Fiber Brillouin Scattering Spectra Analysis Based On Parametric Spectrum Estimation

By using AR spectrum estimator to calculate the power spectral density(PSD),several limitations of FFT algorithm can be improved.This project uses a distributed fiber laser sensor to monitor the temperature change of a long-distance fiber.This thesis focuses on parametric spectral estimator,comparison of it with traditional FFT spectral estimator,the advantages of AR model in spatial resolution and temporal resolution,and builds a linear prediction model for Brillouin backscattering optical signal.The thesis proposes a performance evaluation scheme through the comparison of the respective standard deviations of spectral center frequency estimation obtained by AR algorithm and FFT algorithm relative to calibrated temperature.In addition,this thesis also presents some research on Lorentz curve fitting and hardware development for optical signal processing of Brillouin spectrum.The main contents of this article are as follows:First,the theoretical model of Brillouin scattered light center frequency shift is introduced,and the reason that the Brillouin center frequency changes with temperature is explained.In this thesis,we discuss the advantages of parametric spectral estimation of Brillouin spectrum from two aspects: signal demodulation through parametric spectral estimator and Brillouin distributed sensing hardware design.Secondly,theoretical model of parametric spectrum and calculation method are introduced in detail,and three methods of calculating Brillouin spectrum,AR spectrum estimator,FFT spectrum estimator and FFT with zero-padding,are proposed,analyzed and compared.Brillouin scattering spectrum of BOTDR system analyzed by Burg algorithm of AR model shows a standard deviation of about 0.75 MHz when the data length is greater than 128.In comparison,the standard deviation obtained by FFT algorithm is about 2 MHz.Thus,AR algorithm can provide an improvement on frequency resolution of 3 times better than the FFT algorithm at medium spatial resolution.Thirdly,the curve model obtained in previous work is improved,and more rigorous spectrum fitting and center frequency measurement are obtained by using Lorentz curve fitting model.The thesis proposes a solution to apply AR algorithm to BOTDR data processing module through implementation of embedded hardware.This solution can well improve processing speed and resource consumption of practical engineering applications.Finally,the influence and application of parametric spectral estimation on Brillouin distributed sensing is studied.A set of LABVIEW virtual instrument programs is designed to realize fiber-optic sensing measurement of Brillouin Optical Time-Domain Sensing(BOTDR).