Optimization of extraction of dregs of Saussurea involucrata after ethanol extraction based on Box-Behnken response surface methodology and BP neural network
Objective To optimize the extraction of dregs of Saussurea involucrata after ethanol extraction.Methods Based on the content of chlorogenic acid,rutin and polysaccharides in the extract,the analytic hierarchy process(AHP)was used to assign weights to obtain a comprehensive score.The extraction for the chemical constituents of dregs was optimized with a composite score as a preference index.Material liquid ratio,extraction time,and extraction times served as factors.To accomplish this,the Box-Behnken response surface design and BP neural network were used.Results The extraction conditions were optimized by the Box-Behnken response surface methodol as follows:reflux extraction being conducted with 9 times of water for 42 min,repeating 3 times.Seventeen sets of data were selected from the response surface test for training and validation.The BP neural network predicted the preferred extraction to be adding six times of water to reflux the extraction for 3 times,followed by decoction for 30 min each time.The validation test demonstrated that the BP neural network optimization achieved a higher actual comprehensive score than that of the response surface method,indicating a more reasonable extraction progress by the BP neural network.Conclusion The extraction progress by the BP neural network outlined in this paper is reasonable,stable and viable.It serves as a useful reference for the secondary development and utilisation of dregs of Saussurea involucrata.
NETL
NSTL
万方数据
