[资源分享] 一种基于数据挖掘的计算机自动识别饱和脂肪烃的方法

一种基于数据挖掘的计算机自动识别饱和脂肪烃的方法
  张良晓;梁逸曾;曾茂茂;曾映旭;陈爱明;高海燕;王亚敏;吴海
  　　通过对Nist107标准物质中饱和脂肪烃的质谱数据进行挖掘,发现了新的饱和脂肪烃的谱图特征。利用这些特征设置分步问答,使用MatLab编程实现并优化建立了自动识别饱和脂肪烃的方法。利用库内分子量小于450的标准质谱数据和本实验室测定的齿叶铁线莲(根)挥发油数据进行检验。库内检验结果表明:第一类错误(弃真错误)率为0.503%,第二类错误(存伪错误)率为0.008%。正构烷烃的误识率为0。实验数据验证结果表明:此法能准确识别饱和脂肪烃,用于识别正构烷烃时可克服因夹角余弦相似匹配中相似度接近而无法正确定性的缺点,为植物挥发成分测定、汽油样品的PONA分析和空气颗粒物中饱和脂肪烃的识别和总量的测定提供了一种新的快速的方法。
【作者单位】：中南大学化学化工学院中药现代化研究中心 长沙410083
【关键词】：低分辨率质谱;烷烃;数据挖掘;识别
【基金】：科技部国际合作项目(Nos.2007DFA40680、2006DAF41090)资助
【分类号】：O657.63
【DOI】：CNKI:SUN:FXHX.0.2008-05-017
【正文快照】：
　　1引言尽管以软电离技术(例如,电喷雾电离质谱和基质辅助激光解吸电离质谱)为依托的生物质谱已成为有机质谱的前沿课题之一[1],电子轰击质谱(EI-MS)仍然广泛使用,尤其与气相色谱和液相色谱联用[2]。传统的质谱图解析的方法主要有利用经典裂解规律的手动解析,基于标准物质质谱数
Novel Computer-assisted Interpretation Method to Recognize Saturated Alkane with the Help of Data Mining
  ZHANG Liang-Xiao;LIANG Yi-Zeng;ZENG Mao-Mao;ZENG Ying-Xu;CHEN Ai-Ming;GAO Hai-Yan;WANG Ya-Min;WU Hai(Research Center of Modernization of Chinese Herbal Medicines;College of Chemistry and Chemical Engineering;Central South University;Changsha 410083)
  　Based on the discovering and summarizing of several new concrete rules of alkanes mass spectra,a novel method for recognizing alkanes was proposed in the style of stepwise question and answer by applying data mining technique in Nist107 database of mass spectra of standard compounds.Two experiments were designed by implementing the presented scheme deliberately,and satisfactory results appeared in those tests.In the former test, the occurrence percentage of type Ⅰ errors is 0.503% while the counterpart of type Ⅱ errors is 0.008%.Meanwhile,the normal alkane could be perfectly recognized by this method.In the later test,a component,whose spectrum is a result of HELP resolution,is accurately determined as n-C11H24 from candidates gained utilizing dot-product similarity algorithm.This technique could be used to recognize and determine alkane in fast detection of essential oil in plant,PONA analysis and solid particulate matter in the air.The presented approach is programmed by MatLab6.5.


 附件: 您所在的用户组无法下载或查看附件