A centerline symmetry and double-line transformation based algorithm for large-scale multi-objective optimization

The search space of large-scale multi-objective optimization problems (LSMOPs) is huge because the hundreds or even thousands decision variables involved. It very challenging to design efficient algorithms for LSMOPs whole effectively and balance convergence diversity at same time. In this paper, tackle challenge, we develop a new algorithm based on weighted framework with two effective strategies. transforms an LSMOP into multiple small-scale problem transformation mechanism reduce dimensionality effectively. To further improve its effectiveness, firstly propose centerline symmetry strategy select reference solutions transform LSMOPs. takes not only some non-dominated but also their symmetric points as solutions, which can enhance population avoid falling local minima. Then, double-line function designed expand range transformed diversity. With strategies, more widely distributed potential areas are provided optimal be found easier. demonstrate effectiveness our proposed algorithm, numerical experiments used benchmarks executed statistical results show that competitive performs better than other state-of-the-art solving