摘要
目的评价LM-BP神经网络模型优化酶催化合成维生素C酯化物条件效果。方法使用Levenberg-Mar-quardt改进BP算法计算了酶催化合成维生素C乳酸酯实验数据,通过LM-BP ANN模型关联预测酶量、温度、底物浓度、反应时间因素对维生素C乳酸酯产率的影响,并且将计算结果与标准多元回归方法比较。结果LM-BPANN模型拟合程度好,预测精度高,优于标准多元回归方法。结论LM-BP ANN可以为酶催化合成维生素C酯化物条件寻优,可有效改善响应曲面分析方法。此方法也可用于其他酶催化实验条件分析。
Objective To evaluate the application of LM-BP ANN for the optimization of lipase-catalyzed synthesis of vitamin C ester.Methods The improved ANN based on Levenberg-Maquardt algorithm was used to calculate the dataset of lipase-catalyzed synthesis of vitamin C lactate and to evaluate the interactive effect of temperature,amount of enzyme,substrate concentration（vitamin C） and reaction time on the percentage yield of vitamin C lactate.Results This approach can forecast the interactive effect of experiment condition on the percentage yield of vitamin C lactate.Compared with multiple regression equation,LM-BP had fewer errors.This method improved response surface methodology.Conclusion LM-BP ANN is useful in the optimization of lipase-catalyzed synthesis of vitamin C ester.
出处
《中南药学》
CAS
2009年第12期884-889,共6页
Central South Pharmacy
