基于ACO-SVM的光谱分类算法研究

为了实现胡萝卜汁品牌的快速无损鉴别,文章以市售两种品牌的胡萝卜汁为研究对象,通过采集拉曼光谱,并结合支持向量机算法,建立了快速判断胡萝卜汁品牌的分类模型。两种品牌的胡萝卜汁光谱主要在1007 cm^-1,1157 cm^-1,1516 cm^-1这3个谱峰存在差异。先对拉曼光谱进行预处理,再选用蚁群优化算法进行特征选择,最后结合SVM构建分类模型。结果显示,相比SVM直接分类,模型ACO-SVM的最优分类准确率为96.67%,提高了2.5%;其分类时间为7.17 s,缩短了163.31 s。研究表明,基于拉曼光谱分析技术和模式识别算法构建的分类模型能够有效地鉴别胡萝卜汁品牌。

Research on spectrum classification algorithm based on ACO-SVM

In order to achieve rapid and non-destructive identification of carrot juice brands,two types of commercially available carrot juice brands were used as research objects in this paper.A classification model for quickly determining carrot juice brands was established by collecting Raman spectra and using support vector machine algorithm.The carrot juice spectra of two brands are mainly different in the three peaks of 1007 m^-1,1157 cm^-1,and 1516 cm^-1.Firstly,preprocessing the Raman spectra,then the ant colony optimization algorithm were used for feature selection,finally,the classification model was built with SVM.The results show that compared with direct classification by SVM,the optimal classification accuracy of the model ACO-SVM is 96.67%,which is an improvement of 2.5%;the classification time is 7.17 s,which is shortened by 163.31 s.Research shows that a classification model based on Raman spectroscopy and pattern recognition algorithms can effectively identify carrot juice brands.