Research On Brillouin Frequency Shift Extraction Technology In BOTDA Optical Fiber Temperature Sensing System

At present,with the application and popularization of big data,artificial intelligence,and Internet of Things,distributed optical fiber sensing technology has received extensive attention.At the same time,distributed optical fiber sensing technology has the advantages of radiation resistance,moisture resistance,small volume,simultaneous temperature stress measurement,etc.Many military organizations and enterprises have begun to use distributed optical fiber sensors to achieve accurate monitoring or measurement of environmental information.Among them,distributed optical fiber sensing technology(BOTDA)based on stimulated Brillouin scattering is widely used in oil pipeline monitoring,building structure detection and other application scenarios due to its long sensing distance and high measurement accuracy.Currently,there is still a problem in the BOTDA system that the measurement time,sensing distance,and measurement accuracy cannot be simultaneously compatible.Since the sensing distance is long,increasing the average number of signals to obtain sufficient accuracy will increase the measurement time;conversely,reducing the measurement time by using a method such as reducing the scanning interval will affect the measurement accuracy.The above-mentioned problems of different performance constraints have greatly hindered the promotion of the technology.Therefore,it is of great significance to study the method of improving the measurement accuracy and distance of the BOTDA sensing system and reducing the measurement time.Aiming at the above problems,this paper studies the BOTDA sensing system based on stimulated Brillouin scattering,and builds a 24.4km BOTDA temperature sensing experimental platform.For the Brillouin gain spectrum(BGS)of BOTDA system,this paper proposes a wavelet-Gaussian filtering algorithm to improve the performance of the system.For the extraction of Brillouin frequency shift(BFS),this paper proposes to use AdaGrad(Adaptive gradient algorithm)algorithm and sparse constrained cross-correlation convolution method to achieve fast,high-precision extraction of BFS extraction.The main work of this thesis is as follows:(1)Research on BGS denoising method based on wavelet-Gaussian filteringFor the BGS in BOTDA sensing system,this paper proposes a wavelet-Gaussian filtering algorithm to improve the signal-noise ratio.Results show that this method improves the signal intensity of BGS by about 2mV and improves the signal-noise ratio.Therefore,this method is beneficial to improve the signal-noise ratio and sensing distance of BOTDA sensing system.(2)Research on Brillouin frequency shift extraction based on AdaGrad algorithmIn this paper,the AdaGrad algorithm is used to extract BFS from BGS,and the speed and initial value dependence performance of this method will be focused on.Results show that compared with the traditional gradient descent method,this method is less dependent on the initial value and has a faster convergence rate.Compared with the LM(Levenberg-Marquard)algorithm,the average time-consuming of each data in this method is reduced about 0.023s.Therefore,the BFS extraction method based on the AdaGrad algorithm can extract BFS quickly and accurately.(3)Research on BFS extraction based on sparse constrained cross-correlation convolution algorithmThis paper proposes a sparse constrained cross-correlation convolution algorithm to extract BFS.This method improves the extraction accuracy of BFS by constraining the frequency distribution of the reference Lorentz curve.Results show that in the 24.4km BOTDA system,the BFS measurement accuracy of this proposed method is improved by about 5MHz compared with the traditional cross-correlation convolution algorithm.Therefore,the method is advantageous for improving the measurement accuracy of the BOTDA sensing system.