BP神经网络结合遗传算法优化黄芪多糖提取工艺

Extraction optimization of astragalus polysaccharides with back-propagation neural networks and genetic algorithm

目的:通过BP神经网络结合遗传算法优化黄芪多糖的提取工艺。方法:以正交试验结果作为初始种群,以液料比、提取时间、提取温度为输入值,以黄芪多糖提取率作为其函数的输出值,通过BP神经网络模型来调试函数的适应度后,再通过遗传算法对黄芪多糖的水提工艺进行全局寻优,获取最佳工艺。结果:黄芪多糖提取工艺的最佳条件为液料比15∶1、提取时间89 min、提取温度100℃。在此条件下,提取率达到4.93%。结论:BP神经网络联合GA的模型,能较好地优化黄芪多糖的提取工艺,为解决非线性模型的工艺优化问题提供了新思路。

OBJECTIVE To optimize the extraction of astragalus polysaccharides with back-propagation neural networks and genetic algorithm（ GA）. METHODS Based on the orthogonal test as the initial population,the liquid-solid ratio,extraction time and extraction temperature as the input values,extraction rate of the astragalus polysaccharides as the output value,the model of BP neural networks was investigated to debug function of fitness,then optimize the water extraction process of astragalus polysaccharides by the genetic algorithms and obtain the optimum extracting parameters. RESULTS The optimal protocol for astragalus polysaccharides extraction was at 100 ℃ with a liquid-to-solid ratio of 15∶ 1 in 89 min,when the extraction yield reached 4. 93%. CONCLUSION The extraction of astragalus polysaccharide can be optimized with BP neural networks and GA model,and the model provides a new way to solve the problem in nonlinear problem.