Research On Noise Suppression Method For BOTDR System Based On Sparse Representation

BOTDR system is widely used in safety monitoring,disaster warning and other engineering fields because of its advantages such as distributed sensing,single-ended accessing,and anti-electromagnetic interference.However,the BOTDR system detects the weak spontaneous Brillouin scattering signal,which has the inherent defect of low signal-to-noise ratio(SNR).At the same time,in the process of engineering practice,when the sensing optical cable is laid in harsh areas in the field,the optical fiber attenuation is relatively large,which further reduces the SNR of the system.Therefore,in the process of improving the performance of BOTDR system,improving the SNR is an urgent problem to be solved.It has important theoretical significance and practical value.In this paper,the sparse representation method in the field of digital image processing is proposed to reduce the noise of BOTDR system.A series of theoretical research and experimental analysis have been carried out,and the specific research contents are as follows.(1)Research on the generation mechanism of Brillouin scattering in optical fiber and BOTDR sensing mechanism.Discuss the main sources of noise in BOTDR system.Analyze the principle that digital image processing algorithms can be used for BOTDR noise reduction.A BOTDR noise reduction scheme based on sparse representation algorithm is proposed.Further,the basic principle of noise reduction of sparse representation algorithm is explained.Defining a metric for algorithmic noise reduction performance.The pre-configured and learning dictionary design methods and OMP algorithm are studied.It lays a theoretical foundation for subsequent simulation and experimental research.(2)Design the scheme of sparse representation algorithm based on pre-configured dictionary and learned dictionary.Build the BOTDR simulation system.Using the sparse representation algorithm based on the pre-configured dictionary to denoise the simulation data,and the fluctuation range of BFS in the variable temperature area is reduced from 8.07 MHz to4.62 MHz.Using the sparse representation algorithm based on learning dictionary to denoise the simulation data,the fluctuation range of RMSE is reduced from 1.103 MHz to 0.164 MHz in the distance of 24.15 km.Further,the simulation results of the sparse representation algorithm based on the pre-configured dictionary and the learned dictionary are compared and analyzed.(3)The BOTDR temperature sensing experimental system is designed and built.Collect experimental data and denoise the experimental data using the sparse representation algorithm based on pre-configured dictionary and learned dictionary.Further,the noise reduction performance of the sparse representation algorithm based on the pre-configured dictionary and the learned dictionary on the experimental data is compared.Quantitatively analyze the noise reduction performance of sparse representation algorithms based on the learned dictionary.The experimental results show that the sparse representation algorithm based on the learning dictionary can effectively denoise the data from 40 to 80℃.The highest measurement accuracy is increased from ±3.35℃ to ±0.59℃.The SNR improvement at the end of the fiber is 8.44 d B.