主成分分析法和Fisher判别方法在汽油分类分析过程中的应用_王荣辉.pdf

3612 中　国　科　学　技　术　大　学　学　报 Vol.36,No. 12 200612 JOURNAL OF UNI VERSIT Y OF SCIENCE AND T ECHNOLOGY OF CHINA Dec. 2 0 0 6 :0253-2778(2006)12-133-105 Fisher * 王荣辉,宗若雯, 王正洲,廖光煊 (, 230027) :应用主成分分析(principal component analysis,PCA)法对从90 #和93 #两种汽油的50 个 实验样所取的特征 据进行降维处理, 再结合Fisher 判别方法对这两种汽油进行分类, 并将分类 结果与不采用PCA 法而直接计算 据所得出的Fisher 判别结果进行比较,前者的分类正确率达 到100%,而后者却只有 50%. 结果说明采用PCA 方法事先对 据处理可以大大的提高汽油分类 的准确性. :主成分分析;Fisher 判别方法;汽油 :928. 7　　　:A Application of principal component analysis and Fisher discrimination method in the classification of gasoline WANG Rong-hui , ZONG Ruo-wen, WANG Zheng-zhou, LIAO Guang-xuan (State Key Laboratory of F ireScience, University of Science and Technology of China, Hef ei 230027, China) Abstract:Principal component analysis (PCA)and Fisher discrimination method were applied to studying the classification of 90 #and 93 #gasoline using gas chromatography-mass spectrometry (GC-MS ) spectral data. It was found that when both methods were used in calculating the dataset, all the samples were classified, while when the Fisher discrimination method alone was used, only 50%of the samples were correctly classified. It show s that when PCA is used in pre-disposal of the dataset, the accuracy of classification of gasoline is improved noticeably. Key words:principal component analysis;Fisher discrimination method;gasoline . 0 　 、、. , [1] ., , . . 1994 , (National Fire ,, Protection Association,NFPA), ., 60 , . *:200-6 0-6 13;:200-6 0-8 28 :(973)(201CB409600). :, , 19