Research On Algorithm For Feature Extraction Of Single-Peak And Multi-Peak Brillouin Scattering Spectra

Accord to the corresponding characteristics of the external environment and the Brillouin scattering spectrum feature quantity,the distributed fiber optic sensing based on Brillouin scattering can realize the continuous measurement of the external environment in space.Distributed fiber optic sensors have been used in different areas.It is very important to improve the extraction accuracy of Brillouin scattering spectrum features.Therefore,a new algorithm should be developed to improve the extraction accuracy of Brillouin scattering spectrum features.Based on the summary of the key techniques in Brillouin scattering spectrum feature extraction,several intelligent algorithms have been analyzed to improve the accuracy of Brillouin scattering spectrum feature extraction,and analyzes the feasibility and applicability of these algorithms for Brillouin scattering spectrum feature extraction.The main contents are summarized as follows.Firstly,the research status of Brillouin scattering spectrum feature extraction for distributed optical fiber sensing technology are discussed.The scattering principle of Brillouin scattering,the fitting principle of Brillouin scattering spectrum,and the mapping relationship between temperature and Brillouin scattering spectrum characteristics are analyzed.And the Brillouin scattering spectrum signal detection technology is also studied.Secondly,the GA-SOA hybrid optimization algorithm is proposed,and the simulation and experimental data of Brillouin scattering spectrum under different conditions are processed by this algorithm and compared with other comparison algorithms.Then the features of multi-peak Brillouin scattering spectrum is extracted by this new hybrid algorithm.Thirdly,the SFLA-LSSVM hybrid optimization algorithm is proposed and applied to the feature extraction of Brillouin scattering spectrum.The multi-peak Brillouin scattering spectrum simulation analysis and experimental verification are performed under different conditions,the single-peak Brillouin scattering spectrum simulation analysis is carried outunder different conditions.Finally,the IGSSL hybrid optimization algorithm is proposed,which combines GA-SOA and SFLA-LSSVM,and applied it to unknown types of Brillouin scattering spectrum feature extraction.The simulation analysis and experimental verification of unknown types of Brillouin scattering spectrum are performed under different conditions.