【摘要】 采用Flash GC全谱数据,通过支持向量机(SVM)技术,较好地识别A品牌和非A品牌卷烟,A品牌一、二和三价类卷烟,以及A品牌Ⅰ,Ⅱ,Ⅲ和Ⅴ风格卷烟,但Ⅳ和Ⅴ风格较为接近,无法完全区分。由于每一种样本均具有特征指纹色谱,全色谱数据能够反映不同样本之间的微小差异。利用支持向量机构建预测模型,采用线性核函数和全色谱数据,获得了较高的预测结果。更多还原

【Abstract】 By utilizing Support Vector Machine( SVM) technique to analyze full chromatographic data of Flash GC,a method was proposed to identify A brand and non-A brand cigarette. Moreover,it can identify cigarettes of different price class,including first,second and third class. Meanwhile,it can also distinguish different price class cigarettes in the same style. It can discriminate A brand cigarettes of Ⅰstyle,Ⅱstyle,Ⅲ style and Ⅴstyle,but Ⅳ style and Ⅴstyle can not be completely distinguished because they are too close. The satisfied predicted results were obtained due to the adoption of liner kernel function and full chromatographic data,because every cigarette sample has its characteristic fingerprint chromatography and full chromatographic data can capture the minute differences among samples.更多还原