基于遗传算法与神经网络的微波浸提过程优化

微波浸提是一个非线性、不确定因素较多的过程。本实验以荞麦皮粉总黄酮的萃取工艺为例,采用神经网络对微波浸提过程进行模拟,并用遗传算法进行工艺参数的优化,得出荞麦皮粉中总黄酮的微波萃取过程的最佳工艺为:液料比6.25、乙醇浓度54%、微波功率456W、作用时间11.3min、搅拌速率483r/min。在该工艺条件下做了验证实验,总黄酮提取得率为0.796%。并通过对比实验证实:在达到相同的提取率时微波浸提在溶剂用量、溶剂回收率等方面均优于传统萃取方法。

Optimization of Microwave Extraction Process by ANN and GA

The process of microwave extraction is a non-linear and undetermined behaviour. In this paper, with the extraction of flavonoids as example, a new method was build: the microwave extraction process was simulated by artificial neural network, and the technical parameters were optimized by genetic algorithms. After optimization, the best technology was the ratio 6. 25 between solvent and material, alcohol density of 54%, 456 microwave power, 11.3 minute extracting time, and stir-rate 483 r/min. Under these conditions, the actual output of flavonoids is 0.796%. Comparison experiment results confirmed that: with the objective of reaching the same rate of extraction, microwave method is better than conventional method on the dosage of impregnant and the recovery rate etc.