一种新的质谱分类变量及在饱和醇醚中的应用  

Development of a new spectra classifier and it's application in saturated alcohol-and ether

作　　者：负栋 梁逸曾[1] 徐承建[1] 胡黔南[1] 梁江龙[1] 

机构地区：[1]中南大学化学计量学及智能仪器研究所,湖南长沙410083

出　　处：《计算机与应用化学》2005年第4期277-281,共5页Computers and Applied Chemistry

基　　金：国家自然科学基金重点资助项目(20235020;20175036)

摘　　要：子结构模式识别一直是低分辨率质谱中的难点问题,为提供更多的分类信息,达到更好的识别和分类效果,将质谱图中被脱去碎片的信息,通过峰簇运算的方法提取出来,建立了一种新的质谱分类变量:FM(Fractional Mass)。联合块变量正交化和典型相关分析方法,以饱和醇醚的模式识别模型对变量进行了验证。结果表明,分类效果较好,无论是总体的模型还是交互检验,预测误差率都在2.5%以下,FM变量是传统分类变量的有益补充。Substructure recognition is always a problem of low resolution Mass spectrometry identification. To serve more classification information and get better classification result, more emphases focus on fragment's information of MASS Spectroscopy. Through peak cluster calculation, a new series of classifier named Fractional Mass (FM) is obtained. Such new variable is tested by pattern recognition for saturated alcohol and ether, combined with block variables' orthogonal computation and canonical correlation analysis. Results show that FM has excellent performance. Error probability of both Least Square model and Crossvalidation estimation are all under 2. 5% . FM is a kind of good supplement to traditional classifiers.

关 键 词：分类变量 模式识别 质谱 块变量 典型相关分析 

分 类 号：O657.31[理学—分析化学]