基于气相色谱-质谱联用与偏最小二乘-判别分析的啤酒爽口性评价

目的：建立一种偏最小二乘-判别分析（partial least squares-discrimination analysis，PLS-DA）评价啤酒爽口性的方法。方法：基于顶空固相微萃取（headspace solid-phase microextraction，HS-SPME）结合气相色谱-质谱（gas chromatography-mass spectrometry，GC-MS）联用检测啤酒中爽口性相关指标转化为数据矩阵，利用SIMCA-P软件进行PLS-DA。结果：不同爽口性的啤酒在得分图中能够明显区分。根据载荷图显示爽口性较差的样品集DA（1）中敏感特征变量为辛酸乙酯、癸酸乙酯及乙酸异戊酯等酯类化合物。选取50 个样本为校准集，建立PLS-DA模型，4 个爽口性不同的啤酒样本模型的回归相关系数分别为0.861、0.798、0.765、0.812，样品辨别率为92%。利用建立的PLS-DA模型对27 个未知样品进行预测，不同爽口性样品的预测均方根误差分别为0.183、0.321、0.523、0.323，准确识别率为74.07%。结论：HS-SPME-GC-MS结合PLS-DA法是一种简单、快速、有效评价啤酒爽口性的方法。

Evaluation of Beer Crispness Using Gas Chromatography-Mass Spectrometry and Partial Least Squares-Discriminant Analysis

Objective: This study aimed to establish a new method for the evaluation of beer crispness using partial least squares-discriminant analysis (PLS-DA). Methods: The concentrations of the chemical constituents associated with beer crispness were determined by headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC-MS), and the obtained data were transformed to a matrix and analyzed by PLS-DA method using SIMCA-P software. Results: Beers with different crispness showed clear discrimination in the score plot of PLS-DA. According to the loading plot, ethyl caprylate, ethyl caprate and isoamyl acetate were the sensitive characteristic variables for the DA (1) set of samples with poor crispness. Fifty samples were selected as a validation set to establish a PLS-DA discrimination model. The correlation coefficients of the model for four beer samples with different crispness were 0.861, 0.798, 0.765 and 0.812, respectively, with a recognition rate of 92%. A total of 27 unknown samples were predicted by the PLS-DA model with root mean square errors of prediction (RMSEP) of 0.183, 0.321, 0.523 and 0.323 and a correct recognition rate of 74.07%. Conclusion: HS-SPME-GC-MS combined with PLS-DA is a simple and useful method for the evaluation of beer crispness.