Modelling of a Fed-Batch Culture Applying Simulated Annealing

In this paper the metaheuristic Simulated Annealing (SA) is applied for parameter identification of non-linear model of cultivation process. SA algorithm is a stochastic relaxation technique, using the Metropolis algorithm based on the Boltzmann distribution in statistical mechanics, for solving nonconvex optimization problems. A real E. coli MC4110 fed-batch cultivation process is considered. The mathematical model is presented by a system of five ordinary differential equations, describing the regarded cultivation process variables - biomass, substrate, acetate, dissolved oxygen and bioreactor volume increasing. The obtained criteria values show that the developed model is adequate and has a high degree of accuracy. The presented results are a confirmation of successful application of the SA algorithm and of the choice of SA algorithm parameters