基于改进非劣分类遗传算法的多目标药物制备工艺优化分析

Optimization Analysis of Multi-objective Drug Preparation Process based on NSGA-Ⅱ

目的研究改进非劣分类遗传算法(nondominated sorting genetic algorithmⅡ,NSGA-Ⅱ)对葛根素亚微乳制备工艺的优化效果,并与多目标遗传算法(multiple objective genetic algorithm,MOGA)以及原文采用的效应面法的优化效果进行比较。方法利用制备葛根素亚微乳中心试验的数据,采用改进非劣分类遗传算法寻找平均粒径和跨距均最小、而包封率最大的工艺条件。结果采用NSGA-Ⅱ随机搜索30次,得到平均粒径为(220.173±18.153)nm;跨距的平均水平为(0.623±0.137)μm;包封率的平均水平为(83.873±2.176)%,与MOGA搜索结果的平均水平相比,NSGA-Ⅱ搜索精度更高,变异度更小。当乳化时间、搅拌转速、超声时间分别为12.97min、1613r·min-1、31.15min时,对应的目标值平均粒径、跨距、包封率分别为229.70nm、0.54μm、85.92%,优化效果满意,比MOGA及原文采用的效应面法找到的最优方案更理想。结论在确保多个目标值都达到最优的前提下,NSGA-Ⅱ搜索得到的Pareto非劣解是合理的,达到了较满意的效果,为试验设计最优条件的选择提供了合理的方法。同时研究人员可根据实际情况,从Pareto非劣解集中确定可行、合理且最优的工艺优化方案。

Objective To study the optimization effect of the Nondominated Sorting Genetic AlgorithmⅡ(NSGA-Ⅱ)on the preparation of puerarin submicron emulsion,the optimization results were compared with MOGA under multi-objective genetic algorithm and the three dimensional response surface graphs used in the original text.Methods Using NSGA-Ⅱmulti-objective genetic algorithm to find the optimal process conditions with the minimum mean diameter,minimum span of dispersity and the maximum entrapment efficiency based on the data of the central test for the preparation of puerarin submicron emulsion.Results NSGA-Ⅱmulti-objective genetic algorithm was randomly searched for 30 times,and the average level of mean diameter was(220.173±18.153)nm;the average level of span of dispersity is(0.623±0.137)μm;the average level of entrapment efficiency was(83.873±2.176)%.Compared with the average level of MOGA search results,NSGA-Ⅱmulti-objective genetic algorithm has higher search accuracy and less variability.When the emulsification time,stirring velocity and ultrasonic time were 12.97min,1613r·min-1 and 31.15min respectively,the mean diameter,span of dispersity and entrapment efficiency of the corresponding target values were 229.70nm,0.54μm and 85.92%respectively.The optimization effect is satisfactory,which is more ideal than the optimal scheme found by MOGA and the three dimensional response surface graphs used in the original text.Conclusion On the premise of ensuring that multiple target values are optimal,the Pareto non inferior solution obtained by NSGA-Ⅱsearch is reasonable and achieves satisfactory results,which provides a reasonable method for the selection of optimal conditions for experimental design.At the same time,researchers can determine the feasible,reasonable and optimal process optimization scheme from Pareto non-inferior solution set according to the actual situation.