基于最小二乘支持向量机的秦皮提取工艺的研究

目的：应用最小二乘支持向量机（LS-SVM）技术优化秦皮的提取工艺。方法：以秦皮甲素和秦皮乙素为指标，采用均匀设计安排提取试验，并用最小二乘支持向量机建立关系模型。结果：最小二乘支持向量机对秦皮甲素和秦皮乙素的拟和相关系数分别为0．9997与0．9999；得到的最优工艺条件为提取温度100℃、乙醇浓度50％、液固比11、提取时间70min，机器模型在此条件下的预测值为秦皮甲素提取量为9．291mg·g^-1，秦皮乙素提取量为2．241mg·g^-1，和实际测量值的相对误差仅为-2．97％和2．66％，具有较好的预测性。结论：最小二乘支持向量机可用优化秦皮提取工艺。

Application of Least Squares Support Vector Machine for Optimization of Extraction Technology of Cortex Fraxini

Objective： To make use of least squares support vector machine （LS-SVM） for optimization of extraction technology of cortex fraxini. Method： Least squares support vector machine was employed to establish relational model of extraction technology of cortex fraxini with uniform design. Result： The coefficient correlations of aesculin and aesculetin were 0.999 7 and 0.999 9 respectively after least squares support vector machine trained. The optimization of extraction was as follows： extraction temperature was 100℃, concentration of alcohol was 50%, liquid-sohd ratio was 11, extraction time was 70min. The proportional error between predictive value and practical measured value was just -2.97% and 2.66%. Conclusion： Least squares support vector machine can be used for optimization of extraction technology of cortex fraxini.