主 办:北 京 中 医 药 大 学
ISSN 1006-2157 CN 11-3574/R

JOURNAL OF BEIJIGN UNIVERSITY OF TRADITIONAL CHINE ›› 2017, Vol. 40 ›› Issue (4): 334-338.doi: 10.3969/j.issn.1006-2157.2017.04.013

Previous Articles     Next Articles

Multi-level optimal extraction technique of licorice using R language application*

YU Li1, JIN Weifeng2, LI Min1, FAN Hongjing1, LI Xiaohong2, ZHANG Yuyan1#   

  1. 1 College of Life Science, Zhejiang Chinese Medical University, Zhejiang 310053;
    2 College of Pharmaceutical Science, Zhejiang Chinese Medical University
  • Received:2016-10-27 Online:2017-04-30 Published:2017-04-30

Abstract: Objective To optimize the simultaneous extracting technique of saponins and total flavonoids from licorice (gancao).Methods Ammonia concentration (A), ethanol concentration (B), reflux time(C), and liquid/solid ratio(D) were set as the independent variables in this single factor experiment. Four factors and five levels of central composite design (CCD) in response surface methods were used to determine the content of saponins and total flavonoids in licorice. This study used ultraviolet spectrophotometric method to measure saponins and total flavonoids in licorice at the wave length of 252 nm and 510 nm respectively. The entropy weight method in the R language application was used to assign weight to the above two parameters. The three-layered model of BP neural network was established to test the effect of the number of hidden neurons (size). Finally, genetic algorithm was established to optimize the extraction techniques with real-coded program of R language. Results This method achieved the objective of testing requirements.There was a good linear relationship between saponins at 0.008~0.056 g/L, total flavonoids at 0.024~0.08 g/L, and light absorbance. This method set the neural network model with five hidden layer neurons. After optimizing the parameters of genetic algorithm, the extraction process of saponins and total flavonoids from licorice was optimized. The final optimal parameters were 0.62% ammonia, 64% ethanol, 1.8 h reflux time, and 12:1 of liquid-solid ratio. In this optimal extraction condition, predictive value of this model was 191.65, and experimental average value was 188.90. The relative error was 1.43%, which demonstrated a good predictability of the neural network model and genetic algorithm. Conclusion This mathematical model to optimize the extraction techniques of saponins and total flavonoids from licorice is scientific and feasible. It also provids an innovative reference and approach to the multi-objective extraction techniques for identifying chemical compound and active ingredients of traditional Chinese medicine.

Key words: R language, CCD, liquorice, extraction technology, optimization

CLC Number: 

  • R283