PCA-LS-SVM预测模型在地沟油鉴别中的应用  

Identification of gutter oil based on the model of PCA-LS-SVM

作　　者：彭秀辉[1] 刘飞[1] 陈珺[1] 刘艳[1] 

机构地区：[1]江南大学轻工过程先进控制教育部重点实验室,自动化研究所,江苏无锡214122

出　　处：《计算机与应用化学》2013年第10期1207-1210,共4页Computers and Applied Chemistry

基　　金：2012年教育部国家创新创业训练计划(201210295058)

摘　　要：通过对油脂近红外吸光度的分析研究,提出了一种快速无损鉴别地沟油的新方法。首先运用主成分分析法对预处理后的各油脂光谱数据进行聚类分析并获取油脂的近红外指纹图谱,然后选用前4个主成分作为最小二乘支持向量机的输入变量,建立PCA-LS-SVM预测模型,实现地沟油的鉴别。实验共选用6种油脂,每种选用35个样本建模,20个样本验证,结果表明,近红外光谱技术结合PCA-LS-SVM方法能够定性有效地识别油脂的种类,为地沟油的鉴别开创了新前景。A new method is put forward to identify gutter oil rapidly through the study of near-infrared spectroscopy of oil. First, a clustering analysis on preprocessed spectroscopic is conducted with Principal Component Analysis （PCA）, aim to obtain the near-infrared fingerprints of oil. Then the first four principal components are selected as input variables of the Least Squares Support Vector Machine （LS-SVM）, based on the established PCA-LS-SVM prediction model. The gutter oil can be identified by the proposed technique. In the experiment, six kinds of oil are selected, and for each kind of oil, 35 samples are used to establish model while 20 samples are chosen to test and verify the model. The result shows that near-infrared spectroscopy combined with PCA-LS-SVM prediction model can identify different kinds of oil with high precision, which provides a new prospect for gutter oil identification.

关 键 词：近红外光谱 主成分分析 最小二乘支持向量机 地沟油 

分 类 号：O657.3[理学—分析化学]