Many-Objective Artificial Bee Colony Algorithm Based on Decision Variable Grouping

The artificial bee colony (ABC) algorithm has exhibited commendable performance in addressing single-objective optimization problems (SOPs) and multi-objective optimization problems (MOPs). However, limited research has been conducted in applying ABC to tackle many-objective optimization problems (MaOPs). The primary impediment arises from the inherent inability of traditional Pareto dominance to exert sufficient selection pressure, leading to a sharp increase in the proportion of incomparable solutions within the population. To address these challenges, this study proposes a novel many-objective ABC algorithm based on decision variable grouping (DVGABC) for MaOPs. Firstly, the decision variables are divided into two groups: convergence and diversity, updated during the employed bee stage and the onlooker bee stage, respectively. Then, a new fitness function based on strength Pareto dominance is proposed. Finally, the DTLZ benchmark suite is used to validate the effectiveness of DVGABC. The results obtained demonstrate the superiority of DVGABC over five other well-established algorithms on two performance metrics.