结合小波变换和经验模态分解的拉曼信号去噪研究

De-Noising of Raman Signals Based on Combination of Wavelet Transform and Empirical Mode Decomposition

摘要:

在分布式光纤温度传感器(DTS)的系统分析中, 所探测到的含有温度信息的后向拉曼散射信号的强度十分微弱, 并且所携带的噪声很大, 严重影响了测温精度。针对该系统存在的噪声, 提出了将小波变换和经验模态分解(EMD)相结合的方法对拉曼信号进行信号处理。实验结果表明: 该方法能够有效地去除系统中的噪声, 很好地保留了信号的有用部分, 系统的信噪比由原来的5.7dB提高到15dB, 系统的平均测温误差也由原来的2.8℃降到0.5℃, 从而得到更加精确的温度值。

Abstract:

In an optical fiber distributed temperature sensing (DTS) system, the detected backward Raman scattering signal with the temperature information is extremely weak and accompanied with strong noise, so the accuracy of the temperature measurement is seriously affected. To lower the noise in the system, a new de-noising method of combining wavelet transform with empirical mode decomposition (EMD) is proposed to process the Raman signal. Experimental results show that the noise is eliminated effectively, whereas the desired signal is preserved very well: the signal-to-noise ratio of the system is improved from 5.7dB to 15dB, concurrently the average error of measured temperature is reduced from 2.8℃ to 0.5℃.