[Back-propagation neural network and genetic algorithm for multi-objective optimization of extraction technology of...

Area of Science:

• Pharmacognosy
• Computational Chemistry
• Chemical Engineering

Background:

• Cortex Fraxini is a traditional Chinese medicine with various therapeutic applications.
• Optimizing extraction processes is crucial for maximizing yield and purity of active compounds.
• Existing extraction methods may lack efficiency and multi-objective optimization capabilities.

Purpose of the Study:

• To apply Back-propagation (BP) neural networks and genetic algorithms for the multi-objective optimization of Cortex Fraxini extraction.
• To determine the optimal extraction parameters for Cortex Fraxini using computational methods.

Main Methods:

• BP neural network model was developed and optimized using uniform design.
• Genetic algorithm was employed for multi-objective optimization of the extraction process.
• Extraction parameters investigated included temperature, ethanol concentration, liquid-solid ratio, and extraction time.

Main Results:

• Optimal extraction conditions were identified as: 99°C extraction temperature, 50% ethanol concentration, a liquid-solid ratio of 7, and 94 minutes extraction time.
• The predictive model showed a low proportional error of -1.16% and -5.14% compared to practical measurements.
• The developed method demonstrated high accuracy in predicting optimal extraction parameters.

Conclusions:

• The combined approach of BP neural network and genetic algorithm is effective for multi-objective optimization of Cortex Fraxini extraction.
• This computational strategy offers a reliable and efficient method for optimizing traditional medicine extraction processes.
• The findings suggest this methodology is advisable for enhancing the extraction technology of Cortex Fraxini and potentially other botanical materials.