基于PCA和LS-SVM的傅里叶变换近红外光谱的黄酒酒龄的鉴别模型

Discriminative Model for FTNIS Analysis on Age of Shaoxing Rice Wine Based on PCA and LS-SVM

利用傅里叶变换近红外光谱技术,对黄酒酒龄鉴别的模型进行研究。对绍兴黄酒样本光谱主成分进行提取与分析,并发现前3个主成分具有较明显的聚类特征。其次,利用LS-SVM模型对不同主成分数进行分类和寻优,发现当主成分数为6时达到判别的正确率是100%,此时模型的两个参数γ和2σ分别为61.890和1.769。研究表明,利用傅里叶变换近红外光谱技术并结合主成分分析(PCA)和最小二乘法支持向量机(LS-SVM)可作为一种可靠、准确、快速的检测方法用于黄酒酒龄定性鉴别分析。

The discriminative model for the age of Shaoxing rice wine was investigated by Fourier transform near infrared（FTNIR） spectral technology.The spectral principal components of Shaoxing rice wine sample were extracted and analyzed,and the results showed the first three principal components were found to have relatively obvious clustering features.Then the different principal components were classified and optimized by LS-SVM model,and when the number of principal components reached 6,the identification accuracy was 100%,and the two parameters γ and σ2 of the model were 61.890 and 1.769,respectively.FTNIR spectral technology can be adopted as a reliable,accurate and quick method in the qualitative discriminative analysis of age of Shaoxing rice wine combining with PCA and LS-SVM.