论文部分内容阅读
该文采用中心复合设计(central composite design),以醇沉后的糖去除率、五味子醇甲保留率及可溶性固形物含量为指标考察醇沉乙醇加入量、乙醇浓度、冷藏温度及冷藏时间4个醇沉工艺参数对于醇沉效果的影响。运用贝叶斯网络分析发现乙醇加入量和乙醇浓度为2个重要的醇沉工艺参数,然后运用遗传算法优化的BP人工神经网络建立2输入4输出的网络模型,所得训练集回归模型R~2=0.983 8,MSE=0.001 1,验证集回归模型R~2=0.975 9,MSE=0.001 8,模型拟合精度和预测效果均比较理想。研究表明该方法可有效地用于五味子醇沉过程关键工艺参数辨析与过程建模。
In this paper, central composite design was used to investigate the effects of ethanol-ethanol, ethanol concentration, refrigeration temperature and refrigerated storage time on alcohol-precipitated sugar removal rate, schisandrin retention rate and soluble solids content Effect of alcohol deposition parameters on alcohol precipitation. Using Bayesian network analysis, ethanol addition and ethanol concentration were found to be two important parameters of ethanol precipitation process. Then, BP neural network optimized by genetic algorithm was used to establish 2-input 4-output network model. The training set regression model R ~ 2 = 0.983 8, MSE = 0.001 1. The validation set regression model R 2 = 0.975 9, MSE = 0.001 8, the model fitting accuracy and prediction effect are ideal. The results show that the method can be effectively used in the key process parameters of Schisandra alcohol precipitation process and process modeling.