R语言优化温郁金油的羟丙基-β-环糊精包合工艺

Optimization of hydroxypropyl-β-cyclodextrin of Radix Curcumae oil inclusion process by R language

目的:通过R语言结合BP神经网络和遗传算法优化温郁金挥发油的羟丙基-β-环糊精(HPCD)包合工艺。方法:以挥发油包合率、包合物收得率、含油率为考察指标,运用响应面设计试验,采用溶液搅拌法-真空冷冻干燥等技术,制备温郁金油HPCD包合物。在R语言环境下建立数学模型,编程建立BP神经网络,训练并优化网络模型,再运用遗传算法,进行网络目标寻优,得到包合工艺的最佳条件。结果:温郁金HPCD包合工艺最佳条件为HPCD与温郁金油用量比10∶1,包合温度30℃,包合时间2h,模型综合评价预测值为43.58,与实验所得平均综合评价值43.81相比,相对误差为0.53%,说明网络具有良好的预测性。结论:该数学模型简单、客观、准确,为温郁金油包合工艺优化提供了新的参考。

Objective： To optimize the hydroxypropyl-β-cyclodextrin of Radix Curcumae oil inclusion process by R language combined with BP neural network and genetic algorithm. Methods： With the the volatile oil packing rate, inclusion yield and inclusion oil content as the indexes, the response surface methodololgy（RSM） design was used to design the experiment. The solution stirring method and vacuum freeze drying technology were used to prepare the inclusion compound of hydroxypropyl-β-cyclodextrin of Radix Curcumae oil. With the R language programming environment, the mathematical model was established to programme a BP neural network. Then, to optimize the inclusion process, the BP neural network was trained and optimized with genetic algorithm. Results： The optimum conditions for the inclusion were as follows： the ratio of HPCD to Radix Curcumae oil was 10∶1; the inclusion temperature was 30℃; the inclusion time was 2h, and the predictive value of model comprehensive evaluation was 43.58. The average comprehensive evaluation value was 43.81 from the experimental study. The relative error between the model and the experiments reached 0.53%. Therefore, it domenstrated that network had a good predictability. Conclusion： With its simple, objective and accurate advantages, the mathematical model could provide a new reference for optimizing the the hydroxypropyl-β-cyclodextrin of Radix Curcumae oil inclusion process.