提出了一种多区间主成分交互验证法。 该方法根据目标气体的光谱特征吸收峰, 在全光谱范围内划分出多个特征波数区间, 同时, 对单组分气体构成的参考光谱矩阵在划分出的特征波数区间内做主成分分析, 依据特征波数区间内累计贡献率大于99%的主分数和单组分气体在特征区内光谱最大值的排序, 识别出目标气体可能含有的组分, 再提取这些可能组分的特征波数区, 将这些区间与目标气体的特征波数区间逐一比对, 以确认或排除目标气体实际含有的组分。 由于在多个特征吸收区间上进行主成分分析, 避免了组分“漏选”现象; 由于使用了组分光谱特征波数区与目标气体光谱特征波数区的匹配验证, 避免了组分“多选”的可能。 模拟和实验数据表明, 该方法可对多组分的目标气体进行准确有效的定性。

A method of principal component interactive validation in multi-region is presented.Based on the characteristic peaks of target gas the full spectral region is divided into multiple characteristic regions.In succession the principal component analysis around the pure components spectral matrix in divided characteristic regions is applied.Thus the potential components included in target gas are identified according to the main components number according to 99% larger cumulated contribution and the descending sorting of maximum value of pure component spectrums.After that, the characteristic regions of potential components are compared with the ones of target gas one by one.In this way the real components of target gas are confirmed.Because the principal component analysis was applied in multiple characteristic regions, this method can avoid omitted identification.Owning to the matching validation between the characteristic absorption regions of target gas and the ones of pure component gas this method can avoid additional identification.The simulated and the measured spectra both validate the qualitative ability of this method.