Paper
18 March 2024 LSTM-based signal processing for slow-changing strain trend of Brillouin optical time domain reflectometry
Muping Song, Enxue Cui, Ning Jia
Author Affiliations +
Proceedings Volume 13104, Advanced Fiber Laser Conference (AFL2023); 131043Y (2024) https://doi.org/10.1117/12.3023543
Event: Advanced Fiber Laser Conference (AFL2023), 2023, Shenzhen, China
Abstract
Brillouin optical time domain reflectometry (BOTDR) is frequently used in structural health monitoring due to its ability to measure strain by extracting Brillouin frequency shift (BFS). Accurate strain determination is a problem in practical engineering, because of BOTDR’s limited BFS resolution and its cross sensitivity of temperature and strain. The important slow-changing information of strain is mostly submerged in the noise and periodic temperature variation. To obtain the slowing-changing strain from noisy BFS, a two-step method based on long short-term memory (LSTM) is designed for subsequent processing of BOTDR’s BFS. This LSTM model is specially trained and tested by both numerical simulation and experiment to get the trend and periodic information. Then the BFS of adjacent reference fiber segments is subtracted and the real strain information is obtained. Results show that the interference of noise and periodic temperature variation is successfully reduced and the strain resolution is improved with LSTM method. The mean short-time variance of strain decreases from 36.27 to 2.52.
(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Muping Song, Enxue Cui, and Ning Jia "LSTM-based signal processing for slow-changing strain trend of Brillouin optical time domain reflectometry", Proc. SPIE 13104, Advanced Fiber Laser Conference (AFL2023), 131043Y (18 March 2024); https://doi.org/10.1117/12.3023543
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Reflectometry

Signal processing

Structural health monitoring

Signal detection

Back to Top