程度相似度结合聚类分析评价烟用香精质量

Quality Evaluation of Tobacco Flavor by the Extent Similarity and the Clustering Analysis

[目的]探索烟用香精质量评价的新方法。[方法]采用气相色谱/质谱法(GC/MS)建立烟用香精指纹图谱,对19批次B#烟用香精,计算其相关系数、夹角余弦和程度相似度,并进行聚类分析。[结果]程度相似度算法结合聚类分析能够有效的反映样品间的整体差异。[结论]该方法准确、可行,可作为烟用香精质量评价的一种直观、合理、有效的技术手段。

[Objective]The research aimed to explore the new methods for the quality evaluation of tobacco flavor.[Method]The gas chromatography/mass spectrometry（GC/MS）fingerprint spectum of tobacco flavor was established to calculate correlation coefficient,cosine of the included angle and extent similarity of 19 batches of tobacco flavor B# samples.Then the clustering analysis was adopted.[Result]In combination of the extent similarity and the clustering analysis could effectively reflect the overall differences between samples.[Conclusion]This method was accurate and reliable,and it was an apparent,credible and efficient method for quality evaluation of tobacco flavor.