RESUMO
Objective: Based on the central-composite design (CCD), the genetic neural network (GNN) and genetic algorithm (GA) were applied to optimize the microwave extraction conditions of astragalus saponins. Methods: The HPLC fingerprint of astragaloside was constructed, and seven components (astragaloside I—V, isoastragaloside I, II) were selected to calculate the comprehensive score by the entropy weight method. On the basis of single factor experiment, CCD was used to designed the experimental condition. The quantitative relationship between extraction conditions and comprehensive score was established by GNN, and the optimal microwave extraction parameters of astragalus saponins were optimized by GA. Results: The optimal extraction conditions were obtained by GA-GNN. The extraction time was 260 s, the extraction power was 695 W, the ethanol content was 50%, the ratio of material to liquid was 21.5, and the comprehensive score of seven astragalosides was 1 432.584. Meanwhile, the optimal extraction conditions and comprehensive evaluation scores obtained were by response surface methodology (RSM). The extraction time was 190 s, the extraction power was 880 W, the ethanol content was 70%, the ratio of material to liquid was 18.5, and the comprehensive scores of seven astragaloside were 1 066.236. The experimental results showed that the extraction conditions obtained by GA-GNN can effectively increase the comprehensive score. Conclusion: It is feasible to construct a mathematical model between astragaloside components and microwave extraction conditions by using entropy weight method combined with GNN, which can provide a new scientific method for optimizing the extraction, separation, and purification of effective components of traditional Chinese medicine.
