Memetic Modified Artificial Bee Colony for constrained optimization

This paper presents a memetic approach combining the Modified Artificial Bee Colony algorithm (MABC) and the Hooke-Jeeves method to improve its performance to solve constrained numerical optimization problems. The operator used by the employed bees was modified in such a way that more diverse solutions are generated. For constraint handling, the set of feasibility rules used in the original MABC was replaced by the ε-constrained method. Furthermore, the application frequency of the local search depends on a measure based on diversity of the solutions in the current population. The proposed algorithm is tested in a set of 24 well-known benchmark problems and the results are compared against the original (MABC) and also against one state-of-the-art approach. The overall performance provided by the proposed memetic algorithm outperforms those of the compared algorithms.