基于KPCA和LSSVM的蜂蜜近红外光谱鉴别分析  

Identification of honey by NIR spectroscopy technology based on KPCA and LSSVM

作　　者：谈爱玲[1] 毕卫红[1] 

机构地区：[1]燕山大学信息科学与工程学院光电子工程系,河北秦皇岛066004

出　　处：《激光与红外》2011年第12期1331-1336,共6页Laser & Infrared

基　　金：教育部高等学校科技创新工程重大项目培育基金(No.708025);河北省自然科学基金(No.F2010001268);秦皇岛市科学技术研究与发展计划(No.200901A032)资助

摘　　要：为快速鉴别普通植物源与中草药植物源蜂蜜,提出一种核主成分析和最小二乘支持向量机相结合的蜂蜜近红外光谱定性分析新方法。利用傅里叶变换近红外光谱仪测定普通洋槐蜂蜜和益母草、黄连两种中药植物源蜂蜜样本的近红外光谱并预处理,然后对光谱进行核主成分分析,提取非线性特征,最后设计基于纠错编码最小二乘支持向量机的多类分类器模型。采用网格搜索法确定模型最优参数,利用最优分类模型对未知类别蜂蜜样本进行识别,正确率可达96.67%。结果表明,基于KPCA和LSSVM的近红外光谱定性分析算法鉴别普通植物源与中草药植物源蜂蜜是可行的。For the rapid identification of common plant honey and Chinese medicine nectar plant honey, a novel quali- tative identification method combined Kernel Principal Component Analysis （KPCA）and Least Square Support Vector Machine （LSSVM）is proposed. The presented method uses Fourier transform NIR spectrophotometer to collect the spectral data of locust, leonuri and coptis honey. The KPCA algorithm is used to extract nonlinear features. Then a LSSVM classificaion model based on Error Correcting Output Code （ECOC）is designed and grid search method is used to determine the optimal model parameters. To unknown honey samples, the optimal model has the best identification capability with a accuracy of 96.67%. Experimental results indicate that the proposed qualitative analysis method based on KPCA and LSSVM can distinguish the honey of common and Chinese medicine nectar plant.

关 键 词：近红外光谱 核主成分分析 最小二乘支持向量机 蜂蜜 中草药植物源 

分 类 号：O657.3[理学—分析化学]