Font Size: a A A

Distributed Optical Fiber Sensing Techniques Based On Specialty Optical Fiber And Advanced Signal Processing

Posted on:2022-09-28Degree:DoctorType:Dissertation
Country:ChinaCandidate:L ShenFull Text:PDF
GTID:1488306572476704Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
Distributed optical fiber sensors have the advantages of small,lightweight,resistant to electromagnetic interference,and spatial continuous sensing,and are widely used in large infrastructure detection,geohazard prevention,geophysical exploration and other fields.The Brillouin optical time domain analysis(BOTDA)technology can achieve a sensing distance of up to several hundred kilometers and a spatial resolution of up to millimeter,which has important research and application value.Single-mode fiber-based BOTDA systems are currently widely used for distributed temperature and strain measurements,however,there are many limitations in the measurement of other physical quantities.In distributed curvature measurement,the low curvature sensitivity and high bending loss of single-mode fibers severely limit their application in practical scenarios,and the measurement accuracy at small bending radii is also affected by the spatial resolution of the system.This thesis presents an innovative technique for distributed curvature measurement based on ring-core fibers,which can achieve more than four times the bend sensitivity of single-mode fiber with Brillouin frequency shift.The spatial resolution of BOTDA system based on deconvolution algorithm is proposed to improve the spatial resolution of 60/40 ns differential pulse pair up to 0.5 m.The 1 m and 0.5 m spatial resolution Brillouin frequency shift distributions can be extracted from the 40 ns pump pulse measurement results without changing the hardware conditions.The main research works and innovations of this thesis are summarized as follows.(1)For the problem of low bending sensitivity and high bending loss of single-mode fiber,this thesis proposes a distributed curvature measurement scheme based on ring-core fiber.Since the bending loss of the ring-core fiber is less than 0.01 d B/turn at 0.5 cm bending radius,the proposed scheme is very suitable for distributed curvature measurement under extreme bending conditions.(2)To improve the spatial resolution of the BOTDA system,a high spatial resolution BOTDA system based on the deconvolution algorithm is proposed in this thesis.The dependence between Brillouin gain and detuned frequency is eliminated by the differential pulse pair technique,and thus a distortion-free high spatial resolution Brillouin gain spectrum is obtained using the total variation deconvolution algorithm.Experimental results show that the proposed scheme can recover 1 m and 0.5 m spatial resolution from the 60/40 ns differential pulse pair measurements.At the same time,the deconvolution algorithm achieves 5.9 d B and 7.9 d B signal-to-noise ratio improvement at 1 m and 0.5 m spatial resolution,respectively,compared with the differential pulse pair technique.(3)In order to avoid the pulse differential process in the conventional high spatial resolution BOTDA system,a high spatial resolution Brillouin frequency shift extraction technique based on convolutional neural network is proposed in this thesis.By training the convolutional neural network with the Brillouin gain spectrum obtained from simulation and the corresponding Brillouin frequency shift,the trained network is able to extract the high spatial resolution sensing information directly from the long pump pulse measurement results.In the experiments,the network can extract 1 m and 0.5 m spatial resolution Brillouin frequency shift distributions from 40 ns pump pulse measurements,respectively.Compared with the differential pulse pair technique,the proposed scheme avoids the pulse differential process and thus requires only half of the measurement time.At the same time,the uncertainty of the network output is only 0.7 and 0.49 times of the differential pulse pair technique at 1 m and 0.5 m spatial resolutions.In addition,the convolutional neural network processes 10000 Brillouin gain spectrum in about 0.14 seconds,enabling real-time signal processing.By applying the proposed algorithm to the distributed curvature measurement of ring-core fiber,the spatial resolution is improved from 1.1 m to 0.4 m,which provides a feasible solution to achieve fast and accurate distributed curvature measurement.
Keywords/Search Tags:Distributed Brillouin optical fiber sensing, Distributed curvature sensing, Ring-core fiber, Total variation deconvolution, Convolutional neural network
PDF Full Text Request
Related items