Hybrid differential evolution including geometric mean mutation for optimization of biochemical systems

Many process optimization problems have large parameter search spaces. Evolutionary algorithms generally lack the capability to solve such problems. In this study, we have introduced a geometric mean mutation into the hybrid differential evolution algorithm to replace genes having very small or large values. This operation could avoid generating perturbed individuals that are clustered near the parameter search bounds. The comparison is performed on a fed-batch optimization problem, an inverse problem, and a number of benchmark unconstrained and constrained optimization problems. The results from this study show that the proposed algorithm outperforms the other algorithms.