Online identification and optimization of feed rate profiles for high productivity fed-batch culture of hybridoma cells using genetic algorithms

An online identification and optimization method, based on a series of real-valued genetic algorithms (GAs), is studied for a seventh-order nonlinear model of fed-batch culture of hybridoma cells. The parameters of the model are assumed to be unknown. The online procedure is divided into three stages: 1) GAs are used for identifying the unknown parameters of the model; 2) the best feed rate control profiles of glucose and glutamine are found by GA based on the estimated parameters; and 3) the fermentation is driven by these best feed rate control profiles. The final level of monoclonal antibodies obtained by this method is then compared with the case where all the parameters are assumed to be known. It is found that the final level of monoclonal antibodies obtained by the online identification and optimization method is only about 3% less than the final level of monoclonal antibodies obtained by the case where all the parameters are assumed to be known. The real-valued genetic algorithms proved to be a good alternative method for solving online identification and optimization problems