Mixed Differential Evolution and Genetic Algorithm Hybridization for Solving Global Optimization Problems

Evolutionary Algorithms (EAs) have progressively been proving their appropriacy in solving diverse optimization problems. Nonetheless, scrutiny of new variants of classical EAs is an ongoing active research area. Hybridization of EAs—mixing them with other EAs or with other optimization algorithms—is one of the avenues for this research direction. This paper proposes to appraise hybridization strategies of two EAs—Differential Evolution (DE) and Genetic Algorithm (GA). The chosen algorithms are hybridized by embedding variation operators (mutation and crossover (recombination)), of other algorithms, into their structures. Adopting this strategy, this study designed 7 versions of DE and GA hybrid algorithms. The hybrid versions and the classical versions of DE and GA were implemented to solve a standard benchmarking problem. Their performances were compared, empirically and statistically, based on the value of the best objective function (bOFV) and the execution time (eTime). The results reported that the hybrid versions of the DE-GA algorithms still need to be tuned, to achieve superior performance, or, at least, comparable as to their classical counterparts. The premature convergence and stagnating nature of the hybrid versions of these algorithms were identified as the reasons for their lower performance. Apt remedial measures are suggested along with the results and discussions.