不同储存条件下的蜂王浆中红外光谱判别方法

Identifying royal jelly under different storage conditions based on mid-infrared spectroscopy

为了能够快速、无损地检测不同储存条件下的蜂王浆,探究了一种基于中红外光谱技术结合支持向量机算法(support vector machine,SVM)与正交偏最小二乘判别分析法(orthogonal partial least squares discriminant analysis, OPLS-DA)的蜂王浆定性分析方法。试验以-4 ℃冷冻储存和25 ℃室温储存7、14、21 d的蜂王浆为样品,应用中红外光谱技术采集蜂王浆样本光谱,并建立蜂王浆二分类(冷冻和室温储存)和三分类(室温储存7、14、21 d)定性分析模型。试验结果显示,基于SVM算法建立的蜂王浆二分类定性鉴别模型的预测准确率达到了92.31%,三分类定性模型预测准确率达到了100%。结合OPLS-DA法所建立的蜂王浆二分类模型和三分类模型的预测准确率分别为95.52%和96.97%。结果表明,运用中红外光谱技术结合SVM算法和OPLS-DA法可以有效鉴别出冷冻和室温储存的蜂王浆,为蜂王浆品质的快速、无损鉴别提供了可能性。

In order to efficiently, quickly and non-destructively detect royal jelly under different storage conditions, a qualitative analysis method was studied based on support vector machine (SVM), orthogonal partial least squares discriminant analysis (OPLS-DA) and mid-infrared spectroscopy. Royal jelly stored at 4 ℃ and at room temperature (25 ℃) for 7,14,21 d were tested. The spectra of the samples were collected by mid-infrared spectroscopy, followed by establishing qualitative analysis models for two-class (freezing and room temperature) and three-class (stored at room temperature for 7,14,21 d). The results showed that the predictive accuracy of the two-class and three-class models based on SVM were 92.31% and100%, respectively. Moreover, the predictive accuracy of the two-class and three class models based on OPLS-DA were 95.52% and 96.97%, respectively. Therefore, mid-infrared spectroscopy combined with SVM and OPLS-DA algorithm can effectively identify frozen and room temperature stored royal jelly, which provides a possibility for rapid and non-destructive identification of royal jelly quality.