基于NIRS的食用醋品牌溯源模型的建立与优化

Development and Optimization of the Brand Traceability Model of Vinegar Based on Near-infrared Spectroscopy

摘要:

本文主要探讨了近红外光谱（NIRS）结合模式识别技术应用于食用醋品牌溯源研究。采集了四个品牌（四川保宁香醋、山西东湖老陈醋、镇江恒顺香醋、镇江香醋）共160组食醋样品的近红外漫反射光谱，通过主成分分析（PCA）进行光谱变量压缩及剔除8个异常样本数据后，随机选取其中的114组样品组成训练集用于建立溯源模型，剩余38组样品用作测试集进行模型验证。比较了MSC、SD、SNV等几种不同光谱预处理方法以及它们的不同组合对溯源模型的影响，同时考察了PLS-DA与SIMCA两种建模方法对模型的影响。结果表明：选择MSC与SD相结合的方法对光谱数据进行预处理，并采用SIMCA建模方法所建立的醋品牌溯源模型对四大品牌醋的正确识别率分别可达100%、100%、91.7%、90%。由此说明采用近红外光谱技术结合模式识别技术可有效实现食用醋品牌溯源的目的。

Abstract:

In this paper, near-infrared spectroscopy (NIRS) and pattern recognition technology were used to establish a brand traceability model for vinegar. NIR diffuse reflectance spectra of 160 vinegar samples from four different brands (SICHUAN BAONING vinegar, SHANXI EASTLAKE vinegar, HENGSHUN CHINKIANG vinegar, and CHINKIANG vinegar), were collected. After the spectral variables were compressed and eight abnormal sample data were removed using principal component analysis (PCA), 114 samples were randomly selected to form a training set for the construction of a traceability model, and the remaining 38 samples were used to form a test set for model validation. The effects of different pretreatment methods of the NIR spectra including multiplication signal correction (MSC), second derivative (SD), standard normal variate (SNV) etc., and different combinations of these methods on the traceability model were compared, and the effects of two modeling methods, partial least squares-discriminant analysis (PLS-DA) and soft independent modeling by class analogy (SIMCA)were examined. The results demonstrated that the correct recognition rates of the four different brands of vinegars could reach 100%, 100%, 91.7%, and 90%, respectively, when the spectral data were pretreated by MSC combined with SD and the model was developed using SIMCA. Therefore, the brand traceability of vinegar could be effectively achieved using NIR and a pattern recognition technique.