| 基于改进遗传神经网络的紫花苜蓿总黄酮提取工艺参数优化 |
| |
| 引用本文： | 朱会霞,李彤煜,刘凤超,马巍,王辉暖.基于改进遗传神经网络的紫花苜蓿总黄酮提取工艺参数优化[J].饲料研究,2020,43(3):61-64. |
| |
| 作者姓名： | 朱会霞 李彤煜 刘凤超 马巍 王辉暖 |
| |
| 作者单位： | 辽宁工业大学管理学院,辽宁锦州121001;锦州医科大学畜牧兽医学院,辽宁锦州121001 |
| |
| 基金项目： | 国家自然科学基金项目(项目编号:31601914);辽宁省自然科学基金(项目编号:2019-ZD-0804);辽宁省社会科学规划基金项目(项目编号:L18DGL008);辽宁省教育厅高校基本科研项目(项目编号:JQW201715407)。 |
| |
| 摘 要： | 针对回归分析方法优化紫花苜蓿总黄酮提取工艺参数存在的问题,提出一种改进的遗传神经网络优化方法,该方法利用区间自适应遗传算法自动调整搜索区间的特点,彻底取代神经网络反向传播过程调整权值和阈值。以紫花苜蓿总黄酮提取率为性能指标,选取液料比、提取时间、提取温度和乙醇浓度4个工艺参数为试验因素,依据4因素5水平正交旋转组合试验数据建立学习样本,训练遗传神经网络。当神经网络训练误差达到4.34×10^-9时,遗传神经网络算法的平均相对误差为0%,而回归模型的平均相对误差为0.58%,拟合精度和拟合优度明显优于回归模型。用训练好的遗传神经网络算法优化紫花苜蓿总黄酮提取工艺参数,获得最优工艺参数为:取液料比53.26 mL/g、提取温度70.96℃、提取时间50.32 min、乙醇浓度为60%,紫花苜蓿总黄酮提取率达到最大值6.51 mg/g。
|
| 关 键 词： | 紫花苜蓿 总黄酮 遗传神经网络 参数优化 |
Optimization of extraction process parameters of total flavonoids from Medicago Sativa L. based on improved genetic neural network |
| |
| Abstract: | The fit accuracy and accuracy of the process parameters are not well by regression analysis in optimizing the extraction process parameters of total flavonoids from medicago sativa. A high-precision genetic neural network optimization method is proposed in which the interval adaptive genetic algorithm is used to automatically adjust search interval, and completely replace the adjustment of the weight threshold in in the back propagation of neural networks. Adaptively adjust the interval of the neural network weight threshold. The extraction rate of total flavonoids from Medicago sativa was used as criterion, the four process parameters of ratio of liquid to solid, extraction temperature, extraction time, and ethanol concentration were selected as experimental factors. We used the data of four-factor five-level orthogonal rotation combination test as the learning samples to train the genetic neural network. When the error of the neural network was 4.34×10^-9, the average relative error of the genetic neural network algorithm was 0%, but the average relative error of the regression model was 0.58%. The fitting accuracy and goodness of fit of the genetic neural network algorithm are significantly better than that of the quadratic regression method. If the trained genetic neural network algorithm is used to optimize the extraction parameters of total flavonoids from medicago sativa, the optimal parameters are obtained. When ratio of liquid to solid is 53.26 mL/g, extraction temperature is 70.96 ℃, extraction time is 50.32 min, and ethanol concentration is 60%, the extraction rate of total flavonoids reaches the maximum 6.51 mg/g. |
| |
| Keywords: | Medicago Sativa L flavonoids genetic neural network parameter optimization |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
