An improved decomposition method for large-scale global optimization: bidirectional-detection differential grouping

摘要

Differential grouping (DG) is an efficient decomposition method that is used to solve large-scale global optimization (LSGO) problems. To further reduce the computational cost, a bidirectional-detection differential grouping (BDDG) method is proposed in this paper. By exploiting the bidirectional detection structure (BDS), BDDG is able to spend less computation than other DG-based methods. An adaptive perturbation strategy (APS) is proposed to improve the problem with the BDS decomposition accuracy. Analytical methods are used to demonstrate that the complexity of BDDG is lower than that of other state-of-the-art DG-based methods. Experiments showed that BDDG substantially reduced the computational cost for problem decomposition and that the computational cost used by BDDG grew slowly, as the problem dimension grew compared to other DG-based methods. When BDDG was embedded in the cooperative coevolution (CC) framework, it improved the performance of the CC for solving LSGO problems.