Distributed Brillouin frequency shift extraction via a convolutional neural network

Abstract

Distributed optical fiber Brillouin sensors detect the temperature and strain along a fiber according to the local Brillouin frequency shift, which is usually calculated by the measured Brillouin spectrum using Lorentzian curve fitting. In addition, cross-correlation, principal component analysis, and machine learning methods have been proposed for the more efficient extraction of Brillouin frequency shifts. However, existing methods only process the Brillouin spectrum individually, ignoring the correlation in the time domain, indicating that there is still room for improvement. Here, we propose and experimentally demonstrate a full convolution neural network to extract the distributed Brillouin frequency shift directly from the measured two-dimensional data. Simulated ideal Brillouin spectrum with various parameters are used to train the network. Both the simulation and experimental results show that the extraction accuracy of the network is better than that of the traditional curve fitting algorithm with a much shorter processing time. This network has good universality and robustness and can effectively improve the performances of existing Brillouin sensors.

Time-Efficient Convolutional Neural Network-Assisted Brillouin Optical Frequency Domain Analysis

“…The CNN model was trained and evaluated with real experimental data that were collected using the setup and parameters reported in Section 2.1 . In contrast to synthetic data, where artificial white Gaussian noise is added to ideal BGS in order to increase the generalizability of the model [ 13 , 14 ], the experimental data contains the actual noise that arises from the optical components [ 19 ]. The data were collected from measurements under controlled temperature conditions using a temperature chamber.…”

“…It has been shown that machine learning can provide solutions to many problems related to and enhancing the performance of the distributed fiber optic sensors [ 11 ]. Particularly in BOTDA sensing, machine learning algorithms based on artificial neural networks (ANN) [ 12 , 13 , 14 ] and support vector machines (SVM) [ 15 ] were implemented to extract the Brillouin frequency shift (BFS) outperforming conventional algorithms based on Lorentzian curve fitting (LCF). Because the extraction of temperature or strain necessitates the estimation of the temperature or strain coefficient, respectively, machine learning models were trained to predict the measurand of interest directly from the Brillouin gain spectrum providing a more compact solution [ 16 , 17 , 18 ].…”

Spatial Resolution Enhancement of Brillouin Optical Correlation-Domain Reflectometry Using Convolutional Neural Network: Proof of Concept

“…We also report potential improvement of the shortest strained length that can be realized for BFS extraction, which, throughout this paper, will be referred to as spatial resolution. Although Yao et al [27] and Chang et al [28] also proposed a CNN-based machine learning algorithm to obtain the BFS distribution from the measured BGS distribution, [27] did not report any specific results, whereas [28] reported that the spatial resolution of their system remained independent of the CNN, having used input data with varying BGS traces. Here, we also report an improvement of the spatial resolution by at least five times that of the nominal value.…”

Comparative Analysis on the Deployment of Machine Learning Algorithms in the Distributed Brillouin Optical Time Domain Analysis (BOTDA) Fiber Sensor

“…The prediction is made by comparing the input with a known outcome numerous times while updating the correlation coefficient, until the smallest mean error is obtained. A similar method can also be employed for temperature and strain prediction from BFS in BOTDA technique, such as combining artificial neural network (ANN) and principal component analysis (PCA) algorithms for temperature extraction [22,23], ANN and PCA for strain and temperature discrimination [24][25][26][27], deep learning (DL) [28][29][30][31], convolutional neural network [32], and support vector machine (SVM) [33] for accuracy improvement. We have also previously demonstrated the use of generalized linear model (GLM) in data processing for BOTDA [34].…”