傅立叶红外光谱技术快速检测黄酒中的氨基酸质量浓度

摘要

氨基酸含量是评价黄酒质量和风味的一项重要指标.为了实现对黄酒中氨基酸质量浓度的快速检测,作者探索了将傅立叶红外光谱技术(FT-IR)应用于黄酒中的17种自由氨基酸质量浓度的快速检测的可行性,同时采用协同区间偏最小二乘算法(SiPLS)和遗传算法(GA)选出有效波长以期提高模型的预测精度.实验结果显示与基于全波长光谱建立的经典偏最小二乘算法模型相比,基于SiPLS和GA选出来的有效光谱变量而建立的回归模型的精度有显著提高,尤其是对半胱氨酸(Cys)、精氨酸(Arg)和脯氨酸(Pro).此外,对于所有的氨基酸,GA-SiPLS模型效果最好,交互验证相关系数均大于0.80,残余预测偏差均大于2.00.因此,FT-IR结合特征谱区筛选方法可以作为常规检测的一种替代手段实现对黄酒中的自由氨基酸的快速检测.%The content of amino acid in Chinese rice wine (CRW) is one of the most important indexes to evaluate the quality and flavor of Chinese rice wine.In order to rapidly determine the contents of free amino acids in CRW,the possibility of Fourier transform infrared spectroscopy (FT-IR) for the fast detection of 17 different kinds of free amino acids in CRW wasdiscussed.Synergy interval partial least squares (SiPLS) and genetic algorithm (GA) were used to select the most efficient spectral variables to improve the prediction precision of the classical partial least squares (PLS) model based on the full-spectrum.Compared with the PLS model based on the full-spectrum,the prediction accuracy of model based on the spectral variables selected by SiPLS and GA was significantly improved,especially for cysteine,arginine and proline.In addition,GA-SiPLS model showed the most efficient prediction accuracy to all of the free amino acids,with the correlation coefficient of cross-validation higher than 0.80 and the residual predictive deviation larger than 2.00.The FT-IR combined with efficient variable selection algorithms is confirmed as a useful method to replace the traditional methods for routine analysis of free amino acids in CRW.