1. LTCSO/D: a large-scale tri-particle competitive swarm optimizer based on decomposition for multiobjective optimization.
- Author
-
Deng, Libao, Di, Yuanzhu, Song, Le, and Gong, Wenyin
- Subjects
EVOLUTIONARY algorithms - Abstract
Large-scale multiobjective optimization problems (LSMOPs), whose exponentially expanded search space and conflicting objectives have brought great challenges to traditional multiobjective evolutionary algorithms (MOEAs), frequently appear in engineering applications. To make full use of the positive information carried by all the individuals, thus avoiding the lopsided distribution of the population, a large-scale tri-particle competitive swarm optimizer based on decomposition (LTCSO/D) with better distribution and convergence is proposed for solving LSMOPs properly. In the proposed optimization framework, a population splitting strategy is designed to allocate subpopulations to the decomposed search space, an innovative tri-particle competitive method is employed to enhance the diversity of the offspring, and an environmental selection mechanism based on the maximum angle priority strategy is used to balance distribution and convergence. Finally, extensive computational tests between LTCSO/D and four existing algorithms are carried out on a variety of LSMOPs with different numbers of decision variables and objectives, and the effectiveness of the specific designs in LTCSO/D is verified through the experiments. The comparative results demonstrate the significant advantages of LTCSO/D over several state-of-the-art MOEAs in solving LSMOPs. [ABSTRACT FROM AUTHOR]
- Published
- 2023
- Full Text
- View/download PDF