Comparison of different metaheuristic algorithms based on InterCriteria analysis

Abstract

In this paper InterCriteria analysis (ICrA), based on the apparatus of the Index Matrices and the Intuitionistic Fuzzy Sets, is performed for a model parameters identification using different pure and hybrid metaheuristic techniques. As a case study a non-linear E. coli MC4110 fed-batch cultivation process model is considered. Series of cultivation model identification procedures using metaheuristics as genetic algorithms (GA), ant colony optimization (ACO), firefly algorithm (FA) and simulated annealing (SA) are done. The results are compared with the once obtained by applied hybrid algorithms ACO–GA, ACO–FA and GA–ACO. Further, the ICrA is used to explore the existing relations and dependences of defined cultivation model parameters, namely μ m a x , k S and Y X ∕ S , and considered metaheuristic algorithms outcomes, e.g. computation time T and objective function value J . Applying ICrA on the obtained average results of model parameters estimates, T and J , some relations between the defined criteria are established. The presented results show some dependences relating to the physical meaning of the considered model parameters and to stochastic nature of the applied in this paper metaheuristic techniques.