This paper demonstrates a comparative analysis of five machine learning (ML) algorithms for improving the signal processing time and temperature prediction accuracy in Brillouin optical time domain analysis (BOTDA) fiber sensor. The algorithms analyzed were generalized linear model (GLM), deep learning (DL), random forest (RF), gradient boosted trees (GBT), and support vector machine (SVM). In this proof-of-concept experiment, the performance of each algorithm was investigated by pairing Brillouin gain spectrum (BGS) with its corresponding temperature reading in the training dataset. It was found that all of the ML algorithms have significantly reduced the signal processing time to be between 3.5 and 655 times faster than the conventional Lorentzian curve fitting (LCF) method. Furthermore, the temperature prediction accuracy and temperature measurement precision made by some algorithms were comparable, and some were even better than the conventional LCF method. The results obtained from the experiments would provide some general idea in deploying ML algorithm for characterizing the Brillouin-based fiber sensor signals.
In this paper, we studied the possibility of increasing the Brillouin frequency shift (BFS) detection accuracy in distributed fibre-optic sensors by the separate and joint use of different algorithms for finding the spectral maximum: Lorentzian curve fitting (LCF, including the Levenberg–Marquardt (LM) method), the backward correlation technique (BWC) and a machine learning algorithm, the generalized linear model (GLM). The study was carried out on real spectra subjected to the subsequent addition of extreme digital noise. The precision and accuracy of the LM and BWC methods were studied by varying the signal-to-noise ratios (SNRs) and by incorporating the GLM method into the processing steps. It was found that the use of methods in sequence gives a gain in the accuracy of determining the sensor temperature from tenths to several degrees Celsius (or MHz in BFS scale), which is manifested for signal-to-noise ratios within 0 to 20 dB. We have found out that the double processing (BWC + GLM) is more effective for positive SNR values (in dB): it gives a gain in BFS measurement precision near 0.4 °C (428 kHz or 9.3 με); for BWC + GLM, the difference of precisions between single and double processing for SNRs below 2.6 dB is about 1.5 °C (1.6 MHz or 35 με). In this case, double processing is more effective for all SNRs. The described technique’s potential application in structural health monitoring (SHM) of concrete objects and different areas in metrology and sensing were also discussed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.