基于主成分分析和判别分析的白酒品牌鉴别方法

    Recognition method of liquor brands based on principal component and linear discriminant analysis

    • 摘要: 白酒的香气物质决定了白酒产品的差异。为了实现不同白酒产品的区分鉴别,提出了基于气相色谱分析技术结合模式识别实现白酒区分的方法。采集了7种产品共70个白酒样本的气相色谱数据,定性定量分析了己酸乙酯、乳酸乙酯等10种基本香气物质的含量,并对测定的物质进行主成分分析,验证区分效果,最后利用线性判别法建立判别函数,对不同白酒进行区分。结果表明,2种分析方法均可用于区分不同白酒,主成分分析结果显示,前3个主成分累计贡献率为86.527%,能有效描述香气物质和产品之间的复杂关系;线性判别分析对所有样本均得到准确的判别,正确率为100%,对预测样本的正确判别率达93.9%,建立的判别函数能准确区分不同白酒。研究表明,利用气相色谱技术结合模式识别的方法可用于不同白酒的区分鉴别。

       

      Abstract: Flavor substances determine the difference of the liquors. A recognition method for liquor brands discrimination was developed based on gas chromatography (GC) detection combining with pattern recognition in this study. First, 70 samples’ GC data from seven liquors were collected. Ten kinds of basic flavor substances including ethyl caproate and ethyl lactate were analyzed qualitatively and quantitatively. Then, principal components analysis was used to testify the discrimination ability based on the GC quantitative data of the 10 flavor substances. Finally, linear discriminant analysis was used to establish discriminant functions in order to differentiate the liquors. The results showed that both methods can differentiate liquors. In principal components analysis, the first three principal components accounted for 86.527% of total variation, which can effectively describe the complicated relationship between flavor substances and products. The precision rate of linear discriminant analysis was 100%, which can correctly discriminate all the samples, and the correct discriminate rate for new samples is 93.9%. The results show that established discriminant functions can effectively differentiate liquors. According to the study, the method of GC technique combining with pattern recognition can be used to identify and differentiate liquors.

       

    /

    返回文章
    返回