基于小生境遗传算法的多目标药物提取条件优化分析应用

Multi-objective Optimize Based on Niche Pareto Genetic Algorithms—Uniform Multi-Objective Optimization of Extraction Conditions of Drug Application

目的研究小生境遗传算法在均匀试验设计多目标药物提取条件优化中的应用。方法对微萃取五味子的均匀试验数据建立以浸膏得率、五味子醇甲、五味子总木脂素的子目标模型,采用遗传算法分别对其进行单目标优化,NPGA对其进行多目标优化,搜索最优提取条件,比较搜索结果;利用课题组成员英国Glasgow大学软件工程师陈益编写的Matlab2009a外挂SGALAB工具箱beta5008完成遗传算法寻优。结果单目标遗传算法优化可以得到各目标最大时的最优提取条件,NPGA进行三目标优化时,对各子目标进行了折衷处理使各子目标尽可能获得最大的解,在主要目标上达到了单目标最大函数值的76%以上,确定的最优提取条件的效果高于均匀试验中的任何一个方案。结论 NPGA搜索的Pareto非劣解是合理的,达到了较好的效果,为均匀试验设计最优条件选择提供了合理的方法,可推广到正交试验设计、析因试验设计的最优条件选择。

Objective：Study the application of multi-objective optimization analysis of Niched Pareto Genetic Algorithm in drug extraction.Methods：Using micro-extraction schisandra data in uniform design establish three-objective function.Applying simple genetic algorithm and NPGA explore the optimal extracting conditions.Compare their optimal extracting conditions.Using SGALAB beta5008 of the Matlab2009a plug-in tool-box,which was written by Chen Yi in the United Kingdom University of Glasgow,achieves the genetic algorithm optimization.Results：Single-objective genetic algorithm optimization can obtain the optimal extraction conditions of each objective.NPGA for three-objective optimization,balance all the sub-objectives to obtain the greatest solution,which can achieve 76% of the maximum value of single objective function in the main goals,thus we can get optimal extraction conditions,the effect of which is best.Conclusion：NPGA provides reasonable pareto optimal solutions.,It is a reasonable method for selecting optimal conditions for Uniform Experimental Design.This method can be extended to the selection of the optimal conditions in the orthogonal experimental design and factorial design.