Modelling and optimization of fed-batch fermentation processes using dynamic neural networks and genetic algorithms

Optimization of a fed-batch bioreactor using a cascade recurrent neural network (RNN) model and modified genetic algorithm (GA) is studied in this paper. The complex nonlinear relationship between the manipulated feed rate and the biomass product is described by two recurrent neural sub-models, in which outputs of one sub-model are fed into another sub-model to provide meaningful information for the biomass prediction. The simulation results show that the error of prediction is less than 8 %. Based on the neural network model, a modified GA is employed to determine a smooth optimal feed rate. The evolution of feed rate profiles shows that the algorithm is able to generate a smooth feed rate profile, where the optimality is still maintained. The final biomass quantity that yields from the optimal feed rate profile based on the neural network model reaches 99.8 % of the "real" optimal value obtained based on the mechanistic model. An experimental investigation has also been carried out to verify the feasibility of the proposed technique. (C) 2004 Elsevier B.V. All rights reserved.