Abstract: In order to improve the extraction to the sensing of brillouin scattering spectrum of the accuracy of brillouin frequency deviation，the squeezing factor of using particle swarm optimization algorithm is used to adjust the weights of ＲBFN ( Ｒadial Basis Function Net) network． The proposed algorithm overcomes the shortcoming of traditional ＲBFN neural network which is easy to fall into local extremum． Using PSO ( Particle Swarm Optimization) algorithm after adjust the weights to the transmission network，precision of brillouin scattering spectrum are extracted，ensurimg the solution speed and precision． In the process of numerical analysis，a new algorithm is used to estimate the scattering spectra of different line width and different SNＲ( Signal Noise Ｒatio) at different temperatures． Through the experiment the brillouin scattering spectrum data are obtained． Using YSPSOＲBFN( Particle Swarm Optimization with Shinkage Factor Shirnhage Factor-Ｒadical Basis Function) algorithm to deal with the experimental data，the results show that the algorithm can improve the accuracy of brillouin scattering spectrum feature extraction，the fitting error is 1． 99 MHz，under 25 ℃ when temperature fitting error is reduced．In 85 ℃ the frequency shift of fitting error is 0． 047 MHz． Therefore，when the algorithm is applied to the scattering temperature and strain sensing system of brillouin，it has great application prospect in improving detection accuracy．

Key words: temperature, optical fiber optics, compression factor particle swarm algorithm;radial basis function net ( ＲBFN) neural network, distributed optical fiber sensing