Modified Simple Genetic Algorithms Improving Convergence Time for the Purposes of Fermentation Process Parameter Identification

Fermentation processes are characterized with non-linear and time-dependent parameters that make their parameter identification non-trivial task. Failure of conventional optimization methods to yield a satisfactory solution provokes the idea for some stochastic algorithms to be applied. As such, different modifications of simple genetic algorithms (SGA) have been investigated aiming to improve the model accuracy and the algorithm convergence time. For that purpose two new modifications of SGA are developed here. SGA realizations differ from each other in the sequence of implementation of the main genetic operators selection, crossover and mutation. A comparison of the herewith developed two modifications of SGA and standard SGA towards algorithm convergence time and model accuracy is presented for parameter identification of S. cerevisiae fed-batch cultivation. The influence of the most important genetic algorithm parameters, namely generation gap, crossover and mutation rates has been investigated, too. Both proposed modifications of SGA produce similar values of the optimization criterion, meanwhile being significantly faster than the standard SGA. Among the considered genetic algorithms parameters, generation gap influences the algorithm calculation time most significantly, saving up to 53% of the time without affecting the model accuracy. Key-Words: Genetic algorithms, genetic operators, genetic algorithm parameters, parameter identification, fed-batch fermentation process.