locate perturbation's position remains as a barely addressed study. In this work, a system able to classify perturbations according to the place they have been caused along a multimode optical fiber has been designed. As proof of concept, a multimode optical fiber has been perturbated in different points, recording the videos of the perturbations in the speckle pattern, processing these videos, training with them a machine learning algorithm, and classifying further perturbations based on the spatial locations they were generated. The results show classifications up to 99% when the system has to categorize among three different locations lowering to 71% when the locations rise to ten.