基于主成分指标变化分析水牛乳中掺入普通牛乳的研究

Study on the identification of buffalo milk adulterated by bovine milk with principal component analysis

通过测定掺入不同比例普通牛乳的水牛乳理化指标,并基于这些指标参数采用单因素方差分析、主成分分析和多元逐步线性回归法,以此鉴别掺入不同普通牛乳量的水牛乳。结果表明,水牛乳掺入普通牛乳的量为15%以上时具有显著差异;主成分分析(PCA)中第一主成分(PC1)贡献率达到91.871%,PC1得分与掺假量的线性关系显著;多元逐步线性回归法建立的4个定量模型方程,其相关系数R^(2)分别为0.954、0.970、0.973、0.975,平均绝对误差分别为6.39%、6.34%、3.85%、5.38%,可实现水牛乳掺入普通牛乳的定量鉴别。

The objective of this study was to establish a method to identify buffalo milk adulterated with bovine milk with Principal Component Analysis(PCA). Single factor analysis of variance,principal component analysis and multiple stepwise linear regression were used to identify different adulterated buffalo milk based on the physical and chemical data of buffalo milk adulterated by different proportion of bovine milk. The data showed that there was a significant difference between pure buffalo milk and adulterated buffalo milk when the adulterated rate was 15%.The contribution rate of the first principal component(PC1)in principal component analysis(PCA)reached 91.871%,and there was a significant linear relationship between the PC1 score and the amount of adulteration. The four quantitative model equations established by the multiple stepwise linear regression method had correlation coefficients R^(2) of 0.954,0.970,0.973 and 0.975,and average absolute errors of 6.39%,6.34%,3.85%,and 5.38% respectively,which can achieve the quantitative identification of buffalo milk adulterated by bovine milk.