基于挥发性成分定性判别风味油茶籽油掺伪浸出 油茶籽油PCA模型和逻辑回归模型的对比分析 Comparative analysis between PCA model and logistic regression model for qualitative identification of flavor oil-tea camellia seed oil adulteration with leaching oil-tea camellia seed oil based on volatile components

为了解决风味（原香和烤香）油茶籽油掺伪浸出油茶籽油的定性判别问题，设计高、低两个掺伪梯度，基于挥发性成分构建并对比分析了定性判别风味油茶籽油掺伪浸出油茶籽油的主成分分析（PCA）模型和逻辑回归模型。结果表明：逻辑回归模型定性判别风味油茶籽油掺伪浸出油茶籽油的能力较强，优于PCA模型；高掺伪梯度下定性判别原香和烤香油茶籽油掺伪浸出油茶籽油，PCA模型的最低检出限分别为20%和60%，而逻辑回归模型的最低检出限均为10%；低掺伪梯度下定性判别原香和烤香油茶籽油掺伪浸出油茶籽油，PCA模型的判别不准确，而逻辑回归模型的最低检出限均为4%。逻辑回归模型能很好地定性判别风味油茶籽油掺伪浸出油茶籽油。 Based on volatile components, principal component analysis (PCA) and logistic regression (LR)models were constructed and compared to solve the problem of qualitative identification of flavor (original/roasted) oil-tea camellia seed oil adulterated with leaching oil-tea camellia seed oil under high and low adulteration gradients.The results showed that LR model had a good ability to qualitatively identify the flavor oil-tea camellia seed oil adulterated with leaching oil-tea camellia seed oil, which was better than that of PCA model. Under the high adulteration gradient, the detection limit of the PCA model for original/roasted oil-tea camellia seed oil adulterated with leaching oil-tea camellia seed oil was 20%/60%, and the detection limit of the LS model was 10%/10%, respectively. Under the low adulteration gradient, the detection limit of LR model for original/roasted oil-tea camellia seed oil adulterated with leaching oil-tea camellia seed oil was 4%/4%, but the discriminative ability of the PCA model was not accurate.The LR model can qualitatively identify flavor oil-tea camellia seed oil adulterated with leaching oil-tea camellia seed oil.