ABSTRACT
OBJECTIVE: To explore the feasibility of genetic algorithm (GA) on multiple objective blending technology for extractions of Cortex Fraxini. METHOD: According to that the optimization objective was the combination of fingerprint similarity and the root-mean-square error of multiple key constituents, a new multiple objective optimization model of 10 batches extractions of Cortex Fraxini was built. The blending coefficient was obtained by genetic algorithm. The quality of 10 batches extractions of Cortex Fraxini that after blending was evaluated with the finger print similarity and root-mean-square error as indexes. RESULT: The quality of 10 batches extractions of Cortex Fraxini that after blending was well improved. Comparing with the fingerprint of the control sample, the similarity was up, but the degree of variation is down. The relative deviation of the key constituents was less than 10%. CONCLUSION: It is proved that genetic algorithm works well on multiple objective blending technology for extractions of Cortex Fraxini. This method can be a reference to control the quality of extractions of Cortex Fraxini. Genetic algorithm in blending technology for extractions of Chinese medicines is advisable.
Subject(s)
Drugs, Chinese Herbal/analysis , Plants, Medicinal/chemistry , Aesculus , Algorithms , Chromatography, High Pressure Liquid , Medicine, Chinese Traditional/standards , Plants, Medicinal/genetics , Quality ControlABSTRACT
OBJECTIVE: To introduce Back-propagation (BP) neural network and genetic algorithm for multi-objective optimization of extraction technology of Cortex Fraxini. METHOD: BP neural network was established and optimized with uniform design. Genetic algotithm was used for multi-objective optimization of extraction technology of cortex fraxini. RESULT: the optimization of extraction was as follows: extraction temperature was 99 degrees C, concentration of EtOH was 50%, liquid-solid ratio was 7, extraction time was 94 min. The proportional error between predictive value and practical measured value was just -1.16% and -5.14%. CONCLUSION: Back-propagation neural network and genetic algorithm for multi-objective optimization of extraction technology of cortex fraxini is advisable.