气体拉曼光谱压缩感知方法分析研究

Gas Analysis by Raman Spectroscopy With Compressed Sensing Method

气体监测与我们的生活息息相关,氢气作为一种理想的研究模型更是受到广泛关注。拉曼光谱作为一种气体分析手段,具有无损非接触等优点[1]。本文以氢气和氘气为研究对象,研究气体拉曼光谱压缩感知方法。气体拉曼光谱测量存在的一个主要问题是气体拉曼散射信号弱,实际测得的拉曼信号容易被荧光、射线、CCD暗电流等噪声所掩盖。而在许多实验中,为了观察动态过程,需要采用较短的扫描时间,由于曝光时间短,单个拉曼光谱的噪声很大。

Gas monitoring is closely related to our lives.Hydrogen as an ideal research model has received widespread attention.As a gas analysis method,Raman spectroscopy has the advantages of non-destructive and non-contact.In this paper,hydrogen(H2)and deuterium(D2)are used as research objects to study the compressed sensing method of gas Raman spectroscopy.A major problem in gas Raman spectroscopy is that the gas Raman scattering signal is weak,and the actual Raman signal is easily masked by noise such as fluorescence,radiation and CCD dark current.In many experiments,in order to observe the dynamic process,a short scanning time is needed.Due to the short exposure time,the noise of a single Raman spectrum is very large.Compressed sensing theory can not only sample and compress the signal,but also remove the noise of the weak signal,and more accurately realize the restoration and reconstruction of the original signal.In the experiment,a 532 nm laser was used as the excitation light source to obtain the Raman spectrum of a H2(4 mbar partial pressure)and D2(3 mbar partial pressure)mixed gas.The Raman signal of the H2 and D2 mixed gas was processed based on the CVX toolbox and the orthogonal matching pursuit(OMP)algorithm(see Figure 1 and Figure 2).Preliminary experiments have proved that the compressed sensing method is feasible in mathematical theory.