Artificial Neural Network for Accurate Retrieval of Fiber Brillouin Frequency Shift With Non-Local Effects

作者：Chongyu Lu, Yongxin Liang, Xin-Hong Jia, Yu Fu, Jing Liang, Zinan Wang · 发表于：IEEE Sensors Journal · 年份：2020 · DOI：10.1109/jsen.2020.2985550 · 被引用次数：20 · 研究领域：Advanced Fiber Optic Sensors、Photonic and Optical Devices、Advanced Fiber Laser Technologies

Brillouin optical time-domain analysis (BOTDA) that operates over a long sensing fiber is prone to be affected by the detrimental non-local effects (NLE); since NLE can distort Brillouin gain spectrum (BGS), therefore correctly retrieving Brillouin frequency shift (BFS) is very challenging. Recently, the basic artificial neural networks (B-ANN) has been demonstrated to retrieve BFS effectively, in the case that a distortion-free BGS can be obtained. However, in the more general cases with NLE, a neural network for retrieving BFS in BOTDA data has not been proposed. In this paper, firstly the physical origin of NLE is analyzed theoretically and experimentally in detail. Then a specific ANN (NLE-ANN) is proposed to deal with the BOTDA data affected by NLE for the first time. The experimental verifications show that with the cooperative implementation of NLE-ANN and B-ANN, the effective retrieval of BFS along the whole fiber can be achieved, even though the BGS has been affected by NLE. This work proposes a promising approach in traditional BOTDA, because the upper limit of probe power could be raised by artificial intelligence to boost the sensing performance.