核磁共振氢谱结合PCA-SVM算法分类鉴别食用植物油
Classification of edible vegetable oils based on 1H-NMR spectroscopy and PCA-SVM
-
摘要: 采用核磁共振氢谱(1H-NMR)结合主成分分析-支持向量机法(PCA-SVM)对7种市面上常见的食用植物油进行了分类研究。首先运用PCA法对预处理后的各食用植物油的1H-NMR谱图积分数据进行降维处理,然后选用前2个主成分作为SVM的输入变量,建立预测模型,再对测试集样品进行预测,以实现食用植物油的种类鉴别,并与簇类独立软模式法(SIMCA)的分类结果进行了比较。结果显示:采用网格划分法优化得到模型最优核函数参数值为1.7411,最优惩罚参数值为0.3299,以最优参数建立的PCA-SVM食用植物油分类模型对测试集的分类正确率为100%,高于SIMCA分类法的61.90%。建立的1H-NMR结合PCA-SVM法食用植物油分类模型,可以快速、有效的鉴别食用植物油种类,适合实际食品检测工作中建模样本有限的实际,为食用植物油的品质鉴别和质量控制提供分析方法。Abstract: To establish a method for the classification of edible oils by 1H-NMR spectroscopy and PCA-SVM and to compare its effectiveness with that of SIMCA. First,the PCA method was used to reduce the dimensionality of independent variables. Then the first two principal components were selected as input variables of the support vector machine(SVM),based on the established PCA-SVM prediction model. The seven kinds of oils could be identified by the proposed technique. The results revealed that the value of g and c were 1.7411 and 0.3299,respectively,which were optimized by grid method. The accuracy of prediction could reach to 100% with the PCA-SVM model,while that was only 61.90% with SIMCA model. It was validated by results that the combination of 1H-NMR spectroscopy with PCA-SVM could achieve the classification of edible oils quickly and effectively.
下载: