基于子空间重合判断的混合醇组分光谱识别方法

针对光谱分析中传统的特征峰比对法对物质定性检测存在的不足,采用基于子空间重合判断的混合光谱模式识别方法对9种混合醇组分进行定性检测。此法将测量到的待定性混合醇光谱信息视为向量,通过计算混合醇组分拉曼光谱与标准样品数据库拉曼光谱的子空间夹角,并依据子空间夹角变化,排列筛选出含有最少标准样品数目的子空间,该子空间所包含的标准样品组成即为待定性混合醇组分组成,从而成功实现了9种混合醇组分的定性识别,识别率100%。该法具有测试时间短、测试结果准确、与被测物含量无关、操作简便等优点,适用于多组分混合体系的定性识别。

Spectral Pattern Recognition of Mixed Alcohols by Means of the Method Based on Judging the Subspace Coincidence

Based on the deficiency of qualitative detection of traditional characteristic peak comparison,a new method of the spectral pattern recognition based on judging the subspace coincidence was developed for the qualitative detection of nine mixed alcohols solution with different proportions.The spectra of mixed alcohol were treated as a vector in the method.First,the angle of raman spectra of mixed alcohol and the gather of standards were calculated.Then the least number of subspace,which contains the standard samples composed of the mixed alcohol undetermined,was selected according to the change of the subspace angle.Finally,it is successful to complete the qualitative recognition of all nine mixed alcohols of different proportions,and the detection rate of the method is 100%.With advantages of accurancy,simplicity,sensitivity and time saving,the method was suitable for the qualitative detection of multicomponent mixed system.