SVM回归法在佛手精油β-环糊精包合工艺优化中的应用  

Optimization of Microcapsule Preparation Technology of Bergamot Essential Oil and β-Cyclodextrin Based on Support Vector Machines

作　　者：邓刚[1] 许杭琳[1] 焦聪聪[1] 朱程华[1] 

机构地区：[1]浙江师范大学化学与生命科学学院,金华321004

出　　处：《中国粮油学报》2012年第4期59-63,共5页Journal of the Chinese Cereals and Oils Association

基　　金：国家自然科学基金(21006098)

摘　　要：采用了基于统计学习理论的支持向量机(SVM)回归方法对佛手精油β-环糊精包含物制备的工艺条件进行了预测及优化。通过正交试验为构建SVM回归模型提供了基础数据,并运用交叉验证对模型参数进行了优化。回归结果显示,回收率SVM模型(MSE=0.003,R2=0.958 1)和包埋率SVM模型(MSE=0.007,R2=0.900 8)得到的回归预测值和试验测定值拟合良好,两者相对误差小于0.1%的回归值比例分别为100%和88.0%,表明SVM回归法可精确地预测挥发油β-环糊精包含物制备工艺状况,从而获取更为可靠的最优化工艺条件。Optimization of microcapsule preparation technology of bergamot essential oil and β-cyclodextrin based on support vector machines was studied in the paper.The SVM prediction model was constructed depending on orthogonal experimental data,which parameters were optimized by cross-validation statistical method.SVM analysis results showed that the data percentage of relative error below 0.1% of product yield-SVM model（MSE=0.003,R2=0.958 1）and embedding rate-SVM model（MSE=0.007,R2=0.900 8）are 100.00% and 88.00%,respectively.The SVM prediction agreed so well with the experimental data that constructed SVM regression model could predict microcapsule preparation process accurately,resulting in more reliable optimal production.

关 键 词：支持向量机回归 佛手精油 Β-环糊精 

分 类 号：TS202.3[轻工技术与工程—食品科学]