基于独立成分分析的复杂光谱的定量分析

A Quantitative Analysis for Complicated Spectra Based on Independent Component Analysis

基于独立成分分析(ICA)算法,对超短脉冲激光与气体相互作用所产生的复杂的非线性荧光光谱数据进行了有效的特征提取,进而对空气中含有的3种杂质气体在不同浓度下的光谱进行了预测,得到的结果与实测值相比较误差很小。对这3种气体共计27组光谱数据进行浓度值的定量预测,达到了满意的实验结果。

A quantitative analysis and feature extraction approach based on independent component analysis（ICA） is proposed to analyze the complicated nonlinear fluorescent spectra induced by nonlinear interactions of femto-second（fs） laser pulses with compound gas medium through,which the laser pulses transmit. The spectra of different impurity concentration predicted by using the extracted features show a low error compared with the experimental values. The results of a quantitative analysis for the impurity concentration of twenty seven sets of spectra data are satisfying.