Simultaneous Multiresponse Optimization of the Medium for Submerged Fermenting Cordyceps Gunnii Mycelia Using Genetic Algorithm

An experimental mixture design coupled with data analysis by means of genetic algorithm-artificial neural network (GA-ANN) was applied to optimize the fermentation medium of Cordyceps gunnii Mycelia for enhancing the yields of the intracellular polysaccharide. With the yield rate of intracellular polysaccharide as index, a sequential statistical strategy was investigated during this optimization process, which consisted of Plackett-Burman design (PBD), Box-Behnken design (BBD), Multi-quadratic regression (MQR), artificial neuron networks (ANN) and genetic algorithm (GA). PBD combined with linear modeling method was used for identifying the significant components, BBD was used for further optimization. MQR and ANN were used for modeling the BBD data. While the ANN model was developed, genetic algorithm (GA) was employed to search for the optimum medium which was as follow (g/L): lactose 29.81, beef extract 17.28, KH2PO4¡·5H2O 3.0, MgSO4¡·7H2O 2.0, NaCl 0.003 and VB1 0.271, with expected maximum yield rate of 0.1.213. Under the optimal conditions, the corresponding response value predicted for the intracellular polysaccharide yield rate was 1.201 g×g-1, which was confirmed by validation experiments.