基于SVM对不同品牌阿胶的聚类分析

摘要

提出利用支持向量机(SVM)对不同品牌阿胶进行鉴别,首先对不同品牌阿胶的红外光谱图进行分析,同时系统分析了阿胶谱图的预处理过程,建立3种品牌阿胶的鉴别模型.并对4种不同的聚类分析算法(KNN、Bayes、LDF和SVM)进行方法比对.试验表明,不同品牌的阿胶聚类模型以SVM的识别率和拒绝率最高.试验结果表明该方法能快速对不同品牌阿胶进行鉴别.%An identifying method for different brands of Ejiao by using support vector machine (SVM) was in-troduced. First, SVM analyzed the infrared spectrums of different brands of Ejiao, researching the pre-process-ing of Ejiao spectrums systematically at the same time and the discrimination models of three brands of Ejiao were established. Method comparison was used to discriminate the four different clustering analysis algorithms (KNN, Bayes, LDF and SVM). The experiment proved that the recognition rate and rejection rate of SVM were the highest in the cluster models of different brands of Ejiao. The results showed that this method could identify different types of Ejiao effectively.