研究目的
Investigating the use of extreme learning machine (ELM) network to extract temperature distribution from the measured Brillouin gain spectra (BGSs) along the sensing fiber obtained by Brillouin optical fiber sensors.
研究成果
The ELM network can successfully extract temperatures distribution without determining BFSs by using any curve fitting from BGSs and hence the transforming BFSs to temperatures by calculating. It has higher accuracy and better tolerance of measurement error even using large frequency scanning step compared to conventional Lorentzian CFM.
研究不足
The accuracy of temperature extraction may be affected by the frequency scanning step and the linewidth variation of BGSs. The study focuses on temperature extraction and does not address strain measurement.
1：Experimental Design and Method Selection:
The study employs an ELM network for temperature extraction from BGSs, comparing its performance with conventional curve fitting method (CFM). The ELM network is trained using ideal BGSs constructed with Pseudo-Voigt curves.
2：Sample Selection and Data Sources:
The experiment uses a standard single-mode fiber (SMF) G. 652D, with a total length of 9.5 km, where the last 50 m is heated at different temperatures.
3：5 km, where the last 50 m is heated at different temperatures.
List of Experimental Equipment and Materials:
3. List of Experimental Equipment and Materials: Includes a tunable laser, electro-optic modulator (EOM), erbium-doped fiber amplifiers (EDFA1 and EDFA2), polarization scrambler (PS), fiber Bragg grating filters (FBGF1 and FBGF2), photodetector (PD), and electrical spectrum analyzer (ESA).
4：Experimental Procedures and Operational Workflow:
The pump pulse and local reference light are generated from a tunable laser. The SPBS light is amplified and filtered before being analyzed to reconstruct three-dimensional BGSs.
5：Data Analysis Methods:
The performance of ELM and CFM is compared using root mean square error (RMSE) and processing time.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
