摘要
通过支持向量机数学模型预测了绞股蓝茶浸提液中的总酚含量、总黄酮含量及其自由基清除能力,其中浸提条件包括11个浸提温度、5个浸提时间及3个料液比.在4个涉及的支持向量机基本核函数中,径向基核函数表现出最佳的预测效果.在径向基核函数下,用3种参数优化方法:网格参数优化法、遗传优化算法和粒子群优化算法,分别对其模型进行了参数优化.通过比较发现,用粒子群优化算法的径向基核函数支持向量机模型达到了最佳的预测效果,其测试集的相关系数分别达到了0.962 8(预测总酚含量),0.979 7(预测总黄酮含量)和0.951 3(预测自由基清除能力).
The total phenols, flavonoids, and antioxidant activity of Gynostemma pentaphyllurn water extracts were predicted by using support vector machine （ SVM ）. The extraction conditions included 11 incubation tem-peratures, five incubation times and three different materials to water ratios. Among the four basic kernel func-tions, namely, linear, polynomial, RBF and sigmoid, used in the study, the RBF had the best performance. Three parameter optimization methods, namely, grid search, genetic algorithm, and particle swarm optimiza- tion, were investigated. The results showed that the RBF SVM optimized by particle swarm optimization pro-duced the best correlation coefficients for the testing sets （ 0.962 8 for total phenol content, 0. 979 7 for total flavonoids content, and 0.951 3 for DPPH radical scavenging activity ）.
出处
《浙江师范大学学报（自然科学版）》
CAS
2013年第4期450-458,共9页
Journal of Zhejiang Normal University:Natural Sciences
基金
金华市科技项目(2010-3-078
2010-3-079)
关键词
绞股蓝
支持向量机
总酚
总黄酮
自由基清除能力
Gynostemma pentaphyllum
support vector machine （ SVM ）
phenols
flavonoids
scavenging ac-tivity
