On the development of Intelligent Control System for Recombinant Cell Culture

摘要

Two neural network estimators were developed to identify residual gluocose and galactose concentrations in the fed-batch cultivation of recombinant Saccharomyces cerevisiae for overexpression of heterologous protein. Two multi-variable fuzzy controllers were then used to control the glucose and galactose feed rates during the growth and production phases, respectively. Experimental data shown that under fuzzy control with neural network estimators, the productivity of recombinant protein could be increased by 2.7-fold compared to that obtained in the cultivation under the exponential feeding policy.