Crossover-based artificial bee colony algorithm for constrained optimization problems

Artificial bee colony (ABC) algorithm represents one of the most-studied swarm intelligence algorithms. Since the original ABC has been found to be very effective, today there are a lot of improved variants of ABC algorithm used to solve a wide range of hard optimization problems. This paper describes a novel artificial bee colony algorithm for constrained optimization problems. In the proposed algorithm, five modifications are introduced. Firstly, to improve the exploitation abilities of ABC, two different modified ABC search operators are used in employed and onlooker phases, and crossover operator is used in scout phase instead of random search. Secondly, modifications related to dynamic tolerance for handling equality constraints and improved boundary constraint-handling method are employed. The experimental results, obtained by testing on a set of 24 well-known benchmark functions and four widely used engineering design problems, show that the proposed approach can outperform ABC-based approaches for constrained optimization problems in terms of the quality of the results, robustness and convergence speed. Additionally, it provides better results in most cases compared with other state-of-the-art algorithms.